In article <[EMAIL PROTECTED]>,
Anton Blanchard <[EMAIL PROTECTED]> wrote:
>
>It was not surprising the BKL was one of the main offenders. Looking at the
>stats ext2_get_block was the bad guy (UTIL is % of time lock was busy for,
>WAIT is time spent waiting for lock):
Actually, I find the BKL fairly surprising - we've whittled down all the
major non-lowlevel-FS offenders, and I didn't realize that it's still
there in do_exit().
And the do_exit() case should be _trivial_ to fix: almost none of the
code protected by the kernel lock in the exit path actually needs the
lock. I suspect you could cut down the kernel lock there to much
smaller.
The big case seems to be ext2_get_block(), we'll fix that early in
2.5.x. I think Al already has patches for it.
As to lseek, that one should probably get the inode semaphore, not the
kernel lock.
Linus
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
