Skip to contentSkip to conversation

2.5.33-mm5

DISCUSSION Andrew Morton 4 participants 5 messages Sep 8, 2002 – Sep 9, 2002 Memory management

lore
Share & tools
lore
#0 lore raw

URL: http://www.zip.com.au/~akpm/linux/patches/2.5/2.5.33/2.5.33-mm5/

+refill-rate-fix.patch

Fix a problem in refill_inactive_zone() which could soak a lot of CPU.

+sleeping-release_page.patch

Allow mapped->releasepage() to sleep again. My passing in non-zero gfp_mask.

+filemap-integration-fixes.patch

Some fixes to the readv/writev rework.

Plus a lot of stabilisation, tuning and testing of the new VM latency control code. Including fixing one rarely-occurring infinite loop which might explain Steve Cole's reported failure.

Some testing with no swap has been performed as well. Works OK, and some speedups were made in this area (if there's no swap online, don't bring anon pages onto the inactive list).

It's looking pretty good now - the system is quite responsive under all heavy writeout workloads. It's still very latent under heavy swapout load; that is deliberate. It is latent when overloaded by dirty MAP_SHARED data. We can fix that.

A side-effect of the VM rework is an improvement in many-spindle pagecache writeout. This is the first kernel which can keep four queues saturated. I tested six disks - the LEDs never went out.

I'd appreciate it if people could grab this one, be nasty to it and send a report.

You will probably see increased CPU utilisation by kswapd. I believe that this is not an efficiency problem - it's due to kswapd doing more work that it used to, rather than sleeping on request queues all the time.

Also, pdflush appears to be taking more CPU, but profiling shows that it is not - this may be due to synchronisation with the CPU load accounting.

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

refill-rate-fix.patch
Don't call shrink_zone with a negative nr_pages

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.

sleeping-release_page.patch
Allow a_ops->releasepage() to sleep again

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

filemap-integration.patch
Clean up readv/writev

filemap-integration-fixes.patch
More readv/writev fixes

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

lpp.patch
ia32 huge tlb pages

lpp2.patch
hugetlbpage fixes

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

mmap-fixes.patch
mmap.c cleanup and lock ranking fixes

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-ext2-preread.patch
avoid ext2 inode prereads if the queue is congested

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

nonblocking-vm.patch
Non-blocking page reclaim

#1 lore raw

Hi Andrew!

Quoted 2 lines · Andrew Morton

I'd appreciate it if people could grab this one, be nasty to it and send a report.

What are your favorite tests to run? I'd like to send you some useful test results. But which do you like to see?

Best regards,
Axel Siebenwirth

#2 lore raw
Quoted 10 lines · Axel Siebenwirth

Hi Andrew!

Quoted 2 lines · Andrew Morton

I'd appreciate it if people could grab this one, be nasty to it and send a report.

What are your favorite tests to run? I'd like to send you some useful test results. But which do you like to see?

I've already run my favourite tests ;) The value of external testing is in the extra coverage which it gives - different hardware, different tests. And also different requirements: there may be things which I think are cool, but which you think suck.

So... The real test is of course "daily use". If it works OK in daily use for you, and for everyone else then we ship 2.6. By definition.

Of course, on top of daily use it is best to run additional stress tests to find problems more quickly. Large desktop applications, web and file servers, databases, etc would be interesting. CD burning, funny old PIO-mode IDE drives, stress testing with gigabt NICs, you name it. Coverage.

#3 lore raw
Quoted 6 lines · Andrew Morton
Quoted 5 lines · Axel Siebenwirth

Hi Andrew!

On Sat, 07 Sep 2002, Andrew Morton wrote:

Quoted 7 lines
Quoted 5 lines
Quoted 2 lines

I'd appreciate it if people could grab this one, be nasty to it and send a report.

What are your favorite tests to run? I'd like to send you some useful test results. But which do you like to see?

I've already run my favourite tests ;)

How about some swap-intensive comparisons to 2.4.19?

Signature · 1 line
Daniel
#4 lore raw
In reply to Andrew Morton
Quoted 16 lines · Andrew Morton

URL: http://www.zip.com.au/~akpm/linux/patches/2.5/2.5.33/2.5.33-mm5/

(quoted diff) +3 -0
+refill-rate-fix.patch
 
 Fix a problem in refill_inactive_zone() which could soak a lot of CPU.
 
+sleeping-release_page.patch
 
 Allow mapped->releasepage() to sleep again.  My passing in non-zero
 gfp_mask.
 
+filemap-integration-fixes.patch
 
 Some fixes to the readv/writev rework.
Quoted 3 lines

Plus a lot of stabilisation, tuning and testing of the new VM latency control code. Including fixing one rarely-occurring infinite loop which might explain Steve Cole's reported failure.

This looks pretty good so far. The test box has run up to 112 dbench clients successfully with 2.5.33-mm5, ext3 data=ordered, which is much better than before. Thanks.

...and there was much rejoicing.

Steven