Re: [PATCH] avoid possible overflow in delta size filtering computation

[Nicolas Pitre <nico@cam.org>]

On Wed, 25 Mar 2009, Kjetil Barvik wrote:

> Nicolas Pitre <nico@cam.org> writes:
> 
> > On a 32-bit system, the maximum possible size for an object is less than 
> > 4GB, while 64-bit systems may cope with larger objects.  Due to this 
> > limitation, variables holding object sizes are using an unsigned long 
> > type (32 bits on 32-bit systems, or 64 bits on 64-bit systems).
> >
> > When large objects are encountered, and/or people play with large delta 
> > depth values, it is possible for the maximum allowed delta size 
> > computation to overflow, especially on a 32-bit system.  When this 
> > occurs, surviving result bits may represent a value much smaller than 
> > what it is supposed to be, or even zero.  This prevents some objects 
> > from being deltified although they do get deltified when a smaller depth 
> > limit is used.  Fix this by always performing a 64-bit multiplication.
> >
> > Signed-off-by: Nicolas Pitre <nico@cam.org>
> 
>   I added this patch and rerun the 2 test cases form the table where
>   --depth is 20000 and 95000, and got the following result:
> 
>     --depth=20000 => file size: 19126077  delta: 73814
>     --depth=95000 => file size: 19126087  delta: 73814
> 
>   So, it seems that this patch almost fixed the issue.  But notice that
>   the pack file was 10 bytes larger for the --depth=95000 case.
> 
>   I made a small perl script to compare the output from 'git verify-pack
>   -v' of the 2 idx/pack files, and found the following difference(1)
>   (first line from --depth=20000 case, second from --depth=95000):
> 
>   fe0a6f3e971373590714dbafd087b235ea60ac00  tree   9  19  18921247  731  96a3ec5789504e6d0f90c99fb1937af1ebd58e2d
>   fe0a6f3e971373590714dbafd087b235ea60ac00  tree  20  29  18921247  730  12e560f7fb28558b15e3a2008fba860f9a4b2222

OK.  Apparently, a different base object for that one delta was chosen 
between those two runs.

Is your machine SMP?


Nicolas