Home | History | Annotate | Line # | Download | only in doc
TODO.compat-module revision 1.1.2.2
      1 /* $NetBSD: TODO.compat-module,v 1.1.2.2 2018/09/10 08:49:25 pgoyette Exp $ */
      2 
      3 DONE
      4 ----
      5 1.  Returned the build to use a .a compat library rather than a .o
      6     library.  The original method used was .a but that was changed
      7     (fairly recently) as a work-around to address some support
      8     routines that were not being included when needed.  These support
      9     modules are now included in their own module, and the work-around
     10     is therefore no longer needed.
     11 
     12 2.  Reverted some intentional auto-load breakage for loading the sysv_ipc
     13     module; the breakage was introduced as the fix for the above-mentioned
     14     build breakage.
     15 
     16 3.  Split the sysv_ipc compat routines into their own compat_sysv module.
     17 
     18 4.  Resolved some inter-module dependencies.
     19 
     20 5.  Extracted some net/if.c compat routines into the compat module, and
     21     replaced the originals with indirect (vectored) function calls.
     22 
     23 6.  Reconfirmed existing compat-module dependencies, and update the
     24     defopt/defflag lines in the config files* as needed, to insure that
     25     built-in dependencies get resolved.
     26 
     27 7.  Fixed limits on the number of module depedencies and maximum
     28     recursion level have been removed.  Previous code for reporting
     29     module status to userland has been versioned and moved to the
     30     compat_80 module.
     31 
     32 8.  The old monolithic compat module has been broken into multiple
     33     modules, one for each old NetBSD version.  The monolithic module
     34     is no longer available.
     35 
     36     Similarly, the compat_sysv module has also been split into several
     37     version-specific modules, and the mini-monolithic compat_sysv module
     38     is no longer provided.
     39 
     40 9.  syscalls.master has been updated to autoload the version-specific
     41     compat modules rather than the monolithic modules.
     42 
     43 10. Separated COMPAT_BSDPTY stuff, allowing the COMPAT_60 module to be
     44     built regardless.
     45 
     46 
     47 TODO - Required for branch merge
     48 --------------------------------
     49 1.  Replace version-specific parts of the compat_netbsd32 module with
     50     individual modules.  Update dependencies accordingly.
     51 
     52 
     53 TODO - Not required for branch merge
     54 ------------------------------------
     55 1.  Audit the entire code base for any remaining embedded #ifdef's for
     56     COMPAT_xx.  When found, move the actual compat code into the compat
     57     hierarchy and replace originals with indirect (vectored) calls.
     58 
     59 2.  The rtsock compat code is a disaster, with rtsock_50.c #include-ing
     60     the main rtsock.c code with various manipulations of the COMPAT_50
     61     macro.  Once rtsock is separated, compat_14 references to rtsock_50
     62     routines needs to be verified.
     63 
     64     Currently, this entire code is built for the monolithic COMPAT
     65     module, but there's no way to reach the entry points, so none of
     66     the compat code can be executed, neither on the branch nor on
     67     HEAD.
     68 
     69 3.  The compat_60 module still needs some work for XEN systems.  We
     70     probably need some build infrastructure changes to ensure that
     71     XEN (and, for i386, XEN-PAE) modules are build with the correct
     72     macros defined and with -I directories specified in the same order
     73     as for building kernels. See PR port-xen/53130.  This currently
     74     prevents loading of micro-code updates for amd64 processors running
     75     XEN kernels.  This limitation also exists on HEAD.
     76 
     77 4.  There seems to be quite a bit of MD compat_xx code, in the various
     78     sys/arch/ directories.  I haven't yet looked at any of this.  But it
     79     seems to me that the MI compat build infrastructure should have some
     80     mechanism to "reach over" to the MD code, #include a Makefile.inc file,
     81     and perhaps define something to enable the MI modcmd code to call a
     82     compat_xx_MD_init() routine.
     83 
     84     Note also that there are a few bits of MD code that is COMPAT_44
     85     related.  (The only bit of MI COMPAT_44 code is in the single module
     86     shared by COMPAT_43 and COMPAT_09.)  This affects the cesfic, hp300,
     87     news68k, and x68k platforms, all in their respective machdep.c
     88     source file.  Additionally, the zaurus platform defines COMPAT_44 in
     89     its INSTALL kernel configuration - but no other configuration files!
     90 
     91     As far as I can tell, none of the MD compat code is currently built
     92     into the monolithic COMPAT module on HEAD.  Thus, its absence from
     93     any of the version-specific modules is not a regression.
     94 
     95 5.  For compat_50, in addition to rtsock there are some things in dev/vnd,
     96     dev/gpio, and dev/wscons/wsmux that I haven't been able to cleanly
     97     separate.  These items are not currently included in the monolithic
     98     COMPAT module on HEAD, so lack of integration on the branch is not a
     99     regression.
    100 
    101 6.  Even though the build mechanism has been switched back to using a
    102     .a compat library, it might be useful to make it work with the .o
    103     library.
    104 
    105