Re: [PATCH] Do _not_ call unlink on a directory

From: Linus Torvalds <torvalds@linux-foundation.org>
Date: Jul 16, 2007, 19:58 UTC
Message-ID: <alpine.LFD.0.999.0707161252330.20061@woody.linux-foundation.org>
In-Reply-To: <11846059721204-git-send-email-sithglan@stud.uni-erlangen.de>

On Mon, 16 Jul 2007, Thomas Glanzmann wrote:

>
> Calling unlink on a directory on a Solaris UFS filesystem as root makes it
> inconsistent. Thanks to Johannes Sixt for the obvious fix.

Ack, I think this is the right thing to do.

As pointed out, it doesn't _guarantee_ that git won't call "unlink()" on a directory (race conditions etc), but that's fundamentally true (there is no "funlink()" like there is "fstat()"), and besides, that is in no way git-specific (ie it's true of *any* application that gets run as root).

The theoretical race would only happen if somebody on purpose tries to screw things over, it would never happen under any reasonable usage.

The old ordering of those tests was designed for sane operating systems, so that you could basically do the unlink() without bothering, but switching the order around is certainly not a disaster either, and if it avoids the nasty bug in Solaris it's worth doing.

I have to say that I'm still a bit shocked that Solaris would have that kind of behaviour. And they call that pile of sh*t "enterprise class"..

		Linus