CVS log for dietlibc/i386/dyn_syscalls.S

Up to [DVB] / dietlibc / i386

Request diff between arbitrary revisions

Keyword substitution: kv
Default branch: MAIN

fix dynamic build

fix compiler warnings
make i386 start code slightly smaller (I think profiling was broken,
too, and this should fix it)
add PIC stack smash handler

  use vsyscall for time on x86_64 instead of wrapping gettimeofday (Nikola Vladov)
  CPU affinity support (Indan Zupancic)

fix make dyn

fix dyamic build

add syscalls to dynamic linking

lots of new syscalls

  provide toascii and ustat (Reynir Stefansson)

dynamic x86_64 support (libdl relocator and dyn_syscalls)

settimeofday...

set the elf-type of some symbols

  Fix i386 unified syscall -EMEDIUMTYPE handling (Juergen Sawinski via
    Clifford Wolf)

added missing syscalls (cap{g,s}et)
added missing addmntent

msync was missing from dynamic syscalls on x86 and arm (Thomas Walpuski)

missed to add the missing syscall in this...

sendfile64 was missing in the dynamic libraries.
changing alignment flags to new values

- removing "<arch>/mmap.*" where no arch-specific impementation is needed.
  (added "syscalls.s/mmap.S")

- fixing the vsyscall in the dynamic library... (self made problem)

- removing warning in "res_query.c"

- changing type of stream-mutex to recursive

- sparc/sparc64: shrink of the unified syscall and the special syscalls
  (fork/pipe) to use the "unified_syscall_error_handler"

added SYSENTER support

fixed a "bug"... someone had missed to put in PIC-code *grumble*

work around new x86 2.5 syscall numbers exceeding 255, thus breaking our
%al size optimization by adding a new fixup __unified_syscall_256 that
uses %ax and fixing up the stubs to call this one if the syscall number
is large enough.  Ugh.

add a few more syscall #defines
add epoll header file and syscalls

work around problem caused by other workaround (caused by a workaround
gone bad working around some other problem)

Don' ask.

do some _really_ gross kludgery to get the new i386 socketcalls in the
dyn_syscalls and out of the separate object files so the assembler can
produce short jumps

still doesn't compile, but it's a start

i386/*	reduce number of executed instructions
atexit	added symbol for new pthread implementation
x86_64	hope that profiling and {con,de}structors works again

make signal() a wrapper around sigaction()

add truncate64 and ftruncate64 (with backwards compatibility)

a few patches to make Berkeley DB and Postfix compile

Assembler and PIC code the source of all evel in the dynamic-section....
Added a helper for x86 and modified the assembler sources (now the
libdietc.so is NON textrel again....). S390 fixed a typo.
Fixed diet so dynamic binaries uses the "correct" dynamic-linker...

add ptrace to i386 dyn syscalls

be more alpha and gcc 2.96 for SuSE 7.1 AXP friendly, add fdatasync.

remove vfork and add alarm to dynamic syscalls.
fix diet-dyn to link in dyn_dstop instead of dyn_stop

redo a few changes from the other repository

removed a bug with the largefile back compatible option and dynamic-linking

arg... fixed some names (dyn_syscalls)
and the other weren't dynamic aware...

new syscalls for modutils

even more missing syscalls (all 32-bit uid/gid calls)

"new" adjtimex wasn't added to the dynamic-syscalls

so a LITTLE change of stdio.... (urg...)
added some weak aliases to assembler returns :) that are always present.

ups missed some asm-syms and removed warning

asm-symbol-declarstion and dynlibs *ARG* ....
fixed Makefile for target dyn to create the pic-$(ARCH) dir ...

ups... PIC code. Forgot the PIC "lead-in" in clone....

mini change ...

changed a "little" for dynamic dietlibc(.so)