> Try using %CYGWIN%=error_start to get either gdb or dumper.exe attached at > the exact moment of failure. Check the user guide > http://cygwin.com/cygwin-ug-net/using-cygwinenv.html > for full usage information.
Thanks for the tip. That was enough to get a stack trace with no symbol information -- not too helpful on its own. When I rebuilt make with debug information from the Cygwin source package, I was completely unable to reproduce the segfaults -- four complete builds have now succeeded in a row, compared with previously when each build would show between 1 and about 7 segfaults. I'd say this is beginning to look like a weird compiler and/or optimisation bug, given that it happens frequently with the Cygwin package and not at all with an almost un-optimised build (CFLAGS="-O1 -fno-inline-functions -ggdb3"). Unfortunately I don't have any sort of reduced test case demonstrating it, but if I get a chance I'll try to narrow down what specifically is triggering it. Is there an easy way to find which compiler and optimisation settings were used to build the current Cygwin make package? Accelrys Limited (http://www.accelrys.com) Registered office: 334 Cambridge Science Park, Cambridge, CB4 0WN, UK Registered in England: 2326316 -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Problem reports: http://cygwin.com/problems.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/