Re: bug#7489: [coreutils] over aggressive threads in sort

[Paul Eggert]

On 11/26/2010 06:52 PM, Pádraig Brady wrote:
> Hmm, seems like multiple threads are racing to update the
> static "saved" variable in write_unique() ?

I don't think it's as simple as that.  write_unique
is generating output, and when it is run it is supposed
to have exclusive access to the output file, which means
it also has exclusive access to the "saved" variable.

I suspect the problem is that the line structure is
messed up when you have multiple threads and sufficiently
large input, in that when one merge node is done and
the next one is run, the "saved" variable (which points
to storage managed by the earlier node) is pointing to
storage that is no longer valid.  I haven't had time to
look into the details, though.

Chen, do you have some insight into this?  Here's the thread:

http://lists.gnu.org/archive/html/bug-coreutils/2010-11/msg00209.html

I have now reproduced the bug on RHEL 5.5 x86-64, using
coreutils 8.7 built with "configure CC='cc -m32'".  The bug
is also present in the savannah git master.