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

Re: *Not a* ext2fs file corruption bug, but another toolchain issue


On Mon, 13 Jun 2011 16:05:31 +0200, I wrote:
> On Sat, 11 Jun 2011 19:26:33 +0200, I wrote:
> > spent several hours on something that looks like a ext2fs translator file
> > corruption bug -- or hopefully Git bug, but it unfortunately rather
> > points into the direction of ext2fs...  :-/
> > 
> > Some (a random handful out of a thousand, roughly) files' content is
> > getting duplicated: for these files, the pattern generally is that, once
> > the file regularely ends, its content appears once more.  That is, the
> > files' content appears once (regularely), and then the same again.  Has
> > someone ever seen something like that before?
> It's not ext2fs's fault, but Git's -- or rather: another toolchain
> issues, as it seems.  If I recompile the faulty Debian git
> package on my box, the issue goes away!

Oh no!  It's not gone, but just doesn't occur anymore as often as it did
before...  I'll analyze Git's ``git checkout'' source code.

I've never seen any inconsistencies using and repeating these commands:

    $ rm -rf ar_master*
    $ (cd git/ && git archive master) | (mkdir ar_master && cd ar_master/ && tar -x)
    $ diff -x .git -ru tar_master/ ar_master/


    $ (cd git/ && git archive master) | md5sum

Thus, the ``upper layers'' of Git seem to be fine.

> Additionally,
> <http://www.bddebian.com:8888/gitweb/?p=hurd-web;a=blob;f=open_issues/git-core-2.mdwn;hb=21f7f80cd190b56a14e1faf2ba24c7137d3faef5>
> goes away, too -- which I have filed in December 2008, so this can't
> really have been a recent toolchain hiccup?!  Strange.  (Perhaps I'm
> speaking too early with respect to that one, but I'm not able to
> reproduce it anymore at the moment using the same Git commands that
> quasi-reliably failed earlier on.)

This really didn't show up anymore -- so far...


Attachment: pgpuGVZN5mVb8.pgp
Description: PGP signature

Reply to: