Re: [patch] lowlatency-2.4.0-test6-E2

From: Andrew Morton (andrewm@uow.edu.au)
Date: Sat Aug 12 2000 - 09:55:00 EST


Ingo Molnar wrote:
>
> the newest version of the lowlatency patch can be downloaded from:
>
> http://www.redhat.com/~mingo/lowlatency-patches/lowlatency-2.4.0-test6-E2

Comments and testing results:

- Measured a 2 millisec hit in lat_tcp and 4 millisec in bw_tcp.
  This was in tasklet context :(

- The 5 millisec pc_keyb X startup thing is still there. My vote is
  to leave it alone.

- You need to add a reschedule in ipc/shm.c:shm_free(). Quintela's
  shm-stress will show why.

- mmap002 is killed by the VM when running `amlat' at 1024 Hz to
  create scheduling pressure.

  This is changed behaviour: it does not happen without this patch.

- In swap_out_mm() it is not correct to restart the vma scan after
  scheduling. If the current vma will take over a millisec to scan,
  and there is a process being scheduled once per millisec, swap_out_mm
  will never terminate. This is fairly easy to demonstrate with
  mmap002.

  The fix is to resume the scan from `vma->vm_start'.

- mmap002 sometimes hangs when run under scheduling pressure.
  Probably due to the above problem.

- Running Quintela's ipc001 and then running mmap002 causes 8
  millisec scheduling holes.

- `lilo' hangs when run under scheduling pressure if there are a lot
  of dirty blocks on the device. fsync_dev() changes.

- If you run bonnie++ to create lots of slab cache entries and then
  run mmap002, the resulting call to kmem_cache_reap() causes basically
  unbounded scheduling delays. I observed ~6 millisecs. I didn't fix
  this either.

- The patch significantly worsens bonnie++ figures. The overall
  execution time went from 9:04 to 9:23 when run during 1024 Hz
  scheduling pressure. From 8:44 to 8:52 when run without
  scheduling pressure.

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



This archive was generated by hypermail 2b29 : Tue Aug 15 2000 - 21:00:27 EST