2.5.33-mm2

2.5.33-mm2

[Andrew Morton]

- Linus has merged ia32 NUMA discontigmem support

+ Added a little cleanup patch from various folks.


Threw in the kichen sink:

+writeback-control.patch

  Infrastructure for richer communication between the block layer
  and the VM.

+queue-congestion.patch

  Infrastructure for non-blocking writeout in the block layer.

+nonblocking-pdflush.patch

  Non-blocking background writeback

+nonblocking-vm.patch

  Non-blocking page reclaim.

This is all about reducing latency when the machine is performing heavy
writeback, which has been a significant performance problem for ever.
The code also happens to provide improved scalability in many-spindle
pagecache writeback.

The code is stable, but by no means complete.  Under some loads it will
chew tons of CPU in page reclaim.

But with mem=512m and four instances of `dbench 100' each against a
different disk the machine was 100% responsive and ran a `make -j6
bzImage' in three minutes.  Without these patches the kernel took over
five minutes just to unpack the kernel tarball.



linus.patch
  cset-1.575-to-1.600.txt.gz

scsi_hack.patch
  Fix block-highmem for scsi

ext3-htree.patch
  Indexed directories for ext3

zone-pages-reporting.patch
  Fix the boot-time reporting of each zone's available pages

enospc-recovery-fix.patch
  Fix the __block_write_full_page() error path.

fix-faults.patch
  Back out the initial work for atomic copy_*_user()

spin-lock-check.patch
  spinlock/rwlock checking infrastructure

refill-rate.patch
  refill the inactive list more quickly

copy_user_atomic.patch

kmap_atomic_reads.patch
  Use kmap_atomic() for generic_file_read()

kmap_atomic_writes.patch
  Use kmap_atomic() for generic_file_write()

throttling-fix.patch
  Fix throttling of heavy write()rs.

dirty-state-accounting.patch
  Make the global dirty memory accounting more accurate

rd-cleanup.patch
  Cleanup and fix the ramdisk driver (doesn't work right yet)

discontig-cleanup-1.patch
  i386 discontigmem coding cleanups

discontig-cleanup-2.patch
  i386 discontigmem cleanups

writeback-thresholds.patch
  Downward adjustments to the default dirtymemory thresholds

buffer-strip.patch
  Limit the consumption of ZONE_NORMAL by buffer_heads

rmap-speedup.patch
  rmap pte_chain space and CPU reductions

wli-highpte.patch
  Resurrect CONFIG_HIGHPTE - ia32 pagetables in highmem

readv-writev.patch
  O_DIRECT support for readv/writev

slablru.patch
  age slab pages on the LRU

slablru-speedup.patch
  slablru optimisations

llzpr.patch
  Reduce scheduling latency across zap_page_range

buffermem.patch
  Resurrect buffermem accounting

config-PAGE_OFFSET.patch
  Configurable kenrel/user memory split

lpp.patch
  ia32 huge tlb pages

ext3-sb.patch
  u.ext3_sb -> generic_sbp

oom-fix.patch
  Fix an OOM condition on big highmem machines

tlb-cleanup.patch
  Clean up the tlb gather code

dump-stack.patch
  arch-neutral dump_stack() function

wli-cleanup.patch
  random cleanups

madvise-move.patch
  move mdavise implementation into mm/madvise.c

split-vma.patch
  VMA splitting patch

buffer-ops-move.patch
  Move submit_bh() and ll_rw_block() into fs/buffer.c

writeback-control.patch
  Cleanup and extension of the writeback paths

queue-congestion.patch
  Infrastructure for communicating request queue congestion to the VM

nonblocking-pdflush.patch
  non-blocking writeback infrastructure, use it for pdflush

nonblocking-vm.patch
  Non-blocking page reclaim
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/

Re: 2.5.33-mm2

[Stephen C. Tweedie]

Hi,

On Wed, Sep 04, 2002 at 02:32:51AM -0700, Andrew Morton wrote:
> hm.  I sent this email to lkml and linux-mm.  The one on lkml had
> a URL at the top.  But linux-mm chopped it off.

Not quite --- linux-mm turned it into a header. :-)

--Stephen
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/

Re: 2.5.33-mm2

[Andrew Morton]

hm.  I sent this email to lkml and linux-mm.  The one on lkml had
a URL at the top.  But linux-mm chopped it off.

http://www.zip.com.au/~akpm/linux/patches/2.5/2.5.33/2.5.33-mm2/
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/

Re: 2.5.33-mm2

[Adam Kropelin]

Seems to need this patch to satisfy the compiler gremlins...

--Adam

--- linux-2.5.33-mm2.orig/mm/vmalloc.c	Wed Sep  4 11:42:50 2002
+++ linux-2.5.33-mm2/mm/vmalloc.c	Wed Sep  4 11:38:53 2002
@@ -12,6 +12,7 @@
 #include <linux/slab.h>
 #include <linux/spinlock.h>
 #include <linux/vmalloc.h>
+#include <linux/interrupt.h>
 
 #include <asm/uaccess.h>
 #include <asm/pgalloc.h>

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/

Re: 2.5.33-mm1

[Paul Larson]

I havn't tried this with stock 2.5.33 or 2.5.33-mm2 yet, but I was
trying the old fork07 ltp test and got a problem when I was testing
mm1.  The fork bomb part of that test is now in the fork12 test in LTP
and is not run by runalltests anymore due to the recent kernel changes. 
Here's the ksymoops output for now, and I'll see about trying to
reproduce it.

Thanks,
Paul Larson

Unable to handle kernel NULL pointer dereference at virtual address
00000004
c0131ef0
*pde = 28d4e001
Oops: 0002
CPU:    5
EIP:    0060:[<c0131ef0>]    Not tainted
Using defaults from ksymoops -t elf32-i386 -a i386
EFLAGS: 00010046
eax: 00000000   ebx: c17a67b0   ecx: 00000000   edx: 00000000
esi: e8cd3000   edi: 00000000   ebp: cd0f8660   esp: e8d53d20
ds: 0068   es: 0068   ss: 0068
Stack: 00000001 e8d53de8 c17a67b0 c17a67b0 0000000f c0133b2d c17a67b0
00000001
       00000000 00000020 ed9fea24 00000000 c17ad5d8 c181aaa8 00000000
00000000
       e8dcf320 00000000 00000001 00000005 00000014 00000005 c01115f0
00000000
Call Trace: [<c0133b2d>] [<c01115f0>] [<c0109636>] [<c0107fba>]
[<c0134056>]
   [<c013cdda>] [<c013cd30>] [<c0134687>] [<c01346f1>] [<c013473e>]
[<c01353c3>]
   [<c0135712>] [<c0135790>] [<c0117747>] [<c01181e6>] [<c0107fba>]
[<c0118a43>]
   [<c0115b4d>] [<c01115f0>] [<c0105d47>] [<c01075f3>]
Code: 89 50 04 89 02 c7 06 00 00 00 00 c7 46 04 00 00 00 00 d3 24

>>EIP; c0131ef0 <kmem_shrink_slab+40/b0>   <=====
Trace; c0133b2d <shrink_list+cd/440>
Trace; c01115f0 <smp_apic_timer_interrupt+e0/120>
Trace; c0109636 <do_IRQ+f6/110>
Trace; c0107fba <apic_timer_interrupt+1a/20>
Trace; c0134056 <shrink_cache+1b6/320>
Trace; c013cdda <wakeup_bdflush+1a/20>
Trace; c013cd30 <background_writeout+0/90>
Trace; c0134687 <shrink_zone+87/c0>
Trace; c01346f1 <shrink_caches+31/50>
Trace; c013473e <try_to_free_pages+2e/70>
Trace; c01353c3 <balance_classzone+43/200>
Trace; c0135712 <__alloc_pages+192/200>
Trace; c0135790 <__get_free_pages+10/20>
Trace; c0117747 <dup_task_struct+17/80>
Trace; c01181e6 <copy_process+56/890>
Trace; c0107fba <apic_timer_interrupt+1a/20>
Trace; c0118a43 <do_fork+23/b0>
Trace; c0115b4d <schedule+33d/370>
Trace; c01115f0 <smp_apic_timer_interrupt+e0/120>
Trace; c0105d47 <sys_fork+17/30>
Trace; c01075f3 <syscall_call+7/b>
Code;  c0131ef0 <kmem_shrink_slab+40/b0>
00000000 <_EIP>:
Code;  c0131ef0 <kmem_shrink_slab+40/b0>   <=====
   0:   89 50 04                  mov    %edx,0x4(%eax)   <=====
Code;  c0131ef3 <kmem_shrink_slab+43/b0>
   3:   89 02                     mov    %eax,(%edx)
Code;  c0131ef5 <kmem_shrink_slab+45/b0>
   5:   c7 06 00 00 00 00         movl   $0x0,(%esi)
Code;  c0131efb <kmem_shrink_slab+4b/b0>
   b:   c7 46 04 00 00 00 00      movl   $0x0,0x4(%esi)
Code;  c0131f02 <kmem_shrink_slab+52/b0>
  12:   d3 24 00                  shll   %cl,(%eax,%eax,1)


--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/

Re: 2.5.33-mm1

[Andrew Morton]

Paul Larson wrote:
> 
> I havn't tried this with stock 2.5.33 or 2.5.33-mm2 yet, but I was
> trying the old fork07 ltp test and got a problem when I was testing
> mm1.  The fork bomb part of that test is now in the fork12 test in LTP
> and is not run by runalltests anymore due to the recent kernel changes.
> Here's the ksymoops output for now, and I'll see about trying to
> reproduce it.
> 
> ..
> >>EIP; c0131ef0 <kmem_shrink_slab+40/b0>   <=====


hm.  We seem to have a corrupted slabp->list.  I don't recall any
slab fixes post 2.3.33-mm1.  hm.

Questions, please: how much physical memory, how many CPUs?
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/

Re: 2.5.33-mm1

[Paul Larson]

On Wed, 2002-09-04 at 13:02, Andrew Morton wrote:
> hm.  We seem to have a corrupted slabp->list.  I don't recall any
> slab fixes post 2.3.33-mm1.  hm.
> 
> Questions, please: how much physical memory, how many CPUs?

8 CPU PIII-700
16 GB physical ram


--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/