Re: 2.5.65-mm4

[Andrew Morton]

"Martin J. Bligh" <[e-mail]> wrote:
>
> profile from SDET 64:

SDET is rather irritating because a) nobody has a copy and b) we don't even
know what it does.

> 
> 82303 __down
> 42835 schedule
> 31323 __wake_up
> 26435 .text.lock.sched
> 15924 .text.lock.transaction

But judging by this, it's a rebadged dbench.  The profile is identical.

Note that the lock_kernel() contention has been drastically reduced and we're
now hitting semaphore contention.

Running `dbench 32' on the quad Xeon, this patch took the context switch rate
from 500/sec up to 125,000/sec.

I've asked Alex to put together a patch for spinlock-based locking in the
block allocator (cut-n-paste from ext2).

That will fix up lock_super(), but I suspect the main problem is the
lock_journal() in journal_start().  I haven't thought about that one yet.