[odc] Daily src changes for 2012-04-05

[ODC]

OpenBSD src changes summary for 2012-04-05
==========================================

sbin/iked                               share/man
sys/arch/sgi/hpc                        sys/arch/sgi/sgi
sys/dev/pci                             sys/msdosfs
usr.sbin/ftp-proxy                      usr.sbin/nginx

== sbin ============================================================== 01/04 ==

  http://www.openbsd.org/cgi-bin/cvsweb/src/sbin

iked

  ~ control.c                             ~ iked.h

  > rate-limit accepting of new connections while we are experiencing
  > fd exhaustion.
  > ok mikeb (deraadt@)

== share ============================================================= 02/04 ==

  http://www.openbsd.org/cgi-bin/cvsweb/src/share

man

  ~ man5/port-modules.5                   

  > Mention clang module. (sthen@)

== sys =============================================================== 03/04 ==

  http://www.openbsd.org/cgi-bin/cvsweb/src/sys

arch/sgi/hpc

  ~ hpc.c                                 ~ hpcvar.h
  ~ if_sq.c                               

  > DMA descriptors only need to be aligned on 8-byte boundaries, and the
  > struct
  > layout is enough to enforce this. Don't request DMA page boundary alignment
  > when allocating them. (miod@)

  ~ hpc.c                                 

  > Add an explicit bus_space_barrier() function for revision 3 hpc, which has
  > a
  > PIO write buffer. (miod@)

  ~ zs.c                                  

  > Lower ZS_DELAY() back to what it was, but issue a bus_space_barrier() after
  > every register write. Hinted by IRIX' <sys/z8530.h>.
  > While there, flip the CTS, DCD, RTS and DTR bits in registers #0 and #5.
  > Aforementioned header says they are inverted due to a hardware bug.
  > Tested on IP20, IP22 and IP24. (miod@)

  ~ dsclock.c                             

  > In the neverending tradition of never making things simple, SGI used 1940
  > as
  > the timebase on Indigo 2, but 1970 on Indy (verified with the `date'
  > command
  > at the PROM prompt and checking what values ended up in the DS1286).
  > Indy will no longer be 30 years in the future from an IRIX point of view.
  > (miod@)

arch/sgi/sgi

  ~ bus_dma.c                             

  > Bail out of bus_dmamap_sync() earlier in the non-PREWRITE cases on
  > TGT_COHERENT
  > kernels. No need to walk the whole dma map doing nothing.
  > No functional change. (miod@)

dev/pci

  ~ pcidevs                               

  > add some devices found on a r720
  > ok dlg@ (jsg@)

  ~ pcidevs.h                             ~ pcidevs_data.h

  > regen (jsg@)

msdosfs

  ~ denode.h                              ~ msdosfs_fat.c

  > As it was made evident by Olaf 'Rhialto' Seibert in the NetBSD PR 34583,
  > caching the last cluster in the file before extending it considerably
  > speeds up further lookups.	As a consequence this keeps write speed from
  > gradually decreasing and provides a measurable (up to 2x) performance
  > increase.  Prompted by the diff from Alexander Polakov, ok miod, krw.
  > (mikeb@)

== usr.sbin ========================================================== 04/04 ==

  http://www.openbsd.org/cgi-bin/cvsweb/src/usr.sbin

ftp-proxy

  ~ ftp-proxy.c                           

  > Rate-limit accepting of new connections while we are experiencing
  > fd exhaustion.
  > ok deraadt mikeb (camield@)

nginx

  ~ man/nginx.8                           

  > Mention the -u flag to tell nginx not to chroot.
  > Help with wording from jmc@, thanks! (sthen@)

===============================================================================