[klibc] [PATCH 0/5] resubmitting pending patches

[Thorsten Glaser]

Hi,

I’ve gone through the mailing list archives and hereby want
to resubmit my pending patches. Most are independent of each
other, except the m68k patch which will only be complete if
sigsuspend is also fixed. (It can be applied before that,
though.)

http://www.zytor.com/pipermail/klibc/2012-January/003173.html
	[PATCH] fix m68k support
Resubmitted here as 0005. While there was a question from
Andreas, this WFM as-is and he never responded what else
should be used. (And the form I submitted is used by eglibc, too.)

http://www.zytor.com/pipermail/klibc/2012-January/003171.html
	[PATCH] add mkstemp(3) using entropy from ELF aux vector AT_RANDOM
Dropped because mksh does not use mkstemp/tempnam any more.

http://www.zytor.com/pipermail/klibc/2012-March/003194.html
	[PATCH 1/3] align __le64 (and other) type definitions with linux-libc-dev headers
Resubmitted here as 0001.

http://www.zytor.com/pipermail/klibc/2012-March/003195.html
	[PATCH 2/3] fix missing prototypes for ctype inline functions
Resubmitted here as 0002.

http://www.zytor.com/pipermail/klibc/2012-March/003193.html
	[PATCH 3/3] fix missing prototypes for stdio inline functions
Dropped because hpa recently added prototypes to the headers,
although slightly differently.

http://www.zytor.com/pipermail/klibc/2012-May/003201.html
	[PATCH] fix non-RT sigsuspend()
Cleaned up by hpa’s suggestion and resubmitted here as 0003. Please add
this, definitely, with high urgency (as klibc’s dash is also affected),
until *all* architectures are switched to RT signals.

There’s another, new, patch (0004 here), based on a request by hpa:
http://www.zytor.com/pipermail/klibc/2012-April/003199.html
It changes the way size_t, ssize_t and ptrdiff_t are defined to use
the definition from the compiler (GCC and compatibles, such as Clang)
with a fallback to long/unsigned long if none is given, which is
always correct on LP64 and by-the-size correct on ILP32 platforms.
This is mostly for benefit of GCC’s -Wformat so the fallback not
being int/unsigned int on some platforms does not hurt. (Besides,
klcc invokes gcc, not some-random-compiler. And this is only for
stuff compiled with klibc, anyway.)


I’ve tested the following things:
• compile klibc (amd64, i386, m68k)
• check for header warnings (amd64)
• build mksh against klibc, both statically and dynamically linked
  ‣ amd64, i386: -O2 -g -Wall -Wextra -Wformat
  ‣ m68k: -O2 -Wall -Wextra -Wformat
• test mksh (statically and dynamically linked)

No errors. \o/

bye,
//mirabilos
-- 
I believe no one can invent an algorithm. One just happens to hit upon it
when God enlightens him. Or only God invents algorithms, we merely copy them.
If you don't believe in God, just consider God as Nature if you won't deny
existence.		-- Coywolf Qi Hunt