This is the mail archive of the
mailing list for the glibc project.
Re: [PATCH] Increase fork signal safety for single-threaded processes [BZ #19703]
- From: Florian Weimer <fweimer at redhat dot com>
- To: Tulio Magno Quites Machado Filho <tuliom at linux dot vnet dot ibm dot com>, libc-alpha at sourceware dot org
- Date: Fri, 13 May 2016 16:40:51 +0200
- Subject: Re: [PATCH] Increase fork signal safety for single-threaded processes [BZ #19703]
- Authentication-results: sourceware.org; auth=none
- References: <20160511134121 dot CDFB841B3C727 at oldenburg dot str dot redhat dot com> <87mvnwv00p dot fsf at totoro dot br dot ibm dot com>
On 05/11/2016 09:59 PM, Tulio Magno Quites Machado Filho wrote:
I suggest to fork the SIGUSR2 sender first.
Sometimes the SIGUSR1 sender floods the parent to the point it takes seconds
to fork the second sender.
I believe this is a different issue: sigusr1_sender_pid may be 0 if
SIGUSR1 arrives very early, and we send SIGSTOP to the current process
group, not just the subprocess. This is consistent with no output at
all from the test program (which is what I saw in this case).
I will commit a fix for this test issue shortly.