Re: [PATCH] don't use mmap() to hash files

From: Dmitry Potapov <dpotapov@gmail.com>
Date: Feb 14, 2010, 02:18 UTC
Message-ID: <20100214021847.GA9704@dpotapov.dyndns.org>
In-Reply-To: <7vtytk61im.fsf@alter.siamese.dyndns.org>

On Sat, Feb 13, 2010 at 05:37:21PM -0800, Junio C Hamano wrote:

Show 12 quoted lines

> Dmitry Potapov <dpotapov@gmail.com> writes:
> 
> > If a mmapped file is changed by another program during git-add, it
> > causes the repository corruption. Disabling mmap() in index_fd() does
> > not have any negative impact on the overall speed of Git. In fact, it
> > makes git hash-object to work slightly faster....
> > ...
> > I think more people should test this change to see its impact on
> > performance. For me, it was positive. Here is my results:
> 
> I wasn't particularly impressed by the original problem description, but
> this is a very interesting result.

My initial reaction was to write that the whole problem is due to abuse Git for the purposes that it was not intended. But then I decided to do some testing to see what impact it has. And, because I do not see any negative impact (in fact, slightly improve speed), and I decided to ask other people (who are interested in this patch) to do more testing.

Dmitry