On 05/19/2014 06:49 PM, Ken Brown wrote: > On 5/19/2014 5:47 PM, Daniel Colascione wrote: >> On 05/19/2014 02:45 PM, Ken Brown wrote: >>> On 5/19/2014 5:00 PM, Daniel Colascione wrote: >>>> On 05/19/2014 12:39 PM, Eli Zaretskii wrote: >>>>>> Date: Mon, 19 May 2014 15:19:24 -0400 >>>>>> From: Ken Brown >>>>>> CC: 17526@debbugs.gnu.org >>>>>> >>>>>> On 5/19/2014 11:18 AM, Eli Zaretskii wrote: >>>>>>>> Date: Mon, 19 May 2014 08:30:49 -0400 >>>>>>>> From: Ken Brown >>>>>>>> >>>>>>>> I just got a crash with the following backtrace: >>>> >>>> There goes my BLODA theory (assuming you don't have any). These crashes >>>> (including the indirect-variable one) are all seemingly impossible, but >>>> if there were generalized random memory corruption, I'd expect to see >>>> more severe effects. I have no idea what's wrong. Both trunk and >>>> emacs-24 seem to be affected: did something change in the Cygwin DLL >>>> side? >>> >>> Not that I know of. By the way, all these crashes that are being >>> reported are in 64-bit Cygwin, FWIW. So it may very well be a Cygwin >>> problem rather than an emacs problem. >>> >>> Do you have a 64-bit Cygwin installation on which you could try to >>> reproduce this? The crashes are seemingly random. I've occasionally >>> gotten a crash a few seconds after starting emacs, but other times I've >>> run emacs for days without a problem. So you might have to wait a long >>> time. Also, I don't always get a chance to attach gdb. In other words, >>> emacs just dies with a SEGV, but apparently without calling emacs_abort. >> >> I don't, and it'll be at least a week or so before I'm able to set one >> up. Nobody's reported crashes in the X11 Emacs? What about cygw32 run >> with -nw? > > No, all the recent crash reports have been for cygw32. But I doubt if > anyone has run it with -nw. I'll try that myself. Thanks. I'm just surprised that the crash is specific to the cygw32 configuration, that it happens nowhere near GUI code, and that both regular Cygwin emacs and NT Emacs seem unaffected. Let's encourage people to run --enable-checking for a while.