Re: struct siginfo vs. siginfo_t (was: GNU C Library master sources branch, master, updated. glibc-2.15-229-g4efeffc)
On Thursday 15 March 2012 11:57:00 Carlos O'Donell wrote: > We should be rebuilding *all* of userspace when glibc changes. It > would be nice if we setup an OpenEmbedded system to rebuild as much of > x86-64 userspace as possible against a new glibc and check for > regressions. emerge -e world -mike signature.asc Description: This is a digitally signed message part.
Re: struct siginfo vs. siginfo_t (was: GNU C Library master sources branch, master, updated. glibc-2.15-229-g4efeffc)
On Thu, Mar 15, 2012 at 8:57 AM, Carlos O'Donell
wrote:
> On Thu, Mar 15, 2012 at 11:05 AM, Thomas Schwinge
> wrote:
>> Hi!
>>
>> On 26 Feb 2012 18:17:52 -, drep...@sourceware.org wrote:
>>> http://sources.redhat.com/git/gitweb.cgi?p=glibc.git;a=commitdiff;h=4efeffc1d583597e4f52985b9747269e47b754e2
>>>
>>> commit 4efeffc1d583597e4f52985b9747269e47b754e2
>>> Author: Ulrich Drepper
>>> Date: Sun Feb 26 13:17:27 2012 -0500
>>>
>>> Fix up POSIX testing in conformtest
>>
>>> [...]
>>> + * sysdeps/unix/sysv/linux/bits/siginfo.h: Don't name siginfo_t
>>> + struct. [...]
>>> [...]
>>
>>> diff --git a/sysdeps/unix/sysv/linux/bits/siginfo.h
>>> b/sysdeps/unix/sysv/linux/bits/siginfo.h
>>> index ecef39d..0635e2f 100644
>>> --- a/sysdeps/unix/sysv/linux/bits/siginfo.h
>>> +++ b/sysdeps/unix/sysv/linux/bits/siginfo.h
>>> [...]
>>> @@ -47,7 +47,7 @@ typedef union sigval
>>> # define __SI_PAD_SIZE ((__SI_MAX_SIZE / sizeof (int)) - 3)
>>> # endif
>>>
>>> -typedef struct siginfo
>>> +typedef struct
>>> {
>>> int si_signo; /* Signal number. */
>>> int si_errno; /* If non-zero, an errno value associated with
>>> [...]
>>
>> This change breaks GCC:
>>
>> In file included from
>> /scratch/tschwing/FM_sh-linux-gnu-mk2/src/gcc-mainline/libgcc/unwind-dw2.c:377:0:
>> ./md-unwind-support.h: In function 'sh_fallback_frame_state':
>> ./md-unwind-support.h:182:17: error: field 'info' has incomplete type
>>
>> In my case, this is really libgcc/config/sh/linux-unwind.h:
>>
>> [...]
>> 181 struct rt_sigframe {
>> 182 struct siginfo info;
>> 183 struct ucontext uc;
>> 184 } *rt_ = context->cfa;
>> [...]
>
> POSIX says you get "siginto_t" *not* "struct siginfo," please fix the code.
>
"struct siginfo" may not work correctly for Linux/x32
and siginfo_t does since siginfo_t has an alignment
attribute, which isn't applied to "struct siginfo".
--
H.J.
Re: struct siginfo vs. siginfo_t (was: GNU C Library master sources branch, master, updated. glibc-2.15-229-g4efeffc)
On Thu, Mar 15, 2012 at 11:05 AM, Thomas Schwinge
wrote:
> Hi!
>
> On 26 Feb 2012 18:17:52 -, drep...@sourceware.org wrote:
>> http://sources.redhat.com/git/gitweb.cgi?p=glibc.git;a=commitdiff;h=4efeffc1d583597e4f52985b9747269e47b754e2
>>
>> commit 4efeffc1d583597e4f52985b9747269e47b754e2
>> Author: Ulrich Drepper
>> Date: Sun Feb 26 13:17:27 2012 -0500
>>
>> Fix up POSIX testing in conformtest
>
>> [...]
>> + * sysdeps/unix/sysv/linux/bits/siginfo.h: Don't name siginfo_t
>> + struct. [...]
>> [...]
>
>> diff --git a/sysdeps/unix/sysv/linux/bits/siginfo.h
>> b/sysdeps/unix/sysv/linux/bits/siginfo.h
>> index ecef39d..0635e2f 100644
>> --- a/sysdeps/unix/sysv/linux/bits/siginfo.h
>> +++ b/sysdeps/unix/sysv/linux/bits/siginfo.h
>> [...]
>> @@ -47,7 +47,7 @@ typedef union sigval
>> # define __SI_PAD_SIZE ((__SI_MAX_SIZE / sizeof (int)) - 3)
>> # endif
>>
>> -typedef struct siginfo
>> +typedef struct
>> {
>> int si_signo; /* Signal number. */
>> int si_errno; /* If non-zero, an errno value associated with
>> [...]
>
> This change breaks GCC:
>
> In file included from
> /scratch/tschwing/FM_sh-linux-gnu-mk2/src/gcc-mainline/libgcc/unwind-dw2.c:377:0:
> ./md-unwind-support.h: In function 'sh_fallback_frame_state':
> ./md-unwind-support.h:182:17: error: field 'info' has incomplete type
>
> In my case, this is really libgcc/config/sh/linux-unwind.h:
>
> [...]
> 181 struct rt_sigframe {
> 182 struct siginfo info;
> 183 struct ucontext uc;
> 184 } *rt_ = context->cfa;
> [...]
POSIX says you get "siginto_t" *not* "struct siginfo," please fix the code.
> (Is there really nobody doing nightly testing of GCC against glibc master
> branch on x86, which would have shown this earlier?)
I don't test building GCC against glibc master unless I'm making a
potentially ABI breaking change.
We should be rebuilding *all* of userspace when glibc changes. It
would be nice if we setup an OpenEmbedded system to rebuild as much of
x86-64 userspace as possible against a new glibc and check for
regressions.
> Replacing struct siginfo with siginfo_t in sh/linux-unwind.h makes the
> build pass. Is this the desired approach; shall we apply this for all
> cases listed above (and submit to Boehm GC upstream)? I wonder -- if GCC
> breaks, how much software out in the wild is going to break once this
> glibc change ripples through?
You can't know until you try building a full distribution.
Cheers,
Carlos.
struct siginfo vs. siginfo_t (was: GNU C Library master sources branch, master, updated. glibc-2.15-229-g4efeffc)
Hi!
On 26 Feb 2012 18:17:52 -, drep...@sourceware.org wrote:
> http://sources.redhat.com/git/gitweb.cgi?p=glibc.git;a=commitdiff;h=4efeffc1d583597e4f52985b9747269e47b754e2
>
> commit 4efeffc1d583597e4f52985b9747269e47b754e2
> Author: Ulrich Drepper
> Date: Sun Feb 26 13:17:27 2012 -0500
>
> Fix up POSIX testing in conformtest
> [...]
> + * sysdeps/unix/sysv/linux/bits/siginfo.h: Don't name siginfo_t
> + struct. [...]
> [...]
> diff --git a/sysdeps/unix/sysv/linux/bits/siginfo.h
> b/sysdeps/unix/sysv/linux/bits/siginfo.h
> index ecef39d..0635e2f 100644
> --- a/sysdeps/unix/sysv/linux/bits/siginfo.h
> +++ b/sysdeps/unix/sysv/linux/bits/siginfo.h
> [...]
> @@ -47,7 +47,7 @@ typedef union sigval
> # define __SI_PAD_SIZE ((__SI_MAX_SIZE / sizeof (int)) - 3)
> # endif
>
> -typedef struct siginfo
> +typedef struct
>{
> int si_signo;/* Signal number. */
> int si_errno;/* If non-zero, an errno value associated with
> [...]
This change breaks GCC:
In file included from
/scratch/tschwing/FM_sh-linux-gnu-mk2/src/gcc-mainline/libgcc/unwind-dw2.c:377:0:
./md-unwind-support.h: In function 'sh_fallback_frame_state':
./md-unwind-support.h:182:17: error: field 'info' has incomplete type
In my case, this is really libgcc/config/sh/linux-unwind.h:
[...]
181struct rt_sigframe {
182 struct siginfo info;
183 struct ucontext uc;
184} *rt_ = context->cfa;
[...]
There are more such cases:
boehm-gc/os_dep.c:# define SIGINFO_T struct siginfo
libgcc/config/alpha/linux-unwind.h: struct siginfo info;
libgcc/config/bfin/linux-unwind.h:struct siginfo *pinfo;
libgcc/config/bfin/linux-unwind.h:struct siginfo info;
libgcc/config/i386/linux-unwind.h:struct siginfo *pinfo;
libgcc/config/i386/linux-unwind.h:struct siginfo info;
libgcc/config/ia64/linux-unwind.h:struct siginfo *info;
libgcc/config/ia64/linux-unwind.h:struct siginfo *info;
libgcc/config/mips/linux-unwind.h:struct siginfo info;
libgcc/config/pa/linux-unwind.h:struct siginfo info;
libgcc/config/sh/linux-unwind.h: struct siginfo *pinfo;
libgcc/config/sh/linux-unwind.h: struct siginfo info;
libgcc/config/sh/linux-unwind.h: struct siginfo info;
libgcc/config/tilepro/linux-unwind.h:struct siginfo info;
libgcc/config/xtensa/linux-unwind.h:struct siginfo info;
(Is there really nobody doing nightly testing of GCC against glibc master
branch on x86, which would have shown this earlier?)
Replacing struct siginfo with siginfo_t in sh/linux-unwind.h makes the
build pass. Is this the desired approach; shall we apply this for all
cases listed above (and submit to Boehm GC upstream)? I wonder -- if GCC
breaks, how much software out in the wild is going to break once this
glibc change ripples through?
Grüße,
Thomas
pgpkHs9szr65i.pgp
Description: PGP signature
