2001-03-09 15:42:56

by Chris Mason

Subject: 2.4.2-ac calls FS truncate w/o BKL

The added vmtruncate calls in the ac series trigger calls to the FS
truncate without the BKL held. Easy enough to fix on the reiserfs side,
but if other filesystems care we might want to change vmtruncate to grab
the lock before calling truncate (and update the Locking doc ;-)

-chris

2001-03-09 17:29:19

by Alexander Viro

Subject: Re: 2.4.2-ac calls FS truncate w/o BKL

On Fri, 9 Mar 2001, Chris Mason wrote:

>
> The added vmtruncate calls in the ac series trigger calls to the FS
> truncate without the BKL held. Easy enough to fix on the reiserfs side,
> but if other filesystems care we might want to change vmtruncate to grab
> the lock before calling truncate (and update the Locking doc ;-)

Grr... Thanks for spotting - it's my fault. No, I don't think that
we should move BKL into filesystems right now. Should move it into
vmtruncate(), though...

Alan, could you put lock_kernel()/unlock_kernel() around the call of
->truncate() in vmtruncate()?
Thanks,
Al