On Sep 1 18:14, Pavel Fedin wrote: > Hello! Another update has been published and i still experience random > crashes on x86-64. > At this point i managed to go a little bit farther in my research. The > problem reproduces best with 'sh --version' command. At first glance it > seems to work OK most of the time, but take a look at strace's output: > --- cut --- > 1 1 [main] sh (3292) > ********************************************** > 102 103 [main] sh (3292) Program name: C:\cygwin64\bin\sh.exe (windows > pid 3292) > 41 144 [main] sh (3292) OS version: Windows NT-6.1 > 37 181 [main] sh (3292) > ********************************************** > 110 291 [main] sh (3292) sigprocmask: 0 = sigprocmask (0, 0x0, > 0x1802BED88) > 400 691 [main] sh 3292 open_shared: name shared.5, n 5, shared > 0x180030000 (wanted 0x180030000), h 0x70, *m 6 > 86 777 [main] sh 3292 user_heap_info::init: heap base 0x600000000, > heap top 0x600000000, heap size 0x20000000 (536870912) > 58 835 [main] sh 3292 open_shared: name > S-1-5-21-1454471165-515967899-839522115-3227.1, n 1, shared 0x180020000 > (wanted 0x180020000), h 0x6C, *m 6 > 41 876 [main] sh 3292 user_info::create: opening user shared for > 'S-1-5-21-1454471165-515967899-839522115-3227' at 0x180020000 > 41 917 [main] sh 3292 user_info::create: user shared version AB1FCCE8 > 70 987 [main] sh 3292 fhandler_pipe::create: name > \\.\pipe\cygwin-e022582115c10879-3292-sigwait, size 11440, mode > PIPE_TYPE_MESSAGE > 88 1075 [main] sh 3292 fhandler_pipe::create: pipe read handle 0x84 > 39 1114 [main] sh 3292 fhandler_pipe::create: CreateFile: name > \\.\pipe\cygwin-e022582115c10879-3292-sigwait > 65 1179 [main] sh 3292 fhandler_pipe::create: pipe write handle 0x88 > 47 1226 [main] sh 3292 dll_crt0_0: finished dll_crt0_0 initialization > --- Process 3292, exception c0000005 at 0000000076F3E4E4
I don't get this with 64 bit Cygwin 1.7.32 (tried on W7 and W8.1). Are you sure this exception is triggered by Cygwin and not induced by a BLODA, a virus scanner or whatever? This early in process initialization it could even be the Windows Loader itself. But even then, not all exceptions are fatal. This one is an exception in a Windows DLL, which is probably supposed to be catched by SEH. The OS certainly shouldn't crash due to a NULL pointer or handle, but it's less hassle to catch an exception than to test pointers or handles for validity every time a function is entered. > $ gdb --args sh --version > GNU gdb (GDB) 7.8 > Copyright (C) 2014 Free Software Foundation, Inc. > License GPLv3+: GNU GPL version 3 or later > <http://gnu.org/licenses/gpl.html> > This is free software: you are free to change and redistribute it. > There is NO WARRANTY, to the extent permitted by law. Type "show copying" > and "show warranty" for details. > This GDB was configured as "x86_64-unknown-cygwin". > Type "show configuration" for configuration details. > For bug reporting instructions, please see: > <http://www.gnu.org/software/gdb/bugs/>. > Find the GDB manual and other documentation resources online at: > <http://www.gnu.org/software/gdb/documentation/>. > For help, type "help". > Type "apropos word" to search for commands related to "word"... > Reading symbols from sh...(no debugging symbols found)...done. > (gdb) run > Starting program: /usr/bin/sh --version > [New Thread 2716.0x650] > > Program received signal SIGSEGV, Segmentation fault. > 0x0000000076f3e4e4 in ntdll!RtlDeNormalizeProcessParams () from > /cygdrive/c/Windows/SYSTEM32/ntdll.dll Try `bt' at this point. Maybe the backtrace shows something of interest. Apart from that, try the latest snapshot from https://cygwin.com/snapshots/ I converted the internal exception handling avoiding crashes from system functions to SEH. Perhaps the above doesn't show up with it. Corinna -- Corinna Vinschen Please, send mails regarding Cygwin to Cygwin Maintainer cygwin AT cygwin DOT com Red Hat
pgpey9SH1Wo6I.pgp
Description: PGP signature
