On Jul 7, 2020, at 9:42 AM, Jiri Palecek <jpale...@web.de> wrote: > > Package: checkinstall > Version: 1.6.2+git20170426.d24a630-2 > Severity: important > File: /usr/bin/installwatch > > Dear Maintainer, > > while trying to use checkinstall to create a debianized package from a > cmake based source, the build failed with a segfault. These are linked > to installwatch and don't happen without it: > > $ installwatch make cmake_check_build_system > > INFO : Using a default root directory : /tmp/tmp.JBpq66zd4H > > make: *** [Makefile:10806: cmake_check_build_system] Neoprávněný přístup do > paměti (SIGSEGV) (obraz paměti uložen) > > There is a backtrace of the crash, which indicates it happens early in > the initialization of cmake around a stat call: > > (gdb) bt > #0 0x00000000 in ?? () > #1 0xb6a3fbd3 in stat64 (__statbuf=<optimized out>, __path=0xb6b472bb > "/etc/gnutls/config") at /usr/include/i386-linux-gnu/sys/stat.h:455 > #2 _gnutls_update_system_priorities () at ../../lib/priority.c:1309 > #3 0xb6a534f5 in _gnutls_global_init (constructor=constructor@entry=1) at > ../../lib/global.c:387 > #4 0xb6a25950 in lib_init () at ../../lib/global.c:511 > #5 0xb7f35f5c in call_init (l=<optimized out>, argc=argc@entry=6, > argv=argv@entry=0xbfe33e64, env=0xbfe33e80) at dl-init.c:72 > #6 0xb7f36062 in call_init (env=0xbfe33e80, argv=0xbfe33e64, argc=6, > l=<optimized out>) at dl-init.c:30 > #7 _dl_init (main_map=<optimized out>, argc=6, argv=0xbfe33e64, > env=0xbfe33e80) at dl-init.c:119 > #8 0xb7f270fa in _dl_start_user () from /lib/ld-linux.so.2 > (gdb) frame 1 > #1 0xb6a3fbd3 in stat64 (__statbuf=<optimized out>, __path=0xb6b472bb > "/etc/gnutls/config") at /usr/include/i386-linux-gnu/sys/stat.h:455 > 455 return __xstat (_STAT_VER, __path, __statbuf); > > Why did it end up with EIP=0 I don't know. > > It seems there's some incompatibility between installwatch's LD_PRELOAD > and glibc. > > Could you have a look at it? > > Regards > Jiri Palecek
Jiri, Thanks for the report. In order to help me narrow this down are you able to provide a simple test case to reproduce the problem? Thanks! Stephen