On Tue, Jan 16, 2001 at 12:10:31PM -0800, Theodore Y. Ts'o wrote:
> Actually, looking at the fast path of down_trylock compared to huge mess
> of code that's currently there, I actually suspect that using
> down_trylock() would actually be faster, since in the fast path case
> there would only two
Date: Tue, 16 Jan 2001 20:33:34 +0100
From: Andrea Arcangeli <[EMAIL PROTECTED]>
> Of course, this is utterly unsafe on an SMP machines, since access to
> the "block" variable isn't protected at all. So the first question is
Wrong, it's obviously protected by the inode_lock. And
On Tue, Jan 16, 2001 at 11:04:45AM -0800, Theodore Y. Ts'o wrote:
>
> HJ Lu recently pointed me at a potential locking problem
> try_to_free_inodes(), and when I started proding at it, I found what
> appears to be another set of SMP locking issues in the dcache code.
> (But if that were the case,
HJ Lu recently pointed me at a potential locking problem
try_to_free_inodes(), and when I started proding at it, I found what
appears to be another set of SMP locking issues in the dcache code.
(But if that were the case, why aren't we seeing huge numbers of
complaints? So I'm wondering if I'm m
4 matches
Mail list logo
