Re: [PATCH] VFS bugfix: two read_inode() calles without clear_inode() call between

[David Woodhouse <dwmw2@infradead.org>]

On Thu, 2005-05-05 at 18:18 +0200, Miklos Szeredi wrote:
> Using yield() to wait for a precisely defined event (clear_inode()
> finishing) doesn't seem like a very good idea.  Especially, since
> Artem's patch will probably make it trigger more often.

Agreed. Even before Artem's patch, we're still effectively busy-waiting
for something which calls back into the file system's clear_inode method
and may well sleep and perform I/O.

> How about this (totally untested) patch?  Even if I_LOCK is not set
> initially, wake_up_inode() should do the right thing and wake up the
> waiting task after clear_inode().  It shouldn't cause spurious
> wakeups, since there should be no other reference to the inode.

Since Artem introduced a wake_up_inode() in dispose_list(), your patch
seems reasonable.

-- 
dwmw2

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/