[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#218917: marked as done (apt: Apt dies with a memory corruption error)

Your message dated Fri, 19 Aug 2011 12:05:22 +0200
with message-id <20110819115258.GA4040@debian.org>
and subject line Re: Bug#81829: "Segmentation faulty tree" (#270147) still present in Lenny, broken /var/cache/apt/*.bin available for download
has caused the Debian Bug report #81829,
regarding apt: Apt dies with a memory corruption error
to be marked as done.

This means that you claim that the problem has been dealt with.
If this is not the case it is now your responsibility to reopen the
Bug report if necessary, and/or fix the problem forthwith.

(NB: If you are a system administrator and have no idea what this
message is talking about, this may indicate a serious mail system
misconfiguration somewhere. Please contact owner@bugs.debian.org
immediately.)


-- 
81829: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=81829
Debian Bug Tracking System
Contact owner@bugs.debian.org with problems
--- Begin Message --- 
Package: apt
Version: 0.5.4
Severity: normal

Now for the second time, apt-get dies with
Reading Package Lists... Done
Speicherzugriffsfehleree... 78%

irregardless of the command issued (e.g. dist-upgrade, clean, ..).
(Sorry, I tried changing the german error message to C, but failed,
even and env LC_MESSAGES=C did not help, it means "memory acess
error"). 

I can solve the problem by moving both /var/cache/apt/pkgcache.bin and
/var/cache/apt-old2/srcpkgcache.bin to a new place. Copying them back
"restores" the problem.

If you are interested, I can provide you with those two files. IMHO
apt should complain "corrupted databases" if something has happened,
but not die. While I don't understand how they got corrupted (now the
second time within a few days). 

Also, if desired, I can rebuild apt with debug symbols enabled and run
a backtrace on it.


-- System Information
Debian Release: testing/unstable
Architecture: alpha
Kernel: Linux sixtyfor 2.4.21-pre1 #1 Don Okt 2 18:27:25 CEST 2003 alpha
Locale: LANG=de_DE@euro, LC_CTYPE=de_DE@euro

Versions of packages apt depends on:
ii  libc6.1                2.3.2-7           GNU C Library: Shared libraries an
ii  libstdc++2.10-glibc2.2 1:2.95.4-11woody1 The GNU stdc++ library
-- 
Helge Kreutzmann, Dipl.-Phys.               Helge.Kreutzmann@itp.uni-hannover.de
  gpg signed mail preferred    gpg-key: finger kreutzm@rigel.itp.uni-hannover.de
    64bit GNU powered                  http://www.itp.uni-hannover.de/~kreutzm
       Help keep free software "libre": http://www.freepatents.org/

--- End Message ---
--- Begin Message --- 
Source: apt
Source-Version: 0.8.16~exp4

apt (0.8.16~exp4) experimental; urgency=low

   [ Julian Andres Klode ]
   * apt-pkg/pkgcache.h:
     - [ABI break] Add pkgCache::Header::CacheFileSize, storing the cache size
   * apt-pkg/pkgcachegen.cc:
     - Write the file size to the cache
   * apt-pkg/pkgcache.cc:
     - Check that cache is at least CacheFileSize bytes large (LP: #16467)

On Thu, Aug 18, 2011 at 04:16:05PM -0500, Jonathan Nieder wrote:
> tags 81829 - moreinfo
> quit
> 
> Axel Beckert wrote:
> 
> > I occasionally ran into this bug on Lenny, can't remember on which
> > platform, but never deterministically.
> >
> > But today I reproducibly ran into this bug with both, apt-get and
> > aptitude. Independent of what I did: aptitude; aptitude -u; aptitude
> > upgrade; apt-get upgrade, I always get the "Segmentation faulty
> > tree... 50%" ("Building dependency tree... 50%^MSegmentation fault").
> >
> > Moving /var/lib/apt/extended_states away didn't help.
> >
> > Couldn't even do an apt-get install gdb for generating a backtrace.
> >
> > Moving away pkgcache.bin and srcpkgcache.bin from /var/cache/apt/
> > finally did help (thanks to waldi for that hint), but copying them
> > back after upgrading two packages didn't reproduce the segfault --
> > they always got recreated.
> 
> Thanks!  No promises about being able to take a look soon, but I've
> downloaded them.
I closed the Launchpad bug in 0.8.16~exp4, but forgot to close that 
one. We still cannot detect invalid caches where data changes, but we
can now detect all truncated caches, and reject them.

I could have included a CRC checksum in the header of the remaining
cache, but our experience so far is that 

 (a) most (all?) of these bugs are the result of truncated cache files
 (b) checksumming the cache on opening is much slower than we want,
     especially on ARM systems (200 ms on abel.d.o, 500 ms on an N900,
     12 ms on my Intel Core i5)

That said, if future shows us cases where there are problems with
correctly-sized caches, we can still add a checksum when we break
ABI again, and enable it by default only on amd64 and other fast
architectures.

-- 
Julian Andres Klode  - Debian Developer, Ubuntu Member

See http://wiki.debian.org/JulianAndresKlode and http://jak-linux.org/.

Attachment: pgpmSKTbqec0I.pgp
Description: PGP signature


--- End Message ---
Reply to: