git/list[1] front-page[2] threads[3] people[4] search[5] about
 

Re: Reproducible, corrupt packfile after fresh

From
Stefan Beller <stefanbeller@googlemail.com>
Date
Aug 8, 2013, 12:18 UTC
Message-ID
<52038CA9.5090602@googlemail.com>
In-Reply-To
<f1d8f80d6fa3678ac043bfdb19bebf6bf4261273@localhost>
On 08/08/2013 01:56 PM, gitml.jexpert@recursor.net wrote:
Show 59 quoted lines
> I'm a heavy user of git-svn and experience an issue with one specific
> (git-svn) repository: 'git fsck' reports a corrupt packfile after every
> checkout.
> 
> Now I'm totally puzzled about the cause and what do about it.
> This is what I do:
> 
>     git svn init -s http://svn.foo.com/myproject myproject.git
>     cd myproject.git
>     git svn fetch         # Much more reliable than 'git-svn clone'
> 
> This checks out ~2100 commits and executes 2 git-gc during checkout. The
> final .git repo size is about 940MB. Then I run
> 
>     git fsck
> 
> ➜ myproject.git git:(master) git fsck
> Checking object directories: 100% (256/256), done.
> error: packed 0f5f33639bfc1a781fe080c31a1f076d9a25c1d3 from
> .git/objects/pack/pack-6a6f5355584a5d71215d5fc867ce09602ceab533.pack is
> corrupt
> *** glibc detected *** git: free(): invalid pointer: 0x00007f46a09e9010 ***
> ======= Backtrace: =========
> /lib/x86_64-linux-gnu/libc.so.6(+0x7eb96)[0x7f46d9ebab96]
> git[0x4ddf46]
> git[0x4b4123]
> git[0x431524]
> git[0x405ce8]
> git[0x4050e2]
> /lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xed)[0x7f46d9e5d76d]
> git[0x405529]
> ======= Memory map: ========
> 00400000-0055f000 r-xp 00000000 fc:01 12452043
>  /usr/bin/git
> 0075e000-0075f000 r--p 0015e000 fc:01 12452043
>  /usr/bin/git
> 
> ....
> 
> 
> This only affects this very particular git-svn repo (and I have dozens
> of them). The error happens reproducible on every fresh checkout as
> described above. The backtrace does not appear always. The object is a
> very large blob
> 
>     git show 0f5f33639bfc1a781fe080c31a1f076d9a25c1d3 | wc -c
>     39524691
> 
> 
> Any hints what to do?
> 
> Thanks
> - Ben
> 
> --
> To unsubscribe from this list: send the line "unsubscribe git" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 
Hi,

Regarding the backtrace, would it be possible to install debug symbols, so the backtrace is a little more meaningful? If you distribution doesn't provide debug symbols, compiling git yourself is rather easy (git clone && make && make install [defaults to ~/bin, so not root required for installing apart from missing dependencies])

Stefan
Previous: Stefan Beller
Message 26 of 26 in “Reproducible, corrupt packfile after fresh”
  1. gitml.jexpert@recursor.netAug 8, 2013
  2. Thomas RastAug 8, 2013
  3. gitml.jexpert@recursor.netAug 8, 2013
  4. Thomas RastAug 8, 2013
  5. gitml.jexpert@recursor.netAug 8, 2013
  6. Matthieu MoyAug 8, 2013
  7. gitml.jexpert@recursor.netAug 8, 2013
  8. Matthieu MoyAug 8, 2013
  9. gitml.jexpert@recursor.netAug 8, 2013
  10. Thomas RastAug 8, 2013
  11. Ben TebulinAug 8, 2013
  12. Thomas RastAug 8, 2013
  13. Ben TebulinAug 8, 2013
  14. Matthieu MoyAug 8, 2013
  15. Junio C HamanoAug 8, 2013
  16. Ben TebulinAug 9, 2013
  17. Stefan BellerAug 9, 2013
  18. Reproducible, corrupt packfile after fresh git-svn checkouts - IDENTIFIEDBen Tebulin, Aug 14, 2013
  19. Thomas RastAug 14, 2013
  20. Ben TebulinAug 14, 2013
  21. Ben TebulinAug 15, 2013
  22. Junio C HamanoAug 15, 2013
  23. Thomas RastAug 8, 2013
  24. Matthieu MoyAug 8, 2013
  25. Stefan BellerAug 8, 2013
  26. Stefan BellerAug 8, 2013

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.