Searched hist:1.436 (Results 1 - 25 of 41) sorted by relevance

12

/src/sys/dev/pci/
H A Dfiles.pci1.436 Tue Mar 09 10:03:18 GMT 2021 msaitoh branches: 1.436.4;
Modify some parameters to reduce packet dropping.

- Background: ixgbe doesn't use common MCLGET() interface and use the
driver specific cluster allocation mechanism (jcl). The cluster is
pre-allocated with a fixed number and the current number per queue
is num_rx_desc * 2 (2048*2=4096). It's too small. It also has a problem
that the max length of the pcq which is used in the TX path is big
(4096). Example:

100M <----- [ixg0 ixg1] <----- 1G
2048 TX descs <--- 4096 pcqs <---- 2048 RX descs

If a machine forwards a traffic from 1G interface to 100M interface,
It would require 2048+4096+2048=8192 descriptors, but the current number
is 2048*2=4096. It's too small. Even if the both interface's link speed
is the same and only small number of packet is queued in the pcq, 4096
jcl is small because 2048(RX)+TX(2048)=4096. If jcl is exhausted, not only
forwarding from ixg1 to ixg0 is dropped, but also another forwarding path
from ixg1 to another interface(e.g. wm0) is also dropped. Sockets also
queue packets, so if a lot of sockets are used and/or a socket buffer
size is changed to bigger one, it'll also become a problem. If the jcl
is exhausted, evcnt(9) counter "ixgX qY Rx no jumbo mbuf" is incremented.
Example:
vmstat -ev | grep ixg1 | grep "no jumbo"
ixg1 q0 Rx no jumbo mbuf 0 0 misc
ixg1 q1 Rx no jumbo mbuf 0 0 misc
ixg1 q2 Rx no jumbo mbuf 141326 0 misc
ixg1 q3 Rx no jumbo mbuf 0 0 misc


- To solve this problem:
- Add new config parameter IXGBE_JCLNUM_MULTI and set the default to 3
(2048 * 3). The minimum number is 2. The total number of jcl per queue
is available with hw.ixgN.num_jcl_per_queue sysctl.
- Reduce the max length of the pcq() which is used in the TX path from
4096 to 2048.

- Reviewed by knakahara@ and ozaki-r@.

- TODO: Use MCLGET().
1.436 Tue Mar 09 10:03:18 GMT 2021 msaitoh branches: 1.436.4;
Modify some parameters to reduce packet dropping.

- Background: ixgbe doesn't use common MCLGET() interface and use the
driver specific cluster allocation mechanism (jcl). The cluster is
pre-allocated with a fixed number and the current number per queue
is num_rx_desc * 2 (2048*2=4096). It's too small. It also has a problem
that the max length of the pcq which is used in the TX path is big
(4096). Example:

100M <----- [ixg0 ixg1] <----- 1G
2048 TX descs <--- 4096 pcqs <---- 2048 RX descs

If a machine forwards a traffic from 1G interface to 100M interface,
It would require 2048+4096+2048=8192 descriptors, but the current number
is 2048*2=4096. It's too small. Even if the both interface's link speed
is the same and only small number of packet is queued in the pcq, 4096
jcl is small because 2048(RX)+TX(2048)=4096. If jcl is exhausted, not only
forwarding from ixg1 to ixg0 is dropped, but also another forwarding path
from ixg1 to another interface(e.g. wm0) is also dropped. Sockets also
queue packets, so if a lot of sockets are used and/or a socket buffer
size is changed to bigger one, it'll also become a problem. If the jcl
is exhausted, evcnt(9) counter "ixgX qY Rx no jumbo mbuf" is incremented.
Example:
vmstat -ev | grep ixg1 | grep "no jumbo"
ixg1 q0 Rx no jumbo mbuf 0 0 misc
ixg1 q1 Rx no jumbo mbuf 0 0 misc
ixg1 q2 Rx no jumbo mbuf 141326 0 misc
ixg1 q3 Rx no jumbo mbuf 0 0 misc


- To solve this problem:
- Add new config parameter IXGBE_JCLNUM_MULTI and set the default to 3
(2048 * 3). The minimum number is 2. The total number of jcl per queue
is available with hw.ixgN.num_jcl_per_queue sysctl.
- Reduce the max length of the pcq() which is used in the TX path from
4096 to 2048.

- Reviewed by knakahara@ and ozaki-r@.

- TODO: Use MCLGET().
H A Dpcidevs.h1.436 Mon May 27 08:52:36 GMT 2002 drochner regen
H A Dpcidevs_data.h1.436 Tue May 28 17:46:24 GMT 2002 christos regen
/src/sys/kern/
H A Dinit_main.c1.436 Wed Sep 28 15:52:47 GMT 2011 jruoho branches: 1.436.2;
Initialize cpufreq(9) normally from main().
1.436 Wed Sep 28 15:52:47 GMT 2011 jruoho branches: 1.436.2;
Initialize cpufreq(9) normally from main().
H A Dkern_exec.c1.436 Sat Aug 06 15:13:13 GMT 2016 maxv The way the kernel tries to prevent a userland process from allocating page
zero is hugely flawed. It is easy to demonstrate that one can trick UVM
into chosing a NULL hint after the user_va0_disable check from uvm_map.
Such a bypass allows kernel NULL pointer dereferences to be exploitable on
architectures with a shared userland<->kernel VA, like amd64.

Fix this by increasing the limit of the vm space made available for
userland processes. This way, UVM will never chose a NULL hint, since it
would be outside of the vm space.

The user_va0_disable sysctl still controls this feature.
H A Dvfs_subr.c1.436 Wed Feb 13 14:03:48 GMT 2013 hannken Make the spec_node table implementation private to spec_vnops.c.

To retrieve a spec_node, two new lookup functions (by device or by mount)
are implemented. Both return a referenced vnode, for an opened block device
the opened vnode is returned so further diagnostic checks "vp == ... sd_bdevvp"
will not fire. Otherwise any vnode matching the criteria gets returned.

No objections on tech-kern.

Welcome to 6.99.17
/src/sys/arch/sparc64/sparc64/
H A Dlocore.s1.436 Sat Sep 06 02:53:23 GMT 2025 riastradh paravirt_membar_sync(9): New memory barrier.

For use in paravirtualized drivers which require store-before-load
ordering -- irrespective of whether the kernel is built for a single
processor, or whether the (virtual) machine is booted with a single
processor.

This is even required on architectures that don't even have a
store-before-load ordering barrier, like m68k; adding, e.g., a virtio
bus is _as if_ the architecture has been extended with relaxed memory
ordering when talking with that new bus. Such architectures need
some way to request the hypervisor enforce that ordering -- on m68k,
that's done by issuing a CASL instruction, which qemu maps to an
atomic r/m/w with sequential consistency ordering in the host.

PR kern/59618: occasional virtio block device lock ups/hangs
/src/sys/dev/ata/
H A Dwd.c1.436 Tue Nov 07 04:09:08 GMT 2017 mlelstv Make wddone poll all drives of a channel again.
/src/sys/arch/i386/conf/
H A DALL1.436 Thu Mar 29 18:54:48 GMT 2018 maxv Remove TCPREASS_DEBUG. It was introduced 20 years ago when the reassembler
was being developed, but it's irrelevant today. Makes the code clearer.
/src/sys/arch/arm/arm32/
H A Dpmap.c1.436 Sat Apr 09 23:38:31 GMT 2022 riastradh sys: Use membar_release/acquire around reference drop.

This just goes through my recent reference count membar audit and
changes membar_exit to membar_release and membar_enter to
membar_acquire -- this should make everything cheaper on most CPUs
without hurting correctness, because membar_acquire is generally
cheaper than membar_enter.

/src/distrib/notes/common/
H A Dmain1.436 Sun Feb 22 14:32:54 GMT 2009 apb Under "The following issues need to be resolved manually", say that
unprivileged use of mount(8) now needs the nosuid and nodev flags.
/src/distrib/sets/lists/debug/
H A Dmi1.436 Mon Jul 01 04:03:04 GMT 2024 mrg GCC sanitizers: only build the sanitizers if $MKGCCCMDS != "no".

there's a minor use-case for the configuration of "building sanitizers
for distribution, but not the compiler", but it's pretty uncommon, and
the latest sanitizers don't build at all on m68000 so enhance the
option to not build /usr/bin/gcc (which simply can't modern GCC on a
sun2) to not build the sanitizers either.

(if we were to split these, MKGCCSANITIZER would be good variable.)

update set lists for far too many files, including fixing MKDEBUG for
a couple of builds.
/src/etc/
H A DMakefile1.436 Sat Sep 29 06:10:01 GMT 2018 martin Clarify comment, pointed out by maya: we do not expect bootloaders
to understand xz compression, so the kernel.gz stay as they are, only
the sets are moving to .tar.xz.
/src/share/man/man9/
H A DMakefile1.436 Sat Apr 06 03:06:24 GMT 2019 thorpej Overhaul the API used to fetch and store individual memory cells in
userspace. The old fetch(9) and store(9) APIs (fubyte(), fuword(),
subyte(), suword(), etc.) are retired and replaced with new ufetch(9)
and ustore(9) APIs that can return proper error codes, etc. and are
implemented consistently across all platforms. The interrupt-safe
variants are no longer supported (and several of the existing attempts
at fuswintr(), etc. were buggy and not actually interrupt-safe).

Also augmement the ucas(9) API, making it consistently available on
all plaforms, supporting uniprocessor and multiprocessor systems, even
those that do not have CAS or LL/SC primitives.

Welcome to NetBSD 8.99.37.
/src/share/mk/
H A Dbsd.README1.436 Sat May 28 14:31:12 GMT 2022 nia mk: Add MKFINGER, MKTALK flags for not building
talk, talkd, finger, fingerd.

/src/share/man/man4/
H A Doptions.41.436 Sun Mar 23 00:48:08 GMT 2014 dholland Revise documentation of QUOTA and QUOTA2 options to make it clear that
these are FFS options. (QUOTA2 also works with mfs. AFAIK, traditional
quotas don't.)

Don't refer to QUOTA2 as "the new quotas format" as "new" is not a
good long-term identifier.

Also, don't xref half the quota utilities; there's no need to.
H A DMakefile1.436 Sat Aug 18 11:26:36 GMT 2007 tnn Add spdmem(4) driver. This decodes technical specifications stored in
the eeprom on common types of memory modules. The specifications are
displayed during boot and can later be queried in the hw.spdmemN sysctl
subtree. Stub driver written by Nicolas Joly and greatly improved upon by
Paul Goyette. From PR 36745, with additional improvements by Paul and me.
/src/sys/dev/usb/
H A Dusbdevs1.436 Fri Sep 22 00:43:52 GMT 2006 christos From: Jaime A Fournier: Patch to support SMC 2262WV1
H A Dusbdevs.h1.436 Fri Jun 30 13:56:25 GMT 2006 chap Ankh-Morpork, we have a MIDI driver....

Merge from chap-midi branch, after
~month for review
Comments by thorpej@ drochner@ and Alexandre Ratchov

Incorporated: points by thorpej@ drochner@; preliminary support for
a stats-collecting ioctl suggested by martin@ from comments by A.R.

PR kern/32441 kern/32442 kern/32567 kern/32588 kern/32694 kern/33590
kern/33614 and one instance of kern/32651

ok martin@
H A Dusbdevs_data.h1.436 Thu Jun 15 20:33:33 GMT 2006 reinoud Regen due to addition of MosChip Semiconductors MCS7830 Ethernet adapter
/src/usr.bin/xlint/lint1/
H A Dcgram.y1.436 Mon May 22 17:47:27 GMT 2023 rillig lint: clean up the name of a local variable

No binary change.
H A Dtree.c1.436 Tue Apr 19 23:16:14 GMT 2022 rillig lint: fix integer overflow on seeing -(uint64_t)INT64_MIN

Seen in external/cddl/osnet/dist/lib/libuutil/common/uu_strtoint.c,
which is protected by a NOLINT in ../Makefile.inc.
/src/sys/net/
H A Dif.c1.436 Fri Sep 07 13:24:14 GMT 2018 christos Flip the order of free'ing things to avoid crash (from ozaki-r). Tested
with a month's uptime. Used to crash once a week.
/src/sys/arch/amd64/conf/
H A DGENERIC1.436 Wed Jul 27 07:04:57 GMT 2016 msaitoh Add sdtemp(4)
/src/usr.bin/make/
H A Djob.c1.436 Sun Oct 24 18:45:46 GMT 2021 sjg Do not ignore write failures.

We should not ignore failure to write to cmdFILE,
meta files and filemon.

Reviewed by: rillig

Completed in 895 milliseconds

12