[odc] Daily src changes for 2005-11-23

ODC auto at squish.net
Thu Nov 24 07:00:35 GMT 2005


OpenBSD src changes summary for 2005-11-23
==========================================

lib/libm                                regress/usr.bin
share/man                               share/mk
sys/arch/alpha/alpha                    sys/arch/arm/xscale
sys/arch/hp300/stand/common             sys/arch/hp300/stand/include
sys/arch/i386/i386                      sys/arch/i386/include
sys/arch/i386/isa                       sys/arch/i386/pci
sys/arch/mips64/include                 sys/arch/sparc/dev
sys/dev/eisa                            sys/dev/ic
sys/dev/isa                             sys/dev/pci
sys/dev/pcmcia                          sys/dev/sbus
sys/dev/usb                             usr.bin/awk
usr.bin/rcs                             usr.bin/xlint
usr.sbin/apm                            usr.sbin/apmd
usr.sbin/hostapd                        usr.sbin/tcpdump

== lib =============================================================== 01/06 ==

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

libm

  ~ arch/vax/n_support.S                  

  > avoid ' which confuses make depend (deraadt@)

  + src/n_infnan.c                        

  > an empty stub file for lint: on the vax, lint will chew on this, but avoid
  > it
  > when compiling, and use the .S file instead
  > if someone has a better solution, show it.. (deraadt@)

== regress =========================================================== 02/06 ==

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

usr.bin

  ~ xlint/Makefile                        + xlint/test-2.c
  + xlint/test-2.c.exp                    

  > Add regression test for right shift of N-bit quantity by >= N bits
  > OK millert, deraadt (cloder@)

  ~ xlint/test-1.c.exp                    ~ xlint/test-2.c.exp

  > Update output format to reflect the fact that lint doesn't print out a
  > line with just the filename anymore. (cloder@)

  ~ xlint/Makefile                        + xlint/test-3.c
  + xlint/test-3.c.exp                    

  > Test "inline" keyword (cloder@)

== share ============================================================= 03/06 ==

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

man

  ~ man4/ral.4                            

  > Implement the Adaptive Multi Rate Retry algorithm (AMRR) for BSS mode.
  > This algorithm is particularly well suited for ural since it does not
  > require per-frame retry statistics. (damien@)

  ~ man5/mk.conf.5                        

  > NOLINT -> WANTLINT
  > ok deraadt@ (jmc@)

mk

  ~ bsd.README                            ~ bsd.lib.mk
  ~ bsd.own.mk                            

  > change NOLINT into opposite sense; WANTLINT, since it lets us built
  > lint libraries for those libraries which want them (and can have them)
  > ok millert (deraadt@)

  ~ bsd.README                            

  > NOLINT -> WANTLINT
  > ok deraadt@ (jmc@)

== sys =============================================================== 04/06 ==

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

arch/alpha/alpha

  ~ machdep.c                             

  > fix DEBUG build (martin@)

arch/arm/xscale

  ~ pxa2x0_pcic.c                         

  > use proper bus_size_t for register offsets and not bus_addr_t; uwe@ miod@
  > testing (mickey@)

arch/hp300/stand/common

  ~ ct.c                                  ~ hd.c

  > Pick HP-IB describe structures changes from main kernel code here as well.
  > (miod@)

arch/hp300/stand/include

  ~ hpibvar.h                             

  > Pick HP-IB describe structures changes from main kernel code here as well.
  > (miod@)

arch/i386/i386

  ~ esm.c                                 ~ esmreg.h

  > merge probe and match since one is only ever called from the other. knf too
  > yup deraadt@ ok marco@ (dlg@)

  ~ mpbios.c                              

  > resolve a couple of problems in mpbios-mapped interrupts:
  > - synthesise isa mappings (as 1-1) should those be missing in mpbios;
  > - for rcc osb* firce "special" ints into isa mappings always.
  > niklas@ ok and testing by many since (mickey@)

  ~ machdep.c                             

  > prioritise <16m memory for isa bus_dma'ble memorble memory allocations
  > (mickey@)

  ~ cpu.c                                 

  > assume vtophys(vaddr_t) just what all the other archs expect; no functional
  > change (mickey@)

  ~ genassym.cf                           ~ locore.s

  > finnish the PTDPTDI and APTDPTDI conversion to PDSLOT_PTE and PDSLOT_APTE
  > thus reducing confusion; remove compatibility defines and comments
  > (mickey@)

arch/i386/include

  ~ i82093var.h                           ~ mpbiosreg.h
  ~ mpbiosvar.h                           

  > resolve a couple of problems in mpbios-mapped interrupts:
  > - synthesise isa mappings (as 1-1) should those be missing in mpbios;
  > - for rcc osb* firce "special" ints into isa mappings always.
  > niklas@ ok and testing by many since (mickey@)

  ~ param.h                               

  > kill two old mach macros that are not used anywhere in the tree anymore (or
  > ever) (mickey@)

  ~ pmap.h                                ~ vmparam.h

  > finnish the PTDPTDI and APTDPTDI conversion to PDSLOT_PTE and PDSLOT_APTE
  > thus reducing confusion; remove compatibility defines and comments
  > (mickey@)

arch/i386/isa

  ~ isa_machdep.c                         

  > resolve a couple of problems in mpbios-mapped interrupts:
  > - synthesise isa mappings (as 1-1) should those be missing in mpbios;
  > - for rcc osb* firce "special" ints into isa mappings always.
  > niklas@ ok and testing by many since (mickey@)

  ~ isa_machdep.c                         ~ isa_machdep.h

  > prioritise <16m memory for isa bus_dma'ble memorble memory allocations
  > (mickey@)

arch/i386/pci

  ~ pci_intr_fixup.c                      ~ pci_machdep.c
  ~ pcibiosvar.h                          ~ rccosb4.c
  ~ rccosb4reg.h                          

  > resolve a couple of problems in mpbios-mapped interrupts:
  > - synthesise isa mappings (as 1-1) should those be missing in mpbios;
  > - for rcc osb* firce "special" ints into isa mappings always.
  > niklas@ ok and testing by many since (mickey@)

arch/mips64/include

  ~ pmap.h                                

  > there is no vtophys here (mickey@)

arch/sparc/dev

  ~ ts102.c                               

  > use proper bus_size_t for register offsets and not bus_addr_t; uwe@ miod@
  > testing (mickey@)

dev/eisa

  ~ uha_eisa.c                            

  > assume vtophys(vaddr_t) just what all the other archs expect; no functional
  > change (mickey@)

dev/ic

  ~ pdqvar.h                              ~ uha.c

  > assume vtophys(vaddr_t) just what all the other archs expect; no functional
  > change (mickey@)

  ~ i82365.c                              ~ i82365var.h
  ~ tcic2.c                               ~ tcic2var.h

  > use proper bus_size_t for register offsets and not bus_addr_t; uwe@ miod@
  > testing (mickey@)

  ~ ami.c                                 

  > slight shuffle at the top of the file. dont include biovar.h if bio is not
  > configured (dlg@)

  ~ ciss.c                                ~ cissreg.h

  > hp sa p600 sas; from Srebrenko Sehic <haver at insecure.dkmickey@)

  ~ ral.c                                 

  > Optimize TXTIME and PLCP LENGTH field computation.
  > Avoid modulus operations. (damien@)

  ~ ral.c                                 

  > Be more robust when handling Rx interrupts.  If we can't allocate and DMA
  > map
  > a new mbuf, just discard the received frame and reuse the old mbuf.
  > From NetBSD (joerg@) (damien@)

  ~ ral.c                                 

  > When defragmenting a mbuf chain before transmitting it, don't allocate a
  > mbuf
  > cluster if the payload fits in the header.
  > From NetBSD (scw@) (damien@)

  ~ ami.c                                 ~ amivar.h

  > make this easier for me to read by changing the locking to use the bare
  > spl calls rather than ami's macro wrappers for them.
  > ok krw@ marco@ (dlg@)

dev/isa

  ~ isadma.c                              

  > Correct malloc flag handling.
  > OK deraadt@ mickey@ (mpf@)

  ~ aha.c                                 ~ uha_isa.c
  ~ wds.c                                 

  > assume vtophys(vaddr_t) just what all the other archs expect; no functional
  > change (mickey@)

dev/pci

  ~ if_le_pci.c                           ~ if_lge.c
  ~ if_nge.c                              ~ if_tl.c
  ~ if_wb.c                               

  > assume vtophys(vaddr_t) just what all the other archs expect; no functional
  > change (mickey@)

  ~ pccbb.c                               

  > use proper bus_size_t for register offsets and not bus_addr_t; uwe@ miod@
  > testing (mickey@)

  ~ pcidevs.h                             ~ pcidevs_data.h

  > hp sa p600 sas; from Srebrenko Sehic <haver at insecure.dkmickey@)

  ~ if_iwi.c                              

  > Fix endianness issues.  iwi should now work on big endian architectures.
  > From NetBSD (scw@). (damien@)

  ~ if_iwi.c                              

  > Be more robust when handling Rx interrupts.  If we can't allocate and DMA
  > map
  > a new mbuf, just discard the received frame and reuse the old mbuf.
  > From NetBSD (joerg@) (damien@)

  ~ if_iwi.c                              

  > fix my previous commit.  don't leak an mbuf. (damien@)

  ~ if_ipw.c                              ~ if_iwi.c

  > When defragmenting a mbuf chain before transmitting it, don't allocate a
  > mbuf
  > cluster if the payload fits in the header.
  > From NetBSD (scw@) (damien@)

  ~ if_ipw.c                              

  > Be more robust when handling Rx interrupts.  If we can't allocate and DMA
  > map
  > a new mbuf, just discard the received frame and reuse the old mbuf.
  > From NetBSD (joerg@) (damien@)

dev/pcmcia

  ~ if_ne_pcmcia.c                        ~ if_ray.c
  ~ if_xe.c                               ~ pcmciachip.h
  ~ pcmciavar.h                           

  > use proper bus_size_t for register offsets and not bus_addr_t; uwe@ miod@
  > testing (mickey@)

dev/sbus

  ~ stp4020.c                             

  > use proper bus_size_t for register offsets and not bus_addr_t; uwe@ miod@
  > testing (mickey@)

dev/usb

  ~ if_ral.c                              

  > Optimize TXTIME and PLCP LENGTH field computation.
  > Avoid modulus operations. (damien@)

  ~ if_ral.c                              ~ if_ralvar.h

  > Implement the Adaptive Multi Rate Retry algorithm (AMRR) for BSS mode.
  > This algorithm is particularly well suited for ural since it does not
  > require per-frame retry statistics. (damien@)

  ~ if_ral.c                              

  > use sizeof instead of hard-coded values. (damien@)

  ~ if_ral.c                              

  > Be more robust when receiving frames.  If we can't allocate a new mbuf,
  > just discard the received frame and reuse the old mbuf. (damien@)

== usr.bin =========================================================== 05/06 ==

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

awk

  ~ lib.c                                 

  > Don't try to fclose(infile) when infile == NULL.  This can happen
  > when using nextfile in the BEGIN block.  From Matthias Kilian. (millert@)

rcs

  ~ co.1                                  ~ co.c

  > sync SYNOPSIS with Reality and usage() accordingly; (xsa@)

  ~ co.c                                  

  > add support for `-kmode'; (xsa@)

  ~ ci.c                                  ~ co.c
  ~ rcsclean.c                            ~ rcsdiff.c
  ~ rcsprog.c                             ~ rcsprog.h

  > initial bits for -T support; (xsa@)

  ~ ci.c                                  

  > knf; (xsa@)

  ~ co.c                                  

  > cannot combine -kv and -l; (xsa@)

  ~ ident.c                               

  > output consistency; (xsa@)

  ~ co.1                                  

  > one too many `Ns' (jmc@)

xlint

  ~ lint1/tree.c                          ~ lint1/err.c

  > Warn on a right shift of an N-bit quantity by >= N bits. OK millert,
  > deraadt (cloder@)

  ~ xlint/xlint.c                         

  > Do not pass -Wtraditional to cpp.  If someone wants that, they can
  > force that in other ways.  We want good lint messages, not cpp moaning
  > (deraadt@)

  ~ lint1/cgram.y                         ~ lint1/scan.l

  > support __attribute__ (deraadt@)

  ~ lint1/cgram.y                         ~ lint1/scan.l

  > add a fake C construct __lint_equal__ which will be used to generate
  > alias entries for weak symbols and such
  > also let ; at the top scope be legal (deraadt@)

  ~ lint1/main1.c                         ~ lint2/main2.c
  ~ xlint/xlint.c                         

  > adjust some default options (mostly, that lint always print full paths
  > in errors/warnings, so that compiling in dirs controlled by .PATH'd
  > Makefiles gives useful errors (deraadt@)

  ~ lint1/cgram.y                         ~ lint1/decl.c
  ~ lint1/lint1.h                         ~ lint1/scan.l

  > Grok __restrict__ and __restrict keywords.	OK deraadt (cloder@)

  ~ lint1/scan.l                          

  > "inline" is a c99 keyword, so support it unless tflag is selected.
  > OK and input from millert, earlier version looked at by deraadt (cloder@)

  ~ lint1/cgram.y                         ~ lint1/err.c

  > Deal somewhat with case ranges, which is a gcc extension.
  > "fine" deraadt, with his suggested modifications (cloder@)

  ~ lint1/cgram.y                         

  > Deal with empty array initializers (a gcc-ism), but warn. (cloder@)

== usr.sbin ========================================================== 06/06 ==

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

apm

  ~ apm.8                                 ~ apm.c

  > permit apmd to run on systems without apm support
  > let apmd control hw.setperf manually or dynamically
  > tested by Chris Kuethe and beck@, 'get it in' deraadt@ (sturm@)

apmd

  ~ apm-proto.h                           ~ apmd.8
  ~ apmd.c                                ~ apmsubr.c

  > permit apmd to run on systems without apm support
  > let apmd control hw.setperf manually or dynamically
  > tested by Chris Kuethe and beck@, 'get it in' deraadt@ (sturm@)

  ~ apmd.c                                

  > do not abort if hw.setperf is not there.  warn.  maybe we should even
  > skip that, too. (deraadt@)

hostapd

  ~ handle.c                              ~ hostapd.conf.5
  ~ hostapd.h                             ~ parse.y

  > add optional interface rule for event rules (reyk@)

  ~ iapp.c                                

  > log message consistency (reyk@)

tcpdump

  ~ print-pflog.c                         

  > Replace a C++ comment with a C comment. No code changes. (cloder@)

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



More information about the odc mailing list