aptd crashed with SIGSEGV in pkgProblemResolver::MakeScores()

[kistareddyd]

hi

ProblemType: Crash
DistroRelease: Ubuntu 11.10
Package: aptdaemon 0.43+bzr691
ProcVersionSignature: Ubuntu 3.0.0-10.16-generic 3.0.4
Uname: Linux 3.0.0-10-generic i686
Architecture: i386
Date: Tue Sep 6 11:57:07 2011
DesktopFile: /usr/share/aptdaemon/aptdaemon.desktop
ExecutablePath: /usr/sbin/aptd
InstallationMedia: Ubuntu 11.10 "Oneiric Ocelot" - Beta i386 (20110901)
InterpreterPath: /usr/bin/python2.7
PackageArchitecture: all
ProcCmdline: /usr/bin/python /usr/sbin/aptd
ProcEnviron:

SegvAnalysis:
 Segfault happened at: 0x103dc4d <_ZN18pkgProblemResolver10MakeScoresEv+1645>: imul $0x1c,0x28(%eax),%eax
 PC (0x0103dc4d) ok
 source "$0x1c" ok
 destination "0x28(%eax)" (0xc405a8b8) not located in a known VMA region (needed writable region)!
SegvReason: writing unknown VMA
Signal: 11
SourcePackage: aptdaemon
StacktraceTop:
 pkgProblemResolver::MakeScores() () from /usr/lib/libapt-pkg.so.4.11
 pkgProblemResolver::ResolveByKeepInternal() () from /usr/lib/libapt-pkg.so.4.11
 pkgProblemResolver::ResolveByKeep() () from /usr/lib/libapt-pkg.so.4.11
 pkgAllUpgrade(pkgDepCache&) () from /usr/lib/libapt-pkg.so.4.11
 ?? () from /usr/lib/python2.7/dist-packages/apt_pkg.so
Title: aptd crashed with SIGSEGV in pkgProblemResolver::MakeScores()
UpgradeStatus: No upgrade log present (probably fresh install)
UserGroups:

[Apport retracing service (apport)]

StacktraceTop:
 pkgProblemResolver::MakeScores (this=0xbfb9b108) at algorithms.cc:600
 pkgProblemResolver::ResolveByKeepInternal (this=0xbfb9b108) at algorithms.cc:1258
 pkgProblemResolver::ResolveByKeep (this=0xbfb9b108) at algorithms.cc:1244
 pkgAllUpgrade (Cache=...) at algorithms.cc:418
 PkgDepCacheUpgrade (Self=<apt_pkg.DepCache at remote 0x93f4d2c>, Args=(False,)) at python/depcache.cc:264

[Julian Andres Klode (juliank)]

Two options:

(a) If we release the GIL, and call pkgAllUpgrade(), pkgAllUpgrade() can modify the depCache. If now another (Python)-thread accesses the depcache, we're lost (and that's potentially a real bug, given that APT is almost totally not thread-safe).

(b) The cache file is broken - we rule out too short cache files now, as that is detected since August 2011, so it can basically only be caused by a broken storage or memory.

[Julian Andres Klode (juliank)]

Reassigning to APT, that's APT function

[Julian Andres Klode (juliank)]

Closing this, since this seems to be a broken cache file.