Cross Reference: /src/lib/libpuffs

History log of /src/lib/libpuffs
Revision	Date	Author	Comments
1.30	05-Sep-2023	mrg	apply -Wno-dangling-pointer to callcontext.c. puffs_cc_getcc() uses a mask against a stack variable to find the struct puffs_cc stashed below the stack, triggering the dangling pointer problem.
1.29	31-Mar-2023	rillig	libpuffs: ignore lint warning about nested extern for now This should be properly fixed by someone familiar with puffs.
1.28	28-Mar-2023	rillig	lint: warn about extern declarations outside headers https://mail-index.netbsd.org/tech-userlevel/2023/03/15/msg013727.html
1.27	22-Jan-2022	pho	Allow calling puffs_mount(3) before puffs_daemon(3) puffs_daemon(3) creates a pipe before forking, and the parent process waits for the child to either complete puffs_mount(3) or fail. If a user calls puffs_daemon(3) after puffs_mount(3), the function deadlocks. While this error-reporting functionality is really a nice thing to have, deadlocking is not great. If the filesystem has already been mounted, puffs_mount(3) should just daemonize the process and return. This became an issue because fuse_daemonize(3) in FUSE API had no such requirement and some FUSE filesystems in the wild suffered deadlocks.
1.26	23-Jan-2016	christos	Define _KERNTYPES for things that need it.
1.25	21-Mar-2012	matt	These directories default to WARNS?=5
1.24	06-Dec-2010	pooka	branches: 1.24.6; Remove PUFFS_WITH_THREADS ifdef garbage, just use libc weak symbols.
1.23	05-Dec-2009	pooka	Remove suspension support from here too.
1.22	04-Dec-2007	pooka	Part 3/n of extensive changes to message handling: Get rid of the original puffs_req(3) framework and use puffs_framebuf(3) instead for file system requests. It has the advantage of being suitable for transporting a distributed message passing protocol and therefore us being able to run the file system server on any host. Ok, puffs is not quite here yet: libpuffs needs to grow request routing support and the message contents need to be munged into a host independent format. Saying which format would be telling, but it might begin with an X, end in an L and have the 13th character in the middle. Keep an eye out for the sequels: Parts 3+m/n.
1.21	26-Oct-2007	pooka	Add some very initial support for a threading worker model as an alternative to the (vastly superior ;) continuation model. This is very preliminary stuff and not compiled by default (which it even won't do without some other patches I cannot commit yet). The raison d'commit of the patch is a snippet which ensures proper in-order dispatching of all operations, including those which don't require a response. Previously many of them would be dispatched simultaneosly, e.g. fsync and reclaim on the same node, which obviously isn't all that nice for correct operation.
1.20	18-Jul-2007	pooka	branches: 1.20.4; 1.20.6; get rid of DBG=
1.19	28-May-2007	tls	Add new Makefile knob, USE_FORT, which extends USE_SSP by turning on the FORTIFY_SOURCE feature of libssp, thus checking the size of arguments to various string and memory copy and set functions (as well as a few system calls and other miscellany) where known at function entry. RedHat has evidently built all "core system packages" with this option for some time. This option should be used at the top of Makefiles (or Makefile.inc where this is used for subdirectories) but after any setting of LIB. This is only useful for userland code, and cannot be used in libc or in any code which includes the libc internals, because it overrides certain libc functions with macros. Some effort has been made to make USE_FORT=yes work correctly for a full-system build by having the bsd.sys.mk logic disable the feature where it should not be used (libc, libssp iteself, the kernel) but no attempt has been made to build the entire system with USE_FORT and doing so will doubtless expose numerous bugs and misfeatures. Adjust the system build so that all programs and libraries that are setuid, directly handle network data (including serial comm data), perform authentication, or appear likely to have (or have a history of having) data-driven bugs (e.g. file(1)) are built with USE_FORT=yes by default, with the exception of libc, which cannot use USE_FORT and thus uses only USE_SSP by default. Tested on i386 with no ill results; USE_FORT=no per-directory or in a system build will disable if desired.
1.18	11-May-2007	pooka	move puffs request dispatching code into it's own module. no functional change
1.17	05-May-2007	pooka	add stubby version of puffs_framebuf manual page. proper one forthcoming to an internet near you Real Soon Now(tm)
1.16	05-May-2007	pooka	Add puffs "frame buffers", which are an abstraction of the buffering and event handling mechanisms required in file servers with blocking I/O backends. puffs_framebuf is built on the concept of puffs_cc and uses those to multiplex execution where needed. File systems are required to implement three methods: * read frame * write frame * compare if frame is a response to the given one Memory management is provided by puffs_framebuf, but the file systems must still, of course, interpret the protocol and do e.g. byte order conversion. As always, puffs_framebuf is work in progress. Current users are mount_psshfs and mount_9p.
1.15	16-Apr-2007	pooka	Move file system and node callbacks into their own manual page. Improve markup and descriptions a bit (not completely finished yet).
1.14	22-Mar-2007	pooka	document credentials & access control methods
1.13	26-Jan-2007	pooka	document puffs_fs_suspend()
1.12	26-Jan-2007	pooka	add interfaces for suspending the file system
1.11	20-Jan-2007	pooka	document request handling routines
1.10	19-Jan-2007	pooka	Bring the documentation slightly more into sync with the current reality. It's still really a far cry from proper documentation, though.
1.9	15-Jan-2007	pooka	Checkpoint some more work in progress: for the benefit those file system backends which operate purely based on paths, push out more path management into the library and make path management more abstract: enable a file system to define a bunch of path management callbacks, which are used by the framework. Management of normal /this/is/a/path type paths is provided by the library.
1.8	11-Jan-2007	pooka	Add preliminary code for a nullfs layer, which can be used to mount a directory hierarchy to another point, just like with the kernel nullfs. This is not really a layering scheme yet, but it should evolve into one. Currently it can just be used to do 1:1 mapping.
1.7	09-Jan-2007	pooka	lib interface for invalidation routines
1.6	29-Dec-2006	pooka	checkpoint some experimental work-in-progress, namely: Add support for having multiple outstanding operations. This is done by exposing enough interfaces so that it is convenient to have the main event loop in the implementation itself and by providing a continuation framework for convinient blocking and rescheduling. works fine, but will undergo further cleanup & development
1.5	07-Dec-2006	pooka	outline & export code for interfacing with kernel GET/PUT requests
1.4	09-Nov-2006	pooka	initial documentation for the puffs library interface
1.3	23-Oct-2006	christos	pass lint; fix install.
1.2	22-Oct-2006	dogcow	add the .include <bsd.own.mk>, so .d and .o files don't end up in the CWD. remove INCDIR=/usr/include, as my leenooks system doesn't have puffs .h files.
1.1	22-Oct-2006	pooka	add libpuffs - the userspace programming interface for puffs This work was initially started and completed for Google SoC 2005 and tweaked to work a bit better in the past few weeks. While being far from complete, it is functional enough to be able and stable to host a fairly general-purpose in-memory file system in userspace. Even so, puffs should be considered experimental and no binary compatibility for interfaces or crash-freedom or zero security implications should be relied upon just yet. The GSoC project was mentored by William Studenmund and the final review for the code was done by Christos.
1.20.6.2	18-Jul-2007	pooka	get rid of DBG=
1.20.6.1	18-Jul-2007	pooka	file Makefile was added on branch matt-mips64 on 2007-07-18 22:38:59 +0000
1.20.4.2	09-Jan-2008	matt	sync with HEAD
1.20.4.1	06-Nov-2007	matt	sync with HEAD
1.24.6.1	17-Apr-2012	yamt	sync with head
1.3	19-Jul-2007	pooka	branches: 1.3.16; stuff
1.2	01-Jul-2007	pooka	pid is gone
1.1	19-May-2007	pooka	quick TODO-list. very incomplete
1.3.16.2	19-Jul-2007	pooka	stuff
1.3.16.1	19-Jul-2007	pooka	file TODO was added on branch matt-mips64 on 2007-07-19 07:48:11 +0000
1.1	03-Apr-2013	agc	branches: 1.1.2; file Version.map was initially added on branch agc-symver.
1.1.2.1	03-Apr-2013	agc	Add symbol versioning information for the following libraries: libbluetooth libbsdmalloc libexecinfo libform libintl libipsec libkvm liblwres libnpf libp2k libpci libperfuse libppath libprop libpuffs librefuse librpcsvc librt librumpclient librumphijack librumpuser libukfs libusbhid libwrap
1.30	09-Jul-2018	christos	revert previous now that MAP_ALIGNED has been fixed.
1.29	08-Jul-2018	christos	correct previous cast.
1.28	08-Jul-2018	christos	cast for lint
1.27	06-Dec-2011	skrll	branches: 1.27.38; 1.27.40; Deal with __MACHINE_STACK_GROWS_UP machines. constify the other psize while I'm here. Thanks to Christos for looking.
1.26	02-Nov-2011	yamt	branches: 1.26.2; puffs_cc_yield: fix a cc leak. PR/44679
1.25	04-Mar-2011	yamt	puffs__cc_destroy: clear PCC_HASCALLER. the caller info is not necessarily relevant to the next life of the cc. this also fixes the failure of the assertion in the next line.
1.24	10-Jan-2011	yamt	branches: 1.24.2; typo in a comment
1.23	11-Aug-2008	pooka	clear cached puffs_cc's in puffs_exit()
1.22	11-Aug-2008	pooka	* make sure we come out of mainloop in the main context. otherwise in certain cases the server would jump to hyperspace when unmounted * sprinkle some debug prints
1.21	28-Jan-2008	pooka	branches: 1.21.6; Don't add asserts between testing and commit - especially not if you're going to write "assert(foo = 0)"
1.20	28-Jan-2008	pooka	Implement lazy contexts for file systems using puffs_mainloop(). Previously each request was executed on its own callcontext and switched to every time the request was being processed. Now requests borrow the mainloop context and switch only if/when they yield. This takes two context switches away from a file system request bringing down the typical amounts 2->0 (e.g. dtfs) and 4->2 (e.g. psshfs). The interfaces for manually executing requests changed a bit: puffs_dispatch_create() and puffs_dispatch_exec() must now be used. They are not tested, as nothing in-tree wants them and I doubt anyone else is really interested in them either. Also do some misc code cleanup related to execution contexts. The "work-in-progress checkpoint" committed over a year ago was starting to look slightly weed-infested.
1.19	17-Jan-2008	pooka	Fix fakecc case. Noticed by Pavel Jirout on current-users
1.18	16-Jan-2008	pooka	Cache a number of execution contexts instead of re-initializing them every time. Speeds up pure in-memory file systems such as sysctlfs or dtfs quite a bit. For actual I/O-workhorses the result is of course less tasty.
1.17	16-Jan-2008	pooka	Use PROT_NONE page to protect bottom of stack from overrun
1.16	16-Dec-2007	pooka	* nuke puffs_cc_get{specific,usermount} for good * move prototypes for puffs_docc and puffs_dopufbuf into the public header, as they are should be exposed
1.15	04-Dec-2007	pooka	Part 3/n of extensive changes to message handling: Get rid of the original puffs_req(3) framework and use puffs_framebuf(3) instead for file system requests. It has the advantage of being suitable for transporting a distributed message passing protocol and therefore us being able to run the file system server on any host. Ok, puffs is not quite here yet: libpuffs needs to grow request routing support and the message contents need to be munged into a host independent format. Saying which format would be telling, but it might begin with an X, end in an L and have the 13th character in the middle. Keep an eye out for the sequels: Parts 3+m/n.
1.14	29-Nov-2007	pooka	* allocate pcc as a slice from the stack instead of allocating it separately * provide puffs_cc_getcc() This is in preparation for the removal of you-should-guess-what as an argument to routines here and there and everywhere.
1.13	31-Oct-2007	pooka	Fix a few bugs and memleaks with the threading code.
1.12	26-Oct-2007	pooka	Add some very initial support for a threading worker model as an alternative to the (vastly superior ;) continuation model. This is very preliminary stuff and not compiled by default (which it even won't do without some other patches I cannot commit yet). The raison d'commit of the patch is a snippet which ensures proper in-order dispatching of all operations, including those which don't require a response. Previously many of them would be dispatched simultaneosly, e.g. fsync and reclaim on the same node, which obviously isn't all that nice for correct operation.
1.11	26-Oct-2007	pooka	mmap stacks, makes it possible to align them
1.10	21-Oct-2007	pooka	Add puffs_cc_schedule() which marks a pcc runnable and will pass execution to it when in the main loop the next time.
1.9	21-Oct-2007	pooka	Always provide caller information from the kernel based on curlwp. (but don't deprecate the old puffs_cid interface just yet)
1.8	11-Oct-2007	pooka	Part 1/n of some pretty extensive changes to how the kernel module interacts with the userspace file server: * since the kernel-user communication is not purely request-response anymore (hasn't been since 2006), try to rename some "request" to "message". more similar mangling will take place in the future. * completely rework how messages are allocated. previously most of them were borrowed from the stack (originally all of them), but now always allocate dynamically. this makes the structure of the code much cleaner. also makes it possible to fix a locking order violation. it enables plenty of future enhancements. * start generalizing the transport interface to be independent of puffs * move transport interface to read/write instead of ioctl. the old one had legacy design problems, and besides, ioctl's suck. implement a very generic version for now; this will be worked on later hopefully some day reaching "highly optimized". * implement libpuffs support behind existing library request interfaces. this will change eventually (I hate those interfaces)
1.7	06-Jun-2007	pooka	branches: 1.7.4; Move puffs to a two clause license where it already isn't so. And as agc pointed out, even files with the third clause were already effectively two clause because of a slight bug in the language...
1.6	15-May-2007	pooka	Step 3 and currently final step of a more general purpose event loop: support removal and addition of i/o file descriptors on the fly. * detect closed file descriptors * automatically free waiters of a dead file descriptor * give the file server the possibility to specify a callback which notifies of a dead file descriptor * move loop function to be a property of the mainloop instead of framebuf (doesn't change effective behaviour) * add the possibility to configure a timespec parameter which attempts to call the loop function periodically * move the event loop functions from the puffs_framebuf namespace to puffs_framev to differential between pure memory management functions
1.5	10-May-2007	pooka	assert that yield/continue is done only with a real continuation
1.4	19-Apr-2007	pooka	add puffs_cc_getspecific(pcc), which is shorthand for: puffs_getspecific(puffs_cc_getusermount(pcc))
1.3	17-Feb-2007	pooka	#include <ucontext.h>
1.2	06-Jan-2007	pooka	* get rid of the mount callback; it's no great surprise to the server that it needs to mount the file system backend if it wants to call mount * provide some options for getmntopts(), assume that callers will parse command line (or fstab) args * reorganize the puffs_cc interface just a bit, preparing for a bigger revamp later
1.1	29-Dec-2006	pooka	checkpoint some experimental work-in-progress, namely: Add support for having multiple outstanding operations. This is done by exposing enough interfaces so that it is convenient to have the main event loop in the implementation itself and by providing a continuation framework for convinient blocking and rescheduling. works fine, but will undergo further cleanup & development
1.7.4.3	23-Mar-2008	matt	sync with HEAD
1.7.4.2	09-Jan-2008	matt	sync with HEAD
1.7.4.1	06-Nov-2007	matt	sync with HEAD
1.21.6.1	18-Sep-2008	wrstuden	Sync with wrstuden-revivesa-base-2.
1.24.2.1	05-Mar-2011	bouyer	Sync with HEAD
1.26.2.1	17-Apr-2012	yamt	sync with head
1.27.40.1	10-Jun-2019	christos	Sync with HEAD
1.27.38.1	28-Jul-2018	pgoyette	Sync with HEAD
1.16	15-Mar-2012	njoly	Use puffs_access() return value in the VA_UTIMES_NULL case.
1.15	20-Nov-2009	pooka	branches: 1.15.6; 1.15.8; In getgroups(), copy only up to the number of supplementary groups that actually exist.
1.14	08-Dec-2007	pooka	branches: 1.14.12; 1.14.18; Now that "l" is gone both as an argument to operations and from componentname, remove all vestiges of puffs_cid.
1.13	18-Oct-2007	pooka	* Change type of boolean functions to, well, boolean instead of int. * add puffs_cid_isequal() which can be used to test if the two caller id's refer to the same process/lwp.
1.12	01-Jul-2007	pooka	branches: 1.12.4; Instead of supplying a plain pid, supply an abstract struct puffs_cid *, which can currently be used to query the pid and lwpid.
1.11	01-Jul-2007	pooka	make puffs_cred an opaque type
1.10	06-Jun-2007	pooka	Move puffs to a two clause license where it already isn't so. And as agc pointed out, even files with the third clause were already effectively two clause because of a slight bug in the language...
1.9	22-Mar-2007	pooka	allow kernel creds in addition to superuser creds when checking for access
1.8	22-Mar-2007	pooka	unify the location of pcr in puffs_access_
1.7	22-Mar-2007	pooka	copy ngroups * sizeof(gid_t) groups instead of just ngroups
1.6	22-Mar-2007	pooka	* return -1 and set errno instead of returning errno directly * introduce puffs_cred_isregular() which checks if the cred structure has regular uid/gid (i.e. "person" credentials and not kernel creds)
1.5	21-Mar-2007	pooka	add routiens which check against ufs semantics for permission to chown, chmod and change a/mtime
1.4	20-Mar-2007	pooka	implement counterpart of vaccess()
1.3	23-Oct-2006	christos	fix a lint warning.
1.2	23-Oct-2006	christos	pass lint; fix install.
1.1	22-Oct-2006	pooka	add libpuffs - the userspace programming interface for puffs This work was initially started and completed for Google SoC 2005 and tweaked to work a bit better in the past few weeks. While being far from complete, it is functional enough to be able and stable to host a fairly general-purpose in-memory file system in userspace. Even so, puffs should be considered experimental and no binary compatibility for interfaces or crash-freedom or zero security implications should be relied upon just yet. The GSoC project was mentored by William Studenmund and the final review for the code was done by Christos.
1.12.4.2	09-Jan-2008	matt	sync with HEAD
1.12.4.1	06-Nov-2007	matt	sync with HEAD
1.14.18.1	21-Apr-2010	matt	sync to netbsd-5
1.14.12.1	28-Nov-2009	bouyer	Pull up following revision(s) (requested by pooka in ticket #1155): lib/libpuffs/creds.c: revision 1.15 In getgroups(), copy only up to the number of supplementary groups that actually exist.
1.15.8.1	03-Nov-2014	msaitoh	Pull up following revision(s) (requested by manu in ticket #1140): lib/libperfuse/ops.c 1.63-1.69 lib/libperfuse/perfuse.c 1.32-1.33 lib/libperfuse/perfuse_priv.h 1.32-1.34 lib/libperfuse/subr.c 1.20 lib/libpuffs/creds.c 1.16 lib/libpuffs/dispatcher.c 1.47 lib/libpuffs/puffs.h 1.125 lib/libpuffs/puffs_ops.3 1.37-1.38 lib/libpuffs/requests.c 1.24 sys/fs/puffs/puffs_msgif.h 1.81 sys/fs/puffs/puffs_sys.h 1.85 sys/fs/puffs/puffs_vnops.c 1.183 usr.sbin/perfused/msg.c 1.22 Bring libpuffs, libperfuse and perfused on par with -current: - implement FUSE direct I/O - remove useless code and warnings - fix missing GETATTR bugs - fix exended attribute get and list operations
1.15.6.1	17-Apr-2012	yamt	sync with head
1.50	11-Dec-2023	mlelstv	pathconf needs to return EINVAL when the variable is invalid or cannot be associated with a file. This also needs to be true when the node doesn't implement the pathconf function at all.
1.49	08-Mar-2021	christos	branches: 1.49.6; cast to the proper enums for lint
1.48	31-Oct-2014	manu	libpuffs support for fallocate and fdiscard operations
1.47	16-Aug-2014	manu	Add an open2 method, like open but with an additionnal oflags used by the filesystem to send back information about the file. This is used to implement PUFFS_OPEN_IO_DIRECT by which the filesystem tells the kernel that read/write should bypass the page cache.
1.46	06-Nov-2013	christos	branches: 1.46.4; revert previous, will turn off globally.
1.45	06-Nov-2013	christos	XXX: gcc initialization (mc68k gcc-4.5.x gcc-4.8.x)
1.44	16-Aug-2012	manu	Fix regression that has been introduced when the lookup/reclaim race condition was addressed in libpuffs by counting lookups. The fix assumes that cookies map to struct puffs_cookie, which has not been documented as a requirement for filesystems using libpuffs. As an example, we got burnt by this assumption in libp2k (kern/46734), and we fixed bit by actually mapping libp2k cookies to struct puffs_node. It is unlikely, but there may be third party filesystems that use cookies unmapped to struct puffs_node, and they were left broken for now. - we introduce a puffs_init() flag PUFFS_FLAG_PNCOOKIE that let filesystems inform libpuffs that they map cookies to struct puffs_node. Is that flag is used, the lookup/reclaim race condition fix is enabled. We enable the flag for libp2k. - filesystems that use puffs_pn_new() obviouslty use struct puffs_node and gain PUFFS_FLAG_PNCOOKIE automatically even if they did not specify it in puffs_init(). This include all our PUFFS filesystem in-tree except libp2k. - for filesystems not willing to use struct puffs_node, we introduce a reclaim2 vnop, which is reclaim with an additionnal lookup count argument. This vnop let the filesystem implement the lookup/reclaim race fix on its own.
1.43	10-Aug-2012	manu	Fix race condition between (create\|mknod\|mkdir\|symlino) and reclaim, just like we did it between lookup and reclaim.
1.42	21-Jul-2012	manu	- Improve PUFFS_KFLAG_CACHE_FS_TTL by reclaiming older inactive nodes. The normal kernel behavior is to retain inactive nodes in the freelist until it runs out of vnodes. This has some merit for local filesystems, where the cost of an allocation is about the same as the cost of a lookup. But that situation is not true for distributed filesystems. On the other hand, keeping inactive nodes for a long time hold memory in the file server process, and when the kernel runs out of vnodes, it produce reclaim avalanches that increase lattency for other operations. We do not reclaim inactive vnodes immediatly either, as they may be looked up again shortly. Instead we introduce a grace time and we reclaim nodes that have been inactive beyond the grace time. - Fix lookup/reclaim race condition. The above improvement undercovered a race condition between lookup and reclaim. If we reclaimed a vnode associated with a userland cookie while a lookup returning that same cookiewas inprogress, then the kernel ends up with a vnode associated with a cookie that has been reclaimed in userland. Next operation on the cookie will crash (or at least confuse) the filesystem. We fix this by introducing a lookup count in kernel and userland. On reclaim, the kernel sends the count, which enable userland to detect situation where it initiated a lookup that is not completed in kernel. In such a situation, the reclaim must be ignored, as the node is about to be looked up again.
1.41	27-Jun-2012	manu	Pass the FAF flag to handlers: - setattr_ttl is updated to add a flag argument. Since it was not present in a previous release, we can change its API - write2 is introduced, this is write with an extra flag for FAF. - fsync already has the FAF information in a flag and needs no change - for other operations, FAF is unconditional
1.40	18-Apr-2012	manu	- When using PUFFS_KFLAG_CACHE_FS_TTL, do not use puffs_node to carry attribute and TTL fora newly created node. Instead extend puffs_newinfo and add puffs_newinfo_setva() and puffs_newinfo_setttl() - Remove node_mk_common_final in libperfuse. It used to set uid/gid for a newly created vnode but has been made redundant along time ago since uid and gid are properly set in FUSE header. - In libperfuse, check for corner case where opc = 0 on INACTIVE and RECLAIM (how is it possible? Check for it to avoid a crash anyway) - In libperfuse, make sure we unlimit RLIMIT_AS and RLIMIT_DATA so that we do notrun out of memory because the kernel is lazy at reclaiming vnodes. - In libperfuse, cleanup style of perfuse_destroy_pn()
1.39	08-Apr-2012	manu	Add PUFFS_KFLAG_CACHE_FS_TTL flag to puffs_init(3) to use name and attribute cache with filesystem provided TTL. lookup, create, mknod, mkdir, symlink, getattr and setattr messages have been extended so that attributes and their TTL can be provided by the filesytem. lookup, create, mknod, mkdir, and symlink messages are also extended so that the filesystem can provide name TTL. The filesystem updates attributes and TTL using puffs_pn_getvap(3), puffs_pn_getvattl(3), and puffs_pn_getcnttl(3)
1.38	25-Nov-2011	manu	branches: 1.38.2; Rollback previous change as it breaks autmates tests
1.37	24-Nov-2011	manu	Remove files specified in pcn. When remove files using name from pnode, another link on this file can be unlinked. E.g. "touch 1; ln 1 2; rm 2" will remove file named "1". Thus puffs_null_node_remove should remove directory entry which name is provided by pcn (as said in puffs_ops.3). Caller should provide appropriately initialized pcn. From Evgeniy Ivanov <lolkaantimat@gmail.com>
1.36	04-Jul-2011	manu	branches: 1.36.2; Add a flag to VOP_LISTEXTATTR(9) so that the vnode interface can tell the filesystem in which format extended attribute shall be listed. There are currently two formats: - NUL-terminated strings, used for listxattr(2), this is the default. - one byte length-pprefixed, non NUL-terminated strings, used for extattr_list_file(2), which is obtanined by setting the EXTATTR_LIST_PREFIXLEN flag to VOP_LISTEXTATTR(9) This approach avoid the need for converting the list back and forth, except in libperfuse, since FUSE uses NUL-terminated strings, and the kernel may have requested EXTATTR_LIST_PREFIXLEN.
1.35	06-Dec-2010	pooka	Remove PUFFS_WITH_THREADS ifdef garbage, just use libc weak symbols.
1.34	21-May-2010	pooka	* support extended attributes * bump major due to structure growth * add some spare space * remove ABI sillyness
1.33	17-Oct-2009	pooka	Dispatch puffs_node_abortop(). Note: We use the storage of puffs_cache_write from puffs_ops for this purpose. It's not issued by the kernel and hence currently unused, and this saves us from the trouble of bumping the lib major version.
1.32	12-Aug-2008	pooka	branches: 1.32.4; 1.32.10; Replace void * by puffs_cookie_t where appropriate. No functional change.
1.31	20-Mar-2008	pooka	branches: 1.31.4; Reserve enough memory for ops which return more than they receive.
1.30	28-Jan-2008	pooka	branches: 1.30.2; Implement lazy contexts for file systems using puffs_mainloop(). Previously each request was executed on its own callcontext and switched to every time the request was being processed. Now requests borrow the mainloop context and switch only if/when they yield. This takes two context switches away from a file system request bringing down the typical amounts 2->0 (e.g. dtfs) and 4->2 (e.g. psshfs). The interfaces for manually executing requests changed a bit: puffs_dispatch_create() and puffs_dispatch_exec() must now be used. They are not tested, as nothing in-tree wants them and I doubt anyone else is really interested in them either. Also do some misc code cleanup related to execution contexts. The "work-in-progress checkpoint" committed over a year ago was starting to look slightly weed-infested.
1.29	16-Jan-2008	pooka	Comment out the puffs_executor hacks for now, they have races. The issue should be solved in another fashion anyway.
1.28	16-Jan-2008	pooka	Cache a number of execution contexts instead of re-initializing them every time. Speeds up pure in-memory file systems such as sysctlfs or dtfs quite a bit. For actual I/O-workhorses the result is of course less tasty.
1.27	16-Dec-2007	pooka	* nuke puffs_cc_get{specific,usermount} for good * move prototypes for puffs_docc and puffs_dopufbuf into the public header, as they are should be exposed
1.26	08-Dec-2007	pooka	Now that "l" is gone both as an argument to operations and from componentname, remove all vestiges of puffs_cid.
1.25	04-Dec-2007	pooka	Part 3/n of extensive changes to message handling: Get rid of the original puffs_req(3) framework and use puffs_framebuf(3) instead for file system requests. It has the advantage of being suitable for transporting a distributed message passing protocol and therefore us being able to run the file system server on any host. Ok, puffs is not quite here yet: libpuffs needs to grow request routing support and the message contents need to be munged into a host independent format. Saying which format would be telling, but it might begin with an X, end in an L and have the 13th character in the middle. Keep an eye out for the sequels: Parts 3+m/n.
1.24	30-Nov-2007	pooka	Rototill. Ok, ok, a few more words about it: stop holding puffs_cc as a holy value and passing it around to almost every possible place (popquiz: which kernel variable does this remind you of?). Instead, pass the natural choice, puffs_usermount, and fetch puffs_cc via puffs_cc_getcc() only in routines which actually need it. This not only simplifies code, but (thanks to the introduction of puffs_cc_getcc()) enables constructs which weren't previously sanely possible, say layering as a curious example. There's still a little to do on this front, but this was the major fs interface blast.
1.23	29-Nov-2007	pooka	* allocate pcc as a slice from the stack instead of allocating it separately * provide puffs_cc_getcc() This is in preparation for the removal of you-should-guess-what as an argument to routines here and there and everywhere.
1.22	27-Nov-2007	pooka	Remove "puffs_cid" from the puffs interface following l-removal from the kernel vfs interfaces. puffs_cc_getcaller(pcc) can be used now should the same information be desired.
1.21	31-Oct-2007	pooka	Fix a few bugs and memleaks with the threading code.
1.20	29-Oct-2007	pooka	lock massage, although I don't think there's a happy ending here
1.19	29-Oct-2007	pooka	Sprinkle some hackish locks here and there, just enough to allow libp2k & rump to run multithreaded.
1.18	28-Oct-2007	pooka	Add possibility to set pre- and post callbacks which will be called for all operations before or after executing the specific callback (the api is not be final yet, though).
1.17	26-Oct-2007	pooka	Add some very initial support for a threading worker model as an alternative to the (vastly superior ;) continuation model. This is very preliminary stuff and not compiled by default (which it even won't do without some other patches I cannot commit yet). The raison d'commit of the patch is a snippet which ensures proper in-order dispatching of all operations, including those which don't require a response. Previously many of them would be dispatched simultaneosly, e.g. fsync and reclaim on the same node, which obviously isn't all that nice for correct operation.
1.16	21-Oct-2007	pooka	Always provide caller information from the kernel based on curlwp. (but don't deprecate the old puffs_cid interface just yet)
1.15	11-Oct-2007	pooka	Part 1/n of some pretty extensive changes to how the kernel module interacts with the userspace file server: * since the kernel-user communication is not purely request-response anymore (hasn't been since 2006), try to rename some "request" to "message". more similar mangling will take place in the future. * completely rework how messages are allocated. previously most of them were borrowed from the stack (originally all of them), but now always allocate dynamically. this makes the structure of the code much cleaner. also makes it possible to fix a locking order violation. it enables plenty of future enhancements. * start generalizing the transport interface to be independent of puffs * move transport interface to read/write instead of ioctl. the old one had legacy design problems, and besides, ioctl's suck. implement a very generic version for now; this will be worked on later hopefully some day reaching "highly optimized". * implement libpuffs support behind existing library request interfaces. this will change eventually (I hate those interfaces)
1.14	09-Oct-2007	pooka	g/c #if 0'd code which isn't likely to get used any time soon
1.13	01-Oct-2007	pooka	Pass reason string to pu_errnotify. Print it in the default version before aborting.
1.12	27-Sep-2007	pooka	Add error notifications, which are used to deliver errors from the kernel to the file server for silly things the file server did, e.g. attempting to create a file with size VSIZENOTSET. The file server can handle these as it chooses, but the default action is for it to throw its hands in the air and sing "goodbye, cruel world, it's over, walk on by".
1.11	27-Jul-2007	pooka	branches: 1.11.4; 1.11.6; mmap fflags -> prot
1.10	02-Jul-2007	pooka	Get rid of the "int *refs" parameter to inactive: the same can be accomplished now with puffs_setbacks.
1.9	01-Jul-2007	pooka	Instead of requesting various pieces of information from node-creating methods using "please fill me"-pointers, pass "struct puffs_newinfo *" and provide methods to fill in new node info.
1.8	01-Jul-2007	pooka	Instead of supplying a plain pid, supply an abstract struct puffs_cid *, which can currently be used to query the pid and lwpid.
1.7	01-Jul-2007	pooka	make puffs_cred an opaque type
1.6	06-Jun-2007	pooka	Move puffs to a two clause license where it already isn't so. And as agc pointed out, even files with the third clause were already effectively two clause because of a slight bug in the language...
1.5	18-May-2007	pooka	default refcount to 1 in inactive
1.4	18-May-2007	pooka	dispatch PUFFS_VN_POLL
1.3	15-May-2007	pooka	Step 3 and currently final step of a more general purpose event loop: support removal and addition of i/o file descriptors on the fly. * detect closed file descriptors * automatically free waiters of a dead file descriptor * give the file server the possibility to specify a callback which notifies of a dead file descriptor * move loop function to be a property of the mainloop instead of framebuf (doesn't change effective behaviour) * add the possibility to configure a timespec parameter which attempts to call the loop function periodically * move the event loop functions from the puffs_framebuf namespace to puffs_framev to differential between pure memory management functions
1.2	11-May-2007	pooka	copyright adjustment
1.1	11-May-2007	pooka	move puffs request dispatching code into it's own module. no functional change
1.11.6.2	27-Jul-2007	pooka	mmap fflags -> prot
1.11.6.1	27-Jul-2007	pooka	file dispatcher.c was added on branch matt-mips64 on 2007-07-27 08:28:17 +0000
1.11.4.3	23-Mar-2008	matt	sync with HEAD
1.11.4.2	09-Jan-2008	matt	sync with HEAD
1.11.4.1	06-Nov-2007	matt	sync with HEAD
1.30.2.1	24-Mar-2008	keiichi	sync with head.
1.31.4.1	18-Sep-2008	wrstuden	Sync with wrstuden-revivesa-base-2.
1.32.10.1	21-Apr-2010	matt	sync to netbsd-5
1.32.4.2	17-Jul-2011	riz	Pull up following revision(s) (requested by manu in ticket #1645): lib/libc/sys/Makefile.inc 1.207 via patch lib/libc/sys/extattr_get_file.2 patch lib/libpuffs/dispatcher.c 1.34,1.36 via patch lib/libpuffs/puffs.c 1.107 via patch lib/libpuffs/puffs.h 1.115,1.118 via patch sys/fs/puffs/puffs_msgif.h 1.71,1.76 via patch sys/fs/puffs/puffs_vfsops.c 1.88 via patch sys/fs/puffs/puffs_vnops.c 1.145,1.154 via patch sys/kern/vfs_xattr.c 1.24-1.27 via patch sys/kern/vnode_if.c 1.87 via patch sys/sys/Makefile 1.133 via patch sys/sys/extattr.h 1.6 via patch sys/sys/vnode_if.h 1.81 via patch sys/ufs/ffs/ffs_vnops.c patch sys/ufs/ufs/ufs_extattr.c 1.31,1.34 via patch * support extended attributes * bump major due to structure growth * add some spare space * remove ABI sillyness Support extended attributes. Fix multiple non compliances in our Linux-like extattr API, and make it public so that it can be used. Improve a bit listxattr(2). It attemps to list both system and user extended attributes, and it faled if calling user did not have privilege for reading system EA. Now we just lise user EA and skip system EA in reading them is not allowed. Fix bug introduced in previous commuit: Do not vrele() a vnode we did not obtained. Improve UFS1 extended attributes usability - autocreate attribute backing file for new attributes - autoload attributes when issuing extattrctl start - when autoloading attributes, do not display garbage warning when looking up entries that got ENOENT Add a flag to VOP_LISTEXTATTR(9) so that the vnode interface can tell the filesystem in which format extended attribute shall be listed. There are currently two formats: - NUL-terminated strings, used for listxattr(2), this is the default. - one byte length-pprefixed, non NUL-terminated strings, used for extattr_list_file(2), which is obtanined by setting the EXTATTR_LIST_PREFIXLEN flag to VOP_LISTEXTATTR(9) This approach avoid the need for converting the list back and forth, except in libperfuse, since FUSE uses NUL-terminated strings, and the kernel may have requested EXTATTR_LIST_PREFIXLEN.
1.32.4.1	18-Oct-2009	sborrill	Pull up the following revisions(s) (requested by pooka in ticket #1100): lib/libpuffs/dispatcher.c: revision 1.33 lib/libpuffs/puffs.c: revision 1.99 lib/libpuffs/puffs.h: revision 1.111 sys/fs/puffs/puffs_msgif.h: revision 1.67 via patch sys/fs/puffs/puffs_vnops.c: revision 1.136 Support VOP_ABORTOP() in puffs.
1.36.2.3	30-Oct-2012	yamt	sync with head
1.36.2.2	23-May-2012	yamt	sync with head.
1.36.2.1	17-Apr-2012	yamt	sync with head
1.38.2.5	03-Nov-2014	msaitoh	Pull up following revision(s) (requested by manu in ticket #1140): lib/libperfuse/ops.c 1.63-1.69 lib/libperfuse/perfuse.c 1.32-1.33 lib/libperfuse/perfuse_priv.h 1.32-1.34 lib/libperfuse/subr.c 1.20 lib/libpuffs/creds.c 1.16 lib/libpuffs/dispatcher.c 1.47 lib/libpuffs/puffs.h 1.125 lib/libpuffs/puffs_ops.3 1.37-1.38 lib/libpuffs/requests.c 1.24 sys/fs/puffs/puffs_msgif.h 1.81 sys/fs/puffs/puffs_sys.h 1.85 sys/fs/puffs/puffs_vnops.c 1.183 usr.sbin/perfused/msg.c 1.22 Bring libpuffs, libperfuse and perfused on par with -current: - implement FUSE direct I/O - remove useless code and warnings - fix missing GETATTR bugs - fix exended attribute get and list operations
1.38.2.4	13-Sep-2012	riz	Pull up following revision(s) (requested by manu in ticket #525): lib/libpuffs/dispatcher.c: revision 1.44 lib/libp2k/p2k.c: revision 1.57 lib/libpuffs/puffs_ops.3: revision 1.35 lib/libpuffs/pnode.c: revision 1.13 lib/libpuffs/puffs.3: revision 1.56 lib/libpuffs/puffs.h: revision 1.124 Fix regression that has been introduced when the lookup/reclaim race condition was addressed in libpuffs by counting lookups. The fix assumes that cookies map to struct puffs_cookie, which has not been documented as a requirement for filesystems using libpuffs. As an example, we got burnt by this assumption in libp2k (kern/46734), and we fixed bit by actually mapping libp2k cookies to struct puffs_node. It is unlikely, but there may be third party filesystems that use cookies unmapped to struct puffs_node, and they were left broken for now. - we introduce a puffs_init() flag PUFFS_FLAG_PNCOOKIE that let filesystems inform libpuffs that they map cookies to struct puffs_node. Is that flag is used, the lookup/reclaim race condition fix is enabled. We enable the flag for libp2k. - filesystems that use puffs_pn_new() obviouslty use struct puffs_node and gain PUFFS_FLAG_PNCOOKIE automatically even if they did not specify it in puffs_init(). This include all our PUFFS filesystem in-tree except libp2k. - for filesystems not willing to use struct puffs_node, we introduce a reclaim2 vnop, which is reclaim with an additionnal lookup count argument. This vnop let the filesystem implement the lookup/reclaim race fix on its own.
1.38.2.3	12-Aug-2012	martin	branches: 1.38.2.3.2; Pull up following revision(s) (requested by manu in ticket #438): lib/libperfuse/perfuse_priv.h: revision 1.31 sys/fs/puffs/puffs_msgif.h: revision 1.80 sys/fs/puffs/puffs_vnops.c: revision 1.171 lib/libpuffs/puffs_ops.3: revision 1.31 sys/fs/puffs/puffs_vnops.c: revision 1.172 sys/fs/puffs/puffs_vnops.c: revision 1.173 sys/fs/puffs/puffs_vnops.c: revision 1.174 usr.sbin/perfused/perfused.c: revision 1.24 sys/fs/puffs/puffs_sys.h: revision 1.80 sys/fs/puffs/puffs_sys.h: revision 1.81 sys/fs/puffs/puffs_sys.h: revision 1.82 lib/libperfuse/subr.c: revision 1.19 lib/libperfuse/perfuse.c: revision 1.30 sys/fs/puffs/puffs_msgif.c: revision 1.90 sys/fs/puffs/puffs_msgif.c: revision 1.91 sys/fs/puffs/puffs_msgif.c: revision 1.92 lib/libperfuse/ops.c: revision 1.59 lib/libpuffs/puffs.3: revision 1.53 lib/libperfuse/debug.c: revision 1.12 lib/libpuffs/puffs.3: revision 1.54 sys/fs/puffs/puffs_vnops.c: revision 1.167 sys/fs/puffs/puffs_msgif.h: revision 1.79 usr.sbin/perfused/msg.c: revision 1.21 sys/fs/puffs/puffs_vfsops.c: revision 1.102 sys/fs/puffs/puffs_vfsops.c: revision 1.103 sys/fs/puffs/puffs_vfsops.c: revision 1.105 lib/libpuffs/puffs.h: revision 1.123 lib/libperfuse/perfuse_if.h: revision 1.20 lib/libperfuse/perfuse.c: revision 1.29 lib/libpuffs/dispatcher.c: revision 1.42 lib/libpuffs/dispatcher.c: revision 1.43 - Fix same vnodes associated with multiple cookies The scheme used to retreive known nodes on lookup was flawed, as it only used parent and name. This produced a different cookie for the same file if it was renamed, when looking up ../ or when dealing with multiple files associated with the same name through link(2). We therefore abandon the use of node name and introduce hashed lists of inodes. This causes a huge rewrite of reclaim code, which do not attempt to keep parents allocated until all their children are reclaimed - Fix race conditions in reclaim There are a few situations where we issue multiple FUSE operations for a PUFFS operation. On reclaim, we therefore have to wait for all FUSE operation to complete, not just the current exchanges. We do this by introducing node reference count with node_ref() and node_rele(). - Detect data loss caused by FAF VOP_PUTPAGES causes FAF writes where the kernel does not check the operation result. At least issue a warning on error. - Enjoy FAF shortcut on setattr No need to wait for the result if the kernel does not want it. There is however an exception for setattr that touch the size, we need to wait for completion because we have other operations queued for after the resize. - Fix fchmod() on write-open file fchmod() on a node open with write privilege will send setattr with both mode and size set. This confuses some FUSE filesystem. Therefore we send two FUSE operations, one for mode, and one for size. - Remove node TTL handling for netbsd-5 for simplicity sake. The code still builds on netbsd-5 but does not have the node TTL feature anymore. It works fine with kernel support on netbsd-6. - Improve PUFFS_KFLAG_CACHE_FS_TTL by reclaiming older inactive nodes. The normal kernel behavior is to retain inactive nodes in the freelist until it runs out of vnodes. This has some merit for local filesystems, where the cost of an allocation is about the same as the cost of a lookup. But that situation is not true for distributed filesystems. On the other hand, keeping inactive nodes for a long time hold memory in the file server process, and when the kernel runs out of vnodes, it produce reclaim avalanches that increase lattency for other operations. We do not reclaim inactive vnodes immediatly either, as they may be looked up again shortly. Instead we introduce a grace time and we reclaim nodes that have been inactive beyond the grace time. - Fix lookup/reclaim race condition. The above improvement undercovered a race condition between lookup and reclaim. If we reclaimed a vnode associated with a userland cookie while a lookup returning that same cookiewas inprogress, then the kernel ends up with a vnode associated with a cookie that has been reclaimed in userland. Next operation on the cookie will crash (or at least confuse) the filesystem. We fix this by introducing a lookup count in kernel and userland. On reclaim, the kernel sends the count, which enable userland to detect situation where it initiated a lookup that is not completed in kernel. In such a situation, the reclaim must be ignored, as the node is about to be looked up again. Fix hang unmount bug introduced by last commit. We introduced a slow queue for delayed reclaims, while the existing queue for unmount, flush and exist has been renamed fast queue. Both queues had timestamp for when an operation should be done, but it was useless for the fast queue, which is always used to run an operation ASAP. And the timestamp test had an error that turned ASAP into "at next tick", but nobody what there to wake the thread at next tick, hence the hang. The fix is to remove the useless and buggy timestamp test for fast queue. Rename slow sopreq queue into node sopreq queue, to refet the fact that is only intended for postponed node reclaims. When purging the node sopreq queue, do not call puffs_msg_sendresp(), as it makes no sense. Fix race condition between (create\|mknod\|mkdir\|symlino) and reclaim, just like we did it between lookup and reclaim. Missing bit in previous commit (prevent race between create\|mknod\|mkdir\|symlink and reclaim) Bump date for previous. New sentence, new line; remove trailing whitespace; fix typos; punctuation nits. Add PUFFS_KFLAG_CACHE_DOTDOT so that vnodes hold a reference on their parent, keeping them active, and allowing to lookup .. without sending a request to the filesystem. Enable the featuure for perfused, as this is how FUSE works. Missing bit in previous commit (PUFFS_KFLAG_CACHE_DOTDOT option to avoid looking up ..)
1.38.2.2	05-Jul-2012	riz	Pull up following revision(s) (requested by manu in ticket #392): lib/libpuffs/puffs_ops.3: revision 1.32 lib/libperfuse/ops.c: revision 1.58 lib/libpuffs/puffs.h: revision 1.122 lib/libpuffs/dispatcher.c: revision 1.41 lib/libperfuse/perfuse_priv.h: revision 1.30 Pass the FAF flag to handlers: - setattr_ttl is updated to add a flag argument. Since it was not present in a previous release, we can change its API - write2 is introduced, this is write with an extra flag for FAF. - fsync already has the FAF information in a flag and needs no change - for other operations, FAF is unconditional Fix the build by adding (unused) flags argument to perfuse_node_setattr_ttl
1.38.2.1	23-Apr-2012	riz	Pull up following revision(s) (requested by manu in ticket #195): lib/libskey/skeysubr.c: revision 1.27 lib/libkvm/kvm_getloadavg.c: revision 1.11 lib/libwrap/update.c: revision 1.9 lib/liby/yyerror.c: revision 1.9 lib/libpuffs/puffs_ops.3: revision 1.30 lib/libwrap/misc.c: revision 1.10 lib/libwrap/hosts_access.c: revision 1.20 lib/libpuffs/pnode.c: revision 1.11 lib/libperfuse/subr.c: revision 1.17 lib/libpuffs/pnode.c: revision 1.12 lib/libperfuse/subr.c: revision 1.18 lib/libwrap/options.c: revision 1.15 lib/libwrap/fix_options.c: revision 1.11 lib/libperfuse/ops.c: revision 1.52 lib/libperfuse/ops.c: revision 1.53 lib/libperfuse/ops.c: revision 1.54 lib/libwrap/hosts_ctl.c: revision 1.5 lib/libintl/gettext.c: revision 1.27 lib/libwrap/shell_cmd.c: revision 1.6 lib/libpuffs/dispatcher.c: revision 1.39 lib/libperfuse/perfuse_priv.h: revision 1.27 lib/libwrap/socket.c: revision 1.19 lib/libpuffs/puffs.3: revision 1.50 lib/libperfuse/perfuse_priv.h: revision 1.28 lib/libpuffs/puffs_priv.h: revision 1.45 lib/libpuffs/puffs.3: revision 1.51 lib/libperfuse/perfuse_priv.h: revision 1.29 lib/libwrap/percent_x.c: revision 1.5 lib/libpuffs/puffs.3: revision 1.52 lib/libperfuse/debug.c: revision 1.11 sys/fs/puffs/puffs_vnops.c: revision 1.165 lib/libwrap/tcpd.h: revision 1.13 sys/fs/puffs/puffs_vnops.c: revision 1.166 lib/libwrap/eval.c: revision 1.7 sys/fs/puffs/puffs_msgif.h: revision 1.78 sys/fs/puffs/puffs_vfsops.c: revision 1.101 lib/libwrap/rfc931.c: revision 1.9 lib/libwrap/clean_exit.c: revision 1.5 lib/libpuffs/puffs.h: revision 1.120 lib/libc/stdlib/jemalloc.c: revision 1.27 lib/librmt/rmtlib.c: revision 1.26 lib/libpuffs/puffs.h: revision 1.121 sys/fs/puffs/puffs_sys.h: revision 1.79 lib/librumpclient/rumpclient.c: revision 1.48 lib/libwrap/refuse.c: revision 1.5 lib/libperfuse/perfuse.c: revision 1.26 lib/libperfuse/perfuse.c: revision 1.27 tests/fs/puffs/t_fuzz.c: revision 1.5 lib/libperfuse/perfuse.c: revision 1.28 lib/libpuffs/dispatcher.c: revision 1.40 sys/fs/puffs/puffs_node.c: revision 1.24 lib/libwrap/diag.c: revision 1.9 lib/libintl/textdomain.c: revision 1.13 Use C89 function definition Add name and atttribute cache with filesytem provided TTL. lookup, create, mknod, mkdir, symlink, getattr and setattr messages have been extended so that attributes and their TTL can be provided by the filesytem. lookup, create, mknod, mkdir, and symlink messages are also extended so that the filesystem can provide name TTL. Add PUFFS_KFLAG_CACHE_FS_TTL flag to puffs_init(3) to use name and attribute cache with filesystem provided TTL. lookup, create, mknod, mkdir, symlink, getattr and setattr messages have been extended so that attributes and their TTL can be provided by the filesytem. lookup, create, mknod, mkdir, and symlink messages are also extended so that the filesystem can provide name TTL. The filesystem updates attributes and TTL using puffs_pn_getvap(3), puffs_pn_getvattl(3), and puffs_pn_getcnttl(3) Use new PUFFS_KFLAG_CACHE_FS_TTL option to puffs_init(3) so that FUSE TTL on name and attributes are used. This save many PUFFS operations and improves performances. PUFFS_KFLAG_CACHE_FS_TTL is #ifdef'ed in many places for now so that libperfuse can still be used on netbsd-5. Split file system. Comma fixes. Remove dangling "and". Bump date for previous. - Makesure update_va does not change vnode size when it should not. For instance when doing a fault-issued VOP_GETPAGES within VOP_WRITE, changing size leads to panic: genfs_getpages: past eof. -Handle ticks wrap around for vnode name andattribute timeout - When using PUFFS_KFLAG_CACHE_FS_TTL, do not use puffs_node to carry attribute and TTL fora newly created node. Instead extend puffs_newinfo and add puffs_newinfo_setva() and puffs_newinfo_setttl() - Remove node_mk_common_final in libperfuse. It used to set uid/gid for a newly created vnode but has been made redundant along time ago since uid and gid are properly set in FUSE header. - In libperfuse, check for corner case where opc = 0 on INACTIVE and RECLAIM (how is it possible? Check for it to avoid a crash anyway) - In libperfuse, make sure we unlimit RLIMIT_AS and RLIMIT_DATA so that we do notrun out of memory because the kernel is lazy at reclaiming vnodes. - In libperfuse, cleanup style of perfuse_destroy_pn() Do not set PUFFS_KFLAG_CACHE_FS_TTL for PUFFS tests
1.38.2.3.2.1	01-Nov-2012	matt	sync with netbsd-6-0-RELEASE.
1.46.4.2	05-Nov-2014	snj	Pull up following revision(s) (requested by manu in ticket #181): lib/libperfuse/fuse.h: revision 1.6 lib/libperfuse/ops.c: revision 1.78 lib/libperfuse/perfuse.c: revision 1.35 lib/libperfuse/perfuse_priv.h: revision 1.36 lib/libpuffs/dispatcher.c: revision 1.48 lib/libpuffs/opdump.c: revision 1.37 lib/libpuffs/puffs.c: revision 1.118 lib/libpuffs/puffs.h: revision 1.126 lib/libpuffs/puffs_ops.3: revisions 1.40-1.41 sys/fs/puffs/puffs_msgif.h: revision 1.82-1.83 sys/fs/puffs/puffs_msgif.h: revision 1.82 sys/fs/puffs/puffs_vnops.c: revision 1.196 Add PUFFS support for fallocate and fdiscard operations -- libpuffs support for fallocate and fdiscard operations -- Add PUFFS_HAVE_FALLOCATE in puffs_msgif.h so that filesystem can decide at build time wether fallocate is usable -- FUSE fallocate support There seems to be no fdiscard FUSE operation at the moment, hence that one is left unused.
1.46.4.1	24-Aug-2014	martin	Pull up following revision(s) (requested by manu in ticket #52): lib/libperfuse/perfuse_priv.h: revision 1.33 lib/libperfuse/ops.c: revision 1.67 lib/libpuffs/dispatcher.c: revision 1.47 lib/libperfuse/perfuse_priv.h: revision 1.34 lib/libperfuse/ops.c: revision 1.68 lib/libperfuse/ops.c: revision 1.69 lib/libpuffs/puffs_ops.3: revision 1.37 lib/libpuffs/puffs_ops.3: revision 1.38 lib/libperfuse/perfuse.c: revision 1.32 lib/libperfuse/perfuse.c: revision 1.33 usr.sbin/perfused/msg.c: revision 1.22 lib/libpuffs/puffs.h: revision 1.125 Implement FUSE direct I/O. Remove useless code and warnings
1.49.6.1	11-Sep-2024	martin	Pull up following revision(s) (requested by rin in ticket #833): lib/libpuffs/dispatcher.c: revision 1.50 pathconf needs to return EINVAL when the variable is invalid or cannot be associated with a file. This also needs to be true when the node doesn't implement the pathconf function at all.
1.16	12-Aug-2008	pooka	Replace void * by puffs_cookie_t where appropriate. No functional change.
1.15	05-Dec-2007	pooka	branches: 1.15.8; free flushbuffer after use
1.14	05-Dec-2007	pooka	Send a response message for flush operations from the kernel instead of abusing the return value of write(2).
1.13	12-Nov-2007	pooka	* split the putter header into a kernel version and a userland version + install latter to /usr/include/dev/putter * remove last dependencies to puffs from putter, it's completely independent now
1.12	11-Oct-2007	pooka	Part 1/n of some pretty extensive changes to how the kernel module interacts with the userspace file server: * since the kernel-user communication is not purely request-response anymore (hasn't been since 2006), try to rename some "request" to "message". more similar mangling will take place in the future. * completely rework how messages are allocated. previously most of them were borrowed from the stack (originally all of them), but now always allocate dynamically. this makes the structure of the code much cleaner. also makes it possible to fix a locking order violation. it enables plenty of future enhancements. * start generalizing the transport interface to be independent of puffs * move transport interface to read/write instead of ioctl. the old one had legacy design problems, and besides, ioctl's suck. implement a very generic version for now; this will be worked on later hopefully some day reaching "highly optimized". * implement libpuffs support behind existing library request interfaces. this will change eventually (I hate those interfaces)
1.11	19-Jul-2007	pooka	branches: 1.11.4; 1.11.6; Don't float puffs_kargs needlessly around after mount, saves a couple k of memory (and there was much rejoycing!)
1.10	06-Jun-2007	pooka	Move puffs to a two clause license where it already isn't so. And as agc pointed out, even files with the third clause were already effectively two clause because of a slight bug in the language...
1.9	13-Apr-2007	pooka	Instead of keeping on adding parameters to puffs_mount(), make it only take the bare essentials, which currently means removing "maxreqlen" from the argument list (all current callers I'm aware of set it as 0 anyway). Introduce puffs_init(), which provides a context for setting various parameters and puffs_domount(), which can be used to mount the file system. Keep puffs_mount() as a shortcut for the above two for simple file systems. Bump development ABI version to 13. After all, it's Friday the 13th. Watch out! Bad things can happen on Friday the 13th. --No carrier--
1.8	12-Apr-2007	pooka	Hide struct puffs_usermount from the rest of the world and provide accessors for interesting data in it. Namely, you can now get pu->pu_privdata with puffs_getspecific(), pu->pu_pn_root with puffs_set/getroot() and pu->pu_maxreqlen with puffs_getmaxreqlen().
1.7	06-Apr-2007	pooka	support page cache flushing
1.6	06-Apr-2007	pooka	use range invalidation op also for invalidating entire page cache of a node
1.5	06-Apr-2007	pooka	implement puffs_inval_pagecache_node_range()
1.4	20-Mar-2007	pooka	g/c unused debug variable
1.3	20-Mar-2007	pooka	initial support for cacheops
1.2	09-Jan-2007	pooka	rename name cache invalidation interface to be in sync with the kernel operation names, i.e. contain "namecache" instead of just "name"
1.1	09-Jan-2007	pooka	lib interface for invalidation routines
1.11.6.2	19-Jul-2007	pooka	Don't float puffs_kargs needlessly around after mount, saves a couple k of memory (and there was much rejoycing!)
1.11.6.1	19-Jul-2007	pooka	file flush.c was added on branch matt-mips64 on 2007-07-19 12:52:29 +0000
1.11.4.2	09-Jan-2008	matt	sync with HEAD
1.11.4.1	06-Nov-2007	matt	sync with HEAD
1.15.8.1	18-Sep-2008	wrstuden	Sync with wrstuden-revivesa-base-2.
1.37	19-Apr-2022	rillig	lib: remove CONSTCOND comment Since 2021-01-31, lint doesn't need it anymore for the common pattern of 'do ... while (0)'.
1.36	30-Oct-2021	nia	puffs(3): Replace realloc(x * y) with reallocarr
1.35	14-Jun-2017	christos	remove now unnecessary casts.
1.34	31-May-2017	christos	Adjust to prototype.
1.33	09-May-2017	christos	Only add a write filter for sockets. It is not supported for vnodes, or for 1/2 closed fifos (which we both have now).
1.32	25-Jun-2012	abs	branches: 1.32.22; Update old-style definitions to ANSI, remove a couple of register definitions along the way. Fixed gcc 4.1 build (thank you vax)
1.31	19-Dec-2011	riastradh	Fix error reporting in puffs_framev_enqueue_waitevent and callers. ok christos
1.30	12-Jan-2010	pooka	branches: 1.30.6; Add convenience routine puffs_unmountonsignal(), which does exactly that.
1.29	04-Sep-2008	pooka	Fix double free when i/o descriptor is violently closed. Bug was introduced in rev 1.27. fix provided by Taylor R Campbell in lib/39353
1.28	29-Jan-2008	pooka	branches: 1.28.6; Fix a weirdosity in the framebuf gotfb interface: hand buffer ownership to the called routine instead of keeping it. This means that callees must now destroy the buffer once they are done with it.
1.27	28-Jan-2008	pooka	Implement lazy contexts for file systems using puffs_mainloop(). Previously each request was executed on its own callcontext and switched to every time the request was being processed. Now requests borrow the mainloop context and switch only if/when they yield. This takes two context switches away from a file system request bringing down the typical amounts 2->0 (e.g. dtfs) and 4->2 (e.g. psshfs). The interfaces for manually executing requests changed a bit: puffs_dispatch_create() and puffs_dispatch_exec() must now be used. They are not tested, as nothing in-tree wants them and I doubt anyone else is really interested in them either. Also do some misc code cleanup related to execution contexts. The "work-in-progress checkpoint" committed over a year ago was starting to look slightly weed-infested.
1.26	16-Dec-2007	pooka	* nuke puffs_cc_get{specific,usermount} for good * move prototypes for puffs_docc and puffs_dopufbuf into the public header, as they are should be exposed
1.25	04-Dec-2007	pooka	Part 3/n of extensive changes to message handling: Get rid of the original puffs_req(3) framework and use puffs_framebuf(3) instead for file system requests. It has the advantage of being suitable for transporting a distributed message passing protocol and therefore us being able to run the file system server on any host. Ok, puffs is not quite here yet: libpuffs needs to grow request routing support and the message contents need to be munged into a host independent format. Saying which format would be telling, but it might begin with an X, end in an L and have the 13th character in the middle. Keep an eye out for the sequels: Parts 3+m/n.
1.24	04-Dec-2007	pooka	realloc() len+incr, not offset+incr. Makes space reservation actually work properly.
1.23	06-Sep-2007	pooka	document puffs_framev_framebuf_ccpromote()
1.22	06-Sep-2007	pooka	Give cmpframe() the ability to shortcircuit the comparison loop by signalling back that the buffer under examination is not a response at all.
1.21	01-Sep-2007	pooka	Add gotframe callback, which is called if an incoming frame matches no outstanding request.
1.20	25-Aug-2007	pooka	branches: 1.20.2; Pass usermount to errnotify, it might not be available there through other structures. CID 4496
1.19	21-Jul-2007	pooka	branches: 1.19.4; Return -1 and errno from puffs_framev_enqueue_waitevent instead of a direct error value.
1.18	20-Jul-2007	pooka	simplify send loop
1.17	20-Jul-2007	pooka	Add puffs_framev_enqueue_waitevent, which waits for read/write/error events from an i/o descriptor. It yields until an event happens and continues execution when the event loop notices the event has happened.
1.16	08-Jul-2007	pooka	g/c leftover
1.15	08-Jul-2007	pooka	make notifyfunction optional and do not call removeonclose() if none is given.
1.14	07-Jul-2007	pooka	Make puffs_framebuf better suited from environments with multiple sources: * in addition add/remove, allow enable/disable, which can be used to control events for descriptors without having to remove all the data associated with them * add directsend/receive, which can be used to pass the same buffer from the caller to read/writeframe and back again * add flags to enqueue functions and allow urgent buffers to be processed as the next PDU
1.13	20-May-2007	pooka	Introduce puffs_framev_framebuf_ccpromote, which can be used to turn a reference to puffs_framebuf in the file system from a cb/justsend operation to a cc wait, should the file system find itself desiring the result.
1.12	20-May-2007	pooka	Change the signature of puffs_respcmp to return -1,0,1 for smaller, equal, larger, respectively instead of 0/1 for non/equal. This will allow sorting the buffers for faster matching in libpuffs. While here, change the name from respcmp to framecmp, as that better reflects the purpose. NOTE! there is no obvious way to make compilation fail for file systems which may already be using this feature (although I don't think there are any outside our tree, as the feature is two weeks old). Nevertheless, non-updated file systems will fail very quickly.
1.11	19-May-2007	pooka	Allocate buf in 4k chunks instead of 64k chunks. If we are seriously reclaiming and the network/server is slow, we might have thousands of buffers allocated at the same time causing the process to run out of vm space. Rate limiting the number of outstanding ops would be a nicer choice, but that requires more complex changes.
1.10	16-May-2007	pooka	Add error parameter to puffs_framev_cb to signal if there is a result of if the function should just do resource cleanup.
1.9	16-May-2007	pooka	switch the arbitrary default errno value for network i/o from ECONNABORTED to ECONNRESET, which is at least a little bit better
1.8	15-May-2007	pooka	Step 3 and currently final step of a more general purpose event loop: support removal and addition of i/o file descriptors on the fly. * detect closed file descriptors * automatically free waiters of a dead file descriptor * give the file server the possibility to specify a callback which notifies of a dead file descriptor * move loop function to be a property of the mainloop instead of framebuf (doesn't change effective behaviour) * add the possibility to configure a timespec parameter which attempts to call the loop function periodically * move the event loop functions from the puffs_framebuf namespace to puffs_framev to differential between pure memory management functions
1.7	12-May-2007	pooka	use size_t instead of int to calm over-eager lint
1.6	11-May-2007	pooka	Step 2 of general-purpose eventloop: get rid of puffs_framebuf_eventloop(). puffs_mainloop() now handles that functionality as well.
1.5	11-May-2007	pooka	Step 1 in moving towards a general-purpose eventloop: allow the framebuf event loop to take n i/o fd's as parameters and also allow dynamic add/remove of fd's. (not tested except for one fd, but more changes coming soon)
1.4	09-May-2007	pooka	Remove the puffs_usermount argument from puffs_req_handle(). It can be found from the other arguments and this way puffs_req_handle() is more symmetric with the other puffs_req() routines.
1.3	06-May-2007	pooka	implement the conveniently-already-prototyped puffs_framebuf_remaining(), which is just shorthand for tellsize() - telloff()
1.2	06-May-2007	pooka	Remove the user-specified argument for the framebuf loop callback. It can be generated through other means if it is desired.
1.1	05-May-2007	pooka	Add puffs "frame buffers", which are an abstraction of the buffering and event handling mechanisms required in file servers with blocking I/O backends. puffs_framebuf is built on the concept of puffs_cc and uses those to multiplex execution where needed. File systems are required to implement three methods: * read frame * write frame * compare if frame is a response to the given one Memory management is provided by puffs_framebuf, but the file systems must still, of course, interpret the protocol and do e.g. byte order conversion. As always, puffs_framebuf is work in progress. Current users are mount_psshfs and mount_9p.
1.19.4.2	21-Jul-2007	pooka	Return -1 and errno from puffs_framev_enqueue_waitevent instead of a direct error value.
1.19.4.1	21-Jul-2007	pooka	file framebuf.c was added on branch matt-mips64 on 2007-07-21 09:29:08 +0000
1.20.2.3	23-Mar-2008	matt	sync with HEAD
1.20.2.2	09-Jan-2008	matt	sync with HEAD
1.20.2.1	06-Nov-2007	matt	sync with HEAD
1.28.6.1	18-Sep-2008	wrstuden	Sync with wrstuden-revivesa-base-2.
1.30.6.2	30-Oct-2012	yamt	sync with head
1.30.6.1	17-Apr-2012	yamt	sync with head
1.32.22.1	11-May-2017	pgoyette	Sync with HEAD
1.36	11-Sep-2021	andvar	Add missing double p and d for stopped and overriden accordingly. Fix few more typos along the way, mainly in copy-pasted comments.
1.35	02-Nov-2019	tnn	puffs_null_node_fsync: don't leak directory handle Directory handles returned by opendir must be closed by closedir. Also directory(3) says we mustn't close(2) descriptors returned by dirfd(3)
1.34	23-Sep-2019	christos	Restore binary compatibility by using the statvfs90 structure internally.
1.33	25-Nov-2011	manu	branches: 1.33.40; Rollback previous change as it breaks autmates tests
1.32	24-Nov-2011	manu	Remove files specified in pcn. When remove files using name from pnode, another link on this file can be unlinked. E.g. "touch 1; ln 1 2; rm 2" will remove file named "1". Thus puffs_null_node_remove should remove directory entry which name is provided by pcn (as said in puffs_ops.3). Caller should provide appropriately initialized pcn. From Evgeniy Ivanov <lolkaantimat@gmail.com>
1.31	24-Nov-2011	manu	Remove pnode of overwritten file. When puffs_null_node_rename() overwrites existing file, its pnode must�be removed, because src pnode already represents this file. From Evgeniy Ivanov <lolkaantimat@gmail.com>
1.30	27-Jun-2011	manu	branches: 1.30.2; From Evgeniy Ivanov: puffs_node_readdir() should set eofflag to one, when end-of-directory is reached.
1.29	24-Jun-2011	manu	Commit fix for misc/45029. Approved by pooka@
1.28	18-Oct-2009	pooka	include some headers we use
1.27	08-Jan-2009	lukem	Cast a value to time_t (not unsigned) when comparing to timespec tv_sec
1.26	26-Nov-2008	pooka	support: a) readdir cookies and eofflag b) file handle routines. File handle routines are REALLY ugly and are not stable, i.e. the file handle will be lost if the file server goes byebye. They probably should be redone to not use getfh(), but for nullfs I wanted to make them the same as when the file system is accessed without the null layer. Well, turns out it doesn't pass anyway since the kernel adds the fsid tags for fhandle_t. Anyway, it's useful for some limited hacking purposes, so I'll commit this since it works .. kinda.
1.25	12-Aug-2008	pooka	Replace void * by puffs_cookie_t where appropriate. No functional change.
1.24	30-Nov-2007	pooka	branches: 1.24.8; Rototill. Ok, ok, a few more words about it: stop holding puffs_cc as a holy value and passing it around to almost every possible place (popquiz: which kernel variable does this remind you of?). Instead, pass the natural choice, puffs_usermount, and fetch puffs_cc via puffs_cc_getcc() only in routines which actually need it. This not only simplifies code, but (thanks to the introduction of puffs_cc_getcc()) enables constructs which weren't previously sanely possible, say layering as a curious example. There's still a little to do on this front, but this was the major fs interface blast.
1.23	27-Nov-2007	pooka	Remove "puffs_cid" from the puffs interface following l-removal from the kernel vfs interfaces. puffs_cc_getcaller(pcc) can be used now should the same information be desired.
1.22	05-Nov-2007	pooka	Call puffs_genfs_reclaim directly instead of through an extra jumpgate.
1.21	01-Jul-2007	pooka	branches: 1.21.4; Instead of requesting various pieces of information from node-creating methods using "please fill me"-pointers, pass "struct puffs_newinfo *" and provide methods to fill in new node info.
1.20	01-Jul-2007	pooka	Instead of supplying a plain pid, supply an abstract struct puffs_cid *, which can currently be used to query the pid and lwpid.
1.19	25-Jun-2007	pooka	fold in common code for node creation. no functional change.
1.18	24-Jun-2007	pooka	Since we locate in-memory nodes based on the inode number, stat a file always after creation to cache the inode number given by the backend file system. Otherwise we would not find a newly created node from incore and create another one. In practise this was pretty well hidden by the kernel name cache.
1.17	24-Jun-2007	pooka	Introduce puffs_null_setops(), which initializes the ops vector with puffs nullfs ops.
1.16	24-Jun-2007	pooka	Introduce puffs_pn_remove() to signal that a node has been deleted and destroyed, but not yet reclaimed. This prevents puffs_pn_nodewalk() from returning stale entries. Make nullfs use this (some file systems are a bit too happy with recycling inode numbers).
1.15	24-Jun-2007	pooka	don't return errno from a function which returns an fd - there's a "little" room for confusion. instead, return -1 and let the caller grab errno.
1.14	06-Jun-2007	pooka	Move puffs to a two clause license where it already isn't so. And as agc pointed out, even files with the third clause were already effectively two clause because of a slight bug in the language...
1.13	12-Apr-2007	pooka	Hide struct puffs_usermount from the rest of the world and provide accessors for interesting data in it. Namely, you can now get pu->pu_privdata with puffs_getspecific(), pu->pu_pn_root with puffs_set/getroot() and pu->pu_maxreqlen with puffs_getmaxreqlen().
1.12	11-Apr-2007	pooka	* support nodetofh and fhtonode callbacks * use new signature of node_readdir
1.11	16-Mar-2007	pooka	va->va_mode doesn't contain the full argument to the mknod() system call, so introduce puffs_addvtype2mode() and use that in null.c to generate the proper syscall argument
1.10	16-Mar-2007	pooka	return errno instead of -1 if file is not present in lookup
1.9	13-Mar-2007	pooka	Create in-memory nodes only per unique inode, not unique pathname. Solves problem with e.g. cache coherency of hardlinked files.
1.8	15-Feb-2007	pooka	Readlink should place the length of the link (without terminating nul) in the length field. Make it so.
1.7	15-Feb-2007	pooka	use puffs_path_walkcmp() instead of a homegrown comparison routine
1.6	15-Feb-2007	pooka	* get rid of the holy kernel-given ISDOTDOT, and instead define a macro which does strcmp against ".." and (the untranslated) componentname * make PUFFS_FLAG_BUILDPATH build paths also if dotdot is the case, and adapt the regular path objects to this * make nullfs lookup readable because we can now get rid of dotdot processing there
1.5	27-Jan-2007	agc	If we can't set the attributes when making the directory, use rmdir to remove the directory, not unlink.
1.4	15-Jan-2007	pooka	Checkpoint some more work in progress: for the benefit those file system backends which operate purely based on paths, push out more path management into the library and make path management more abstract: enable a file system to define a bunch of path management callbacks, which are used by the framework. Management of normal /this/is/a/path type paths is provided by the library.
1.3	11-Jan-2007	pooka	implement fsync
1.2	11-Jan-2007	pooka	* do the l-variations of syscalls, since some nodes are symlinks * truncate only regular files to set size * do the chmod()-dance for cache flush to now write-protected files until I can think of a nicer way to solve this
1.1	11-Jan-2007	pooka	Add preliminary code for a nullfs layer, which can be used to mount a directory hierarchy to another point, just like with the kernel nullfs. This is not really a layering scheme yet, but it should evolve into one. Currently it can just be used to do 1:1 mapping.
1.21.4.2	09-Jan-2008	matt	sync with HEAD
1.21.4.1	06-Nov-2007	matt	sync with HEAD
1.24.8.1	18-Sep-2008	wrstuden	Sync with wrstuden-revivesa-base-2.
1.30.2.1	17-Apr-2012	yamt	sync with head
1.33.40.1	13-Apr-2020	martin	Mostly merge changes from HEAD upto 20200411
1.38	08-Mar-2021	christos	cast to the proper enums for lint
1.37	31-Oct-2014	manu	libpuffs support for fallocate and fdiscard operations
1.36	15-Mar-2012	joerg	branches: 1.36.10; Add __printflike attribution to use vprintf and friends with an argument as format string.
1.35	20-Aug-2010	pooka	branches: 1.35.6; print to stderr instead of stdout
1.34	11-Jul-2010	pooka	Export the opid -> opstr tables (for outside debugging use).
1.33	21-May-2010	pooka	adjust for protocol changes
1.32	21-May-2010	pooka	Don't try to print the names of messages we don't know about.
1.31	20-May-2010	pooka	* fix typo * fix "that's not supposed to be there" causing off-by-one in array * make compile regardless of dev_t size
1.30	07-Jan-2010	pooka	Be slightly more forgiving about unknown operation types.
1.29	07-Jan-2010	pooka	Dump contents of struct vattr -- i never remember which members the kernel passes when, so this helps tapping that info.
1.28	07-Jan-2010	pooka	* make indentation a little less ridiculous * note that interfaces provided by <puffsdump.h> are not stable and are to be used only for debug builds * remove a few unimplemented/used dump interfaces
1.27	06-Apr-2009	pooka	dump readdir offset
1.26	06-Apr-2009	pooka	* dump op-specific return info before the generic one * dump readdir response values
1.25	28-Dec-2008	christos	branches: 1.25.2; fix for dev_t 64 bits.
1.24	12-Aug-2008	pooka	Replace void * by puffs_cookie_t where appropriate. No functional change.
1.23	27-Jan-2008	pooka	branches: 1.23.6; For operations with a target (remove, rmdir, link), print the target cookie.
1.22	25-Dec-2007	pooka	* dump r/w return resid * fix RV dump order to be more logical
1.21	09-Dec-2007	pooka	print newnode cookie value for operations which create nodes
1.20	04-Dec-2007	pooka	add flushops
1.19	29-Nov-2007	pooka	print mode in open
1.18	16-Nov-2007	pooka	dump lookup return values
1.17	29-Oct-2007	pooka	Sprinkle some hackish locks here and there, just enough to allow libp2k & rump to run multithreaded.
1.16	26-Oct-2007	pooka	dump pid/lwp of caller
1.15	11-Oct-2007	pooka	Part 1/n of some pretty extensive changes to how the kernel module interacts with the userspace file server: * since the kernel-user communication is not purely request-response anymore (hasn't been since 2006), try to rename some "request" to "message". more similar mangling will take place in the future. * completely rework how messages are allocated. previously most of them were borrowed from the stack (originally all of them), but now always allocate dynamically. this makes the structure of the code much cleaner. also makes it possible to fix a locking order violation. it enables plenty of future enhancements. * start generalizing the transport interface to be independent of puffs * move transport interface to read/write instead of ioctl. the old one had legacy design problems, and besides, ioctl's suck. implement a very generic version for now; this will be worked on later hopefully some day reaching "highly optimized". * implement libpuffs support behind existing library request interfaces. this will change eventually (I hate those interfaces)
1.14	29-Sep-2007	pooka	support error notifications
1.13	22-Aug-2007	pooka	branches: 1.13.2; Dump lookup/componentname specific info in opdump. (sigh, I guess I guess I'll never do opdump properly at this rate)
1.12	07-Aug-2007	pooka	dump further info for read/write
1.11	30-Jun-2007	pooka	print time elapsed between calls
1.10	06-Jun-2007	pooka	Move puffs to a two clause license where it already isn't so. And as agc pointed out, even files with the third clause were already effectively two clause because of a slight bug in the language...
1.9	20-Mar-2007	pooka	initial support for cacheops
1.8	18-Feb-2007	pooka	dump return values for operations
1.7	26-Jan-2007	pooka	add interfaces for suspending the file system
1.6	05-Dec-2006	pooka	adapt to kernel change: Allow multiple requests to be transferred in each GET/PUTOP
1.5	07-Nov-2006	pooka	* support FAF (Fire-And-Forget) class operations * adapt to new lookup
1.4	23-Oct-2006	pooka	dump correct name for vfs operations
1.3	23-Oct-2006	christos	use portable prints
1.2	23-Oct-2006	christos	pass lint; fix install.
1.1	22-Oct-2006	pooka	add libpuffs - the userspace programming interface for puffs This work was initially started and completed for Google SoC 2005 and tweaked to work a bit better in the past few weeks. While being far from complete, it is functional enough to be able and stable to host a fairly general-purpose in-memory file system in userspace. Even so, puffs should be considered experimental and no binary compatibility for interfaces or crash-freedom or zero security implications should be relied upon just yet. The GSoC project was mentored by William Studenmund and the final review for the code was done by Christos.
1.13.2.3	23-Mar-2008	matt	sync with HEAD
1.13.2.2	09-Jan-2008	matt	sync with HEAD
1.13.2.1	06-Nov-2007	matt	sync with HEAD
1.23.6.1	18-Sep-2008	wrstuden	Sync with wrstuden-revivesa-base-2.
1.25.2.1	13-May-2009	jym	Sync with HEAD. Third (and last) commit. See http://mail-index.netbsd.org/source-changes/2009/05/13/msg221222.html
1.35.6.1	17-Apr-2012	yamt	sync with head
1.36.10.1	05-Nov-2014	snj	Pull up following revision(s) (requested by manu in ticket #181): lib/libperfuse/fuse.h: revision 1.6 lib/libperfuse/ops.c: revision 1.78 lib/libperfuse/perfuse.c: revision 1.35 lib/libperfuse/perfuse_priv.h: revision 1.36 lib/libpuffs/dispatcher.c: revision 1.48 lib/libpuffs/opdump.c: revision 1.37 lib/libpuffs/puffs.c: revision 1.118 lib/libpuffs/puffs.h: revision 1.126 lib/libpuffs/puffs_ops.3: revisions 1.40-1.41 sys/fs/puffs/puffs_msgif.h: revision 1.82-1.83 sys/fs/puffs/puffs_msgif.h: revision 1.82 sys/fs/puffs/puffs_vnops.c: revision 1.196 Add PUFFS support for fallocate and fdiscard operations -- libpuffs support for fallocate and fdiscard operations -- Add PUFFS_HAVE_FALLOCATE in puffs_msgif.h so that filesystem can decide at build time wether fallocate is usable -- FUSE fallocate support There seems to be no fdiscard FUSE operation at the moment, hence that one is left unused.
1.8	12-Aug-2008	pooka	Replace void * by puffs_cookie_t where appropriate. No functional change.
1.7	06-Jun-2007	pooka	branches: 1.7.12; Move puffs to a two clause license where it already isn't so. And as agc pointed out, even files with the third clause were already effectively two clause because of a slight bug in the language...
1.6	01-May-2007	pooka	Introduce PUFFS_FLAG_HASHPATH, which for PUFFS_FLAG_BUILDPATH file servers creates a hash of the node's path into po->po_hash to avoid a full pathcmp every time.
1.5	15-Feb-2007	pooka	in case rootpath is "/", make dotdot lookup in "/foo" build a path of "/", not ""
1.4	15-Feb-2007	pooka	Introduce puffs_path_walkcmp(), which is to be called from nodewalk and compares the path of the node against the given pathobject. Also make comparison method take a flag to indicate if it should check if the second path is a true prefix of the first. plus some namespace cleanup
1.3	15-Feb-2007	pooka	* get rid of the holy kernel-given ISDOTDOT, and instead define a macro which does strcmp against ".." and (the untranslated) componentname * make PUFFS_FLAG_BUILDPATH build paths also if dotdot is the case, and adapt the regular path objects to this * make nullfs lookup readable because we can now get rid of dotdot processing there
1.2	06-Feb-2007	pooka	make it possible for the root path to be "/" and the next path to be "/foo" instead of "//foo"
1.1	15-Jan-2007	pooka	Checkpoint some more work in progress: for the benefit those file system backends which operate purely based on paths, push out more path management into the library and make path management more abstract: enable a file system to define a bunch of path management callbacks, which are used by the framework. Management of normal /this/is/a/path type paths is provided by the library.
1.7.12.1	18-Sep-2008	wrstuden	Sync with wrstuden-revivesa-base-2.
1.13	16-Aug-2012	manu	Fix regression that has been introduced when the lookup/reclaim race condition was addressed in libpuffs by counting lookups. The fix assumes that cookies map to struct puffs_cookie, which has not been documented as a requirement for filesystems using libpuffs. As an example, we got burnt by this assumption in libp2k (kern/46734), and we fixed bit by actually mapping libp2k cookies to struct puffs_node. It is unlikely, but there may be third party filesystems that use cookies unmapped to struct puffs_node, and they were left broken for now. - we introduce a puffs_init() flag PUFFS_FLAG_PNCOOKIE that let filesystems inform libpuffs that they map cookies to struct puffs_node. Is that flag is used, the lookup/reclaim race condition fix is enabled. We enable the flag for libp2k. - filesystems that use puffs_pn_new() obviouslty use struct puffs_node and gain PUFFS_FLAG_PNCOOKIE automatically even if they did not specify it in puffs_init(). This include all our PUFFS filesystem in-tree except libp2k. - for filesystems not willing to use struct puffs_node, we introduce a reclaim2 vnop, which is reclaim with an additionnal lookup count argument. This vnop let the filesystem implement the lookup/reclaim race fix on its own.
1.12	18-Apr-2012	manu	- When using PUFFS_KFLAG_CACHE_FS_TTL, do not use puffs_node to carry attribute and TTL fora newly created node. Instead extend puffs_newinfo and add puffs_newinfo_setva() and puffs_newinfo_setttl() - Remove node_mk_common_final in libperfuse. It used to set uid/gid for a newly created vnode but has been made redundant along time ago since uid and gid are properly set in FUSE header. - In libperfuse, check for corner case where opc = 0 on INACTIVE and RECLAIM (how is it possible? Check for it to avoid a crash anyway) - In libperfuse, make sure we unlimit RLIMIT_AS and RLIMIT_DATA so that we do notrun out of memory because the kernel is lazy at reclaiming vnodes. - In libperfuse, cleanup style of perfuse_destroy_pn()
1.11	08-Apr-2012	manu	Add PUFFS_KFLAG_CACHE_FS_TTL flag to puffs_init(3) to use name and attribute cache with filesystem provided TTL. lookup, create, mknod, mkdir, symlink, getattr and setattr messages have been extended so that attributes and their TTL can be provided by the filesytem. lookup, create, mknod, mkdir, and symlink messages are also extended so that the filesystem can provide name TTL. The filesystem updates attributes and TTL using puffs_pn_getvap(3), puffs_pn_getvattl(3), and puffs_pn_getcnttl(3)
1.10	12-Aug-2008	pooka	branches: 1.10.2; 1.10.20; Replace void * by puffs_cookie_t where appropriate. No functional change.
1.9	17-Aug-2007	pooka	branches: 1.9.10; Add accessors for struct puffs_node to prepare for removal from exported headers.
1.8	01-Jul-2007	pooka	Instead of requesting various pieces of information from node-creating methods using "please fill me"-pointers, pass "struct puffs_newinfo *" and provide methods to fill in new node info.
1.7	24-Jun-2007	pooka	Introduce puffs_pn_remove() to signal that a node has been deleted and destroyed, but not yet reclaimed. This prevents puffs_pn_nodewalk() from returning stale entries. Make nullfs use this (some file systems are a bit too happy with recycling inode numbers).
1.6	06-Jun-2007	pooka	Move puffs to a two clause license where it already isn't so. And as agc pointed out, even files with the third clause were already effectively two clause because of a slight bug in the language...
1.5	12-Apr-2007	pooka	Hide struct puffs_usermount from the rest of the world and provide accessors for interesting data in it. Namely, you can now get pu->pu_privdata with puffs_getspecific(), pu->pu_pn_root with puffs_set/getroot() and pu->pu_maxreqlen with puffs_getmaxreqlen().
1.4	29-Mar-2007	pooka	Remove check against null parameter along with the XXX comment wondering why the check was there. CID 4408
1.3	15-Jan-2007	pooka	Checkpoint some more work in progress: for the benefit those file system backends which operate purely based on paths, push out more path management into the library and make path management more abstract: enable a file system to define a bunch of path management callbacks, which are used by the framework. Management of normal /this/is/a/path type paths is provided by the library.
1.2	11-Jan-2007	pooka	Add preliminary code for a nullfs layer, which can be used to mount a directory hierarchy to another point, just like with the kernel nullfs. This is not really a layering scheme yet, but it should evolve into one. Currently it can just be used to do 1:1 mapping.
1.1	29-Dec-2006	pooka	checkpoint some experimental work-in-progress, namely: Add support for having multiple outstanding operations. This is done by exposing enough interfaces so that it is convenient to have the main event loop in the implementation itself and by providing a continuation framework for convinient blocking and rescheduling. works fine, but will undergo further cleanup & development
1.9.10.1	18-Sep-2008	wrstuden	Sync with wrstuden-revivesa-base-2.
1.10.20.2	13-Sep-2012	riz	Pull up following revision(s) (requested by manu in ticket #525): lib/libpuffs/dispatcher.c: revision 1.44 lib/libp2k/p2k.c: revision 1.57 lib/libpuffs/puffs_ops.3: revision 1.35 lib/libpuffs/pnode.c: revision 1.13 lib/libpuffs/puffs.3: revision 1.56 lib/libpuffs/puffs.h: revision 1.124 Fix regression that has been introduced when the lookup/reclaim race condition was addressed in libpuffs by counting lookups. The fix assumes that cookies map to struct puffs_cookie, which has not been documented as a requirement for filesystems using libpuffs. As an example, we got burnt by this assumption in libp2k (kern/46734), and we fixed bit by actually mapping libp2k cookies to struct puffs_node. It is unlikely, but there may be third party filesystems that use cookies unmapped to struct puffs_node, and they were left broken for now. - we introduce a puffs_init() flag PUFFS_FLAG_PNCOOKIE that let filesystems inform libpuffs that they map cookies to struct puffs_node. Is that flag is used, the lookup/reclaim race condition fix is enabled. We enable the flag for libp2k. - filesystems that use puffs_pn_new() obviouslty use struct puffs_node and gain PUFFS_FLAG_PNCOOKIE automatically even if they did not specify it in puffs_init(). This include all our PUFFS filesystem in-tree except libp2k. - for filesystems not willing to use struct puffs_node, we introduce a reclaim2 vnop, which is reclaim with an additionnal lookup count argument. This vnop let the filesystem implement the lookup/reclaim race fix on its own.
1.10.20.1	23-Apr-2012	riz	branches: 1.10.20.1.2; Pull up following revision(s) (requested by manu in ticket #195): lib/libskey/skeysubr.c: revision 1.27 lib/libkvm/kvm_getloadavg.c: revision 1.11 lib/libwrap/update.c: revision 1.9 lib/liby/yyerror.c: revision 1.9 lib/libpuffs/puffs_ops.3: revision 1.30 lib/libwrap/misc.c: revision 1.10 lib/libwrap/hosts_access.c: revision 1.20 lib/libpuffs/pnode.c: revision 1.11 lib/libperfuse/subr.c: revision 1.17 lib/libpuffs/pnode.c: revision 1.12 lib/libperfuse/subr.c: revision 1.18 lib/libwrap/options.c: revision 1.15 lib/libwrap/fix_options.c: revision 1.11 lib/libperfuse/ops.c: revision 1.52 lib/libperfuse/ops.c: revision 1.53 lib/libperfuse/ops.c: revision 1.54 lib/libwrap/hosts_ctl.c: revision 1.5 lib/libintl/gettext.c: revision 1.27 lib/libwrap/shell_cmd.c: revision 1.6 lib/libpuffs/dispatcher.c: revision 1.39 lib/libperfuse/perfuse_priv.h: revision 1.27 lib/libwrap/socket.c: revision 1.19 lib/libpuffs/puffs.3: revision 1.50 lib/libperfuse/perfuse_priv.h: revision 1.28 lib/libpuffs/puffs_priv.h: revision 1.45 lib/libpuffs/puffs.3: revision 1.51 lib/libperfuse/perfuse_priv.h: revision 1.29 lib/libwrap/percent_x.c: revision 1.5 lib/libpuffs/puffs.3: revision 1.52 lib/libperfuse/debug.c: revision 1.11 sys/fs/puffs/puffs_vnops.c: revision 1.165 lib/libwrap/tcpd.h: revision 1.13 sys/fs/puffs/puffs_vnops.c: revision 1.166 lib/libwrap/eval.c: revision 1.7 sys/fs/puffs/puffs_msgif.h: revision 1.78 sys/fs/puffs/puffs_vfsops.c: revision 1.101 lib/libwrap/rfc931.c: revision 1.9 lib/libwrap/clean_exit.c: revision 1.5 lib/libpuffs/puffs.h: revision 1.120 lib/libc/stdlib/jemalloc.c: revision 1.27 lib/librmt/rmtlib.c: revision 1.26 lib/libpuffs/puffs.h: revision 1.121 sys/fs/puffs/puffs_sys.h: revision 1.79 lib/librumpclient/rumpclient.c: revision 1.48 lib/libwrap/refuse.c: revision 1.5 lib/libperfuse/perfuse.c: revision 1.26 lib/libperfuse/perfuse.c: revision 1.27 tests/fs/puffs/t_fuzz.c: revision 1.5 lib/libperfuse/perfuse.c: revision 1.28 lib/libpuffs/dispatcher.c: revision 1.40 sys/fs/puffs/puffs_node.c: revision 1.24 lib/libwrap/diag.c: revision 1.9 lib/libintl/textdomain.c: revision 1.13 Use C89 function definition Add name and atttribute cache with filesytem provided TTL. lookup, create, mknod, mkdir, symlink, getattr and setattr messages have been extended so that attributes and their TTL can be provided by the filesytem. lookup, create, mknod, mkdir, and symlink messages are also extended so that the filesystem can provide name TTL. Add PUFFS_KFLAG_CACHE_FS_TTL flag to puffs_init(3) to use name and attribute cache with filesystem provided TTL. lookup, create, mknod, mkdir, symlink, getattr and setattr messages have been extended so that attributes and their TTL can be provided by the filesytem. lookup, create, mknod, mkdir, and symlink messages are also extended so that the filesystem can provide name TTL. The filesystem updates attributes and TTL using puffs_pn_getvap(3), puffs_pn_getvattl(3), and puffs_pn_getcnttl(3) Use new PUFFS_KFLAG_CACHE_FS_TTL option to puffs_init(3) so that FUSE TTL on name and attributes are used. This save many PUFFS operations and improves performances. PUFFS_KFLAG_CACHE_FS_TTL is #ifdef'ed in many places for now so that libperfuse can still be used on netbsd-5. Split file system. Comma fixes. Remove dangling "and". Bump date for previous. - Makesure update_va does not change vnode size when it should not. For instance when doing a fault-issued VOP_GETPAGES within VOP_WRITE, changing size leads to panic: genfs_getpages: past eof. -Handle ticks wrap around for vnode name andattribute timeout - When using PUFFS_KFLAG_CACHE_FS_TTL, do not use puffs_node to carry attribute and TTL fora newly created node. Instead extend puffs_newinfo and add puffs_newinfo_setva() and puffs_newinfo_setttl() - Remove node_mk_common_final in libperfuse. It used to set uid/gid for a newly created vnode but has been made redundant along time ago since uid and gid are properly set in FUSE header. - In libperfuse, check for corner case where opc = 0 on INACTIVE and RECLAIM (how is it possible? Check for it to avoid a crash anyway) - In libperfuse, make sure we unlimit RLIMIT_AS and RLIMIT_DATA so that we do notrun out of memory because the kernel is lazy at reclaiming vnodes. - In libperfuse, cleanup style of perfuse_destroy_pn() Do not set PUFFS_KFLAG_CACHE_FS_TTL for PUFFS tests
1.10.20.1.2.1	01-Nov-2012	matt	sync with netbsd-6-0-RELEASE.
1.10.2.3	30-Oct-2012	yamt	sync with head
1.10.2.2	23-May-2012	yamt	sync with head.
1.10.2.1	17-Apr-2012	yamt	sync with head
1.67	22-Jan-2022	pho	Allow calling puffs_mount(3) before puffs_daemon(3) puffs_daemon(3) creates a pipe before forking, and the parent process waits for the child to either complete puffs_mount(3) or fail. If a user calls puffs_daemon(3) after puffs_mount(3), the function deadlocks. While this error-reporting functionality is really a nice thing to have, deadlocking is not great. If the filesystem has already been mounted, puffs_mount(3) should just daemonize the process and return. This became an issue because fuse_daemonize(3) in FUSE API had no such requirement and some FUSE filesystems in the wild suffered deadlocks.
1.66	03-Dec-2021	pho	Use size_t in place of voff_t in puffs_setrootinfo(3) Chuck Silvers pointed out that voff_t was also supposed to be kernel-only. The correct type to use in userland would be off_t, but since changing vsize_t to either voff_t or off_t is an ABI change on 32-bit platforms, we use size_t knowing that it is technically incorrect.
1.65	03-Dec-2021	pho	Avoid using register_t or vsize_t in <puffs.h> <puffs.h> is a user-space header, and should not use any of kernel-only types. It's not reasonable to require user-land filesystems to #define _KERNTYPES.
1.64	27-Dec-2019	msaitoh	s/conjuction/conjunction/
1.63	03-Jul-2017	wiz	branches: 1.63.6; Remove workaround for ancient HTML generation code.
1.62	11-Apr-2016	christos	there is no pflags, just flags.
1.61	16-Feb-2015	wiz	Bump date for previous. filesystem -> file system
1.60	15-Feb-2015	manu	Add PUFFS_KFLAG_NOFLUSH_META to prevent sending metadata flush to FUSE FUSE filesystems do not expect to get metadata updates for [amc]time and size, they updates the value on their own after operations. The PUFFS PUFFS_KFLAG_NOFLUSH_META option prevents regular metadata cache flushes to the filesystem , and libperfuse uses it to match Linux FUSE behavior. While there, fix a bug in SETATTR: do not update kernel metadata cache from SETATTR reply when the request is asynchronous, as we do not have the reply yet.
1.59	18-Mar-2014	riastradh	branches: 1.59.4; Merge riastradh-drm2 to HEAD.
1.58	20-Jul-2013	wiz	Use Mt for email addresses.
1.57	16-Aug-2012	wiz	branches: 1.57.2; 1.57.4; Bump date for previous.
1.56	16-Aug-2012	manu	Fix regression that has been introduced when the lookup/reclaim race condition was addressed in libpuffs by counting lookups. The fix assumes that cookies map to struct puffs_cookie, which has not been documented as a requirement for filesystems using libpuffs. As an example, we got burnt by this assumption in libp2k (kern/46734), and we fixed bit by actually mapping libp2k cookies to struct puffs_node. It is unlikely, but there may be third party filesystems that use cookies unmapped to struct puffs_node, and they were left broken for now. - we introduce a puffs_init() flag PUFFS_FLAG_PNCOOKIE that let filesystems inform libpuffs that they map cookies to struct puffs_node. Is that flag is used, the lookup/reclaim race condition fix is enabled. We enable the flag for libp2k. - filesystems that use puffs_pn_new() obviouslty use struct puffs_node and gain PUFFS_FLAG_PNCOOKIE automatically even if they did not specify it in puffs_init(). This include all our PUFFS filesystem in-tree except libp2k. - for filesystems not willing to use struct puffs_node, we introduce a reclaim2 vnop, which is reclaim with an additionnal lookup count argument. This vnop let the filesystem implement the lookup/reclaim race fix on its own.
1.55	10-Aug-2012	wiz	Use more markup. New sentence, new line. Bump date for previous.
1.54	10-Aug-2012	manu	Add PUFFS_KFLAG_CACHE_DOTDOT so that vnodes hold a reference on their parent, keeping them active, and allowing to lookup .. without sending a request to the filesystem. Enable the featuure for perfused, as this is how FUSE works.
1.53	18-Apr-2012	wiz	Bump date for previous. New sentence, new line; remove trailing whitespace; fix typos; punctuation nits.
1.52	18-Apr-2012	manu	- When using PUFFS_KFLAG_CACHE_FS_TTL, do not use puffs_node to carry attribute and TTL fora newly created node. Instead extend puffs_newinfo and add puffs_newinfo_setva() and puffs_newinfo_setttl() - Remove node_mk_common_final in libperfuse. It used to set uid/gid for a newly created vnode but has been made redundant along time ago since uid and gid are properly set in FUSE header. - In libperfuse, check for corner case where opc = 0 on INACTIVE and RECLAIM (how is it possible? Check for it to avoid a crash anyway) - In libperfuse, make sure we unlimit RLIMIT_AS and RLIMIT_DATA so that we do notrun out of memory because the kernel is lazy at reclaiming vnodes. - In libperfuse, cleanup style of perfuse_destroy_pn()
1.51	08-Apr-2012	wiz	Split file system. Comma fixes. Remove dangling "and". Bump date for previous.
1.50	08-Apr-2012	manu	Add PUFFS_KFLAG_CACHE_FS_TTL flag to puffs_init(3) to use name and attribute cache with filesystem provided TTL. lookup, create, mknod, mkdir, symlink, getattr and setattr messages have been extended so that attributes and their TTL can be provided by the filesytem. lookup, create, mknod, mkdir, and symlink messages are also extended so that the filesystem can provide name TTL. The filesystem updates attributes and TTL using puffs_pn_getvap(3), puffs_pn_getvattl(3), and puffs_pn_getcnttl(3)
1.49	26-Dec-2011	njoly	branches: 1.49.2; Remove obsolete puffs_suspend(3) xref.
1.48	24-Nov-2011	manu	Documentation fix Evgeniy Ivanov <lolkaantimat@gmail.com>
1.47	12-Jan-2010	pooka	branches: 1.47.6; Add convenience routine puffs_unmountonsignal(), which does exactly that.
1.46	20-Feb-2009	pooka	Remove the paragraph saying interfaces will change without warning. Note that a stable version of puffs first appeared in NetBSD 5.0.
1.45	12-Dec-2008	pooka	branches: 1.45.2; add: puffs_setspecific(pu, private) Can be used to set the specific data after the call to puffs_init().
1.44	14-Nov-2008	pooka	note that puffs_daemon() must be called before puffs_mount()
1.43	14-Nov-2008	pooka	fix typographical snafu
1.42	06-Sep-2008	pooka	branches: 1.42.4; Add AsiaBSDCon 2008 paper to SEE ALSO.
1.41	12-Aug-2008	pooka	Replace void * by puffs_cookie_t where appropriate. No functional change.
1.40	25-May-2008	wiz	Remove trailing whitespace.
1.39	28-Jan-2008	pooka	branches: 1.39.4; 1.39.6; Implement lazy contexts for file systems using puffs_mainloop(). Previously each request was executed on its own callcontext and switched to every time the request was being processed. Now requests borrow the mainloop context and switch only if/when they yield. This takes two context switches away from a file system request bringing down the typical amounts 2->0 (e.g. dtfs) and 4->2 (e.g. psshfs). The interfaces for manually executing requests changed a bit: puffs_dispatch_create() and puffs_dispatch_exec() must now be used. They are not tested, as nothing in-tree wants them and I doubt anyone else is really interested in them either. Also do some misc code cleanup related to execution contexts. The "work-in-progress checkpoint" committed over a year ago was starting to look slightly weed-infested.
1.38	14-Jan-2008	pooka	Introduce PUFFS_STACKSIZE_MIN to avoid overloading 0 and getting a warning from the library when really wanting to say "just the minimum, please".
1.37	15-Dec-2007	pooka	document puffs_dopufbuf
1.36	04-Dec-2007	pooka	Part 3/n of extensive changes to message handling: Get rid of the original puffs_req(3) framework and use puffs_framebuf(3) instead for file system requests. It has the advantage of being suitable for transporting a distributed message passing protocol and therefore us being able to run the file system server on any host. Ok, puffs is not quite here yet: libpuffs needs to grow request routing support and the message contents need to be munged into a host independent format. Saying which format would be telling, but it might begin with an X, end in an L and have the 13th character in the middle. Keep an eye out for the sequels: Parts 3+m/n.
1.35	01-Dec-2007	wiz	Fix a typo that has migrated across files.
1.34	01-Dec-2007	wiz	Typo fixes.
1.33	16-Nov-2007	pooka	Introduce puffs_daemon(), which works like daemon(3) except that it also synchronizes with puffs_mount() and does not return (exit) in the parent process until the file system has been mounted. This makes it possible to reliably run e.g. mount_foo jippi /kai ; cd /kai/ee
1.32	06-Nov-2007	pooka	Open the kernel descriptor as part of mount(), not init(). Then it doesn't matter if someone fork()s or does other tricks between init() and mount() (and besides, now it's where it logically should be).
1.31	05-Nov-2007	pooka	Pull the daemonizing code out of the library mainloop into the file servers. Calling daemon() (i.e. fork()ing) inside a library can cause nice surprises for e.g. threaded programs. As discussed with Greg Oster & others.
1.30	05-Oct-2007	pooka	SEE ALSO paper
1.29	24-Sep-2007	pooka	Add refuse paper to SEE ALSO
1.28	03-Sep-2007	pooka	xref refuse, don't bump date
1.27	18-Jul-2007	pooka	branches: 1.27.4; 1.27.6; update
1.26	05-Jul-2007	pooka	document PUFFS_KFLAG_LOOKUP_FULLPNBUF
1.25	24-Jun-2007	pooka	document KFLAG_NOCACHE{_PAGE,_NAME,}
1.24	17-May-2007	pooka	reflect today's mount changes
1.23	15-May-2007	pooka	Step 3 and currently final step of a more general purpose event loop: support removal and addition of i/o file descriptors on the fly. * detect closed file descriptors * automatically free waiters of a dead file descriptor * give the file server the possibility to specify a callback which notifies of a dead file descriptor * move loop function to be a property of the mainloop instead of framebuf (doesn't change effective behaviour) * add the possibility to configure a timespec parameter which attempts to call the loop function periodically * move the event loop functions from the puffs_framebuf namespace to puffs_framev to differential between pure memory management functions
1.22	09-May-2007	pooka	List all current ways to write a file system. Also, add a reference to puffs_req(3) for the manual operation case.
1.21	09-May-2007	pooka	document PUFFS_KFLAG_WTCACHE and PUFFS_KFLAG_IAONDEMAND
1.20	01-May-2007	pooka	Introduce PUFFS_FLAG_HASHPATH, which for PUFFS_FLAG_BUILDPATH file servers creates a hash of the node's path into po->po_hash to avoid a full pathcmp every time.
1.19	16-Apr-2007	pooka	Move file system and node callbacks into their own manual page. Improve markup and descriptions a bit (not completely finished yet).
1.18	16-Apr-2007	pooka	document new interfaces
1.17	12-Apr-2007	pooka	document accessor routines added when making struct puffs_usermount implementation private
1.16	22-Mar-2007	pooka	document credentials & access control methods
1.15	20-Mar-2007	pooka	match puffs_node_access prototype with reality
1.14	13-Mar-2007	pooka	add puffs paper presented at AsiaBSDCon 2007 to "SEE ALSO"
1.13	08-Feb-2007	pooka	Nm the manual puffs instead of libpuffs
1.12	26-Jan-2007	pooka	document puffs_fs_suspend()
1.11	20-Jan-2007	pooka	xref puffs_req
1.10	20-Jan-2007	pooka	document puffs_getstate() and puffs_setstacksize()
1.9	20-Jan-2007	wiz	Sort SEE ALSO.
1.8	19-Jan-2007	pooka	Bring the documentation slightly more into sync with the current reality. It's still really a far cry from proper documentation, though.
1.7	01-Dec-2006	pooka	sauce catchup
1.6	30-Nov-2006	pooka	update with reality
1.5	23-Nov-2006	pooka	* document open, close and access * nomenclature improvements
1.4	19-Nov-2006	wiz	Some mdoc and grammar nits.
1.3	13-Nov-2006	pooka	add note to BUGS about the questionable nature of puffs_node
1.2	13-Nov-2006	pooka	sync with change to lookup
1.1	09-Nov-2006	pooka	initial documentation for the puffs library interface
1.27.6.2	18-Jul-2007	pooka	update
1.27.6.1	18-Jul-2007	pooka	file puffs.3 was added on branch matt-mips64 on 2007-07-18 22:23:38 +0000
1.27.4.4	23-Mar-2008	matt	sync with HEAD
1.27.4.3	09-Jan-2008	matt	sync with HEAD
1.27.4.2	08-Nov-2007	matt	sync with -HEAD
1.27.4.1	06-Nov-2007	matt	sync with HEAD
1.39.6.2	18-Sep-2008	wrstuden	Sync with wrstuden-revivesa-base-2.
1.39.6.1	23-Jun-2008	wrstuden	Sync w/ -current. 34 merge conflicts to follow.
1.39.4.1	04-Jun-2008	yamt	sync with head
1.42.4.2	16-Oct-2009	sborrill	Pull up the following revisions(s) (requested by pooka in ticket #1094): lib/libpuffs/puffs.3: revision 1.45 lib/libpuffs/puffs.c: revision 1.93 lib/libpuffs/puffs.h: revision 1.109 Add function puffs_setspecific(pu, private) which can be used to set the specific data after the call to puffs_init().
1.42.4.1	24-Feb-2009	snj	branches: 1.42.4.1.4; Pull up following revision(s) (requested by pooka in ticket #484): lib/libpuffs/puffs.3: revision 1.46 lib/libpuffs/puffs_cc.3: revision 1.13 lib/libpuffs/puffs_cred.3: revision 1.4 lib/libpuffs/puffs_flush.3: revision 1.8 lib/libpuffs/puffs_framebuf.3: revision 1.26 lib/libpuffs/puffs_node.3: revision 1.4 lib/libpuffs/puffs_ops.3: revision 1.23 lib/libpuffs/puffs_path.3: revision 1.4 lib/libpuffs/puffs_suspend.3: revision 1.3 Remove the paragraph saying interfaces will change without warning. Note that a stable version of puffs first appeared in NetBSD 5.0.
1.42.4.1.4.1	21-Apr-2010	matt	sync to netbsd-5
1.45.2.1	13-May-2009	jym	Sync with HEAD. Third (and last) commit. See http://mail-index.netbsd.org/source-changes/2009/05/13/msg221222.html
1.47.6.4	22-May-2014	yamt	sync with head. for a reference, the tree before this commit was tagged as yamt-pagecache-tag8. this commit was splitted into small chunks to avoid a limitation of cvs. ("Protocol error: too many arguments")
1.47.6.3	30-Oct-2012	yamt	sync with head
1.47.6.2	23-May-2012	yamt	sync with head.
1.47.6.1	17-Apr-2012	yamt	sync with head
1.49.2.4	27-Feb-2015	martin	Pull up following revision(s) (requested by manu in ticket #1260): lib/libpuffs/puffs.3: revision 1,55,1.60 sys/fs/puffs/puffs_msgif.h: revision 1.84 lib/libperfuse/ops.c: revision 1.83 sys/fs/puffs/puffs_sys.h: revision 1.89 sys/fs/puffs/puffs_vfsops.c: revision 1.116 lib/libperfuse/perfuse.c: revision 1.36 sys/fs/puffs/puffs_vnops.c: revision 1.200-1.202 Use more markup. New sentence, new line. Bump date for previous. Add PUFFS_KFLAG_NOFLUSH_META to prevent sending metadata flush to FUSE FUSE filesystems do not expect to get metadata updates for [amc]time and size, they updates the value on their own after operations. The PUFFS PUFFS_KFLAG_NOFLUSH_META option prevents regular metadata cache flushes to the filesystem , and libperfuse uses it to match Linux FUSE behavior. While there, fix a bug in SETATTR: do not update kernel metadata cache from SETATTR reply when the request is asynchronous, as we do not have the reply yet. Update file size after write without metadata flush If we do not use metadata flush, we must make sure the size is updated in the filesystem after a write, otherwise the next GETATTR will get us a stale value and the file will be truncated.
1.49.2.3	13-Sep-2012	riz	Pull up following revision(s) (requested by manu in ticket #525): lib/libpuffs/dispatcher.c: revision 1.44 lib/libp2k/p2k.c: revision 1.57 lib/libpuffs/puffs_ops.3: revision 1.35 lib/libpuffs/pnode.c: revision 1.13 lib/libpuffs/puffs.3: revision 1.56 lib/libpuffs/puffs.h: revision 1.124 Fix regression that has been introduced when the lookup/reclaim race condition was addressed in libpuffs by counting lookups. The fix assumes that cookies map to struct puffs_cookie, which has not been documented as a requirement for filesystems using libpuffs. As an example, we got burnt by this assumption in libp2k (kern/46734), and we fixed bit by actually mapping libp2k cookies to struct puffs_node. It is unlikely, but there may be third party filesystems that use cookies unmapped to struct puffs_node, and they were left broken for now. - we introduce a puffs_init() flag PUFFS_FLAG_PNCOOKIE that let filesystems inform libpuffs that they map cookies to struct puffs_node. Is that flag is used, the lookup/reclaim race condition fix is enabled. We enable the flag for libp2k. - filesystems that use puffs_pn_new() obviouslty use struct puffs_node and gain PUFFS_FLAG_PNCOOKIE automatically even if they did not specify it in puffs_init(). This include all our PUFFS filesystem in-tree except libp2k. - for filesystems not willing to use struct puffs_node, we introduce a reclaim2 vnop, which is reclaim with an additionnal lookup count argument. This vnop let the filesystem implement the lookup/reclaim race fix on its own.
1.49.2.2	12-Aug-2012	martin	branches: 1.49.2.2.2; Pull up following revision(s) (requested by manu in ticket #438): lib/libperfuse/perfuse_priv.h: revision 1.31 sys/fs/puffs/puffs_msgif.h: revision 1.80 sys/fs/puffs/puffs_vnops.c: revision 1.171 lib/libpuffs/puffs_ops.3: revision 1.31 sys/fs/puffs/puffs_vnops.c: revision 1.172 sys/fs/puffs/puffs_vnops.c: revision 1.173 sys/fs/puffs/puffs_vnops.c: revision 1.174 usr.sbin/perfused/perfused.c: revision 1.24 sys/fs/puffs/puffs_sys.h: revision 1.80 sys/fs/puffs/puffs_sys.h: revision 1.81 sys/fs/puffs/puffs_sys.h: revision 1.82 lib/libperfuse/subr.c: revision 1.19 lib/libperfuse/perfuse.c: revision 1.30 sys/fs/puffs/puffs_msgif.c: revision 1.90 sys/fs/puffs/puffs_msgif.c: revision 1.91 sys/fs/puffs/puffs_msgif.c: revision 1.92 lib/libperfuse/ops.c: revision 1.59 lib/libpuffs/puffs.3: revision 1.53 lib/libperfuse/debug.c: revision 1.12 lib/libpuffs/puffs.3: revision 1.54 sys/fs/puffs/puffs_vnops.c: revision 1.167 sys/fs/puffs/puffs_msgif.h: revision 1.79 usr.sbin/perfused/msg.c: revision 1.21 sys/fs/puffs/puffs_vfsops.c: revision 1.102 sys/fs/puffs/puffs_vfsops.c: revision 1.103 sys/fs/puffs/puffs_vfsops.c: revision 1.105 lib/libpuffs/puffs.h: revision 1.123 lib/libperfuse/perfuse_if.h: revision 1.20 lib/libperfuse/perfuse.c: revision 1.29 lib/libpuffs/dispatcher.c: revision 1.42 lib/libpuffs/dispatcher.c: revision 1.43 - Fix same vnodes associated with multiple cookies The scheme used to retreive known nodes on lookup was flawed, as it only used parent and name. This produced a different cookie for the same file if it was renamed, when looking up ../ or when dealing with multiple files associated with the same name through link(2). We therefore abandon the use of node name and introduce hashed lists of inodes. This causes a huge rewrite of reclaim code, which do not attempt to keep parents allocated until all their children are reclaimed - Fix race conditions in reclaim There are a few situations where we issue multiple FUSE operations for a PUFFS operation. On reclaim, we therefore have to wait for all FUSE operation to complete, not just the current exchanges. We do this by introducing node reference count with node_ref() and node_rele(). - Detect data loss caused by FAF VOP_PUTPAGES causes FAF writes where the kernel does not check the operation result. At least issue a warning on error. - Enjoy FAF shortcut on setattr No need to wait for the result if the kernel does not want it. There is however an exception for setattr that touch the size, we need to wait for completion because we have other operations queued for after the resize. - Fix fchmod() on write-open file fchmod() on a node open with write privilege will send setattr with both mode and size set. This confuses some FUSE filesystem. Therefore we send two FUSE operations, one for mode, and one for size. - Remove node TTL handling for netbsd-5 for simplicity sake. The code still builds on netbsd-5 but does not have the node TTL feature anymore. It works fine with kernel support on netbsd-6. - Improve PUFFS_KFLAG_CACHE_FS_TTL by reclaiming older inactive nodes. The normal kernel behavior is to retain inactive nodes in the freelist until it runs out of vnodes. This has some merit for local filesystems, where the cost of an allocation is about the same as the cost of a lookup. But that situation is not true for distributed filesystems. On the other hand, keeping inactive nodes for a long time hold memory in the file server process, and when the kernel runs out of vnodes, it produce reclaim avalanches that increase lattency for other operations. We do not reclaim inactive vnodes immediatly either, as they may be looked up again shortly. Instead we introduce a grace time and we reclaim nodes that have been inactive beyond the grace time. - Fix lookup/reclaim race condition. The above improvement undercovered a race condition between lookup and reclaim. If we reclaimed a vnode associated with a userland cookie while a lookup returning that same cookiewas inprogress, then the kernel ends up with a vnode associated with a cookie that has been reclaimed in userland. Next operation on the cookie will crash (or at least confuse) the filesystem. We fix this by introducing a lookup count in kernel and userland. On reclaim, the kernel sends the count, which enable userland to detect situation where it initiated a lookup that is not completed in kernel. In such a situation, the reclaim must be ignored, as the node is about to be looked up again. Fix hang unmount bug introduced by last commit. We introduced a slow queue for delayed reclaims, while the existing queue for unmount, flush and exist has been renamed fast queue. Both queues had timestamp for when an operation should be done, but it was useless for the fast queue, which is always used to run an operation ASAP. And the timestamp test had an error that turned ASAP into "at next tick", but nobody what there to wake the thread at next tick, hence the hang. The fix is to remove the useless and buggy timestamp test for fast queue. Rename slow sopreq queue into node sopreq queue, to refet the fact that is only intended for postponed node reclaims. When purging the node sopreq queue, do not call puffs_msg_sendresp(), as it makes no sense. Fix race condition between (create\|mknod\|mkdir\|symlino) and reclaim, just like we did it between lookup and reclaim. Missing bit in previous commit (prevent race between create\|mknod\|mkdir\|symlink and reclaim) Bump date for previous. New sentence, new line; remove trailing whitespace; fix typos; punctuation nits. Add PUFFS_KFLAG_CACHE_DOTDOT so that vnodes hold a reference on their parent, keeping them active, and allowing to lookup .. without sending a request to the filesystem. Enable the featuure for perfused, as this is how FUSE works. Missing bit in previous commit (PUFFS_KFLAG_CACHE_DOTDOT option to avoid looking up ..)
1.49.2.1	23-Apr-2012	riz	Pull up following revision(s) (requested by manu in ticket #195): lib/libskey/skeysubr.c: revision 1.27 lib/libkvm/kvm_getloadavg.c: revision 1.11 lib/libwrap/update.c: revision 1.9 lib/liby/yyerror.c: revision 1.9 lib/libpuffs/puffs_ops.3: revision 1.30 lib/libwrap/misc.c: revision 1.10 lib/libwrap/hosts_access.c: revision 1.20 lib/libpuffs/pnode.c: revision 1.11 lib/libperfuse/subr.c: revision 1.17 lib/libpuffs/pnode.c: revision 1.12 lib/libperfuse/subr.c: revision 1.18 lib/libwrap/options.c: revision 1.15 lib/libwrap/fix_options.c: revision 1.11 lib/libperfuse/ops.c: revision 1.52 lib/libperfuse/ops.c: revision 1.53 lib/libperfuse/ops.c: revision 1.54 lib/libwrap/hosts_ctl.c: revision 1.5 lib/libintl/gettext.c: revision 1.27 lib/libwrap/shell_cmd.c: revision 1.6 lib/libpuffs/dispatcher.c: revision 1.39 lib/libperfuse/perfuse_priv.h: revision 1.27 lib/libwrap/socket.c: revision 1.19 lib/libpuffs/puffs.3: revision 1.50 lib/libperfuse/perfuse_priv.h: revision 1.28 lib/libpuffs/puffs_priv.h: revision 1.45 lib/libpuffs/puffs.3: revision 1.51 lib/libperfuse/perfuse_priv.h: revision 1.29 lib/libwrap/percent_x.c: revision 1.5 lib/libpuffs/puffs.3: revision 1.52 lib/libperfuse/debug.c: revision 1.11 sys/fs/puffs/puffs_vnops.c: revision 1.165 lib/libwrap/tcpd.h: revision 1.13 sys/fs/puffs/puffs_vnops.c: revision 1.166 lib/libwrap/eval.c: revision 1.7 sys/fs/puffs/puffs_msgif.h: revision 1.78 sys/fs/puffs/puffs_vfsops.c: revision 1.101 lib/libwrap/rfc931.c: revision 1.9 lib/libwrap/clean_exit.c: revision 1.5 lib/libpuffs/puffs.h: revision 1.120 lib/libc/stdlib/jemalloc.c: revision 1.27 lib/librmt/rmtlib.c: revision 1.26 lib/libpuffs/puffs.h: revision 1.121 sys/fs/puffs/puffs_sys.h: revision 1.79 lib/librumpclient/rumpclient.c: revision 1.48 lib/libwrap/refuse.c: revision 1.5 lib/libperfuse/perfuse.c: revision 1.26 lib/libperfuse/perfuse.c: revision 1.27 tests/fs/puffs/t_fuzz.c: revision 1.5 lib/libperfuse/perfuse.c: revision 1.28 lib/libpuffs/dispatcher.c: revision 1.40 sys/fs/puffs/puffs_node.c: revision 1.24 lib/libwrap/diag.c: revision 1.9 lib/libintl/textdomain.c: revision 1.13 Use C89 function definition Add name and atttribute cache with filesytem provided TTL. lookup, create, mknod, mkdir, symlink, getattr and setattr messages have been extended so that attributes and their TTL can be provided by the filesytem. lookup, create, mknod, mkdir, and symlink messages are also extended so that the filesystem can provide name TTL. Add PUFFS_KFLAG_CACHE_FS_TTL flag to puffs_init(3) to use name and attribute cache with filesystem provided TTL. lookup, create, mknod, mkdir, symlink, getattr and setattr messages have been extended so that attributes and their TTL can be provided by the filesytem. lookup, create, mknod, mkdir, and symlink messages are also extended so that the filesystem can provide name TTL. The filesystem updates attributes and TTL using puffs_pn_getvap(3), puffs_pn_getvattl(3), and puffs_pn_getcnttl(3) Use new PUFFS_KFLAG_CACHE_FS_TTL option to puffs_init(3) so that FUSE TTL on name and attributes are used. This save many PUFFS operations and improves performances. PUFFS_KFLAG_CACHE_FS_TTL is #ifdef'ed in many places for now so that libperfuse can still be used on netbsd-5. Split file system. Comma fixes. Remove dangling "and". Bump date for previous. - Makesure update_va does not change vnode size when it should not. For instance when doing a fault-issued VOP_GETPAGES within VOP_WRITE, changing size leads to panic: genfs_getpages: past eof. -Handle ticks wrap around for vnode name andattribute timeout - When using PUFFS_KFLAG_CACHE_FS_TTL, do not use puffs_node to carry attribute and TTL fora newly created node. Instead extend puffs_newinfo and add puffs_newinfo_setva() and puffs_newinfo_setttl() - Remove node_mk_common_final in libperfuse. It used to set uid/gid for a newly created vnode but has been made redundant along time ago since uid and gid are properly set in FUSE header. - In libperfuse, check for corner case where opc = 0 on INACTIVE and RECLAIM (how is it possible? Check for it to avoid a crash anyway) - In libperfuse, make sure we unlimit RLIMIT_AS and RLIMIT_DATA so that we do notrun out of memory because the kernel is lazy at reclaiming vnodes. - In libperfuse, cleanup style of perfuse_destroy_pn() Do not set PUFFS_KFLAG_CACHE_FS_TTL for PUFFS tests
1.49.2.2.2.1	01-Nov-2012	matt	sync with netbsd-6-0-RELEASE.
1.57.4.1	23-Jul-2013	riastradh	sync with HEAD
1.57.2.1	20-Aug-2014	tls	Rebase to HEAD as of a few days ago.
1.59.4.1	27-Feb-2015	martin	Pull up following revision(s) (requested by manu in ticket #555): lib/libpuffs/puffs.3: revision 1.60 sys/fs/puffs/puffs_msgif.h: revision 1.84 lib/libperfuse/ops.c: revision 1.83 sys/fs/puffs/puffs_sys.h: revision 1.89 sys/fs/puffs/puffs_vfsops.c: revision 1.116 lib/libperfuse/perfuse.c: revision 1.36 sys/fs/puffs/puffs_vnops.c: revision 1.200-1.202 Add PUFFS_KFLAG_NOFLUSH_META to prevent sending metadata flush to FUSE FUSE filesystems do not expect to get metadata updates for [amc]time and size, they updates the value on their own after operations. The PUFFS PUFFS_KFLAG_NOFLUSH_META option prevents regular metadata cache flushes to the filesystem , and libperfuse uses it to match Linux FUSE behavior. While there, fix a bug in SETATTR: do not update kernel metadata cache from SETATTR reply when the request is asynchronous, as we do not have the reply yet. Update file size after write without metadata flush If we do not use metadata flush, we must make sure the size is updated in the filesystem after a write, otherwise the next GETATTR will get us a stale value and the file will be truncated.
1.63.6.1	08-Apr-2020	martin	Merge changes from current as of 20200406
1.129	19-Apr-2022	rillig	lib: remove CONSTCOND comment Since 2021-01-31, lint doesn't need it anymore for the common pattern of 'do ... while (0)'.
1.128	22-Jan-2022	pho	Allow calling puffs_mount(3) before puffs_daemon(3) puffs_daemon(3) creates a pipe before forking, and the parent process waits for the child to either complete puffs_mount(3) or fail. If a user calls puffs_daemon(3) after puffs_mount(3), the function deadlocks. While this error-reporting functionality is really a nice thing to have, deadlocking is not great. If the filesystem has already been mounted, puffs_mount(3) should just daemonize the process and return. This became an issue because fuse_daemonize(3) in FUSE API had no such requirement and some FUSE filesystems in the wild suffered deadlocks.
1.127	03-Dec-2021	pho	Use size_t in place of voff_t in puffs_setrootinfo(3) Chuck Silvers pointed out that voff_t was also supposed to be kernel-only. The correct type to use in userland would be off_t, but since changing vsize_t to either voff_t or off_t is an ABI change on 32-bit platforms, we use size_t knowing that it is technically incorrect.
1.126	03-Dec-2021	pho	Avoid using register_t or vsize_t in <puffs.h> <puffs.h> is a user-space header, and should not use any of kernel-only types. It's not reasonable to require user-land filesystems to #define _KERNTYPES.
1.125	30-Oct-2021	nia	puffs(3): Replace realloc(x * y) with reallocarr
1.124	30-Jun-2018	christos	use a switch.
1.123	08-Feb-2018	dholland	branches: 1.123.2; 1.123.4; Typos.
1.122	05-Nov-2017	christos	fix error messages (use __func__, quotes) also differentiate between canonical and non-relative.
1.121	31-May-2017	christos	Adjust to prototype.
1.120	17-Jun-2015	christos	fix error messages
1.119	22-Dec-2014	manu	Do not warn about relative path because of trailing slash libpuffs calls realpath() to obtain an absolute path to use for mounting. If the obtained path is different from the one given by the caller, a warning is issued. This included the situation where the path passed by the caller just have trailing slashes, a situation where we just want them to be striped without a warning.
1.118	31-Oct-2014	manu	libpuffs support for fallocate and fdiscard operations
1.117	14-Nov-2011	chs	branches: 1.117.18; fix crashes caused by using the results of getcontext() after the caller returns.
1.116	03-May-2011	manu	branches: 1.116.4; Call advlock method if supplied
1.115	06-Dec-2010	pooka	Remove PUFFS_WITH_THREADS ifdef garbage, just use libc weak symbols.
1.114	06-Jul-2010	pooka	unwhine lint
1.113	06-Jul-2010	pooka	stat mount target path only if we're actually goint to use it. This avoids errors from the comfd case when the mount target does not exist in our local namespace.
1.112	06-Jul-2010	pooka	set time_t compat where necessary
1.111	06-Jul-2010	pooka	When using the top secret PUFFS_COMFD model, write out size of the puffs kernel args before the contents. This allows parties which have no clue about the contents to proxy the protocol easily.
1.110	07-Jun-2010	pooka	Reenable pathconf now that the retval argument is register_t.
1.109	06-Jun-2010	pooka	Actually, disable pathconf for a while longer: the retval needs a little finetuning (not that everyone is jumping right to using it, but fixing will have to wait until tomorrow, and who knows what i'll forget during the night and how long it will eventually take).
1.108	06-Jun-2010	pooka	signal pathconf in the vector of supported ops
1.107	21-May-2010	pooka	* support extended attributes * bump major due to structure growth * add some spare space * remove ABI sillyness
1.106	19-May-2010	pooka	remove three years out-of-date comment
1.105	12-Jan-2010	pooka	Add convenience routine puffs_unmountonsignal(), which does exactly that.
1.104	07-Jan-2010	pooka	Use typecast instead of ULL for a less whiny lint on LP64.
1.103	07-Jan-2010	pooka	Rewrite puffs_exit() in terms of the PUFFSOP_UNMOUNT kernel request. Now server exit is a lot quite cleaner than previously.
1.102	07-Jan-2010	pooka	Set high bit on server->kernel requests to better differentiate between the direction of the request.
1.101	05-Dec-2009	pooka	Add an error handler which syslogs but does not abort() and make it the default. Most servers just shovel bits from their backend to the kernel without checking the contents and dumping core in case the backend gives garbage is a bit harsh.
1.100	18-Oct-2009	pooka	Make it possible to supply parameters to puffs_init() later (but before puffs_mount()).
1.99	17-Oct-2009	pooka	Dispatch puffs_node_abortop(). Note: We use the storage of puffs_cache_write from puffs_ops for this purpose. It's not issued by the kernel and hence currently unused, and this saves us from the trouble of bumping the lib major version.
1.98	08-Jan-2009	lukem	fix some sign comparison warnings
1.97	13-Dec-2008	dsl	Actually we need 'long int' to avoif a stupid 'truncating long to int' warning.
1.96	13-Dec-2008	dsl	Change type of 'n' to 'int' so that it can be printed with %d. The domain of the value is [-1..sizeof (int)] so a type of 'int' is fine.
1.95	12-Dec-2008	pooka	Close write-end of daemonpipe in reader to get EOF if the writer goes awol.
1.94	12-Dec-2008	pooka	Add puffs_cancel(), which allows to back down after puffs_init() and before puffs_mount(). (this should actually be integrated with puffs_exit(), but that routine is currently so non-sensible that I didn't bother now)
1.93	12-Dec-2008	pooka	add: puffs_setspecific(pu, private) Can be used to set the specific data after the call to puffs_init().
1.92	12-Aug-2008	pooka	branches: 1.92.4; 1.92.8; Replace void * by puffs_cookie_t where appropriate. No functional change.
1.91	11-Aug-2008	pooka	clear cached puffs_cc's in puffs_exit()
1.90	11-Aug-2008	pooka	* make sure we come out of mainloop in the main context. otherwise in certain cases the server would jump to hyperspace when unmounted * sprinkle some debug prints
1.89	28-Jan-2008	pooka	branches: 1.89.6; Implement lazy contexts for file systems using puffs_mainloop(). Previously each request was executed on its own callcontext and switched to every time the request was being processed. Now requests borrow the mainloop context and switch only if/when they yield. This takes two context switches away from a file system request bringing down the typical amounts 2->0 (e.g. dtfs) and 4->2 (e.g. psshfs). The interfaces for manually executing requests changed a bit: puffs_dispatch_create() and puffs_dispatch_exec() must now be used. They are not tested, as nothing in-tree wants them and I doubt anyone else is really interested in them either. Also do some misc code cleanup related to execution contexts. The "work-in-progress checkpoint" committed over a year ago was starting to look slightly weed-infested.
1.88	16-Jan-2008	pooka	Cache a number of execution contexts instead of re-initializing them every time. Speeds up pure in-memory file systems such as sysctlfs or dtfs quite a bit. For actual I/O-workhorses the result is of course less tasty.
1.87	16-Jan-2008	pooka	Use PROT_NONE page to protect bottom of stack from overrun
1.86	14-Jan-2008	pooka	Introduce PUFFS_STACKSIZE_MIN to avoid overloading 0 and getting a warning from the library when really wanting to say "just the minimum, please".
1.85	08-Jan-2008	pooka	liberate gas-huffin crack-puffin' assert
1.84	02-Jan-2008	pooka	fix pasto
1.83	02-Jan-2008	pooka	Also pass fromname in the "distributed" case.
1.82	02-Jan-2008	pooka	Provide some transparent "support" for "distributed" file systems. If the env variable PUFFS_COMFD is set, the descriptor value contained in it is used for communication instead of opening /dev/puffs and doing mount(2). This feature is obviously very undocumented and should not be used without adult supervision.
1.81	25-Dec-2007	pooka	Add a public interface for setting a cookie mapper.
1.80	05-Dec-2007	pooka	Send a response message for flush operations from the kernel instead of abusing the return value of write(2).
1.79	04-Dec-2007	pooka	Part 3/n of extensive changes to message handling: Get rid of the original puffs_req(3) framework and use puffs_framebuf(3) instead for file system requests. It has the advantage of being suitable for transporting a distributed message passing protocol and therefore us being able to run the file system server on any host. Ok, puffs is not quite here yet: libpuffs needs to grow request routing support and the message contents need to be munged into a host independent format. Saying which format would be telling, but it might begin with an X, end in an L and have the 13th character in the middle. Keep an eye out for the sequels: Parts 3+m/n.
1.78	29-Nov-2007	pooka	* allocate pcc as a slice from the stack instead of allocating it separately * provide puffs_cc_getcc() This is in preparation for the removal of you-should-guess-what as an argument to routines here and there and everywhere.
1.77	26-Nov-2007	pooka	print more info when receiving an abort
1.76	17-Nov-2007	pooka	puffs_mount(): * initialize values better * always go through "fail" to see if puffs_daemon() needs to be woken up * open device r/w instead of r/o
1.75	16-Nov-2007	pooka	Introduce puffs_daemon(), which works like daemon(3) except that it also synchronizes with puffs_mount() and does not return (exit) in the parent process until the file system has been mounted. This makes it possible to reliably run e.g. mount_foo jippi /kai ; cd /kai/ee
1.74	06-Nov-2007	pooka	Open the kernel descriptor as part of mount(), not init(). Then it doesn't matter if someone fork()s or does other tricks between init() and mount() (and besides, now it's where it logically should be).
1.73	05-Nov-2007	pooka	Pull the daemonizing code out of the library mainloop into the file servers. Calling daemon() (i.e. fork()ing) inside a library can cause nice surprises for e.g. threaded programs. As discussed with Greg Oster & others.
1.72	29-Oct-2007	pooka	Sprinkle some hackish locks here and there, just enough to allow libp2k & rump to run multithreaded.
1.71	28-Oct-2007	pooka	Add possibility to set pre- and post callbacks which will be called for all operations before or after executing the specific callback (the api is not be final yet, though).
1.70	26-Oct-2007	pooka	Add some very initial support for a threading worker model as an alternative to the (vastly superior ;) continuation model. This is very preliminary stuff and not compiled by default (which it even won't do without some other patches I cannot commit yet). The raison d'commit of the patch is a snippet which ensures proper in-order dispatching of all operations, including those which don't require a response. Previously many of them would be dispatched simultaneosly, e.g. fsync and reclaim on the same node, which obviously isn't all that nice for correct operation.
1.69	26-Oct-2007	pooka	mmap stacks, makes it possible to align them
1.68	25-Oct-2007	pooka	Make pu_haskq a flag instead of its own variable.
1.67	21-Oct-2007	pooka	Add puffs_cc_schedule() which marks a pcc runnable and will pass execution to it when in the main loop the next time.
1.66	11-Oct-2007	pooka	Part 1/n of some pretty extensive changes to how the kernel module interacts with the userspace file server: * since the kernel-user communication is not purely request-response anymore (hasn't been since 2006), try to rename some "request" to "message". more similar mangling will take place in the future. * completely rework how messages are allocated. previously most of them were borrowed from the stack (originally all of them), but now always allocate dynamically. this makes the structure of the code much cleaner. also makes it possible to fix a locking order violation. it enables plenty of future enhancements. * start generalizing the transport interface to be independent of puffs * move transport interface to read/write instead of ioctl. the old one had legacy design problems, and besides, ioctl's suck. implement a very generic version for now; this will be worked on later hopefully some day reaching "highly optimized". * implement libpuffs support behind existing library request interfaces. this will change eventually (I hate those interfaces)
1.65	01-Oct-2007	pooka	Pass reason string to pu_errnotify. Print it in the default version before aborting.
1.64	27-Sep-2007	pooka	silence lint. also noticed by xtraeme
1.63	27-Sep-2007	pooka	Add error notifications, which are used to deliver errors from the kernel to the file server for silly things the file server did, e.g. attempting to create a file with size VSIZENOTSET. The file server can handle these as it chooses, but the default action is for it to throw its hands in the air and sing "goodbye, cruel world, it's over, walk on by".
1.62	11-Aug-2007	pooka	branches: 1.62.2; run realpath() on the mountpath
1.61	20-Jul-2007	pooka	branches: 1.61.4; Add puffs_framev_enqueue_waitevent, which waits for read/write/error events from an i/o descriptor. It yields until an event happens and continues execution when the event loop notices the event has happened.
1.60	19-Jul-2007	pooka	Don't float puffs_kargs needlessly around after mount, saves a couple k of memory (and there was much rejoycing!)
1.59	19-Jul-2007	pooka	Use ioctl(PUFFSREQSIZEOP) instead of mount(MNT_GETARGS) to fetch the maximum request size after mount. Calling mount(MNT_GETARGS) from the file server is currently not kosher, as it vrele()s the root vnode, potentially causing an inactive, which the file server cannot handle while it itself is inactive in the kernel (deadlock).
1.58	17-Jul-2007	pooka	* add mntfromname parameter to puffs_init() * set it in file servers
1.57	14-Jul-2007	dsl	Add new 'data_len' parameter to mount system call. Add an additional call with MNT_GETARGS to retrieve the modified information instead of relying on the kernel code abusing the mount interface during mount.
1.56	07-Jul-2007	pooka	Make puffs_framebuf better suited from environments with multiple sources: * in addition add/remove, allow enable/disable, which can be used to control events for descriptors without having to remove all the data associated with them * add directsend/receive, which can be used to pass the same buffer from the caller to read/writeframe and back again * add flags to enqueue functions and allow urgent buffers to be processed as the next PDU
1.55	05-Jul-2007	pooka	* reorder calls to get results processed even if our I/O descriptors go back to zero * don't give up on the read-side of an fd if there is still unreceived data
1.54	02-Jul-2007	pooka	Get rid of the "int *refs" parameter to inactive: the same can be accomplished now with puffs_setbacks.
1.53	11-Jun-2007	pooka	print a specific warning if /dev/puffs cannot be opened in init
1.52	06-Jun-2007	pooka	Move puffs to a two clause license where it already isn't so. And as agc pointed out, even files with the third clause were already effectively two clause because of a slight bug in the language...
1.51	21-May-2007	pooka	remember to put the results of the final operations (including unmount) to the kernel in mainloop
1.50	18-May-2007	pooka	* stuff results back to kernel before calling kevent - the file server loop function might generate some results. and this is still "after" event handling (except for the first call, but I'm not too keen on optimizing for that) * don't be such a baby about EINTR from kevent(). if we get it, suck it up and continue instead of quitting
1.49	17-May-2007	pooka	Reflect kernel changes for supplying root node information in mount instead of puffs_start(). Get completely rid of puffs_start(), as everything it used to do is now handled by the mount routine. Introduce an optional pre-mount call puffs_setrootinfo() for setting non-default root node information. As the old puffs_mount() is now virtually useless, say byebye to it and rename the old puffs_domount() to puffs_mount(), but add a root cookie parameter to compensate for the late puffs_start().
1.48	16-May-2007	pooka	switch the arbitrary default errno value for network i/o from ECONNABORTED to ECONNRESET, which is at least a little bit better
1.47	15-May-2007	pooka	memset freshly allocated puffs_usermount to 0
1.46	15-May-2007	pooka	Step 3 and currently final step of a more general purpose event loop: support removal and addition of i/o file descriptors on the fly. * detect closed file descriptors * automatically free waiters of a dead file descriptor * give the file server the possibility to specify a callback which notifies of a dead file descriptor * move loop function to be a property of the mainloop instead of framebuf (doesn't change effective behaviour) * add the possibility to configure a timespec parameter which attempts to call the loop function periodically * move the event loop functions from the puffs_framebuf namespace to puffs_framev to differential between pure memory management functions
1.45	11-May-2007	pooka	move puffs request dispatching code into it's own module. no functional change
1.44	11-May-2007	pooka	Step 2 of general-purpose eventloop: get rid of puffs_framebuf_eventloop(). puffs_mainloop() now handles that functionality as well.
1.43	10-May-2007	pooka	Add some kludges to be able to execute requests on the regular program stack instead of the continuation stack. This is for lib/36011, where pthread gets confused since we aren't running on the regular stack. I'm not really sure which direction to go to with this quite yet, so make the hack hard to enable on purpose. The whole request dispatch code needs cleaning anyway.
1.42	09-May-2007	pooka	Remove the puffs_usermount argument from puffs_req_handle(). It can be found from the other arguments and this way puffs_req_handle() is more symmetric with the other puffs_req() routines.
1.41	09-May-2007	pooka	Return 0 from puffs_mainloop() if no error occurs. pointed out by Phil Nelson
1.40	07-May-2007	pooka	Add puffs_setback(), which can be used to set setbacks for nodes before the call returns to the kernel. Currently it can be called only from open(), mmap(), remove() and rmdir(). (But that might change)
1.39	20-Apr-2007	pooka	assert that the file server doesn't try to return more cookies than it was supposed to
1.38	16-Apr-2007	pooka	alloc to set flags when setting fhsize
1.37	16-Apr-2007	pooka	some routines for setting mount parameters
1.36	13-Apr-2007	pooka	Instead of keeping on adding parameters to puffs_mount(), make it only take the bare essentials, which currently means removing "maxreqlen" from the argument list (all current callers I'm aware of set it as 0 anyway). Introduce puffs_init(), which provides a context for setting various parameters and puffs_domount(), which can be used to mount the file system. Keep puffs_mount() as a shortcut for the above two for simple file systems. Bump development ABI version to 13. After all, it's Friday the 13th. Watch out! Bad things can happen on Friday the 13th. --No carrier--
1.35	12-Apr-2007	pooka	Hide struct puffs_usermount from the rest of the world and provide accessors for interesting data in it. Namely, you can now get pu->pu_privdata with puffs_getspecific(), pu->pu_pn_root with puffs_set/getroot() and pu->pu_maxreqlen with puffs_getmaxreqlen().
1.34	11-Apr-2007	pooka	* support nodetofh and fhtonode callbacks * use new signature of node_readdir
1.33	20-Mar-2007	pooka	initial support for cacheops
1.32	18-Feb-2007	pooka	Until someone can sort non-priviledged mounts and kauth, force MNT_NOSUID \| MNT_NODEV for mounts with geteuid() != 0. I'm tired of typing them on the command line every time I test a file system.
1.31	18-Feb-2007	pooka	dump return values for operations
1.30	15-Feb-2007	pooka	Introduce puffs_path_walkcmp(), which is to be called from nodewalk and compares the path of the node against the given pathobject. Also make comparison method take a flag to indicate if it should check if the second path is a true prefix of the first. plus some namespace cleanup
1.29	15-Feb-2007	pooka	* get rid of the holy kernel-given ISDOTDOT, and instead define a macro which does strcmp against ".." and (the untranslated) componentname * make PUFFS_FLAG_BUILDPATH build paths also if dotdot is the case, and adapt the regular path objects to this * make nullfs lookup readable because we can now get rid of dotdot processing there
1.28	26-Jan-2007	pooka	add interfaces for suspending the file system
1.27	20-Jan-2007	pooka	Be consistent with the rest of the framework about the argument order of puffs_docc and puffs_dopreq
1.26	20-Jan-2007	pooka	be consistent in naming request handlers
1.25	16-Jan-2007	pooka	revoke revoke
1.24	15-Jan-2007	pooka	Checkpoint some more work in progress: for the benefit those file system backends which operate purely based on paths, push out more path management into the library and make path management more abstract: enable a file system to define a bunch of path management callbacks, which are used by the framework. Management of normal /this/is/a/path type paths is provided by the library.
1.23	11-Jan-2007	pooka	For directory renames we need to rename all the child nodes for the given directory if the file system wants paths (PUFFS_FLAG_BUILDPATH). Do this by walking the nodelist and adjusting the path prefix of each matching node.
1.22	10-Jan-2007	pooka	* don't chdir in possible daemon() call for the benefit of file systems which specify a relative path as the root * if (buildpath), build link source full pathname for the duration of the operation
1.21	06-Jan-2007	pooka	* get rid of the mount callback; it's no great surprise to the server that it needs to mount the file system backend if it wants to call mount * provide some options for getmntopts(), assume that callers will parse command line (or fstab) args * reorganize the puffs_cc interface just a bit, preparing for a bigger revamp later
1.20	02-Jan-2007	pooka	* inform kernel of our version * homegrown lib ABI check (to avoid bumping major a ludicrillion times before reaching stability)
1.19	29-Dec-2006	pooka	checkpoint some experimental work-in-progress, namely: Add support for having multiple outstanding operations. This is done by exposing enough interfaces so that it is convenient to have the main event loop in the implementation itself and by providing a continuation framework for convinient blocking and rescheduling. works fine, but will undergo further cleanup & development
1.18	14-Dec-2006	alc	CID4360: don't leak `buf' upon return ok pooka@
1.17	07-Dec-2006	pooka	outline & export code for interfacing with kernel GET/PUT requests
1.16	07-Dec-2006	pooka	g/c out-of-date comment
1.15	07-Dec-2006	pooka	support mmap
1.14	07-Dec-2006	pooka	cleanup generated stuff from comments
1.13	07-Dec-2006	pooka	rototill the operation callbacks: unify the ops under struct puffs_ops and namespace them
1.12	05-Dec-2006	pooka	adapt to kernel change: Allow multiple requests to be transferred in each GET/PUTOP
1.11	01-Dec-2006	pooka	PUFFSFLAG -> PUFFS_FLAG to be consistent with the kernel
1.10	01-Dec-2006	pooka	build & send implemented operations list to kernel
1.9	30-Nov-2006	pooka	summon daemon(3) in mainloop unless the nodaemon flag is given
1.8	18-Nov-2006	pooka	* call statvfs() at mount() time, pass that info along * start is gone
1.7	17-Nov-2006	pooka	shuffle flags a bit
1.6	09-Nov-2006	pooka	* catch up with kernel changes * better error handling when mounting
1.5	07-Nov-2006	pooka	* support FAF (Fire-And-Forget) class operations * adapt to new lookup
1.4	26-Oct-2006	pooka	pass rdev for specfs support
1.3	25-Oct-2006	pooka	handle PUFFS_VN_INACTIVE
1.2	23-Oct-2006	christos	pass lint; fix install.
1.1	22-Oct-2006	pooka	add libpuffs - the userspace programming interface for puffs This work was initially started and completed for Google SoC 2005 and tweaked to work a bit better in the past few weeks. While being far from complete, it is functional enough to be able and stable to host a fairly general-purpose in-memory file system in userspace. Even so, puffs should be considered experimental and no binary compatibility for interfaces or crash-freedom or zero security implications should be relied upon just yet. The GSoC project was mentored by William Studenmund and the final review for the code was done by Christos.
1.61.4.2	20-Jul-2007	pooka	Add puffs_framev_enqueue_waitevent, which waits for read/write/error events from an i/o descriptor. It yields until an event happens and continues execution when the event loop notices the event has happened.
1.61.4.1	20-Jul-2007	pooka	file puffs.c was added on branch matt-mips64 on 2007-07-20 13:14:56 +0000
1.62.2.4	23-Mar-2008	matt	sync with HEAD
1.62.2.3	09-Jan-2008	matt	sync with HEAD
1.62.2.2	08-Nov-2007	matt	sync with -HEAD
1.62.2.1	06-Nov-2007	matt	sync with HEAD
1.89.6.1	18-Sep-2008	wrstuden	Sync with wrstuden-revivesa-base-2.
1.92.8.1	21-Apr-2010	matt	sync to netbsd-5
1.92.4.7	25-Jan-2012	riz	Pull up following revision(s) (requested by manu in ticket #1706): lib/libpuffs/puffs.c: revision 1.108 signal pathconf in the vector of supported ops
1.92.4.6	17-Jul-2011	riz	Pull up following revision(s) (requested by manu in ticket #1645): lib/libc/sys/Makefile.inc 1.207 via patch lib/libc/sys/extattr_get_file.2 patch lib/libpuffs/dispatcher.c 1.34,1.36 via patch lib/libpuffs/puffs.c 1.107 via patch lib/libpuffs/puffs.h 1.115,1.118 via patch sys/fs/puffs/puffs_msgif.h 1.71,1.76 via patch sys/fs/puffs/puffs_vfsops.c 1.88 via patch sys/fs/puffs/puffs_vnops.c 1.145,1.154 via patch sys/kern/vfs_xattr.c 1.24-1.27 via patch sys/kern/vnode_if.c 1.87 via patch sys/sys/Makefile 1.133 via patch sys/sys/extattr.h 1.6 via patch sys/sys/vnode_if.h 1.81 via patch sys/ufs/ffs/ffs_vnops.c patch sys/ufs/ufs/ufs_extattr.c 1.31,1.34 via patch * support extended attributes * bump major due to structure growth * add some spare space * remove ABI sillyness Support extended attributes. Fix multiple non compliances in our Linux-like extattr API, and make it public so that it can be used. Improve a bit listxattr(2). It attemps to list both system and user extended attributes, and it faled if calling user did not have privilege for reading system EA. Now we just lise user EA and skip system EA in reading them is not allowed. Fix bug introduced in previous commuit: Do not vrele() a vnode we did not obtained. Improve UFS1 extended attributes usability - autocreate attribute backing file for new attributes - autoload attributes when issuing extattrctl start - when autoloading attributes, do not display garbage warning when looking up entries that got ENOENT Add a flag to VOP_LISTEXTATTR(9) so that the vnode interface can tell the filesystem in which format extended attribute shall be listed. There are currently two formats: - NUL-terminated strings, used for listxattr(2), this is the default. - one byte length-pprefixed, non NUL-terminated strings, used for extattr_list_file(2), which is obtanined by setting the EXTATTR_LIST_PREFIXLEN flag to VOP_LISTEXTATTR(9) This approach avoid the need for converting the list back and forth, except in libperfuse, since FUSE uses NUL-terminated strings, and the kernel may have requested EXTATTR_LIST_PREFIXLEN.
1.92.4.5	18-Jun-2011	bouyer	Pull up following revision(s) (requested by manu in ticket #1623): lib/libpuffs/puffs.c: revision 1.116 via patch sys/fs/puffs/puffs_vnops.c: revision 1.151 via patch Call advlock method if supplied
1.92.4.4	27-Oct-2009	bouyer	Pull up following revision(s) (requested by pooka in ticket #1105): lib/libpuffs/puffs.h: revision 1.112 lib/libpuffs/puffs.c: revision 1.100 distrib/sets/lists/base/md.amd64: patch distrib/sets/lists/base/md.sparc64: patch distrib/sets/lists/base/shl.mi: patch lib/libpuffs/shlib_version: patch Make it possible to supply parameters to puffs_init() later (but before puffs_mount()).
1.92.4.3	18-Oct-2009	sborrill	Pull up the following revisions(s) (requested by pooka in ticket #1100): lib/libpuffs/dispatcher.c: revision 1.33 lib/libpuffs/puffs.c: revision 1.99 lib/libpuffs/puffs.h: revision 1.111 sys/fs/puffs/puffs_msgif.h: revision 1.67 via patch sys/fs/puffs/puffs_vnops.c: revision 1.136 Support VOP_ABORTOP() in puffs.
1.92.4.2	16-Oct-2009	sborrill	Pull up the following revisions(s) (requested by pooka in ticket #1095): lib/libpuffs/puffs.c: revision 1.94 lib/libpuffs/puffs.h: revision 1.110 Add puffs_cancel(), which allows to back down after puffs_init() and before puffs_mount().
1.92.4.1	16-Oct-2009	sborrill	Pull up the following revisions(s) (requested by pooka in ticket #1094): lib/libpuffs/puffs.3: revision 1.45 lib/libpuffs/puffs.c: revision 1.93 lib/libpuffs/puffs.h: revision 1.109 Add function puffs_setspecific(pu, private) which can be used to set the specific data after the call to puffs_init().
1.116.4.1	17-Apr-2012	yamt	sync with head
1.117.18.1	05-Nov-2014	snj	Pull up following revision(s) (requested by manu in ticket #181): lib/libperfuse/fuse.h: revision 1.6 lib/libperfuse/ops.c: revision 1.78 lib/libperfuse/perfuse.c: revision 1.35 lib/libperfuse/perfuse_priv.h: revision 1.36 lib/libpuffs/dispatcher.c: revision 1.48 lib/libpuffs/opdump.c: revision 1.37 lib/libpuffs/puffs.c: revision 1.118 lib/libpuffs/puffs.h: revision 1.126 lib/libpuffs/puffs_ops.3: revisions 1.40-1.41 sys/fs/puffs/puffs_msgif.h: revision 1.82-1.83 sys/fs/puffs/puffs_msgif.h: revision 1.82 sys/fs/puffs/puffs_vnops.c: revision 1.196 Add PUFFS support for fallocate and fdiscard operations -- libpuffs support for fallocate and fdiscard operations -- Add PUFFS_HAVE_FALLOCATE in puffs_msgif.h so that filesystem can decide at build time wether fallocate is usable -- FUSE fallocate support There seems to be no fdiscard FUSE operation at the moment, hence that one is left unused.
1.123.4.1	10-Jun-2019	christos	Sync with HEAD
1.123.2.1	28-Jul-2018	pgoyette	Sync with HEAD
1.1	21-Nov-2024	riastradh	branches: 1.1.4; libpuffs: Add expected symbols list. NOTE: This exports a symbol pu_lock which is only mentioned in libpuffs/puffs_priv.h which looks very internal. Intentional? PR lib/58838: shared libraries in base should all have expsym lists
1.1.4.2	02-Aug-2025	perseant	Sync with HEAD
1.1.4.1	21-Nov-2024	perseant	file puffs.expsym was added on branch perseant-exfatfs on 2025-08-02 05:54:56 +0000
1.130	03-Dec-2021	pho	Use size_t in place of voff_t in puffs_setrootinfo(3) Chuck Silvers pointed out that voff_t was also supposed to be kernel-only. The correct type to use in userland would be off_t, but since changing vsize_t to either voff_t or off_t is an ABI change on 32-bit platforms, we use size_t knowing that it is technically incorrect.
1.129	03-Dec-2021	pho	Avoid using register_t or vsize_t in <puffs.h> <puffs.h> is a user-space header, and should not use any of kernel-only types. It's not reasonable to require user-land filesystems to #define _KERNTYPES.
1.128	23-Sep-2019	christos	Restore binary compatibility by using the statvfs90 structure internally.
1.127	12-Nov-2015	christos	branches: 1.127.16; consistency fixes.
1.126	31-Oct-2014	manu	libpuffs support for fallocate and fdiscard operations
1.125	16-Aug-2014	manu	Add an open2 method, like open but with an additionnal oflags used by the filesystem to send back information about the file. This is used to implement PUFFS_OPEN_IO_DIRECT by which the filesystem tells the kernel that read/write should bypass the page cache.
1.124	16-Aug-2012	manu	branches: 1.124.10; Fix regression that has been introduced when the lookup/reclaim race condition was addressed in libpuffs by counting lookups. The fix assumes that cookies map to struct puffs_cookie, which has not been documented as a requirement for filesystems using libpuffs. As an example, we got burnt by this assumption in libp2k (kern/46734), and we fixed bit by actually mapping libp2k cookies to struct puffs_node. It is unlikely, but there may be third party filesystems that use cookies unmapped to struct puffs_node, and they were left broken for now. - we introduce a puffs_init() flag PUFFS_FLAG_PNCOOKIE that let filesystems inform libpuffs that they map cookies to struct puffs_node. Is that flag is used, the lookup/reclaim race condition fix is enabled. We enable the flag for libp2k. - filesystems that use puffs_pn_new() obviouslty use struct puffs_node and gain PUFFS_FLAG_PNCOOKIE automatically even if they did not specify it in puffs_init(). This include all our PUFFS filesystem in-tree except libp2k. - for filesystems not willing to use struct puffs_node, we introduce a reclaim2 vnop, which is reclaim with an additionnal lookup count argument. This vnop let the filesystem implement the lookup/reclaim race fix on its own.
1.123	21-Jul-2012	manu	- Improve PUFFS_KFLAG_CACHE_FS_TTL by reclaiming older inactive nodes. The normal kernel behavior is to retain inactive nodes in the freelist until it runs out of vnodes. This has some merit for local filesystems, where the cost of an allocation is about the same as the cost of a lookup. But that situation is not true for distributed filesystems. On the other hand, keeping inactive nodes for a long time hold memory in the file server process, and when the kernel runs out of vnodes, it produce reclaim avalanches that increase lattency for other operations. We do not reclaim inactive vnodes immediatly either, as they may be looked up again shortly. Instead we introduce a grace time and we reclaim nodes that have been inactive beyond the grace time. - Fix lookup/reclaim race condition. The above improvement undercovered a race condition between lookup and reclaim. If we reclaimed a vnode associated with a userland cookie while a lookup returning that same cookiewas inprogress, then the kernel ends up with a vnode associated with a cookie that has been reclaimed in userland. Next operation on the cookie will crash (or at least confuse) the filesystem. We fix this by introducing a lookup count in kernel and userland. On reclaim, the kernel sends the count, which enable userland to detect situation where it initiated a lookup that is not completed in kernel. In such a situation, the reclaim must be ignored, as the node is about to be looked up again.
1.122	27-Jun-2012	manu	Pass the FAF flag to handlers: - setattr_ttl is updated to add a flag argument. Since it was not present in a previous release, we can change its API - write2 is introduced, this is write with an extra flag for FAF. - fsync already has the FAF information in a flag and needs no change - for other operations, FAF is unconditional
1.121	18-Apr-2012	manu	- When using PUFFS_KFLAG_CACHE_FS_TTL, do not use puffs_node to carry attribute and TTL fora newly created node. Instead extend puffs_newinfo and add puffs_newinfo_setva() and puffs_newinfo_setttl() - Remove node_mk_common_final in libperfuse. It used to set uid/gid for a newly created vnode but has been made redundant along time ago since uid and gid are properly set in FUSE header. - In libperfuse, check for corner case where opc = 0 on INACTIVE and RECLAIM (how is it possible? Check for it to avoid a crash anyway) - In libperfuse, make sure we unlimit RLIMIT_AS and RLIMIT_DATA so that we do notrun out of memory because the kernel is lazy at reclaiming vnodes. - In libperfuse, cleanup style of perfuse_destroy_pn()
1.120	08-Apr-2012	manu	Add PUFFS_KFLAG_CACHE_FS_TTL flag to puffs_init(3) to use name and attribute cache with filesystem provided TTL. lookup, create, mknod, mkdir, symlink, getattr and setattr messages have been extended so that attributes and their TTL can be provided by the filesytem. lookup, create, mknod, mkdir, and symlink messages are also extended so that the filesystem can provide name TTL. The filesystem updates attributes and TTL using puffs_pn_getvap(3), puffs_pn_getvattl(3), and puffs_pn_getcnttl(3)
1.119	29-Aug-2011	joerg	branches: 1.119.2; 1.119.4; Use __dead
1.118	04-Jul-2011	manu	Add a flag to VOP_LISTEXTATTR(9) so that the vnode interface can tell the filesystem in which format extended attribute shall be listed. There are currently two formats: - NUL-terminated strings, used for listxattr(2), this is the default. - one byte length-pprefixed, non NUL-terminated strings, used for extattr_list_file(2), which is obtanined by setting the EXTATTR_LIST_PREFIXLEN flag to VOP_LISTEXTATTR(9) This approach avoid the need for converting the list back and forth, except in libperfuse, since FUSE uses NUL-terminated strings, and the kernel may have requested EXTATTR_LIST_PREFIXLEN.
1.117	07-Jun-2010	pooka	Make retval argument for pathconf a register_t to match VOP_PATHCONF. This makes the size the same on 64bit archs. Don't bother bumping any version, since you'd have explicitly had to jump through some hoops to use pathconf before.
1.116	21-May-2010	pooka	add option string for no attribute cache (foreseeing the odd event I might actually implement one some day)
1.115	21-May-2010	pooka	* support extended attributes * bump major due to structure growth * add some spare space * remove ABI sillyness
1.114	12-Jan-2010	pooka	Add convenience routine puffs_unmountonsignal(), which does exactly that.
1.113	05-Dec-2009	pooka	Add an error handler which syslogs but does not abort() and make it the default. Most servers just shovel bits from their backend to the kernel without checking the contents and dumping core in case the backend gives garbage is a bit harsh.
1.112	18-Oct-2009	pooka	Make it possible to supply parameters to puffs_init() later (but before puffs_mount()).
1.111	17-Oct-2009	pooka	Dispatch puffs_node_abortop(). Note: We use the storage of puffs_cache_write from puffs_ops for this purpose. It's not issued by the kernel and hence currently unused, and this saves us from the trouble of bumping the lib major version.
1.110	12-Dec-2008	pooka	Add puffs_cancel(), which allows to back down after puffs_init() and before puffs_mount(). (this should actually be integrated with puffs_exit(), but that routine is currently so non-sensible that I didn't bother now)
1.109	12-Dec-2008	pooka	add: puffs_setspecific(pu, private) Can be used to set the specific data after the call to puffs_init().
1.108	12-Aug-2008	pooka	branches: 1.108.4; 1.108.8; Replace void * by puffs_cookie_t where appropriate. No functional change.
1.107	29-Jan-2008	pooka	branches: 1.107.6; Fix a weirdosity in the framebuf gotfb interface: hand buffer ownership to the called routine instead of keeping it. This means that callees must now destroy the buffer once they are done with it.
1.106	28-Jan-2008	pooka	Bump version for lazy contexts (probably not necessary, but it's been a long time since I bumbed the version ;)
1.105	28-Jan-2008	pooka	Implement lazy contexts for file systems using puffs_mainloop(). Previously each request was executed on its own callcontext and switched to every time the request was being processed. Now requests borrow the mainloop context and switch only if/when they yield. This takes two context switches away from a file system request bringing down the typical amounts 2->0 (e.g. dtfs) and 4->2 (e.g. psshfs). The interfaces for manually executing requests changed a bit: puffs_dispatch_create() and puffs_dispatch_exec() must now be used. They are not tested, as nothing in-tree wants them and I doubt anyone else is really interested in them either. Also do some misc code cleanup related to execution contexts. The "work-in-progress checkpoint" committed over a year ago was starting to look slightly weed-infested.
1.104	14-Jan-2008	pooka	Introduce PUFFS_STACKSIZE_MIN to avoid overloading 0 and getting a warning from the library when really wanting to say "just the minimum, please".
1.103	25-Dec-2007	pooka	Add a public interface for setting a cookie mapper.
1.102	19-Dec-2007	pooka	Add generic getattr routine.
1.101	16-Dec-2007	pooka	* nuke puffs_cc_get{specific,usermount} for good * move prototypes for puffs_docc and puffs_dopufbuf into the public header, as they are should be exposed
1.100	11-Dec-2007	pooka	nuke #define whose purpose has ceased to be
1.99	08-Dec-2007	pooka	Now that "l" is gone both as an argument to operations and from componentname, remove all vestiges of puffs_cid.
1.98	04-Dec-2007	pooka	Part 3/n of extensive changes to message handling: Get rid of the original puffs_req(3) framework and use puffs_framebuf(3) instead for file system requests. It has the advantage of being suitable for transporting a distributed message passing protocol and therefore us being able to run the file system server on any host. Ok, puffs is not quite here yet: libpuffs needs to grow request routing support and the message contents need to be munged into a host independent format. Saying which format would be telling, but it might begin with an X, end in an L and have the 13th character in the middle. Keep an eye out for the sequels: Parts 3+m/n.
1.97	30-Nov-2007	pooka	Rototill. Ok, ok, a few more words about it: stop holding puffs_cc as a holy value and passing it around to almost every possible place (popquiz: which kernel variable does this remind you of?). Instead, pass the natural choice, puffs_usermount, and fetch puffs_cc via puffs_cc_getcc() only in routines which actually need it. This not only simplifies code, but (thanks to the introduction of puffs_cc_getcc()) enables constructs which weren't previously sanely possible, say layering as a curious example. There's still a little to do on this front, but this was the major fs interface blast.
1.96	29-Nov-2007	pooka	* allocate pcc as a slice from the stack instead of allocating it separately * provide puffs_cc_getcc() This is in preparation for the removal of you-should-guess-what as an argument to routines here and there and everywhere.
1.95	27-Nov-2007	pooka	Remove "puffs_cid" from the puffs interface following l-removal from the kernel vfs interfaces. puffs_cc_getcaller(pcc) can be used now should the same information be desired.
1.94	16-Nov-2007	pooka	Introduce puffs_daemon(), which works like daemon(3) except that it also synchronizes with puffs_mount() and does not return (exit) in the parent process until the file system has been mounted. This makes it possible to reliably run e.g. mount_foo jippi /kai ; cd /kai/ee
1.93	05-Nov-2007	pooka	remember to bump version for puffs_mainloop() change
1.92	05-Nov-2007	pooka	Pull the daemonizing code out of the library mainloop into the file servers. Calling daemon() (i.e. fork()ing) inside a library can cause nice surprises for e.g. threaded programs. As discussed with Greg Oster & others.
1.91	28-Oct-2007	pooka	Add possibility to set pre- and post callbacks which will be called for all operations before or after executing the specific callback (the api is not be final yet, though).
1.90	26-Oct-2007	pooka	mmap stacks, makes it possible to align them
1.89	21-Oct-2007	pooka	Add puffs_cc_schedule() which marks a pcc runnable and will pass execution to it when in the main loop the next time.
1.88	21-Oct-2007	pooka	Always provide caller information from the kernel based on curlwp. (but don't deprecate the old puffs_cid interface just yet)
1.87	18-Oct-2007	pooka	* Change type of boolean functions to, well, boolean instead of int. * add puffs_cid_isequal() which can be used to test if the two caller id's refer to the same process/lwp.
1.86	11-Oct-2007	pooka	Part 1/n of some pretty extensive changes to how the kernel module interacts with the userspace file server: * since the kernel-user communication is not purely request-response anymore (hasn't been since 2006), try to rename some "request" to "message". more similar mangling will take place in the future. * completely rework how messages are allocated. previously most of them were borrowed from the stack (originally all of them), but now always allocate dynamically. this makes the structure of the code much cleaner. also makes it possible to fix a locking order violation. it enables plenty of future enhancements. * start generalizing the transport interface to be independent of puffs * move transport interface to read/write instead of ioctl. the old one had legacy design problems, and besides, ioctl's suck. implement a very generic version for now; this will be worked on later hopefully some day reaching "highly optimized". * implement libpuffs support behind existing library request interfaces. this will change eventually (I hate those interfaces)
1.85	01-Oct-2007	pooka	Pass reason string to pu_errnotify. Print it in the default version before aborting.
1.84	27-Sep-2007	pooka	Add error notifications, which are used to deliver errors from the kernel to the file server for silly things the file server did, e.g. attempting to create a file with size VSIZENOTSET. The file server can handle these as it chooses, but the default action is for it to throw its hands in the air and sing "goodbye, cruel world, it's over, walk on by".
1.83	06-Sep-2007	pooka	Give cmpframe() the ability to shortcircuit the comparison loop by signalling back that the buffer under examination is not a response at all.
1.82	01-Sep-2007	pooka	Add gotframe callback, which is called if an incoming frame matches no outstanding request.
1.81	22-Aug-2007	pooka	branches: 1.81.2; Mimic namei structure changes for puffs. bump both kernel & lib version.
1.80	18-Aug-2007	pooka	puffs_cc_create/destroy aren't supposed to be exported to the user
1.79	17-Aug-2007	pooka	Add accessors for struct puffs_node to prepare for removal from exported headers.
1.78	15-Aug-2007	pooka	Nuke PUFFSLOOKUP_FOO and move to NAMEI_FOO
1.77	27-Jul-2007	pooka	branches: 1.77.4; mmap fflags -> prot
1.76	20-Jul-2007	pooka	Add puffs_framev_enqueue_waitevent, which waits for read/write/error events from an i/o descriptor. It yields until an event happens and continues execution when the event loop notices the event has happened.
1.75	17-Jul-2007	pooka	* add mntfromname parameter to puffs_init() * set it in file servers
1.74	09-Jul-2007	pooka	add IO_NDELAY
1.73	08-Jul-2007	pooka	g/c leftover
1.72	08-Jul-2007	pooka	make notifyfunction optional and do not call removeonclose() if none is given.
1.71	07-Jul-2007	pooka	Make puffs_framebuf better suited from environments with multiple sources: * in addition add/remove, allow enable/disable, which can be used to control events for descriptors without having to remove all the data associated with them * add directsend/receive, which can be used to pass the same buffer from the caller to read/writeframe and back again * add flags to enqueue functions and allow urgent buffers to be processed as the next PDU
1.70	02-Jul-2007	pooka	Get rid of the "int *refs" parameter to inactive: the same can be accomplished now with puffs_setbacks.
1.69	01-Jul-2007	pooka	Instead of requesting various pieces of information from node-creating methods using "please fill me"-pointers, pass "struct puffs_newinfo *" and provide methods to fill in new node info.
1.68	01-Jul-2007	pooka	Instead of supplying a plain pid, supply an abstract struct puffs_cid *, which can currently be used to query the pid and lwpid.
1.67	01-Jul-2007	pooka	make puffs_cred an opaque type
1.66	24-Jun-2007	pooka	Actually, keep PUFFS_KFLAG_NOCACHE and -o cache around as shorthand to neither page- nor namecache.
1.65	24-Jun-2007	pooka	PUFFS_KFLAG_NOCACHE became two, so introduce the command line options -o nonamecache and -o nopagecache and adapt file systems where necessary.
1.64	24-Jun-2007	pooka	Introduce puffs_null_setops(), which initializes the ops vector with puffs nullfs ops.
1.63	24-Jun-2007	pooka	Introduce puffs_pn_remove() to signal that a node has been deleted and destroyed, but not yet reclaimed. This prevents puffs_pn_nodewalk() from returning stale entries. Make nullfs use this (some file systems are a bit too happy with recycling inode numbers).
1.62	06-Jun-2007	pooka	Move puffs to a two clause license where it already isn't so. And as agc pointed out, even files with the third clause were already effectively two clause because of a slight bug in the language...
1.61	20-May-2007	pooka	Introduce puffs_framev_framebuf_ccpromote, which can be used to turn a reference to puffs_framebuf in the file system from a cb/justsend operation to a cc wait, should the file system find itself desiring the result.
1.60	20-May-2007	pooka	Change the signature of puffs_respcmp to return -1,0,1 for smaller, equal, larger, respectively instead of 0/1 for non/equal. This will allow sorting the buffers for faster matching in libpuffs. While here, change the name from respcmp to framecmp, as that better reflects the purpose. NOTE! there is no obvious way to make compilation fail for file systems which may already be using this feature (although I don't think there are any outside our tree, as the feature is two weeks old). Nevertheless, non-updated file systems will fail very quickly.
1.59	18-May-2007	pooka	dispatch PUFFS_VN_POLL
1.58	17-May-2007	pooka	Reflect kernel changes for supplying root node information in mount instead of puffs_start(). Get completely rid of puffs_start(), as everything it used to do is now handled by the mount routine. Introduce an optional pre-mount call puffs_setrootinfo() for setting non-default root node information. As the old puffs_mount() is now virtually useless, say byebye to it and rename the old puffs_domount() to puffs_mount(), but add a root cookie parameter to compensate for the late puffs_start().
1.57	16-May-2007	pooka	Add error parameter to puffs_framev_cb to signal if there is a result of if the function should just do resource cleanup.
1.56	15-May-2007	pooka	Step 3 and currently final step of a more general purpose event loop: support removal and addition of i/o file descriptors on the fly. * detect closed file descriptors * automatically free waiters of a dead file descriptor * give the file server the possibility to specify a callback which notifies of a dead file descriptor * move loop function to be a property of the mainloop instead of framebuf (doesn't change effective behaviour) * add the possibility to configure a timespec parameter which attempts to call the loop function periodically * move the event loop functions from the puffs_framebuf namespace to puffs_framev to differential between pure memory management functions
1.55	11-May-2007	pooka	Step 2 of general-purpose eventloop: get rid of puffs_framebuf_eventloop(). puffs_mainloop() now handles that functionality as well.
1.54	11-May-2007	pooka	Step 1 in moving towards a general-purpose eventloop: allow the framebuf event loop to take n i/o fd's as parameters and also allow dynamic add/remove of fd's. (not tested except for one fd, but more changes coming soon)
1.53	09-May-2007	pooka	Remove the puffs_usermount argument from puffs_req_handle(). It can be found from the other arguments and this way puffs_req_handle() is more symmetric with the other puffs_req() routines.
1.52	07-May-2007	pooka	Add puffs_setback(), which can be used to set setbacks for nodes before the call returns to the kernel. Currently it can be called only from open(), mmap(), remove() and rmdir(). (But that might change)
1.51	06-May-2007	pooka	Remove the user-specified argument for the framebuf loop callback. It can be generated through other means if it is desired.
1.50	05-May-2007	pooka	Add puffs "frame buffers", which are an abstraction of the buffering and event handling mechanisms required in file servers with blocking I/O backends. puffs_framebuf is built on the concept of puffs_cc and uses those to multiplex execution where needed. File systems are required to implement three methods: * read frame * write frame * compare if frame is a response to the given one Memory management is provided by puffs_framebuf, but the file systems must still, of course, interpret the protocol and do e.g. byte order conversion. As always, puffs_framebuf is work in progress. Current users are mount_psshfs and mount_9p.
1.49	01-May-2007	pooka	Introduce PUFFS_FLAG_HASHPATH, which for PUFFS_FLAG_BUILDPATH file servers creates a hash of the node's path into po->po_hash to avoid a full pathcmp every time.
1.48	22-Apr-2007	pooka	use more accurate flag masking
1.47	19-Apr-2007	pooka	add puffs_cc_getspecific(pcc), which is shorthand for: puffs_getspecific(puffs_cc_getusermount(pcc))
1.46	17-Apr-2007	pooka	Introduce PUFFS_STORE_DCOOKIE, which stores a directory cookie for a given offset iff the directory read in question wants cookies.
1.45	16-Apr-2007	pooka	alloc to set flags when setting fhsize
1.44	16-Apr-2007	pooka	some routines for setting mount parameters
1.43	13-Apr-2007	pooka	Instead of keeping on adding parameters to puffs_mount(), make it only take the bare essentials, which currently means removing "maxreqlen" from the argument list (all current callers I'm aware of set it as 0 anyway). Introduce puffs_init(), which provides a context for setting various parameters and puffs_domount(), which can be used to mount the file system. Keep puffs_mount() as a shortcut for the above two for simple file systems. Bump development ABI version to 13. After all, it's Friday the 13th. Watch out! Bad things can happen on Friday the 13th. --No carrier--
1.42	12-Apr-2007	pooka	Hide struct puffs_usermount from the rest of the world and provide accessors for interesting data in it. Namely, you can now get pu->pu_privdata with puffs_getspecific(), pu->pu_pn_root with puffs_set/getroot() and pu->pu_maxreqlen with puffs_getmaxreqlen().
1.41	11-Apr-2007	pooka	* support nodetofh and fhtonode callbacks * use new signature of node_readdir
1.40	06-Apr-2007	pooka	support page cache flushing
1.39	06-Apr-2007	pooka	implement puffs_inval_pagecache_node_range()
1.38	22-Mar-2007	pooka	unify the location of pcr in puffs_access_
1.37	22-Mar-2007	pooka	* return -1 and set errno instead of returning errno directly * introduce puffs_cred_isregular() which checks if the cred structure has regular uid/gid (i.e. "person" credentials and not kernel creds)
1.36	22-Mar-2007	pooka	remove variable names from header
1.35	21-Mar-2007	pooka	add routiens which check against ufs semantics for permission to chown, chmod and change a/mtime
1.34	20-Mar-2007	pooka	implement counterpart of vaccess()
1.33	20-Mar-2007	pooka	initial support for cacheops
1.32	16-Mar-2007	pooka	va->va_mode doesn't contain the full argument to the mknod() system call, so introduce puffs_addvtype2mode() and use that in null.c to generate the proper syscall argument
1.31	15-Feb-2007	pooka	Introduce puffs_path_walkcmp(), which is to be called from nodewalk and compares the path of the node against the given pathobject. Also make comparison method take a flag to indicate if it should check if the second path is a true prefix of the first. plus some namespace cleanup
1.30	15-Feb-2007	pooka	* get rid of the holy kernel-given ISDOTDOT, and instead define a macro which does strcmp against ".." and (the untranslated) componentname * make PUFFS_FLAG_BUILDPATH build paths also if dotdot is the case, and adapt the regular path objects to this * make nullfs lookup readable because we can now get rid of dotdot processing there
1.29	26-Jan-2007	pooka	add interfaces for suspending the file system
1.28	20-Jan-2007	pooka	Be consistent with the rest of the framework about the argument order of puffs_docc and puffs_dopreq
1.27	20-Jan-2007	pooka	be consistent in naming request handlers
1.26	16-Jan-2007	pooka	revoke revoke
1.25	15-Jan-2007	pooka	Checkpoint some more work in progress: for the benefit those file system backends which operate purely based on paths, push out more path management into the library and make path management more abstract: enable a file system to define a bunch of path management callbacks, which are used by the framework. Management of normal /this/is/a/path type paths is provided by the library.
1.24	11-Jan-2007	pooka	implement fsync
1.23	11-Jan-2007	pooka	Add preliminary code for a nullfs layer, which can be used to mount a directory hierarchy to another point, just like with the kernel nullfs. This is not really a layering scheme yet, but it should evolve into one. Currently it can just be used to do 1:1 mapping.
1.22	10-Jan-2007	pooka	Be less shy about the default stack size, it's a only virtual memory space. TODO: put a guard page at the end
1.21	09-Jan-2007	pooka	rename name cache invalidation interface to be in sync with the kernel operation names, i.e. contain "namecache" instead of just "name"
1.20	09-Jan-2007	pooka	lib interface for invalidation routines
1.19	06-Jan-2007	pooka	* get rid of the mount callback; it's no great surprise to the server that it needs to mount the file system backend if it wants to call mount * provide some options for getmntopts(), assume that callers will parse command line (or fstab) args * reorganize the puffs_cc interface just a bit, preparing for a bigger revamp later
1.18	02-Jan-2007	pooka	* inform kernel of our version * homegrown lib ABI check (to avoid bumping major a ludicrillion times before reaching stability)
1.17	29-Dec-2006	pooka	checkpoint some experimental work-in-progress, namely: Add support for having multiple outstanding operations. This is done by exposing enough interfaces so that it is convenient to have the main event loop in the implementation itself and by providing a continuation framework for convinient blocking and rescheduling. works fine, but will undergo further cleanup & development
1.16	07-Dec-2006	pooka	outline & export code for interfacing with kernel GET/PUT requests
1.15	07-Dec-2006	pooka	support mmap
1.14	07-Dec-2006	pooka	rototill the operation callbacks: unify the ops under struct puffs_ops and namespace them
1.13	01-Dec-2006	pooka	PUFFSFLAG -> PUFFS_FLAG to be consistent with the kernel
1.12	30-Nov-2006	pooka	summon daemon(3) in mainloop unless the nodaemon flag is given
1.11	23-Nov-2006	pooka	const coherence
1.10	21-Nov-2006	pooka	mode_t to enum vtype conversion
1.9	18-Nov-2006	pooka	* call statvfs() at mount() time, pass that info along * start is gone
1.8	17-Nov-2006	pooka	shuffle flags a bit
1.7	14-Nov-2006	pooka	namespace previous change under puffs_vfsnop to avoid confusions
1.6	14-Nov-2006	pooka	implement unmount, sync and statvfs as dummies for file system which couldn't care less about implementing them
1.5	09-Nov-2006	pooka	* catch up with kernel changes * better error handling when mounting
1.4	07-Nov-2006	pooka	* support FAF (Fire-And-Forget) class operations * adapt to new lookup
1.3	26-Oct-2006	pooka	pass rdev for specfs support
1.2	25-Oct-2006	pooka	handle PUFFS_VN_INACTIVE
1.1	22-Oct-2006	pooka	add libpuffs - the userspace programming interface for puffs This work was initially started and completed for Google SoC 2005 and tweaked to work a bit better in the past few weeks. While being far from complete, it is functional enough to be able and stable to host a fairly general-purpose in-memory file system in userspace. Even so, puffs should be considered experimental and no binary compatibility for interfaces or crash-freedom or zero security implications should be relied upon just yet. The GSoC project was mentored by William Studenmund and the final review for the code was done by Christos.
1.77.4.2	27-Jul-2007	pooka	mmap fflags -> prot
1.77.4.1	27-Jul-2007	pooka	file puffs.h was added on branch matt-mips64 on 2007-07-27 08:28:17 +0000
1.81.2.3	23-Mar-2008	matt	sync with HEAD
1.81.2.2	09-Jan-2008	matt	sync with HEAD
1.81.2.1	06-Nov-2007	matt	sync with HEAD
1.107.6.1	18-Sep-2008	wrstuden	Sync with wrstuden-revivesa-base-2.
1.108.8.1	21-Apr-2010	matt	sync to netbsd-5
1.108.4.5	17-Jul-2011	riz	Pull up following revision(s) (requested by manu in ticket #1645): lib/libc/sys/Makefile.inc 1.207 via patch lib/libc/sys/extattr_get_file.2 patch lib/libpuffs/dispatcher.c 1.34,1.36 via patch lib/libpuffs/puffs.c 1.107 via patch lib/libpuffs/puffs.h 1.115,1.118 via patch sys/fs/puffs/puffs_msgif.h 1.71,1.76 via patch sys/fs/puffs/puffs_vfsops.c 1.88 via patch sys/fs/puffs/puffs_vnops.c 1.145,1.154 via patch sys/kern/vfs_xattr.c 1.24-1.27 via patch sys/kern/vnode_if.c 1.87 via patch sys/sys/Makefile 1.133 via patch sys/sys/extattr.h 1.6 via patch sys/sys/vnode_if.h 1.81 via patch sys/ufs/ffs/ffs_vnops.c patch sys/ufs/ufs/ufs_extattr.c 1.31,1.34 via patch * support extended attributes * bump major due to structure growth * add some spare space * remove ABI sillyness Support extended attributes. Fix multiple non compliances in our Linux-like extattr API, and make it public so that it can be used. Improve a bit listxattr(2). It attemps to list both system and user extended attributes, and it faled if calling user did not have privilege for reading system EA. Now we just lise user EA and skip system EA in reading them is not allowed. Fix bug introduced in previous commuit: Do not vrele() a vnode we did not obtained. Improve UFS1 extended attributes usability - autocreate attribute backing file for new attributes - autoload attributes when issuing extattrctl start - when autoloading attributes, do not display garbage warning when looking up entries that got ENOENT Add a flag to VOP_LISTEXTATTR(9) so that the vnode interface can tell the filesystem in which format extended attribute shall be listed. There are currently two formats: - NUL-terminated strings, used for listxattr(2), this is the default. - one byte length-pprefixed, non NUL-terminated strings, used for extattr_list_file(2), which is obtanined by setting the EXTATTR_LIST_PREFIXLEN flag to VOP_LISTEXTATTR(9) This approach avoid the need for converting the list back and forth, except in libperfuse, since FUSE uses NUL-terminated strings, and the kernel may have requested EXTATTR_LIST_PREFIXLEN.
1.108.4.4	27-Oct-2009	bouyer	Pull up following revision(s) (requested by pooka in ticket #1105): lib/libpuffs/puffs.h: revision 1.112 lib/libpuffs/puffs.c: revision 1.100 distrib/sets/lists/base/md.amd64: patch distrib/sets/lists/base/md.sparc64: patch distrib/sets/lists/base/shl.mi: patch lib/libpuffs/shlib_version: patch Make it possible to supply parameters to puffs_init() later (but before puffs_mount()).
1.108.4.3	18-Oct-2009	sborrill	Pull up the following revisions(s) (requested by pooka in ticket #1100): lib/libpuffs/dispatcher.c: revision 1.33 lib/libpuffs/puffs.c: revision 1.99 lib/libpuffs/puffs.h: revision 1.111 sys/fs/puffs/puffs_msgif.h: revision 1.67 via patch sys/fs/puffs/puffs_vnops.c: revision 1.136 Support VOP_ABORTOP() in puffs.
1.108.4.2	16-Oct-2009	sborrill	Pull up the following revisions(s) (requested by pooka in ticket #1095): lib/libpuffs/puffs.c: revision 1.94 lib/libpuffs/puffs.h: revision 1.110 Add puffs_cancel(), which allows to back down after puffs_init() and before puffs_mount().
1.108.4.1	16-Oct-2009	sborrill	Pull up the following revisions(s) (requested by pooka in ticket #1094): lib/libpuffs/puffs.3: revision 1.45 lib/libpuffs/puffs.c: revision 1.93 lib/libpuffs/puffs.h: revision 1.109 Add function puffs_setspecific(pu, private) which can be used to set the specific data after the call to puffs_init().
1.119.4.5	03-Nov-2014	msaitoh	Pull up following revision(s) (requested by manu in ticket #1140): lib/libperfuse/ops.c 1.63-1.69 lib/libperfuse/perfuse.c 1.32-1.33 lib/libperfuse/perfuse_priv.h 1.32-1.34 lib/libperfuse/subr.c 1.20 lib/libpuffs/creds.c 1.16 lib/libpuffs/dispatcher.c 1.47 lib/libpuffs/puffs.h 1.125 lib/libpuffs/puffs_ops.3 1.37-1.38 lib/libpuffs/requests.c 1.24 sys/fs/puffs/puffs_msgif.h 1.81 sys/fs/puffs/puffs_sys.h 1.85 sys/fs/puffs/puffs_vnops.c 1.183 usr.sbin/perfused/msg.c 1.22 Bring libpuffs, libperfuse and perfused on par with -current: - implement FUSE direct I/O - remove useless code and warnings - fix missing GETATTR bugs - fix exended attribute get and list operations
1.119.4.4	13-Sep-2012	riz	Pull up following revision(s) (requested by manu in ticket #525): lib/libpuffs/dispatcher.c: revision 1.44 lib/libp2k/p2k.c: revision 1.57 lib/libpuffs/puffs_ops.3: revision 1.35 lib/libpuffs/pnode.c: revision 1.13 lib/libpuffs/puffs.3: revision 1.56 lib/libpuffs/puffs.h: revision 1.124 Fix regression that has been introduced when the lookup/reclaim race condition was addressed in libpuffs by counting lookups. The fix assumes that cookies map to struct puffs_cookie, which has not been documented as a requirement for filesystems using libpuffs. As an example, we got burnt by this assumption in libp2k (kern/46734), and we fixed bit by actually mapping libp2k cookies to struct puffs_node. It is unlikely, but there may be third party filesystems that use cookies unmapped to struct puffs_node, and they were left broken for now. - we introduce a puffs_init() flag PUFFS_FLAG_PNCOOKIE that let filesystems inform libpuffs that they map cookies to struct puffs_node. Is that flag is used, the lookup/reclaim race condition fix is enabled. We enable the flag for libp2k. - filesystems that use puffs_pn_new() obviouslty use struct puffs_node and gain PUFFS_FLAG_PNCOOKIE automatically even if they did not specify it in puffs_init(). This include all our PUFFS filesystem in-tree except libp2k. - for filesystems not willing to use struct puffs_node, we introduce a reclaim2 vnop, which is reclaim with an additionnal lookup count argument. This vnop let the filesystem implement the lookup/reclaim race fix on its own.
1.119.4.3	12-Aug-2012	martin	branches: 1.119.4.3.2; Pull up following revision(s) (requested by manu in ticket #438): lib/libperfuse/perfuse_priv.h: revision 1.31 sys/fs/puffs/puffs_msgif.h: revision 1.80 sys/fs/puffs/puffs_vnops.c: revision 1.171 lib/libpuffs/puffs_ops.3: revision 1.31 sys/fs/puffs/puffs_vnops.c: revision 1.172 sys/fs/puffs/puffs_vnops.c: revision 1.173 sys/fs/puffs/puffs_vnops.c: revision 1.174 usr.sbin/perfused/perfused.c: revision 1.24 sys/fs/puffs/puffs_sys.h: revision 1.80 sys/fs/puffs/puffs_sys.h: revision 1.81 sys/fs/puffs/puffs_sys.h: revision 1.82 lib/libperfuse/subr.c: revision 1.19 lib/libperfuse/perfuse.c: revision 1.30 sys/fs/puffs/puffs_msgif.c: revision 1.90 sys/fs/puffs/puffs_msgif.c: revision 1.91 sys/fs/puffs/puffs_msgif.c: revision 1.92 lib/libperfuse/ops.c: revision 1.59 lib/libpuffs/puffs.3: revision 1.53 lib/libperfuse/debug.c: revision 1.12 lib/libpuffs/puffs.3: revision 1.54 sys/fs/puffs/puffs_vnops.c: revision 1.167 sys/fs/puffs/puffs_msgif.h: revision 1.79 usr.sbin/perfused/msg.c: revision 1.21 sys/fs/puffs/puffs_vfsops.c: revision 1.102 sys/fs/puffs/puffs_vfsops.c: revision 1.103 sys/fs/puffs/puffs_vfsops.c: revision 1.105 lib/libpuffs/puffs.h: revision 1.123 lib/libperfuse/perfuse_if.h: revision 1.20 lib/libperfuse/perfuse.c: revision 1.29 lib/libpuffs/dispatcher.c: revision 1.42 lib/libpuffs/dispatcher.c: revision 1.43 - Fix same vnodes associated with multiple cookies The scheme used to retreive known nodes on lookup was flawed, as it only used parent and name. This produced a different cookie for the same file if it was renamed, when looking up ../ or when dealing with multiple files associated with the same name through link(2). We therefore abandon the use of node name and introduce hashed lists of inodes. This causes a huge rewrite of reclaim code, which do not attempt to keep parents allocated until all their children are reclaimed - Fix race conditions in reclaim There are a few situations where we issue multiple FUSE operations for a PUFFS operation. On reclaim, we therefore have to wait for all FUSE operation to complete, not just the current exchanges. We do this by introducing node reference count with node_ref() and node_rele(). - Detect data loss caused by FAF VOP_PUTPAGES causes FAF writes where the kernel does not check the operation result. At least issue a warning on error. - Enjoy FAF shortcut on setattr No need to wait for the result if the kernel does not want it. There is however an exception for setattr that touch the size, we need to wait for completion because we have other operations queued for after the resize. - Fix fchmod() on write-open file fchmod() on a node open with write privilege will send setattr with both mode and size set. This confuses some FUSE filesystem. Therefore we send two FUSE operations, one for mode, and one for size. - Remove node TTL handling for netbsd-5 for simplicity sake. The code still builds on netbsd-5 but does not have the node TTL feature anymore. It works fine with kernel support on netbsd-6. - Improve PUFFS_KFLAG_CACHE_FS_TTL by reclaiming older inactive nodes. The normal kernel behavior is to retain inactive nodes in the freelist until it runs out of vnodes. This has some merit for local filesystems, where the cost of an allocation is about the same as the cost of a lookup. But that situation is not true for distributed filesystems. On the other hand, keeping inactive nodes for a long time hold memory in the file server process, and when the kernel runs out of vnodes, it produce reclaim avalanches that increase lattency for other operations. We do not reclaim inactive vnodes immediatly either, as they may be looked up again shortly. Instead we introduce a grace time and we reclaim nodes that have been inactive beyond the grace time. - Fix lookup/reclaim race condition. The above improvement undercovered a race condition between lookup and reclaim. If we reclaimed a vnode associated with a userland cookie while a lookup returning that same cookiewas inprogress, then the kernel ends up with a vnode associated with a cookie that has been reclaimed in userland. Next operation on the cookie will crash (or at least confuse) the filesystem. We fix this by introducing a lookup count in kernel and userland. On reclaim, the kernel sends the count, which enable userland to detect situation where it initiated a lookup that is not completed in kernel. In such a situation, the reclaim must be ignored, as the node is about to be looked up again. Fix hang unmount bug introduced by last commit. We introduced a slow queue for delayed reclaims, while the existing queue for unmount, flush and exist has been renamed fast queue. Both queues had timestamp for when an operation should be done, but it was useless for the fast queue, which is always used to run an operation ASAP. And the timestamp test had an error that turned ASAP into "at next tick", but nobody what there to wake the thread at next tick, hence the hang. The fix is to remove the useless and buggy timestamp test for fast queue. Rename slow sopreq queue into node sopreq queue, to refet the fact that is only intended for postponed node reclaims. When purging the node sopreq queue, do not call puffs_msg_sendresp(), as it makes no sense. Fix race condition between (create\|mknod\|mkdir\|symlino) and reclaim, just like we did it between lookup and reclaim. Missing bit in previous commit (prevent race between create\|mknod\|mkdir\|symlink and reclaim) Bump date for previous. New sentence, new line; remove trailing whitespace; fix typos; punctuation nits. Add PUFFS_KFLAG_CACHE_DOTDOT so that vnodes hold a reference on their parent, keeping them active, and allowing to lookup .. without sending a request to the filesystem. Enable the featuure for perfused, as this is how FUSE works. Missing bit in previous commit (PUFFS_KFLAG_CACHE_DOTDOT option to avoid looking up ..)
1.119.4.2	05-Jul-2012	riz	Pull up following revision(s) (requested by manu in ticket #392): lib/libpuffs/puffs_ops.3: revision 1.32 lib/libperfuse/ops.c: revision 1.58 lib/libpuffs/puffs.h: revision 1.122 lib/libpuffs/dispatcher.c: revision 1.41 lib/libperfuse/perfuse_priv.h: revision 1.30 Pass the FAF flag to handlers: - setattr_ttl is updated to add a flag argument. Since it was not present in a previous release, we can change its API - write2 is introduced, this is write with an extra flag for FAF. - fsync already has the FAF information in a flag and needs no change - for other operations, FAF is unconditional Fix the build by adding (unused) flags argument to perfuse_node_setattr_ttl
1.119.4.1	23-Apr-2012	riz	Pull up following revision(s) (requested by manu in ticket #195): lib/libskey/skeysubr.c: revision 1.27 lib/libkvm/kvm_getloadavg.c: revision 1.11 lib/libwrap/update.c: revision 1.9 lib/liby/yyerror.c: revision 1.9 lib/libpuffs/puffs_ops.3: revision 1.30 lib/libwrap/misc.c: revision 1.10 lib/libwrap/hosts_access.c: revision 1.20 lib/libpuffs/pnode.c: revision 1.11 lib/libperfuse/subr.c: revision 1.17 lib/libpuffs/pnode.c: revision 1.12 lib/libperfuse/subr.c: revision 1.18 lib/libwrap/options.c: revision 1.15 lib/libwrap/fix_options.c: revision 1.11 lib/libperfuse/ops.c: revision 1.52 lib/libperfuse/ops.c: revision 1.53 lib/libperfuse/ops.c: revision 1.54 lib/libwrap/hosts_ctl.c: revision 1.5 lib/libintl/gettext.c: revision 1.27 lib/libwrap/shell_cmd.c: revision 1.6 lib/libpuffs/dispatcher.c: revision 1.39 lib/libperfuse/perfuse_priv.h: revision 1.27 lib/libwrap/socket.c: revision 1.19 lib/libpuffs/puffs.3: revision 1.50 lib/libperfuse/perfuse_priv.h: revision 1.28 lib/libpuffs/puffs_priv.h: revision 1.45 lib/libpuffs/puffs.3: revision 1.51 lib/libperfuse/perfuse_priv.h: revision 1.29 lib/libwrap/percent_x.c: revision 1.5 lib/libpuffs/puffs.3: revision 1.52 lib/libperfuse/debug.c: revision 1.11 sys/fs/puffs/puffs_vnops.c: revision 1.165 lib/libwrap/tcpd.h: revision 1.13 sys/fs/puffs/puffs_vnops.c: revision 1.166 lib/libwrap/eval.c: revision 1.7 sys/fs/puffs/puffs_msgif.h: revision 1.78 sys/fs/puffs/puffs_vfsops.c: revision 1.101 lib/libwrap/rfc931.c: revision 1.9 lib/libwrap/clean_exit.c: revision 1.5 lib/libpuffs/puffs.h: revision 1.120 lib/libc/stdlib/jemalloc.c: revision 1.27 lib/librmt/rmtlib.c: revision 1.26 lib/libpuffs/puffs.h: revision 1.121 sys/fs/puffs/puffs_sys.h: revision 1.79 lib/librumpclient/rumpclient.c: revision 1.48 lib/libwrap/refuse.c: revision 1.5 lib/libperfuse/perfuse.c: revision 1.26 lib/libperfuse/perfuse.c: revision 1.27 tests/fs/puffs/t_fuzz.c: revision 1.5 lib/libperfuse/perfuse.c: revision 1.28 lib/libpuffs/dispatcher.c: revision 1.40 sys/fs/puffs/puffs_node.c: revision 1.24 lib/libwrap/diag.c: revision 1.9 lib/libintl/textdomain.c: revision 1.13 Use C89 function definition Add name and atttribute cache with filesytem provided TTL. lookup, create, mknod, mkdir, symlink, getattr and setattr messages have been extended so that attributes and their TTL can be provided by the filesytem. lookup, create, mknod, mkdir, and symlink messages are also extended so that the filesystem can provide name TTL. Add PUFFS_KFLAG_CACHE_FS_TTL flag to puffs_init(3) to use name and attribute cache with filesystem provided TTL. lookup, create, mknod, mkdir, symlink, getattr and setattr messages have been extended so that attributes and their TTL can be provided by the filesytem. lookup, create, mknod, mkdir, and symlink messages are also extended so that the filesystem can provide name TTL. The filesystem updates attributes and TTL using puffs_pn_getvap(3), puffs_pn_getvattl(3), and puffs_pn_getcnttl(3) Use new PUFFS_KFLAG_CACHE_FS_TTL option to puffs_init(3) so that FUSE TTL on name and attributes are used. This save many PUFFS operations and improves performances. PUFFS_KFLAG_CACHE_FS_TTL is #ifdef'ed in many places for now so that libperfuse can still be used on netbsd-5. Split file system. Comma fixes. Remove dangling "and". Bump date for previous. - Makesure update_va does not change vnode size when it should not. For instance when doing a fault-issued VOP_GETPAGES within VOP_WRITE, changing size leads to panic: genfs_getpages: past eof. -Handle ticks wrap around for vnode name andattribute timeout - When using PUFFS_KFLAG_CACHE_FS_TTL, do not use puffs_node to carry attribute and TTL fora newly created node. Instead extend puffs_newinfo and add puffs_newinfo_setva() and puffs_newinfo_setttl() - Remove node_mk_common_final in libperfuse. It used to set uid/gid for a newly created vnode but has been made redundant along time ago since uid and gid are properly set in FUSE header. - In libperfuse, check for corner case where opc = 0 on INACTIVE and RECLAIM (how is it possible? Check for it to avoid a crash anyway) - In libperfuse, make sure we unlimit RLIMIT_AS and RLIMIT_DATA so that we do notrun out of memory because the kernel is lazy at reclaiming vnodes. - In libperfuse, cleanup style of perfuse_destroy_pn() Do not set PUFFS_KFLAG_CACHE_FS_TTL for PUFFS tests
1.119.4.3.2.1	01-Nov-2012	matt	sync with netbsd-6-0-RELEASE.
1.119.2.3	30-Oct-2012	yamt	sync with head
1.119.2.2	23-May-2012	yamt	sync with head.
1.119.2.1	17-Apr-2012	yamt	sync with head
1.124.10.2	05-Nov-2014	snj	Pull up following revision(s) (requested by manu in ticket #181): lib/libperfuse/fuse.h: revision 1.6 lib/libperfuse/ops.c: revision 1.78 lib/libperfuse/perfuse.c: revision 1.35 lib/libperfuse/perfuse_priv.h: revision 1.36 lib/libpuffs/dispatcher.c: revision 1.48 lib/libpuffs/opdump.c: revision 1.37 lib/libpuffs/puffs.c: revision 1.118 lib/libpuffs/puffs.h: revision 1.126 lib/libpuffs/puffs_ops.3: revisions 1.40-1.41 sys/fs/puffs/puffs_msgif.h: revision 1.82-1.83 sys/fs/puffs/puffs_msgif.h: revision 1.82 sys/fs/puffs/puffs_vnops.c: revision 1.196 Add PUFFS support for fallocate and fdiscard operations -- libpuffs support for fallocate and fdiscard operations -- Add PUFFS_HAVE_FALLOCATE in puffs_msgif.h so that filesystem can decide at build time wether fallocate is usable -- FUSE fallocate support There seems to be no fdiscard FUSE operation at the moment, hence that one is left unused.
1.124.10.1	24-Aug-2014	martin	Pull up following revision(s) (requested by manu in ticket #52): lib/libperfuse/perfuse_priv.h: revision 1.33 lib/libperfuse/ops.c: revision 1.67 lib/libpuffs/dispatcher.c: revision 1.47 lib/libperfuse/perfuse_priv.h: revision 1.34 lib/libperfuse/ops.c: revision 1.68 lib/libperfuse/ops.c: revision 1.69 lib/libpuffs/puffs_ops.3: revision 1.37 lib/libpuffs/puffs_ops.3: revision 1.38 lib/libperfuse/perfuse.c: revision 1.32 lib/libperfuse/perfuse.c: revision 1.33 usr.sbin/perfused/msg.c: revision 1.22 lib/libpuffs/puffs.h: revision 1.125 Implement FUSE direct I/O. Remove useless code and warnings
1.127.16.1	13-Apr-2020	martin	Mostly merge changes from HEAD upto 20200411
1.14	11-Apr-2009	wiz	New sentence, new line.
1.13	20-Feb-2009	pooka	Remove the paragraph saying interfaces will change without warning. Note that a stable version of puffs first appeared in NetBSD 5.0.
1.12	25-May-2008	wiz	branches: 1.12.4; 1.12.6; Punctuation markup nit.
1.11	28-Jan-2008	pooka	branches: 1.11.4; 1.11.6; Implement lazy contexts for file systems using puffs_mainloop(). Previously each request was executed on its own callcontext and switched to every time the request was being processed. Now requests borrow the mainloop context and switch only if/when they yield. This takes two context switches away from a file system request bringing down the typical amounts 2->0 (e.g. dtfs) and 4->2 (e.g. psshfs). The interfaces for manually executing requests changed a bit: puffs_dispatch_create() and puffs_dispatch_exec() must now be used. They are not tested, as nothing in-tree wants them and I doubt anyone else is really interested in them either. Also do some misc code cleanup related to execution contexts. The "work-in-progress checkpoint" committed over a year ago was starting to look slightly weed-infested.
1.10	15-Dec-2007	pooka	update to match reality
1.9	07-Dec-2007	pooka	puffs_cc_getusermount(), puffs_cc_getspecific() and puffs_docc() are dead.
1.8	01-Dec-2007	wiz	Fix a typo that has migrated across files.
1.7	21-Oct-2007	pooka	Add puffs_cc_schedule() which marks a pcc runnable and will pass execution to it when in the main loop the next time.
1.6	06-May-2007	pooka	branches: 1.6.4; Xr puffs_framebuf
1.5	19-Apr-2007	pooka	improve descriptions and markup
1.4	19-Apr-2007	pooka	add puffs_cc_getspecific(pcc), which is shorthand for: puffs_getspecific(puffs_cc_getusermount(pcc))
1.3	20-Jan-2007	pooka	document puffs_docc
1.2	20-Jan-2007	wiz	Sort SEE ALSO.
1.1	19-Jan-2007	pooka	Bring the documentation slightly more into sync with the current reality. It's still really a far cry from proper documentation, though.
1.6.4.3	23-Mar-2008	matt	sync with HEAD
1.6.4.2	09-Jan-2008	matt	sync with HEAD
1.6.4.1	06-Nov-2007	matt	sync with HEAD
1.11.6.1	23-Jun-2008	wrstuden	Sync w/ -current. 34 merge conflicts to follow.
1.11.4.1	04-Jun-2008	yamt	sync with head
1.12.6.1	13-May-2009	jym	Sync with HEAD. Third (and last) commit. See http://mail-index.netbsd.org/source-changes/2009/05/13/msg221222.html
1.12.4.1	24-Feb-2009	snj	Pull up following revision(s) (requested by pooka in ticket #484): lib/libpuffs/puffs.3: revision 1.46 lib/libpuffs/puffs_cc.3: revision 1.13 lib/libpuffs/puffs_cred.3: revision 1.4 lib/libpuffs/puffs_flush.3: revision 1.8 lib/libpuffs/puffs_framebuf.3: revision 1.26 lib/libpuffs/puffs_node.3: revision 1.4 lib/libpuffs/puffs_ops.3: revision 1.23 lib/libpuffs/puffs_path.3: revision 1.4 lib/libpuffs/puffs_suspend.3: revision 1.3 Remove the paragraph saying interfaces will change without warning. Note that a stable version of puffs first appeared in NetBSD 5.0.
1.5	11-Apr-2009	joerg	Fix missing quote.
1.4	20-Feb-2009	pooka	Remove the paragraph saying interfaces will change without warning. Note that a stable version of puffs first appeared in NetBSD 5.0.
1.3	01-Dec-2007	wiz	branches: 1.3.12; 1.3.14; Fix a typo that has migrated across files.
1.2	18-Oct-2007	pooka	* Change type of boolean functions to, well, boolean instead of int. * add puffs_cid_isequal() which can be used to test if the two caller id's refer to the same process/lwp.
1.1	22-Mar-2007	pooka	branches: 1.1.4; document credentials & access control methods
1.1.4.2	09-Jan-2008	matt	sync with HEAD
1.1.4.1	06-Nov-2007	matt	sync with HEAD
1.3.14.1	13-May-2009	jym	Sync with HEAD. Third (and last) commit. See http://mail-index.netbsd.org/source-changes/2009/05/13/msg221222.html
1.3.12.1	24-Feb-2009	snj	Pull up following revision(s) (requested by pooka in ticket #484): lib/libpuffs/puffs.3: revision 1.46 lib/libpuffs/puffs_cc.3: revision 1.13 lib/libpuffs/puffs_cred.3: revision 1.4 lib/libpuffs/puffs_flush.3: revision 1.8 lib/libpuffs/puffs_framebuf.3: revision 1.26 lib/libpuffs/puffs_node.3: revision 1.4 lib/libpuffs/puffs_ops.3: revision 1.23 lib/libpuffs/puffs_path.3: revision 1.4 lib/libpuffs/puffs_suspend.3: revision 1.3 Remove the paragraph saying interfaces will change without warning. Note that a stable version of puffs first appeared in NetBSD 5.0.
1.8	20-Feb-2009	pooka	Remove the paragraph saying interfaces will change without warning. Note that a stable version of puffs first appeared in NetBSD 5.0.
1.7	12-Aug-2008	pooka	branches: 1.7.4; 1.7.6; Replace void * by puffs_cookie_t where appropriate. No functional change.
1.6	01-Dec-2007	wiz	branches: 1.6.8; Fix another persistent spello.
1.5	01-Dec-2007	wiz	Fix a typo that has migrated across files.
1.4	17-May-2007	pooka	branches: 1.4.4; struct puffs_usermoint does sound more lively, but be serious and fix the typos to puffs_usermount
1.3	06-Apr-2007	pooka	document the flush routines
1.2	06-Apr-2007	pooka	document puffs_inval_pagecache_node{,_range}
1.1	19-Jan-2007	pooka	Bring the documentation slightly more into sync with the current reality. It's still really a far cry from proper documentation, though.
1.4.4.1	09-Jan-2008	matt	sync with HEAD
1.6.8.1	18-Sep-2008	wrstuden	Sync with wrstuden-revivesa-base-2.
1.7.6.1	13-May-2009	jym	Sync with HEAD. Third (and last) commit. See http://mail-index.netbsd.org/source-changes/2009/05/13/msg221222.html
1.7.4.1	24-Feb-2009	snj	Pull up following revision(s) (requested by pooka in ticket #484): lib/libpuffs/puffs.3: revision 1.46 lib/libpuffs/puffs_cc.3: revision 1.13 lib/libpuffs/puffs_cred.3: revision 1.4 lib/libpuffs/puffs_flush.3: revision 1.8 lib/libpuffs/puffs_framebuf.3: revision 1.26 lib/libpuffs/puffs_node.3: revision 1.4 lib/libpuffs/puffs_ops.3: revision 1.23 lib/libpuffs/puffs_path.3: revision 1.4 lib/libpuffs/puffs_suspend.3: revision 1.3 Remove the paragraph saying interfaces will change without warning. Note that a stable version of puffs first appeared in NetBSD 5.0.
1.30	16-Feb-2015	wiz	Sort sections.
1.29	01-Apr-2010	pooka	fix a few pastos
1.28	04-Jan-2010	pooka	fix typography
1.27	22-Nov-2009	mbalmer	s/the the/the/
1.26	20-Feb-2009	pooka	Remove the paragraph saying interfaces will change without warning. Note that a stable version of puffs first appeared in NetBSD 5.0.
1.25	06-Sep-2008	wiz	branches: 1.25.4; 1.25.6; Drop trailing whitespace.
1.24	06-Sep-2008	pooka	Add AsiaBSDCon 2008 paper to SEE ALSO.
1.23	29-Jan-2008	pooka	branches: 1.23.6; whoops, bump date - pretty significant change
1.22	29-Jan-2008	pooka	Fix a weirdosity in the framebuf gotfb interface: hand buffer ownership to the called routine instead of keeping it. This means that callees must now destroy the buffer once they are done with it.
1.21	08-Dec-2007	wiz	Fix typo, remove superfluous minus and describe "fd" as file descriptor in two places.
1.20	04-Dec-2007	pooka	tyop: reserver -> reserve
1.19	02-Dec-2007	wiz	Add comma in enumeration. Punctuation nits. Fix a typo.
1.18	01-Dec-2007	wiz	Fix another persistent spello.
1.17	01-Dec-2007	wiz	Fix a typo that has migrated across files.
1.16	01-Dec-2007	pooka	fix gotframe signature to match code
1.15	05-Oct-2007	pooka	SEE ALSO paper
1.14	02-Oct-2007	pooka	minor updates/clarifications
1.13	06-Sep-2007	pooka	document puffs_framev_framebuf_ccpromote()
1.12	06-Sep-2007	pooka	Give cmpframe() the ability to shortcircuit the comparison loop by signalling back that the buffer under examination is not a response at all.
1.11	01-Sep-2007	pooka	Add gotframe callback, which is called if an incoming frame matches no outstanding request.
1.10	21-Jul-2007	pooka	branches: 1.10.4; 1.10.6; document puffs_framev_enqueue_waitevent
1.9	08-Jul-2007	pooka	bring up-to-date
1.8	06-Jul-2007	pooka	make few function types match reality
1.7	20-May-2007	pooka	Change the signature of puffs_respcmp to return -1,0,1 for smaller, equal, larger, respectively instead of 0/1 for non/equal. This will allow sorting the buffers for faster matching in libpuffs. While here, change the name from respcmp to framecmp, as that better reflects the purpose. NOTE! there is no obvious way to make compilation fail for file systems which may already be using this feature (although I don't think there are any outside our tree, as the feature is two weeks old). Nevertheless, non-updated file systems will fail very quickly.
1.6	16-May-2007	pooka	remember to document puffs_framev_cb error parameter addition
1.5	16-May-2007	pooka	switch the arbitrary default errno value for network i/o from ECONNABORTED to ECONNRESET, which is at least a little bit better
1.4	16-May-2007	pooka	update .Nd to reflect current reality better
1.3	15-May-2007	pooka	Step 3 and currently final step of a more general purpose event loop: support removal and addition of i/o file descriptors on the fly. * detect closed file descriptors * automatically free waiters of a dead file descriptor * give the file server the possibility to specify a callback which notifies of a dead file descriptor * move loop function to be a property of the mainloop instead of framebuf (doesn't change effective behaviour) * add the possibility to configure a timespec parameter which attempts to call the loop function periodically * move the event loop functions from the puffs_framebuf namespace to puffs_framev to differential between pure memory management functions
1.2	06-May-2007	pooka	document puffs_framebuf
1.1	05-May-2007	pooka	add stubby version of puffs_framebuf manual page. proper one forthcoming to an internet near you Real Soon Now(tm)
1.10.6.2	21-Jul-2007	pooka	document puffs_framev_enqueue_waitevent
1.10.6.1	21-Jul-2007	pooka	file puffs_framebuf.3 was added on branch matt-mips64 on 2007-07-21 09:36:50 +0000
1.10.4.3	23-Mar-2008	matt	sync with HEAD
1.10.4.2	09-Jan-2008	matt	sync with HEAD
1.10.4.1	06-Nov-2007	matt	sync with HEAD
1.23.6.1	18-Sep-2008	wrstuden	Sync with wrstuden-revivesa-base-2.
1.25.6.1	13-May-2009	jym	Sync with HEAD. Third (and last) commit. See http://mail-index.netbsd.org/source-changes/2009/05/13/msg221222.html
1.25.4.1	24-Feb-2009	snj	Pull up following revision(s) (requested by pooka in ticket #484): lib/libpuffs/puffs.3: revision 1.46 lib/libpuffs/puffs_cc.3: revision 1.13 lib/libpuffs/puffs_cred.3: revision 1.4 lib/libpuffs/puffs_flush.3: revision 1.8 lib/libpuffs/puffs_framebuf.3: revision 1.26 lib/libpuffs/puffs_node.3: revision 1.4 lib/libpuffs/puffs_ops.3: revision 1.23 lib/libpuffs/puffs_path.3: revision 1.4 lib/libpuffs/puffs_suspend.3: revision 1.3 Remove the paragraph saying interfaces will change without warning. Note that a stable version of puffs first appeared in NetBSD 5.0.
1.6	04-Aug-2021	andvar	fix typos in puffs man pages.
1.5	13-May-2009	wiz	No Pp needed before El.
1.4	20-Feb-2009	pooka	Remove the paragraph saying interfaces will change without warning. Note that a stable version of puffs first appeared in NetBSD 5.0.
1.3	01-Dec-2007	wiz	branches: 1.3.12; 1.3.14; Fix a typo that has migrated across files.
1.2	24-Jun-2007	pooka	branches: 1.2.4; * document puffs_pn_remove * all-around improve the manual page
1.1	19-Jan-2007	pooka	Bring the documentation slightly more into sync with the current reality. It's still really a far cry from proper documentation, though.
1.2.4.1	09-Jan-2008	matt	sync with HEAD
1.3.14.1	13-May-2009	jym	Sync with HEAD. Third (and last) commit. See http://mail-index.netbsd.org/source-changes/2009/05/13/msg221222.html
1.3.12.1	24-Feb-2009	snj	Pull up following revision(s) (requested by pooka in ticket #484): lib/libpuffs/puffs.3: revision 1.46 lib/libpuffs/puffs_cc.3: revision 1.13 lib/libpuffs/puffs_cred.3: revision 1.4 lib/libpuffs/puffs_flush.3: revision 1.8 lib/libpuffs/puffs_framebuf.3: revision 1.26 lib/libpuffs/puffs_node.3: revision 1.4 lib/libpuffs/puffs_ops.3: revision 1.23 lib/libpuffs/puffs_path.3: revision 1.4 lib/libpuffs/puffs_suspend.3: revision 1.3 Remove the paragraph saying interfaces will change without warning. Note that a stable version of puffs first appeared in NetBSD 5.0.
1.48	03-Dec-2021	pho	Avoid using register_t or vsize_t in <puffs.h> <puffs.h> is a user-space header, and should not use any of kernel-only types. It's not reasonable to require user-land filesystems to #define _KERNTYPES.
1.47	04-Aug-2021	andvar	fix typos in puffs man pages.
1.46	03-Jul-2017	wiz	Remove workaround for ancient HTML generation code.
1.45	29-Aug-2016	tnn	mention puffs_node_pathconf and puffs_node_advlock
1.44	12-Nov-2015	christos	consistency fixes.
1.43	09-Nov-2015	christos	fix return type
1.42	16-Feb-2015	wiz	filesystem -> file system
1.41	31-Oct-2014	wiz	Bump date for previous. Remove trailing space.
1.40	31-Oct-2014	manu	libpuffs support for fallocate and fdiscard operations
1.39	18-Oct-2014	snj	src is too big these days to tolerate superfluous apostrophes. It's "its", people!
1.38	16-Aug-2014	wiz	Bump date for previous.
1.37	16-Aug-2014	manu	Add an open2 method, like open but with an additionnal oflags used by the filesystem to send back information about the file. This is used to implement PUFFS_OPEN_IO_DIRECT by which the filesystem tells the kernel that read/write should bypass the page cache.
1.36	16-Aug-2012	wiz	branches: 1.36.10; Whitespace fixes. Bump date for previous.
1.35	16-Aug-2012	manu	Fix regression that has been introduced when the lookup/reclaim race condition was addressed in libpuffs by counting lookups. The fix assumes that cookies map to struct puffs_cookie, which has not been documented as a requirement for filesystems using libpuffs. As an example, we got burnt by this assumption in libp2k (kern/46734), and we fixed bit by actually mapping libp2k cookies to struct puffs_node. It is unlikely, but there may be third party filesystems that use cookies unmapped to struct puffs_node, and they were left broken for now. - we introduce a puffs_init() flag PUFFS_FLAG_PNCOOKIE that let filesystems inform libpuffs that they map cookies to struct puffs_node. Is that flag is used, the lookup/reclaim race condition fix is enabled. We enable the flag for libp2k. - filesystems that use puffs_pn_new() obviouslty use struct puffs_node and gain PUFFS_FLAG_PNCOOKIE automatically even if they did not specify it in puffs_init(). This include all our PUFFS filesystem in-tree except libp2k. - for filesystems not willing to use struct puffs_node, we introduce a reclaim2 vnop, which is reclaim with an additionnal lookup count argument. This vnop let the filesystem implement the lookup/reclaim race fix on its own.
1.34	28-Jul-2012	njoly	Kill extra Fc macro.
1.33	27-Jun-2012	wiz	Bump date for previous, remove trailing whitespace.
1.32	27-Jun-2012	manu	Pass the FAF flag to handlers: - setattr_ttl is updated to add a flag argument. Since it was not present in a previous release, we can change its API - write2 is introduced, this is write with an extra flag for FAF. - fsync already has the FAF information in a flag and needs no change - for other operations, FAF is unconditional
1.31	18-Apr-2012	wiz	Bump date for previous. New sentence, new line; remove trailing whitespace; fix typos; punctuation nits.
1.30	18-Apr-2012	manu	- When using PUFFS_KFLAG_CACHE_FS_TTL, do not use puffs_node to carry attribute and TTL fora newly created node. Instead extend puffs_newinfo and add puffs_newinfo_setva() and puffs_newinfo_setttl() - Remove node_mk_common_final in libperfuse. It used to set uid/gid for a newly created vnode but has been made redundant along time ago since uid and gid are properly set in FUSE header. - In libperfuse, check for corner case where opc = 0 on INACTIVE and RECLAIM (how is it possible? Check for it to avoid a crash anyway) - In libperfuse, make sure we unlimit RLIMIT_AS and RLIMIT_DATA so that we do notrun out of memory because the kernel is lazy at reclaiming vnodes. - In libperfuse, cleanup style of perfuse_destroy_pn()
1.29	04-Jul-2011	manu	branches: 1.29.2; 1.29.4; Add a flag to VOP_LISTEXTATTR(9) so that the vnode interface can tell the filesystem in which format extended attribute shall be listed. There are currently two formats: - NUL-terminated strings, used for listxattr(2), this is the default. - one byte length-pprefixed, non NUL-terminated strings, used for extattr_list_file(2), which is obtanined by setting the EXTATTR_LIST_PREFIXLEN flag to VOP_LISTEXTATTR(9) This approach avoid the need for converting the list back and forth, except in libperfuse, since FUSE uses NUL-terminated strings, and the kernel may have requested EXTATTR_LIST_PREFIXLEN.
1.28	19-Apr-2011	manu	In node_lookup() ops, pcn->pcn_nameiop contains a NAMEI_* constant, not PUFFSLOOKUP_*
1.27	13-Dec-2010	pooka	document abortop. part of PR kern/44208
1.26	21-May-2010	pooka	mention extended attribute ops
1.25	11-Apr-2009	wiz	New sentence, new line.
1.24	11-Apr-2009	joerg	Fix markup.
1.23	20-Feb-2009	pooka	Remove the paragraph saying interfaces will change without warning. Note that a stable version of puffs first appeared in NetBSD 5.0.
1.22	25-Nov-2008	pooka	branches: 1.22.2; i'm almost certain that puffs_cooie_t is supposed to be puffs_cookie_t
1.21	12-Aug-2008	pooka	branches: 1.21.4; Replace void * by puffs_cookie_t where appropriate. No functional change.
1.20	16-Dec-2007	pooka	branches: 1.20.6; Clarify that link_target in node_symlink is null-terminated.
1.19	02-Dec-2007	wiz	Add comma in enumeration.
1.18	01-Dec-2007	wiz	Fix a typo that has migrated across files.
1.17	30-Nov-2007	pooka	Rototill. Ok, ok, a few more words about it: stop holding puffs_cc as a holy value and passing it around to almost every possible place (popquiz: which kernel variable does this remind you of?). Instead, pass the natural choice, puffs_usermount, and fetch puffs_cc via puffs_cc_getcc() only in routines which actually need it. This not only simplifies code, but (thanks to the introduction of puffs_cc_getcc()) enables constructs which weren't previously sanely possible, say layering as a curious example. There's still a little to do on this front, but this was the major fs interface blast.
1.16	27-Nov-2007	pooka	Remove "puffs_cid" from the puffs interface following l-removal from the kernel vfs interfaces. puffs_cc_getcaller(pcc) can be used now should the same information be desired.
1.15	31-Jul-2007	pooka	branches: 1.15.2; 1.15.4; Per rumble's lead in vnodeops, clarify the bit about readdir cookies.
1.14	25-Jul-2007	pooka	* add missing credential handle to fsync args * and now with more cosmetic attack
1.13	25-Jul-2007	pooka	pure cosmetics: minor facelift
1.12	20-Jul-2007	pooka	note caveat with puffs_node_poll
1.11	19-Jul-2007	wiz	Use HTML escapes. Drop trailing whitespace. Punctuation nits.
1.10	19-Jul-2007	pooka	wizd reminded me that if I provide the synopsis for puffs_node_seek(), I would do well to also describe what it does. so make it so.
1.9	18-Jul-2007	pooka	uncomment puffs_node_seek - it's supported and in use
1.8	18-Jul-2007	pooka	puffs_cid* is always const
1.7	05-Jul-2007	pooka	apply ketchup to recent source changes in libpuffs
1.6	18-May-2007	pooka	document puffs_node_poll
1.5	09-May-2007	pooka	document puffs_setback()
1.4	18-Apr-2007	pooka	fix readdir prototype
1.3	17-Apr-2007	pooka	* describe fhtonode, nodetofh and suspend * update description of readdir to match reality
1.2	17-Apr-2007	pooka	improve markup and descriptions
1.1	16-Apr-2007	pooka	Move file system and node callbacks into their own manual page. Improve markup and descriptions a bit (not completely finished yet).
1.15.4.2	31-Jul-2007	pooka	Per rumble's lead in vnodeops, clarify the bit about readdir cookies.
1.15.4.1	31-Jul-2007	pooka	file puffs_ops.3 was added on branch matt-mips64 on 2007-07-31 21:48:19 +0000
1.15.2.1	09-Jan-2008	matt	sync with HEAD
1.20.6.1	18-Sep-2008	wrstuden	Sync with wrstuden-revivesa-base-2.
1.21.4.2	12-Apr-2009	snj	Pull up following revision(s) (requested by joerg in ticket #691): lib/libpuffs/puffs_ops.3: revision 1.24 Fix markup.
1.21.4.1	24-Feb-2009	snj	Pull up following revision(s) (requested by pooka in ticket #484): lib/libpuffs/puffs.3: revision 1.46 lib/libpuffs/puffs_cc.3: revision 1.13 lib/libpuffs/puffs_cred.3: revision 1.4 lib/libpuffs/puffs_flush.3: revision 1.8 lib/libpuffs/puffs_framebuf.3: revision 1.26 lib/libpuffs/puffs_node.3: revision 1.4 lib/libpuffs/puffs_ops.3: revision 1.23 lib/libpuffs/puffs_path.3: revision 1.4 lib/libpuffs/puffs_suspend.3: revision 1.3 Remove the paragraph saying interfaces will change without warning. Note that a stable version of puffs first appeared in NetBSD 5.0.
1.22.2.1	13-May-2009	jym	Sync with HEAD. Third (and last) commit. See http://mail-index.netbsd.org/source-changes/2009/05/13/msg221222.html
1.29.4.5	03-Nov-2014	msaitoh	Pull up following revision(s) (requested by manu in ticket #1140): lib/libperfuse/ops.c 1.63-1.69 lib/libperfuse/perfuse.c 1.32-1.33 lib/libperfuse/perfuse_priv.h 1.32-1.34 lib/libperfuse/subr.c 1.20 lib/libpuffs/creds.c 1.16 lib/libpuffs/dispatcher.c 1.47 lib/libpuffs/puffs.h 1.125 lib/libpuffs/puffs_ops.3 1.37-1.38 lib/libpuffs/requests.c 1.24 sys/fs/puffs/puffs_msgif.h 1.81 sys/fs/puffs/puffs_sys.h 1.85 sys/fs/puffs/puffs_vnops.c 1.183 usr.sbin/perfused/msg.c 1.22 Bring libpuffs, libperfuse and perfused on par with -current: - implement FUSE direct I/O - remove useless code and warnings - fix missing GETATTR bugs - fix exended attribute get and list operations
1.29.4.4	13-Sep-2012	riz	Pull up following revision(s) (requested by manu in ticket #525): lib/libpuffs/dispatcher.c: revision 1.44 lib/libp2k/p2k.c: revision 1.57 lib/libpuffs/puffs_ops.3: revision 1.35 lib/libpuffs/pnode.c: revision 1.13 lib/libpuffs/puffs.3: revision 1.56 lib/libpuffs/puffs.h: revision 1.124 Fix regression that has been introduced when the lookup/reclaim race condition was addressed in libpuffs by counting lookups. The fix assumes that cookies map to struct puffs_cookie, which has not been documented as a requirement for filesystems using libpuffs. As an example, we got burnt by this assumption in libp2k (kern/46734), and we fixed bit by actually mapping libp2k cookies to struct puffs_node. It is unlikely, but there may be third party filesystems that use cookies unmapped to struct puffs_node, and they were left broken for now. - we introduce a puffs_init() flag PUFFS_FLAG_PNCOOKIE that let filesystems inform libpuffs that they map cookies to struct puffs_node. Is that flag is used, the lookup/reclaim race condition fix is enabled. We enable the flag for libp2k. - filesystems that use puffs_pn_new() obviouslty use struct puffs_node and gain PUFFS_FLAG_PNCOOKIE automatically even if they did not specify it in puffs_init(). This include all our PUFFS filesystem in-tree except libp2k. - for filesystems not willing to use struct puffs_node, we introduce a reclaim2 vnop, which is reclaim with an additionnal lookup count argument. This vnop let the filesystem implement the lookup/reclaim race fix on its own.
1.29.4.3	12-Aug-2012	martin	branches: 1.29.4.3.2; Pull up following revision(s) (requested by manu in ticket #438): lib/libperfuse/perfuse_priv.h: revision 1.31 sys/fs/puffs/puffs_msgif.h: revision 1.80 sys/fs/puffs/puffs_vnops.c: revision 1.171 lib/libpuffs/puffs_ops.3: revision 1.31 sys/fs/puffs/puffs_vnops.c: revision 1.172 sys/fs/puffs/puffs_vnops.c: revision 1.173 sys/fs/puffs/puffs_vnops.c: revision 1.174 usr.sbin/perfused/perfused.c: revision 1.24 sys/fs/puffs/puffs_sys.h: revision 1.80 sys/fs/puffs/puffs_sys.h: revision 1.81 sys/fs/puffs/puffs_sys.h: revision 1.82 lib/libperfuse/subr.c: revision 1.19 lib/libperfuse/perfuse.c: revision 1.30 sys/fs/puffs/puffs_msgif.c: revision 1.90 sys/fs/puffs/puffs_msgif.c: revision 1.91 sys/fs/puffs/puffs_msgif.c: revision 1.92 lib/libperfuse/ops.c: revision 1.59 lib/libpuffs/puffs.3: revision 1.53 lib/libperfuse/debug.c: revision 1.12 lib/libpuffs/puffs.3: revision 1.54 sys/fs/puffs/puffs_vnops.c: revision 1.167 sys/fs/puffs/puffs_msgif.h: revision 1.79 usr.sbin/perfused/msg.c: revision 1.21 sys/fs/puffs/puffs_vfsops.c: revision 1.102 sys/fs/puffs/puffs_vfsops.c: revision 1.103 sys/fs/puffs/puffs_vfsops.c: revision 1.105 lib/libpuffs/puffs.h: revision 1.123 lib/libperfuse/perfuse_if.h: revision 1.20 lib/libperfuse/perfuse.c: revision 1.29 lib/libpuffs/dispatcher.c: revision 1.42 lib/libpuffs/dispatcher.c: revision 1.43 - Fix same vnodes associated with multiple cookies The scheme used to retreive known nodes on lookup was flawed, as it only used parent and name. This produced a different cookie for the same file if it was renamed, when looking up ../ or when dealing with multiple files associated with the same name through link(2). We therefore abandon the use of node name and introduce hashed lists of inodes. This causes a huge rewrite of reclaim code, which do not attempt to keep parents allocated until all their children are reclaimed - Fix race conditions in reclaim There are a few situations where we issue multiple FUSE operations for a PUFFS operation. On reclaim, we therefore have to wait for all FUSE operation to complete, not just the current exchanges. We do this by introducing node reference count with node_ref() and node_rele(). - Detect data loss caused by FAF VOP_PUTPAGES causes FAF writes where the kernel does not check the operation result. At least issue a warning on error. - Enjoy FAF shortcut on setattr No need to wait for the result if the kernel does not want it. There is however an exception for setattr that touch the size, we need to wait for completion because we have other operations queued for after the resize. - Fix fchmod() on write-open file fchmod() on a node open with write privilege will send setattr with both mode and size set. This confuses some FUSE filesystem. Therefore we send two FUSE operations, one for mode, and one for size. - Remove node TTL handling for netbsd-5 for simplicity sake. The code still builds on netbsd-5 but does not have the node TTL feature anymore. It works fine with kernel support on netbsd-6. - Improve PUFFS_KFLAG_CACHE_FS_TTL by reclaiming older inactive nodes. The normal kernel behavior is to retain inactive nodes in the freelist until it runs out of vnodes. This has some merit for local filesystems, where the cost of an allocation is about the same as the cost of a lookup. But that situation is not true for distributed filesystems. On the other hand, keeping inactive nodes for a long time hold memory in the file server process, and when the kernel runs out of vnodes, it produce reclaim avalanches that increase lattency for other operations. We do not reclaim inactive vnodes immediatly either, as they may be looked up again shortly. Instead we introduce a grace time and we reclaim nodes that have been inactive beyond the grace time. - Fix lookup/reclaim race condition. The above improvement undercovered a race condition between lookup and reclaim. If we reclaimed a vnode associated with a userland cookie while a lookup returning that same cookiewas inprogress, then the kernel ends up with a vnode associated with a cookie that has been reclaimed in userland. Next operation on the cookie will crash (or at least confuse) the filesystem. We fix this by introducing a lookup count in kernel and userland. On reclaim, the kernel sends the count, which enable userland to detect situation where it initiated a lookup that is not completed in kernel. In such a situation, the reclaim must be ignored, as the node is about to be looked up again. Fix hang unmount bug introduced by last commit. We introduced a slow queue for delayed reclaims, while the existing queue for unmount, flush and exist has been renamed fast queue. Both queues had timestamp for when an operation should be done, but it was useless for the fast queue, which is always used to run an operation ASAP. And the timestamp test had an error that turned ASAP into "at next tick", but nobody what there to wake the thread at next tick, hence the hang. The fix is to remove the useless and buggy timestamp test for fast queue. Rename slow sopreq queue into node sopreq queue, to refet the fact that is only intended for postponed node reclaims. When purging the node sopreq queue, do not call puffs_msg_sendresp(), as it makes no sense. Fix race condition between (create\|mknod\|mkdir\|symlino) and reclaim, just like we did it between lookup and reclaim. Missing bit in previous commit (prevent race between create\|mknod\|mkdir\|symlink and reclaim) Bump date for previous. New sentence, new line; remove trailing whitespace; fix typos; punctuation nits. Add PUFFS_KFLAG_CACHE_DOTDOT so that vnodes hold a reference on their parent, keeping them active, and allowing to lookup .. without sending a request to the filesystem. Enable the featuure for perfused, as this is how FUSE works. Missing bit in previous commit (PUFFS_KFLAG_CACHE_DOTDOT option to avoid looking up ..)
1.29.4.2	05-Jul-2012	riz	Pull up following revision(s) (requested by manu in ticket #392): lib/libpuffs/puffs_ops.3: revision 1.32 lib/libperfuse/ops.c: revision 1.58 lib/libpuffs/puffs.h: revision 1.122 lib/libpuffs/dispatcher.c: revision 1.41 lib/libperfuse/perfuse_priv.h: revision 1.30 Pass the FAF flag to handlers: - setattr_ttl is updated to add a flag argument. Since it was not present in a previous release, we can change its API - write2 is introduced, this is write with an extra flag for FAF. - fsync already has the FAF information in a flag and needs no change - for other operations, FAF is unconditional Fix the build by adding (unused) flags argument to perfuse_node_setattr_ttl
1.29.4.1	23-Apr-2012	riz	Pull up following revision(s) (requested by manu in ticket #195): lib/libskey/skeysubr.c: revision 1.27 lib/libkvm/kvm_getloadavg.c: revision 1.11 lib/libwrap/update.c: revision 1.9 lib/liby/yyerror.c: revision 1.9 lib/libpuffs/puffs_ops.3: revision 1.30 lib/libwrap/misc.c: revision 1.10 lib/libwrap/hosts_access.c: revision 1.20 lib/libpuffs/pnode.c: revision 1.11 lib/libperfuse/subr.c: revision 1.17 lib/libpuffs/pnode.c: revision 1.12 lib/libperfuse/subr.c: revision 1.18 lib/libwrap/options.c: revision 1.15 lib/libwrap/fix_options.c: revision 1.11 lib/libperfuse/ops.c: revision 1.52 lib/libperfuse/ops.c: revision 1.53 lib/libperfuse/ops.c: revision 1.54 lib/libwrap/hosts_ctl.c: revision 1.5 lib/libintl/gettext.c: revision 1.27 lib/libwrap/shell_cmd.c: revision 1.6 lib/libpuffs/dispatcher.c: revision 1.39 lib/libperfuse/perfuse_priv.h: revision 1.27 lib/libwrap/socket.c: revision 1.19 lib/libpuffs/puffs.3: revision 1.50 lib/libperfuse/perfuse_priv.h: revision 1.28 lib/libpuffs/puffs_priv.h: revision 1.45 lib/libpuffs/puffs.3: revision 1.51 lib/libperfuse/perfuse_priv.h: revision 1.29 lib/libwrap/percent_x.c: revision 1.5 lib/libpuffs/puffs.3: revision 1.52 lib/libperfuse/debug.c: revision 1.11 sys/fs/puffs/puffs_vnops.c: revision 1.165 lib/libwrap/tcpd.h: revision 1.13 sys/fs/puffs/puffs_vnops.c: revision 1.166 lib/libwrap/eval.c: revision 1.7 sys/fs/puffs/puffs_msgif.h: revision 1.78 sys/fs/puffs/puffs_vfsops.c: revision 1.101 lib/libwrap/rfc931.c: revision 1.9 lib/libwrap/clean_exit.c: revision 1.5 lib/libpuffs/puffs.h: revision 1.120 lib/libc/stdlib/jemalloc.c: revision 1.27 lib/librmt/rmtlib.c: revision 1.26 lib/libpuffs/puffs.h: revision 1.121 sys/fs/puffs/puffs_sys.h: revision 1.79 lib/librumpclient/rumpclient.c: revision 1.48 lib/libwrap/refuse.c: revision 1.5 lib/libperfuse/perfuse.c: revision 1.26 lib/libperfuse/perfuse.c: revision 1.27 tests/fs/puffs/t_fuzz.c: revision 1.5 lib/libperfuse/perfuse.c: revision 1.28 lib/libpuffs/dispatcher.c: revision 1.40 sys/fs/puffs/puffs_node.c: revision 1.24 lib/libwrap/diag.c: revision 1.9 lib/libintl/textdomain.c: revision 1.13 Use C89 function definition Add name and atttribute cache with filesytem provided TTL. lookup, create, mknod, mkdir, symlink, getattr and setattr messages have been extended so that attributes and their TTL can be provided by the filesytem. lookup, create, mknod, mkdir, and symlink messages are also extended so that the filesystem can provide name TTL. Add PUFFS_KFLAG_CACHE_FS_TTL flag to puffs_init(3) to use name and attribute cache with filesystem provided TTL. lookup, create, mknod, mkdir, symlink, getattr and setattr messages have been extended so that attributes and their TTL can be provided by the filesytem. lookup, create, mknod, mkdir, and symlink messages are also extended so that the filesystem can provide name TTL. The filesystem updates attributes and TTL using puffs_pn_getvap(3), puffs_pn_getvattl(3), and puffs_pn_getcnttl(3) Use new PUFFS_KFLAG_CACHE_FS_TTL option to puffs_init(3) so that FUSE TTL on name and attributes are used. This save many PUFFS operations and improves performances. PUFFS_KFLAG_CACHE_FS_TTL is #ifdef'ed in many places for now so that libperfuse can still be used on netbsd-5. Split file system. Comma fixes. Remove dangling "and". Bump date for previous. - Makesure update_va does not change vnode size when it should not. For instance when doing a fault-issued VOP_GETPAGES within VOP_WRITE, changing size leads to panic: genfs_getpages: past eof. -Handle ticks wrap around for vnode name andattribute timeout - When using PUFFS_KFLAG_CACHE_FS_TTL, do not use puffs_node to carry attribute and TTL fora newly created node. Instead extend puffs_newinfo and add puffs_newinfo_setva() and puffs_newinfo_setttl() - Remove node_mk_common_final in libperfuse. It used to set uid/gid for a newly created vnode but has been made redundant along time ago since uid and gid are properly set in FUSE header. - In libperfuse, check for corner case where opc = 0 on INACTIVE and RECLAIM (how is it possible? Check for it to avoid a crash anyway) - In libperfuse, make sure we unlimit RLIMIT_AS and RLIMIT_DATA so that we do notrun out of memory because the kernel is lazy at reclaiming vnodes. - In libperfuse, cleanup style of perfuse_destroy_pn() Do not set PUFFS_KFLAG_CACHE_FS_TTL for PUFFS tests
1.29.4.3.2.1	01-Nov-2012	matt	sync with netbsd-6-0-RELEASE.
1.29.2.2	30-Oct-2012	yamt	sync with head
1.29.2.1	23-May-2012	yamt	sync with head.
1.36.10.2	05-Nov-2014	snj	Pull up following revision(s) (requested by manu in ticket #181): lib/libperfuse/fuse.h: revision 1.6 lib/libperfuse/ops.c: revision 1.78 lib/libperfuse/perfuse.c: revision 1.35 lib/libperfuse/perfuse_priv.h: revision 1.36 lib/libpuffs/dispatcher.c: revision 1.48 lib/libpuffs/opdump.c: revision 1.37 lib/libpuffs/puffs.c: revision 1.118 lib/libpuffs/puffs.h: revision 1.126 lib/libpuffs/puffs_ops.3: revisions 1.40-1.41 sys/fs/puffs/puffs_msgif.h: revision 1.82-1.83 sys/fs/puffs/puffs_msgif.h: revision 1.82 sys/fs/puffs/puffs_vnops.c: revision 1.196 Add PUFFS support for fallocate and fdiscard operations -- libpuffs support for fallocate and fdiscard operations -- Add PUFFS_HAVE_FALLOCATE in puffs_msgif.h so that filesystem can decide at build time wether fallocate is usable -- FUSE fallocate support There seems to be no fdiscard FUSE operation at the moment, hence that one is left unused.
1.36.10.1	24-Aug-2014	martin	Pull up following revision(s) (requested by manu in ticket #52): lib/libperfuse/perfuse_priv.h: revision 1.33 lib/libperfuse/ops.c: revision 1.67 lib/libpuffs/dispatcher.c: revision 1.47 lib/libperfuse/perfuse_priv.h: revision 1.34 lib/libperfuse/ops.c: revision 1.68 lib/libperfuse/ops.c: revision 1.69 lib/libpuffs/puffs_ops.3: revision 1.37 lib/libpuffs/puffs_ops.3: revision 1.38 lib/libperfuse/perfuse.c: revision 1.32 lib/libperfuse/perfuse.c: revision 1.33 usr.sbin/perfused/msg.c: revision 1.22 lib/libpuffs/puffs.h: revision 1.125 Implement FUSE direct I/O. Remove useless code and warnings
1.4	20-Feb-2009	pooka	Remove the paragraph saying interfaces will change without warning. Note that a stable version of puffs first appeared in NetBSD 5.0.
1.3	27-Dec-2007	pooka	branches: 1.3.10; 1.3.12; Flesh this out a bit more. Unfortunately it's still unfinished, and I don't know when I'll get a chance to finish it.
1.2	01-Dec-2007	wiz	Fix a typo that has migrated across files.
1.1	19-Jan-2007	pooka	branches: 1.1.4; Bring the documentation slightly more into sync with the current reality. It's still really a far cry from proper documentation, though.
1.1.4.1	09-Jan-2008	matt	sync with HEAD
1.3.12.1	13-May-2009	jym	Sync with HEAD. Third (and last) commit. See http://mail-index.netbsd.org/source-changes/2009/05/13/msg221222.html
1.3.10.1	24-Feb-2009	snj	Pull up following revision(s) (requested by pooka in ticket #484): lib/libpuffs/puffs.3: revision 1.46 lib/libpuffs/puffs_cc.3: revision 1.13 lib/libpuffs/puffs_cred.3: revision 1.4 lib/libpuffs/puffs_flush.3: revision 1.8 lib/libpuffs/puffs_framebuf.3: revision 1.26 lib/libpuffs/puffs_node.3: revision 1.4 lib/libpuffs/puffs_ops.3: revision 1.23 lib/libpuffs/puffs_path.3: revision 1.4 lib/libpuffs/puffs_suspend.3: revision 1.3 Remove the paragraph saying interfaces will change without warning. Note that a stable version of puffs first appeared in NetBSD 5.0.
1.46	22-Jan-2022	pho	Allow calling puffs_mount(3) before puffs_daemon(3) puffs_daemon(3) creates a pipe before forking, and the parent process waits for the child to either complete puffs_mount(3) or fail. If a user calls puffs_daemon(3) after puffs_mount(3), the function deadlocks. While this error-reporting functionality is really a nice thing to have, deadlocking is not great. If the filesystem has already been mounted, puffs_mount(3) should just daemonize the process and return. This became an issue because fuse_daemonize(3) in FUSE API had no such requirement and some FUSE filesystems in the wild suffered deadlocks.
1.45	18-Apr-2012	manu	- When using PUFFS_KFLAG_CACHE_FS_TTL, do not use puffs_node to carry attribute and TTL fora newly created node. Instead extend puffs_newinfo and add puffs_newinfo_setva() and puffs_newinfo_setttl() - Remove node_mk_common_final in libperfuse. It used to set uid/gid for a newly created vnode but has been made redundant along time ago since uid and gid are properly set in FUSE header. - In libperfuse, check for corner case where opc = 0 on INACTIVE and RECLAIM (how is it possible? Check for it to avoid a crash anyway) - In libperfuse, make sure we unlimit RLIMIT_AS and RLIMIT_DATA so that we do notrun out of memory because the kernel is lazy at reclaiming vnodes. - In libperfuse, cleanup style of perfuse_destroy_pn()
1.44	20-Jun-2011	mrg	branches: 1.44.2; 1.44.4; various build fixes for gcc 4.5. from chuq. XXX i'm not sure all of these work properly wtf pointer aliasing, but there are no casts at least... the lib/libpuffs/puffs_priv.h is definately a real bug fix. from chuq.
1.43	06-Dec-2010	pooka	branches: 1.43.4; Remove PUFFS_WITH_THREADS ifdef garbage, just use libc weak symbols.
1.42	12-Jan-2010	pooka	Add convenience routine puffs_unmountonsignal(), which does exactly that.
1.41	11-Aug-2008	pooka	clear cached puffs_cc's in puffs_exit()
1.40	11-Aug-2008	pooka	* make sure we come out of mainloop in the main context. otherwise in certain cases the server would jump to hyperspace when unmounted * sprinkle some debug prints
1.39	28-Jan-2008	pooka	branches: 1.39.6; Implement lazy contexts for file systems using puffs_mainloop(). Previously each request was executed on its own callcontext and switched to every time the request was being processed. Now requests borrow the mainloop context and switch only if/when they yield. This takes two context switches away from a file system request bringing down the typical amounts 2->0 (e.g. dtfs) and 4->2 (e.g. psshfs). The interfaces for manually executing requests changed a bit: puffs_dispatch_create() and puffs_dispatch_exec() must now be used. They are not tested, as nothing in-tree wants them and I doubt anyone else is really interested in them either. Also do some misc code cleanup related to execution contexts. The "work-in-progress checkpoint" committed over a year ago was starting to look slightly weed-infested.
1.38	16-Jan-2008	pooka	Cache a number of execution contexts instead of re-initializing them every time. Speeds up pure in-memory file systems such as sysctlfs or dtfs quite a bit. For actual I/O-workhorses the result is of course less tasty.
1.37	25-Dec-2007	pooka	Add a public interface for setting a cookie mapper.
1.36	16-Dec-2007	pooka	* nuke puffs_cc_get{specific,usermount} for good * move prototypes for puffs_docc and puffs_dopufbuf into the public header, as they are should be exposed
1.35	08-Dec-2007	pooka	Now that "l" is gone both as an argument to operations and from componentname, remove all vestiges of puffs_cid.
1.34	05-Dec-2007	pooka	Send a response message for flush operations from the kernel instead of abusing the return value of write(2).
1.33	04-Dec-2007	pooka	Part 3/n of extensive changes to message handling: Get rid of the original puffs_req(3) framework and use puffs_framebuf(3) instead for file system requests. It has the advantage of being suitable for transporting a distributed message passing protocol and therefore us being able to run the file system server on any host. Ok, puffs is not quite here yet: libpuffs needs to grow request routing support and the message contents need to be munged into a host independent format. Saying which format would be telling, but it might begin with an X, end in an L and have the 13th character in the middle. Keep an eye out for the sequels: Parts 3+m/n.
1.32	16-Nov-2007	pooka	Introduce puffs_daemon(), which works like daemon(3) except that it also synchronizes with puffs_mount() and does not return (exit) in the parent process until the file system has been mounted. This makes it possible to reliably run e.g. mount_foo jippi /kai ; cd /kai/ee
1.31	31-Oct-2007	pooka	Fix a few bugs and memleaks with the threading code.
1.30	29-Oct-2007	pooka	Sprinkle some hackish locks here and there, just enough to allow libp2k & rump to run multithreaded.
1.29	28-Oct-2007	pooka	Add possibility to set pre- and post callbacks which will be called for all operations before or after executing the specific callback (the api is not be final yet, though).
1.28	26-Oct-2007	pooka	Add some very initial support for a threading worker model as an alternative to the (vastly superior ;) continuation model. This is very preliminary stuff and not compiled by default (which it even won't do without some other patches I cannot commit yet). The raison d'commit of the patch is a snippet which ensures proper in-order dispatching of all operations, including those which don't require a response. Previously many of them would be dispatched simultaneosly, e.g. fsync and reclaim on the same node, which obviously isn't all that nice for correct operation.
1.27	26-Oct-2007	pooka	mmap stacks, makes it possible to align them
1.26	25-Oct-2007	pooka	Make pu_haskq a flag instead of its own variable.
1.25	21-Oct-2007	pooka	Add puffs_cc_schedule() which marks a pcc runnable and will pass execution to it when in the main loop the next time.
1.24	21-Oct-2007	pooka	Always provide caller information from the kernel based on curlwp. (but don't deprecate the old puffs_cid interface just yet)
1.23	11-Oct-2007	pooka	Part 1/n of some pretty extensive changes to how the kernel module interacts with the userspace file server: * since the kernel-user communication is not purely request-response anymore (hasn't been since 2006), try to rename some "request" to "message". more similar mangling will take place in the future. * completely rework how messages are allocated. previously most of them were borrowed from the stack (originally all of them), but now always allocate dynamically. this makes the structure of the code much cleaner. also makes it possible to fix a locking order violation. it enables plenty of future enhancements. * start generalizing the transport interface to be independent of puffs * move transport interface to read/write instead of ioctl. the old one had legacy design problems, and besides, ioctl's suck. implement a very generic version for now; this will be worked on later hopefully some day reaching "highly optimized". * implement libpuffs support behind existing library request interfaces. this will change eventually (I hate those interfaces)
1.22	27-Sep-2007	pooka	Add error notifications, which are used to deliver errors from the kernel to the file server for silly things the file server did, e.g. attempting to create a file with size VSIZENOTSET. The file server can handle these as it chooses, but the default action is for it to throw its hands in the air and sing "goodbye, cruel world, it's over, walk on by".
1.21	01-Sep-2007	pooka	Add gotframe callback, which is called if an incoming frame matches no outstanding request.
1.20	18-Aug-2007	pooka	branches: 1.20.2; puffs_cc_create/destroy aren't supposed to be exported to the user
1.19	20-Jul-2007	pooka	branches: 1.19.4; Add puffs_framev_enqueue_waitevent, which waits for read/write/error events from an i/o descriptor. It yields until an event happens and continues execution when the event loop notices the event has happened.
1.18	19-Jul-2007	pooka	Don't float puffs_kargs needlessly around after mount, saves a couple k of memory (and there was much rejoycing!)
1.17	07-Jul-2007	pooka	Make puffs_framebuf better suited from environments with multiple sources: * in addition add/remove, allow enable/disable, which can be used to control events for descriptors without having to remove all the data associated with them * add directsend/receive, which can be used to pass the same buffer from the caller to read/writeframe and back again * add flags to enqueue functions and allow urgent buffers to be processed as the next PDU
1.16	01-Jul-2007	pooka	Instead of requesting various pieces of information from node-creating methods using "please fill me"-pointers, pass "struct puffs_newinfo *" and provide methods to fill in new node info.
1.15	01-Jul-2007	pooka	cast to void * instead of actual (albeit incomplete) type to silence gcc about type punning with -fstrict-alias
1.14	01-Jul-2007	pooka	Instead of supplying a plain pid, supply an abstract struct puffs_cid *, which can currently be used to query the pid and lwpid.
1.13	01-Jul-2007	pooka	make puffs_cred an opaque type
1.12	06-Jun-2007	pooka	Move puffs to a two clause license where it already isn't so. And as agc pointed out, even files with the third clause were already effectively two clause because of a slight bug in the language...
1.11	20-May-2007	pooka	Change the signature of puffs_respcmp to return -1,0,1 for smaller, equal, larger, respectively instead of 0/1 for non/equal. This will allow sorting the buffers for faster matching in libpuffs. While here, change the name from respcmp to framecmp, as that better reflects the purpose. NOTE! there is no obvious way to make compilation fail for file systems which may already be using this feature (although I don't think there are any outside our tree, as the feature is two weeks old). Nevertheless, non-updated file systems will fail very quickly.
1.10	15-May-2007	pooka	Step 3 and currently final step of a more general purpose event loop: support removal and addition of i/o file descriptors on the fly. * detect closed file descriptors * automatically free waiters of a dead file descriptor * give the file server the possibility to specify a callback which notifies of a dead file descriptor * move loop function to be a property of the mainloop instead of framebuf (doesn't change effective behaviour) * add the possibility to configure a timespec parameter which attempts to call the loop function periodically * move the event loop functions from the puffs_framebuf namespace to puffs_framev to differential between pure memory management functions
1.9	11-May-2007	pooka	Step 2 of general-purpose eventloop: get rid of puffs_framebuf_eventloop(). puffs_mainloop() now handles that functionality as well.
1.8	10-May-2007	pooka	Add some kludges to be able to execute requests on the regular program stack instead of the continuation stack. This is for lib/36011, where pthread gets confused since we aren't running on the regular stack. I'm not really sure which direction to go to with this quite yet, so make the hack hard to enable on purpose. The whole request dispatch code needs cleaning anyway.
1.7	05-May-2007	pooka	Add puffs "frame buffers", which are an abstraction of the buffering and event handling mechanisms required in file servers with blocking I/O backends. puffs_framebuf is built on the concept of puffs_cc and uses those to multiplex execution where needed. File systems are required to implement three methods: * read frame * write frame * compare if frame is a response to the given one Memory management is provided by puffs_framebuf, but the file systems must still, of course, interpret the protocol and do e.g. byte order conversion. As always, puffs_framebuf is work in progress. Current users are mount_psshfs and mount_9p.
1.6	13-Apr-2007	pooka	Instead of keeping on adding parameters to puffs_mount(), make it only take the bare essentials, which currently means removing "maxreqlen" from the argument list (all current callers I'm aware of set it as 0 anyway). Introduce puffs_init(), which provides a context for setting various parameters and puffs_domount(), which can be used to mount the file system. Keep puffs_mount() as a shortcut for the above two for simple file systems. Bump development ABI version to 13. After all, it's Friday the 13th. Watch out! Bad things can happen on Friday the 13th. --No carrier--
1.5	12-Apr-2007	pooka	Hide struct puffs_usermount from the rest of the world and provide accessors for interesting data in it. Namely, you can now get pu->pu_privdata with puffs_getspecific(), pu->pu_pn_root with puffs_set/getroot() and pu->pu_maxreqlen with puffs_getmaxreqlen().
1.4	15-Jan-2007	pooka	Checkpoint some more work in progress: for the benefit those file system backends which operate purely based on paths, push out more path management into the library and make path management more abstract: enable a file system to define a bunch of path management callbacks, which are used by the framework. Management of normal /this/is/a/path type paths is provided by the library.
1.3	06-Jan-2007	pooka	* get rid of the mount callback; it's no great surprise to the server that it needs to mount the file system backend if it wants to call mount * provide some options for getmntopts(), assume that callers will parse command line (or fstab) args * reorganize the puffs_cc interface just a bit, preparing for a bigger revamp later
1.2	29-Dec-2006	pooka	checkpoint some experimental work-in-progress, namely: Add support for having multiple outstanding operations. This is done by exposing enough interfaces so that it is convenient to have the main event loop in the implementation itself and by providing a continuation framework for convinient blocking and rescheduling. works fine, but will undergo further cleanup & development
1.1	07-Dec-2006	pooka	outline & export code for interfacing with kernel GET/PUT requests
1.19.4.2	20-Jul-2007	pooka	Add puffs_framev_enqueue_waitevent, which waits for read/write/error events from an i/o descriptor. It yields until an event happens and continues execution when the event loop notices the event has happened.
1.19.4.1	20-Jul-2007	pooka	file puffs_priv.h was added on branch matt-mips64 on 2007-07-20 13:14:57 +0000
1.20.2.3	23-Mar-2008	matt	sync with HEAD
1.20.2.2	09-Jan-2008	matt	sync with HEAD
1.20.2.1	06-Nov-2007	matt	sync with HEAD
1.39.6.1	18-Sep-2008	wrstuden	Sync with wrstuden-revivesa-base-2.
1.43.4.1	23-Jun-2011	cherry	Catchup with rmind-uvmplock merge.
1.44.4.1	23-Apr-2012	riz	Pull up following revision(s) (requested by manu in ticket #195): lib/libskey/skeysubr.c: revision 1.27 lib/libkvm/kvm_getloadavg.c: revision 1.11 lib/libwrap/update.c: revision 1.9 lib/liby/yyerror.c: revision 1.9 lib/libpuffs/puffs_ops.3: revision 1.30 lib/libwrap/misc.c: revision 1.10 lib/libwrap/hosts_access.c: revision 1.20 lib/libpuffs/pnode.c: revision 1.11 lib/libperfuse/subr.c: revision 1.17 lib/libpuffs/pnode.c: revision 1.12 lib/libperfuse/subr.c: revision 1.18 lib/libwrap/options.c: revision 1.15 lib/libwrap/fix_options.c: revision 1.11 lib/libperfuse/ops.c: revision 1.52 lib/libperfuse/ops.c: revision 1.53 lib/libperfuse/ops.c: revision 1.54 lib/libwrap/hosts_ctl.c: revision 1.5 lib/libintl/gettext.c: revision 1.27 lib/libwrap/shell_cmd.c: revision 1.6 lib/libpuffs/dispatcher.c: revision 1.39 lib/libperfuse/perfuse_priv.h: revision 1.27 lib/libwrap/socket.c: revision 1.19 lib/libpuffs/puffs.3: revision 1.50 lib/libperfuse/perfuse_priv.h: revision 1.28 lib/libpuffs/puffs_priv.h: revision 1.45 lib/libpuffs/puffs.3: revision 1.51 lib/libperfuse/perfuse_priv.h: revision 1.29 lib/libwrap/percent_x.c: revision 1.5 lib/libpuffs/puffs.3: revision 1.52 lib/libperfuse/debug.c: revision 1.11 sys/fs/puffs/puffs_vnops.c: revision 1.165 lib/libwrap/tcpd.h: revision 1.13 sys/fs/puffs/puffs_vnops.c: revision 1.166 lib/libwrap/eval.c: revision 1.7 sys/fs/puffs/puffs_msgif.h: revision 1.78 sys/fs/puffs/puffs_vfsops.c: revision 1.101 lib/libwrap/rfc931.c: revision 1.9 lib/libwrap/clean_exit.c: revision 1.5 lib/libpuffs/puffs.h: revision 1.120 lib/libc/stdlib/jemalloc.c: revision 1.27 lib/librmt/rmtlib.c: revision 1.26 lib/libpuffs/puffs.h: revision 1.121 sys/fs/puffs/puffs_sys.h: revision 1.79 lib/librumpclient/rumpclient.c: revision 1.48 lib/libwrap/refuse.c: revision 1.5 lib/libperfuse/perfuse.c: revision 1.26 lib/libperfuse/perfuse.c: revision 1.27 tests/fs/puffs/t_fuzz.c: revision 1.5 lib/libperfuse/perfuse.c: revision 1.28 lib/libpuffs/dispatcher.c: revision 1.40 sys/fs/puffs/puffs_node.c: revision 1.24 lib/libwrap/diag.c: revision 1.9 lib/libintl/textdomain.c: revision 1.13 Use C89 function definition Add name and atttribute cache with filesytem provided TTL. lookup, create, mknod, mkdir, symlink, getattr and setattr messages have been extended so that attributes and their TTL can be provided by the filesytem. lookup, create, mknod, mkdir, and symlink messages are also extended so that the filesystem can provide name TTL. Add PUFFS_KFLAG_CACHE_FS_TTL flag to puffs_init(3) to use name and attribute cache with filesystem provided TTL. lookup, create, mknod, mkdir, symlink, getattr and setattr messages have been extended so that attributes and their TTL can be provided by the filesytem. lookup, create, mknod, mkdir, and symlink messages are also extended so that the filesystem can provide name TTL. The filesystem updates attributes and TTL using puffs_pn_getvap(3), puffs_pn_getvattl(3), and puffs_pn_getcnttl(3) Use new PUFFS_KFLAG_CACHE_FS_TTL option to puffs_init(3) so that FUSE TTL on name and attributes are used. This save many PUFFS operations and improves performances. PUFFS_KFLAG_CACHE_FS_TTL is #ifdef'ed in many places for now so that libperfuse can still be used on netbsd-5. Split file system. Comma fixes. Remove dangling "and". Bump date for previous. - Makesure update_va does not change vnode size when it should not. For instance when doing a fault-issued VOP_GETPAGES within VOP_WRITE, changing size leads to panic: genfs_getpages: past eof. -Handle ticks wrap around for vnode name andattribute timeout - When using PUFFS_KFLAG_CACHE_FS_TTL, do not use puffs_node to carry attribute and TTL fora newly created node. Instead extend puffs_newinfo and add puffs_newinfo_setva() and puffs_newinfo_setttl() - Remove node_mk_common_final in libperfuse. It used to set uid/gid for a newly created vnode but has been made redundant along time ago since uid and gid are properly set in FUSE header. - In libperfuse, check for corner case where opc = 0 on INACTIVE and RECLAIM (how is it possible? Check for it to avoid a crash anyway) - In libperfuse, make sure we unlimit RLIMIT_AS and RLIMIT_DATA so that we do notrun out of memory because the kernel is lazy at reclaiming vnodes. - In libperfuse, cleanup style of perfuse_destroy_pn() Do not set PUFFS_KFLAG_CACHE_FS_TTL for PUFFS tests
1.44.2.1	23-May-2012	yamt	sync with head.
1.6	04-Dec-2007	pooka	Part 3/n of extensive changes to message handling: Get rid of the original puffs_req(3) framework and use puffs_framebuf(3) instead for file system requests. It has the advantage of being suitable for transporting a distributed message passing protocol and therefore us being able to run the file system server on any host. Ok, puffs is not quite here yet: libpuffs needs to grow request routing support and the message contents need to be munged into a host independent format. Saying which format would be telling, but it might begin with an X, end in an L and have the 13th character in the middle. Keep an eye out for the sequels: Parts 3+m/n.
1.5	01-Dec-2007	wiz	Fix another persistent spello.
1.4	01-Dec-2007	wiz	Fix a typo that has migrated across files.
1.3	09-May-2007	pooka	branches: 1.3.4; Remove the puffs_usermount argument from puffs_req_handle(). It can be found from the other arguments and this way puffs_req_handle() is more symmetric with the other puffs_req() routines.
1.2	20-Jan-2007	wiz	New sentence, new line.
1.1	20-Jan-2007	pooka	document request handling routines
1.3.4.1	23-Mar-2008	matt	sync with HEAD
1.4	05-Dec-2009	pooka	Remove suspension support from here too.
1.3	20-Feb-2009	pooka	Remove the paragraph saying interfaces will change without warning. Note that a stable version of puffs first appeared in NetBSD 5.0.
1.2	28-Jan-2007	wiz	branches: 1.2.18; 1.2.20; Fix spellos.
1.1	26-Jan-2007	pooka	document puffs_fs_suspend()
1.2.20.1	13-May-2009	jym	Sync with HEAD. Third (and last) commit. See http://mail-index.netbsd.org/source-changes/2009/05/13/msg221222.html
1.2.18.1	24-Feb-2009	snj	Pull up following revision(s) (requested by pooka in ticket #484): lib/libpuffs/puffs.3: revision 1.46 lib/libpuffs/puffs_cc.3: revision 1.13 lib/libpuffs/puffs_cred.3: revision 1.4 lib/libpuffs/puffs_flush.3: revision 1.8 lib/libpuffs/puffs_framebuf.3: revision 1.26 lib/libpuffs/puffs_node.3: revision 1.4 lib/libpuffs/puffs_ops.3: revision 1.23 lib/libpuffs/puffs_path.3: revision 1.4 lib/libpuffs/puffs_suspend.3: revision 1.3 Remove the paragraph saying interfaces will change without warning. Note that a stable version of puffs first appeared in NetBSD 5.0.
1.15	11-Jul-2010	pooka	Export the opid -> opstr tables (for outside debugging use).
1.14	07-Jan-2010	pooka	Dump contents of struct vattr -- i never remember which members the kernel passes when, so this helps tapping that info.
1.13	07-Jan-2010	pooka	* make indentation a little less ridiculous * note that interfaces provided by <puffsdump.h> are not stable and are to be used only for debug builds * remove a few unimplemented/used dump interfaces
1.12	06-Apr-2009	pooka	dump readdir offset
1.11	06-Apr-2009	pooka	* dump op-specific return info before the generic one * dump readdir response values
1.10	12-Aug-2008	pooka	branches: 1.10.6; Replace void * by puffs_cookie_t where appropriate. No functional change.
1.9	27-Jan-2008	pooka	branches: 1.9.6; For operations with a target (remove, rmdir, link), print the target cookie.
1.8	25-Dec-2007	pooka	* dump r/w return resid * fix RV dump order to be more logical
1.7	09-Dec-2007	pooka	print newnode cookie value for operations which create nodes
1.6	29-Nov-2007	pooka	print mode in open
1.5	22-Aug-2007	pooka	branches: 1.5.2; Dump lookup/componentname specific info in opdump. (sigh, I guess I guess I'll never do opdump properly at this rate)
1.4	07-Aug-2007	pooka	dump further info for read/write
1.3	06-Jun-2007	pooka	Move puffs to a two clause license where it already isn't so. And as agc pointed out, even files with the third clause were already effectively two clause because of a slight bug in the language...
1.2	18-Feb-2007	pooka	dump return values for operations
1.1	22-Oct-2006	pooka	add libpuffs - the userspace programming interface for puffs This work was initially started and completed for Google SoC 2005 and tweaked to work a bit better in the past few weeks. While being far from complete, it is functional enough to be able and stable to host a fairly general-purpose in-memory file system in userspace. Even so, puffs should be considered experimental and no binary compatibility for interfaces or crash-freedom or zero security implications should be relied upon just yet. The GSoC project was mentored by William Studenmund and the final review for the code was done by Christos.
1.5.2.2	23-Mar-2008	matt	sync with HEAD
1.5.2.1	09-Jan-2008	matt	sync with HEAD
1.9.6.1	18-Sep-2008	wrstuden	Sync with wrstuden-revivesa-base-2.
1.10.6.1	13-May-2009	jym	Sync with HEAD. Third (and last) commit. See http://mail-index.netbsd.org/source-changes/2009/05/13/msg221222.html
1.24	23-Jan-2013	riastradh	Assert equality, not assignment, in puffs__fsframe_cmp.
1.23	29-Jan-2008	pooka	branches: 1.23.26; 1.23.28; 1.23.32; When reading a request from the kernel, only allocate as much memory as needed to read a request instead of the maximum message size. Insanely enough, this makes "cheap backend I/O" file systems such dtfs and sysctlfs perform 10-30% better (depending on the average size of incoming requests).
1.22	29-Jan-2008	pooka	Fix a weirdosity in the framebuf gotfb interface: hand buffer ownership to the called routine instead of keeping it. This means that callees must now destroy the buffer once they are done with it.
1.21	28-Jan-2008	pooka	Implement lazy contexts for file systems using puffs_mainloop(). Previously each request was executed on its own callcontext and switched to every time the request was being processed. Now requests borrow the mainloop context and switch only if/when they yield. This takes two context switches away from a file system request bringing down the typical amounts 2->0 (e.g. dtfs) and 4->2 (e.g. psshfs). The interfaces for manually executing requests changed a bit: puffs_dispatch_create() and puffs_dispatch_exec() must now be used. They are not tested, as nothing in-tree wants them and I doubt anyone else is really interested in them either. Also do some misc code cleanup related to execution contexts. The "work-in-progress checkpoint" committed over a year ago was starting to look slightly weed-infested.
1.20	05-Dec-2007	pooka	inverse inverse logic
1.19	05-Dec-2007	pooka	apply previous in a more readable way
1.18	05-Dec-2007	dogcow	"fix" complaints about type-punned pointers with some icky casts.
1.17	04-Dec-2007	pooka	Part 3/n of extensive changes to message handling: Get rid of the original puffs_req(3) framework and use puffs_framebuf(3) instead for file system requests. It has the advantage of being suitable for transporting a distributed message passing protocol and therefore us being able to run the file system server on any host. Ok, puffs is not quite here yet: libpuffs needs to grow request routing support and the message contents need to be munged into a host independent format. Saying which format would be telling, but it might begin with an X, end in an L and have the 13th character in the middle. Keep an eye out for the sequels: Parts 3+m/n.
1.16	12-Nov-2007	pooka	* split the putter header into a kernel version and a userland version + install latter to /usr/include/dev/putter * remove last dependencies to puffs from putter, it's completely independent now
1.15	31-Oct-2007	pooka	Fix a few bugs and memleaks with the threading code.
1.14	29-Oct-2007	pooka	Sprinkle some hackish locks here and there, just enough to allow libp2k & rump to run multithreaded.
1.13	26-Oct-2007	pooka	Add some very initial support for a threading worker model as an alternative to the (vastly superior ;) continuation model. This is very preliminary stuff and not compiled by default (which it even won't do without some other patches I cannot commit yet). The raison d'commit of the patch is a snippet which ensures proper in-order dispatching of all operations, including those which don't require a response. Previously many of them would be dispatched simultaneosly, e.g. fsync and reclaim on the same node, which obviously isn't all that nice for correct operation.
1.12	21-Oct-2007	he	To work around lint problems, add a LINTED comment on pooka's request.
1.11	19-Oct-2007	pooka	When doing a read operation, don't copy the whole kernel buffer to userspace, since it doesn't contain any information yet. I should still rework this more so this is just a quickie to get the read/write style interface more up to speed with the ioctl version.
1.10	11-Oct-2007	pooka	Part 1/n of some pretty extensive changes to how the kernel module interacts with the userspace file server: * since the kernel-user communication is not purely request-response anymore (hasn't been since 2006), try to rename some "request" to "message". more similar mangling will take place in the future. * completely rework how messages are allocated. previously most of them were borrowed from the stack (originally all of them), but now always allocate dynamically. this makes the structure of the code much cleaner. also makes it possible to fix a locking order violation. it enables plenty of future enhancements. * start generalizing the transport interface to be independent of puffs * move transport interface to read/write instead of ioctl. the old one had legacy design problems, and besides, ioctl's suck. implement a very generic version for now; this will be worked on later hopefully some day reaching "highly optimized". * implement libpuffs support behind existing library request interfaces. this will change eventually (I hate those interfaces)
1.9	19-Jul-2007	pooka	branches: 1.9.4; 1.9.6; Don't float puffs_kargs needlessly around after mount, saves a couple k of memory (and there was much rejoycing!)
1.8	06-Jun-2007	pooka	Move puffs to a two clause license where it already isn't so. And as agc pointed out, even files with the third clause were already effectively two clause because of a slight bug in the language...
1.7	11-May-2007	pooka	Step 2 of general-purpose eventloop: get rid of puffs_framebuf_eventloop(). puffs_mainloop() now handles that functionality as well.
1.6	09-May-2007	pooka	Remove the puffs_usermount argument from puffs_req_handle(). It can be found from the other arguments and this way puffs_req_handle() is more symmetric with the other puffs_req() routines.
1.5	13-Apr-2007	pooka	Instead of keeping on adding parameters to puffs_mount(), make it only take the bare essentials, which currently means removing "maxreqlen" from the argument list (all current callers I'm aware of set it as 0 anyway). Introduce puffs_init(), which provides a context for setting various parameters and puffs_domount(), which can be used to mount the file system. Keep puffs_mount() as a shortcut for the above two for simple file systems. Bump development ABI version to 13. After all, it's Friday the 13th. Watch out! Bad things can happen on Friday the 13th. --No carrier--
1.4	20-Jan-2007	pooka	Be consistent with the rest of the framework about the argument order of puffs_docc and puffs_dopreq
1.3	20-Jan-2007	pooka	be consistent in naming request handlers
1.2	29-Dec-2006	pooka	checkpoint some experimental work-in-progress, namely: Add support for having multiple outstanding operations. This is done by exposing enough interfaces so that it is convenient to have the main event loop in the implementation itself and by providing a continuation framework for convinient blocking and rescheduling. works fine, but will undergo further cleanup & development
1.1	07-Dec-2006	pooka	outline & export code for interfacing with kernel GET/PUT requests
1.9.6.2	19-Jul-2007	pooka	Don't float puffs_kargs needlessly around after mount, saves a couple k of memory (and there was much rejoycing!)
1.9.6.1	19-Jul-2007	pooka	file requests.c was added on branch matt-mips64 on 2007-07-19 12:52:29 +0000
1.9.4.3	23-Mar-2008	matt	sync with HEAD
1.9.4.2	09-Jan-2008	matt	sync with HEAD
1.9.4.1	06-Nov-2007	matt	sync with HEAD
1.23.32.1	25-Feb-2013	tls	resync with head
1.23.28.1	03-Nov-2014	msaitoh	Pull up following revision(s) (requested by manu in ticket #1140): lib/libperfuse/ops.c 1.63-1.69 lib/libperfuse/perfuse.c 1.32-1.33 lib/libperfuse/perfuse_priv.h 1.32-1.34 lib/libperfuse/subr.c 1.20 lib/libpuffs/creds.c 1.16 lib/libpuffs/dispatcher.c 1.47 lib/libpuffs/puffs.h 1.125 lib/libpuffs/puffs_ops.3 1.37-1.38 lib/libpuffs/requests.c 1.24 sys/fs/puffs/puffs_msgif.h 1.81 sys/fs/puffs/puffs_sys.h 1.85 sys/fs/puffs/puffs_vnops.c 1.183 usr.sbin/perfused/msg.c 1.22 Bring libpuffs, libperfuse and perfused on par with -current: - implement FUSE direct I/O - remove useless code and warnings - fix missing GETATTR bugs - fix exended attribute get and list operations
1.23.26.1	22-May-2014	yamt	sync with head. for a reference, the tree before this commit was tagged as yamt-pagecache-tag8. this commit was splitted into small chunks to avoid a limitation of cvs. ("Protocol error: too many arguments")
1.5	21-May-2010	pooka	* support extended attributes * bump major due to structure growth * add some spare space * remove ABI sillyness
1.4	18-Oct-2009	pooka	bump minor for puffs_setmntinfo()
1.3	17-Oct-2009	pooka	Add some comments on how to act when major is bumped the next time.
1.2	11-Jan-2009	christos	bump shared libraries.
1.1	22-Oct-2006	pooka	branches: 1.1.24; 1.1.28; add libpuffs - the userspace programming interface for puffs This work was initially started and completed for Google SoC 2005 and tweaked to work a bit better in the past few weeks. While being far from complete, it is functional enough to be able and stable to host a fairly general-purpose in-memory file system in userspace. Even so, puffs should be considered experimental and no binary compatibility for interfaces or crash-freedom or zero security implications should be relied upon just yet. The GSoC project was mentored by William Studenmund and the final review for the code was done by Christos.
1.1.28.1	21-Apr-2010	matt	sync to netbsd-5
1.1.24.1	27-Oct-2009	bouyer	Pull up following revision(s) (requested by pooka in ticket #1105): lib/libpuffs/puffs.h: revision 1.112 lib/libpuffs/puffs.c: revision 1.100 distrib/sets/lists/base/md.amd64: patch distrib/sets/lists/base/md.sparc64: patch distrib/sets/lists/base/shl.mi: patch lib/libpuffs/shlib_version: patch Make it possible to supply parameters to puffs_init() later (but before puffs_mount()).
1.28	23-Sep-2019	christos	Restore binary compatibility by using the statvfs90 structure internally.
1.27	17-Feb-2011	pooka	branches: 1.27.44; Support rest of file types in vtype -> s_ifmt
1.26	17-Feb-2011	pooka	st_blocks is in units of DEV_BSIZE, not st_blksize. from Derrik Pates
1.25	15-Jul-2010	pooka	branches: 1.25.2; Default f_namemax to MAXNAMLEN, mostly for initial statvfs struct passed to kernel in mount. XXX: there's currently no way for a file server to override the one passed in mount.
1.24	28-Dec-2008	christos	fix for dev_t 64 bits.
1.23	12-Aug-2008	pooka	Replace void * by puffs_cookie_t where appropriate. No functional change.
1.22	19-Dec-2007	pooka	branches: 1.22.6; Add generic getattr routine.
1.21	30-Nov-2007	pooka	Rototill. Ok, ok, a few more words about it: stop holding puffs_cc as a holy value and passing it around to almost every possible place (popquiz: which kernel variable does this remind you of?). Instead, pass the natural choice, puffs_usermount, and fetch puffs_cc via puffs_cc_getcc() only in routines which actually need it. This not only simplifies code, but (thanks to the introduction of puffs_cc_getcc()) enables constructs which weren't previously sanely possible, say layering as a curious example. There's still a little to do on this front, but this was the major fs interface blast.
1.20	27-Nov-2007	pooka	Remove "puffs_cid" from the puffs interface following l-removal from the kernel vfs interfaces. puffs_cc_getcaller(pcc) can be used now should the same information be desired.
1.19	21-Oct-2007	pooka	typo in comment
1.18	01-Jul-2007	pooka	branches: 1.18.4; Instead of supplying a plain pid, supply an abstract struct puffs_cid *, which can currently be used to query the pid and lwpid.
1.17	06-Jun-2007	pooka	Move puffs to a two clause license where it already isn't so. And as agc pointed out, even files with the third clause were already effectively two clause because of a slight bug in the language...
1.16	20-Mar-2007	pooka	implement counterpart of vaccess()
1.15	16-Mar-2007	pooka	va->va_mode doesn't contain the full argument to the mknod() system call, so introduce puffs_addvtype2mode() and use that in null.c to generate the proper syscall argument
1.14	15-Feb-2007	pooka	generic node reclaim, which just puts a puffs_node
1.13	11-Jan-2007	pooka	Add preliminary code for a nullfs layer, which can be used to mount a directory hierarchy to another point, just like with the kernel nullfs. This is not really a layering scheme yet, but it should evolve into one. Currently it can just be used to do 1:1 mapping.
1.12	06-Jan-2007	pooka	* get rid of the mount callback; it's no great surprise to the server that it needs to mount the file system backend if it wants to call mount * provide some options for getmntopts(), assume that callers will parse command line (or fstab) args * reorganize the puffs_cc interface just a bit, preparing for a bigger revamp later
1.11	29-Dec-2006	pooka	checkpoint some experimental work-in-progress, namely: Add support for having multiple outstanding operations. This is done by exposing enough interfaces so that it is convenient to have the main event loop in the implementation itself and by providing a continuation framework for convinient blocking and rescheduling. works fine, but will undergo further cleanup & development
1.10	07-Dec-2006	pooka	rototill the operation callbacks: unify the ops under struct puffs_ops and namespace them
1.9	21-Nov-2006	pooka	mode_t to enum vtype conversion
1.8	14-Nov-2006	pooka	namespace previous change under puffs_vfsnop to avoid confusions
1.7	14-Nov-2006	pooka	implement unmount, sync and statvfs as dummies for file system which couldn't care less about implementing them
1.6	23-Oct-2006	mrg	use casts to properly truncate some assignments. (sparc64 gcc complained.)
1.5	23-Oct-2006	christos	put back the comparison in a lint friendly way to appease pooka.
1.4	23-Oct-2006	dogcow	comment out comparison of unsigned expression to >= 0
1.3	23-Oct-2006	pooka	still check for array bounds, fix off-by-one
1.2	23-Oct-2006	christos	pass lint; fix install.
1.1	22-Oct-2006	pooka	add libpuffs - the userspace programming interface for puffs This work was initially started and completed for Google SoC 2005 and tweaked to work a bit better in the past few weeks. While being far from complete, it is functional enough to be able and stable to host a fairly general-purpose in-memory file system in userspace. Even so, puffs should be considered experimental and no binary compatibility for interfaces or crash-freedom or zero security implications should be relied upon just yet. The GSoC project was mentored by William Studenmund and the final review for the code was done by Christos.
1.18.4.2	09-Jan-2008	matt	sync with HEAD
1.18.4.1	06-Nov-2007	matt	sync with HEAD
1.22.6.1	18-Sep-2008	wrstuden	Sync with wrstuden-revivesa-base-2.
1.25.2.1	05-Mar-2011	bouyer	Sync with HEAD
1.27.44.1	13-Apr-2020	martin	Mostly merge changes from HEAD upto 20200411
1.10	05-Dec-2009	pooka	Remove suspension support from here too.
1.9	17-Nov-2007	pooka	fix request size
1.8	12-Nov-2007	pooka	remove header that was never committed
1.7	12-Nov-2007	pooka	* split the putter header into a kernel version and a userland version + install latter to /usr/include/dev/putter * remove last dependencies to puffs from putter, it's completely independent now
1.6	11-Oct-2007	pooka	Part 1/n of some pretty extensive changes to how the kernel module interacts with the userspace file server: * since the kernel-user communication is not purely request-response anymore (hasn't been since 2006), try to rename some "request" to "message". more similar mangling will take place in the future. * completely rework how messages are allocated. previously most of them were borrowed from the stack (originally all of them), but now always allocate dynamically. this makes the structure of the code much cleaner. also makes it possible to fix a locking order violation. it enables plenty of future enhancements. * start generalizing the transport interface to be independent of puffs * move transport interface to read/write instead of ioctl. the old one had legacy design problems, and besides, ioctl's suck. implement a very generic version for now; this will be worked on later hopefully some day reaching "highly optimized". * implement libpuffs support behind existing library request interfaces. this will change eventually (I hate those interfaces)
1.5	19-Jul-2007	pooka	branches: 1.5.4; 1.5.6; Don't float puffs_kargs needlessly around after mount, saves a couple k of memory (and there was much rejoycing!)
1.4	06-Jun-2007	pooka	Move puffs to a two clause license where it already isn't so. And as agc pointed out, even files with the third clause were already effectively two clause because of a slight bug in the language...
1.3	13-Apr-2007	pooka	Instead of keeping on adding parameters to puffs_mount(), make it only take the bare essentials, which currently means removing "maxreqlen" from the argument list (all current callers I'm aware of set it as 0 anyway). Introduce puffs_init(), which provides a context for setting various parameters and puffs_domount(), which can be used to mount the file system. Keep puffs_mount() as a shortcut for the above two for simple file systems. Bump development ABI version to 13. After all, it's Friday the 13th. Watch out! Bad things can happen on Friday the 13th. --No carrier--
1.2	12-Apr-2007	pooka	Hide struct puffs_usermount from the rest of the world and provide accessors for interesting data in it. Namely, you can now get pu->pu_privdata with puffs_getspecific(), pu->pu_pn_root with puffs_set/getroot() and pu->pu_maxreqlen with puffs_getmaxreqlen().
1.1	26-Jan-2007	pooka	add interfaces for suspending the file system
1.5.6.2	19-Jul-2007	pooka	Don't float puffs_kargs needlessly around after mount, saves a couple k of memory (and there was much rejoycing!)
1.5.6.1	19-Jul-2007	pooka	file suspend.c was added on branch matt-mips64 on 2007-07-19 12:52:29 +0000
1.5.4.2	09-Jan-2008	matt	sync with HEAD
1.5.4.1	06-Nov-2007	matt	sync with HEAD

OpenGrok