This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH][BZ #14782] Do not enable asynchronous cancelation in system
- From: "Carlos O'Donell" <carlos at redhat dot com>
- To: OndÅej BÃlka <neleai at seznam dot cz>
- Cc: Rich Felker <dalias at aerifal dot cx>, libc-alpha at sourceware dot org
- Date: Tue, 14 Jan 2014 09:44:38 -0500
- Subject: Re: [PATCH][BZ #14782] Do not enable asynchronous cancelation in system
- Authentication-results: sourceware.org; auth=none
- References: <20140112121310 dot GA28961 at domone dot podge> <20140112152839 dot GY24286 at brightrain dot aerifal dot cx> <52D2D2F7 dot 8080301 at redhat dot com> <20140114124328 dot GB15164 at domone dot podge> <52D53ED9 dot 9050307 at redhat dot com> <20140114144131 dot GA17590 at domone dot podge>
On 01/14/2014 09:41 AM, OndÅej BÃlka wrote:
> On Tue, Jan 14, 2014 at 08:42:49AM -0500, Carlos O'Donell wrote:
>> On 01/14/2014 07:43 AM, OndÅej BÃlka wrote:
>>>> This looks good go me... however!
>>>>
>>>> OK to commit as long as you prove waitpid is actually a
>>>> cancellation point as required by POSIX.
>>>>
>>>
>>> It in list here:
>>>
>>> http://pubs.opengroup.org/onlinepubs/000095399/functions/xsh_chap02_09.html#tag_02_09_05_02
>>
>> My apologies. I wasn't clear enough on the requirement.
>>
>> I want you to look at the glibc waitpid implementation
>> and prove it actually has a cancellation point.
>>
> It does, linux waitpid implementation follows same pattern as system:
Thanks. Then your patch is OK.
Please don't do the refactoring, that can be done in 2.20.
We also all agree that the AC enable/disable pattern around
a syscall is wrong (see bug 12683).
Cheers,
Carlos.