On May 21 14:50, Eric Blake wrote: > On 05/21/2012 10:42 AM, Corinna Vinschen wrote: > > >> The crash occurs after echo exited, so bash wakes up from the wait4 > >> call. However, the problem is that the crash does not occur in Cygwin, > >> but in bash itself. > >> > >> 147 350775 [main] bash 3548 wait4: 2320 = wait4(-1, 0x0, 0, 0x0) > >> --- Process 3548, exception C0000005 at 00422B0A > >> > >> Eric, can you reproduce this and see where it happens? I'm pretty sure > >> it's a bug in Cygwin, not in bash, but it would be interesting to learn > >> what bash did at the time the crash happened. > > > > Incidentally I built bash without -O2 option for better debugging and > > the problem vanished. Then I built bash again with default optimization > > and the crash still didn't occur. I built from the latest bash src > > package.4.1.10-4 using cygport. > > Uggh. This sounds familiar to another bash bug that I investigated some > time ago, where bash was abusing longjmp() and miscompiled under -O2 but > compiled correctly at -O0 due to the undefined behavior from that abuse, > but I just verified that my patch from back then is still present in my > latest build of bash for cygwin. I'll have to find more time to look > into this. > https://lists.gnu.org/archive/html/bug-bash/2011-02/msg00060.html
In my testing it doesn't matter if I build execute_cmd.c or, FWIW, any of bash's source files with -O0, -O1, or -O2. My self-built bash never crashes in this scenario. Corinna -- Corinna Vinschen Please, send mails regarding Cygwin to Cygwin Project Co-Leader cygwin AT cygwin DOT com Red Hat -- Problem reports: http://cygwin.com/problems.html FAQ: http://cygwin.com/faq/ Documentation: http://cygwin.com/docs.html Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
