* 2.5.66-mm2
@ 2003-04-01 8:01 Andrew Morton
2003-04-01 8:48 ` 2.5.66-mm2 Andrew Morton
2003-04-02 1:10 ` 2.5.66-mm2 Martin J. Bligh
0 siblings, 2 replies; 6+ messages in thread
From: Andrew Morton @ 2003-04-01 8:01 UTC (permalink / raw)
To: linux-kernel, linux-mm
ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.5/2.5.66/2.5.66-mm2/
. Just lots of little fixes, cleanups, late-but-promised minor features,
etc.
. There is a small patch from Ingo here against the CPU scheduler which we
hope will fix the new starvation problems which people have been reporting.
I this is you, please test and report.
. It turns out that a recent change to the anticipatory scheduler
accidentally made it quite ineffective on SCSI (performance is similar to
deadline). The patch which did that has been disabled, but we expect that
the seeky OLTP loads will suffer until this is fixed up for real.
Changes since 2.5.66-mm1:
linus.patch
Latest from Linus
-initcall-debug.patch
-posix-timers-64-bit-fix.patch
-slab-off-by-one-fix.patch
-install_page-flush_cache_page.patch
-ext3-fsync-speedup.patch
-pcmcia-2.patch
-pcmcia-3b.patch
-pcmcia-3.patch
-pcmcia-4.patch
-pcmcia-5.patch
-pcmcia-6.patch
-pcmcia-7b.patch
-pcmcia-7.patch
-pcmcia-8.patch
-pcmcia-9.patch
-pcmcia-10.patch
-posix-timer-double-expiration-fix.patch
-hugh-01-no-SWAP_ERROR.patch
-hugh-02-try_to_unmap-CONFIG_SWAP.patch
-hugh-03-add_to_swap_cache.patch
-hugh-09-tmpfs-truncation.patch
-hugh-10-tmpfs-atomics.patch
-hugh-12-vm_enough_memory-double-counts.patch
-ext3-max-file-size-fix.patch
-linear-oops-fix-1.patch
-dev_t-drm-warnings.patch
-x86-clock-override-option.patch
-tty_io-cleanup.patch
-page_to_pfn-in-blk_queue_bounce.patch
-init_inode_once-bloat-fix.patch
-VM_DONTEXPAND-fix.patch
-compound-page-warning-fix.patch
-cdevname-irq-safety-fix.patch
-register_chrdev_region-leak-fix.patch
-slab-cache-sizes-cleanup.patch
-sync_blockdev-on-final-close.patch
Merged
+as-queue_notready-cleanup.patch
Anticipatory scheduler cleanups.
+as-disable-thinktime.patch
Anticipatory scheduler simply doesn't work on scsi. Disable the thinktime
heuristic to make it go again (this will cause slowdowns for OLTP loads).
+cdrom-stack-usage.patch
Reduced stack usage
+sched-interactivity-backboost-revert.patch
Revert the CPU scheduler backboost heuristic.
+tmpfs-blk_congestion_wait-fix.patch
tmpfs fixlet.
+page_convert_anon-locking-fix.patch
Fix locking for objrmap.
+monotonic-clock-hangcheck.patch
get_cycles() clock source, used in the hangcheck timer.
+module_load_notification.patch
Provide notification of module load/unload to other places in the kernel.
+put_task_struct-debug.patch
Try to find out who is doing put_task_struct() on a freed task_struct.
+remove-kdev_name.patch
Remove kdevname()
+percpu_counter.patch
+blockgroup_lock.patch
Broken out from the ext2 speedup patches.
+kill-TIOCTTYGSTRUCT.patch
Remove the TIOCTTYGSTRUCT ioctl.
+misc.patch
Misc fixes
+sony-apm-fix.patch
Fix APM for Sony notebooks.
+init-sections-in-kallsyms.patch
Make kallsyms aware of __init symbols.
+3c59x-980-support.patch
Extra PCI IDs for 3c59x
+fadvise-flush-data.patch
Teach fadvise(FADV_DONTNEED) to gently flush out dirty pages.
+console-scrollback.patch
Another TIOCLINUX command for scrolling the console down.
+usb-disconnect-crash-fix.patch
Fix some USB crash (this isn't right)
+devfs-rescan_partitions-fix.patch
Fix a devfs problem
+umsdos-fixes.patch
Fixes
+exp_parent-locking-fix.patch
Fix nfsd locking
+real_lookup-race-fix.patch
Fix a fixed race differently
+remove-dparent_lock.patch
Remove the global dparent_lock
+PCI-aliases-fix.patch
Stuff from Rusty.
+jbd_expect.patch
+jbd-assert-io-failure-fix.patch
Forward-port from Stephen's recent batch of ext3 patches.
+jbd-handle-journal-io-errors.patch
+jbd-handle-journal-io-errors-fix.patch
Propagate IO errors against the ext3 journal back to user syscalls, mainly
fsync().
+ext3-concurrent-block-allocation-hashed.patch
Change the ext3 BKL-removal code to use the hashed locking and
percpu_counters from ext2.
All 99 patches:
linus.patch
mm.patch
add -mmN to EXTRAVERSION
kgdb-ga.patch
kgdb stub for ia32 (George Anzinger's one)
ppa-null-pointer-fix.patch
config_spinline.patch
uninline spinlocks for profiling accuracy.
ppc64-reloc_hide.patch
ppc64-pci-patch.patch
Subject: pci patch
ppc64-aio-32bit-emulation.patch
32/64bit emulation for aio
ppc64-scruffiness.patch
Fix some PPC64 compile warnings
sym-do-160.patch
make the SYM driver do 160 MB/sec
config-PAGE_OFFSET.patch
Configurable kenrel/user memory split
ptrace-flush.patch
cache flushing in the ptrace code
buffer-debug.patch
buffer.c debugging
warn-null-wakeup.patch
ext3-truncate-ordered-pages.patch
ext3: explicitly free truncated pages
reiserfs_file_write-5.patch
rcu-stats.patch
RCU statistics reporting
ext3-journalled-data-assertion-fix.patch
Remove incorrect assertion from ext3
nfs-speedup.patch
nfs-oom-fix.patch
nfs oom fix
sk-allocation.patch
Subject: Re: nfs oom
nfs-more-oom-fix.patch
rpciod-atomic-allocations.patch
Make rcpiod use atomic allocations
linux-isp.patch
isp-update-1.patch
kblockd.patch
Create `kblockd' workqueue
as-iosched.patch
anticipatory I/O scheduler
as-np-reads-1.patch
AS: read-vs-read fixes
as-np-reads-2.patch
AS: more read-vs-read fixes
as-predict-data-direction.patch
as: predict direction of next IO
as-remove-frontmerge.patch
AS: remove frontmerge tunable
as-misc-cleanups.patch
AS: misc cleanups
as-minor-tweaks.patch
AS: tuning and tweaks
as-remove-stats.patch
AS: remove statistics
as-queue_notready-cleanup.patch
don't dispatch request on elv_queue_empty
as-disable-thinktime.patch
cfq-2.patch
CFQ scheduler, #2
unplug-use-kblockd.patch
Use kblockd for running request queues
fremap-all-mappings.patch
Make all executable mappings be nonlinear
cdrom-stack-usage.patch
reduce stack in cdrom/optcd.c
objrmap-2.5.62-5.patch
object-based rmap
sched-2.5.64-D3.patch
sched-2.5.64-D3, more interactivity changes
scheduler-tunables.patch
scheduler tunables
show_task-free-stack-fix.patch
show_task() fix and cleanup
yellowfin-set_bit-fix.patch
yellowfin driver set_bit fix
htree-nfs-fix.patch
Fix ext3 htree / NFS compatibility problems
task_prio-fix.patch
simple task_prio() fix
slab_store_user-large-objects.patch
slab debug: perform redzoning against larger objects
htree-nfs-fix-2.patch
htree nfs fix
hugh-04-page_convert_anon-ENOMEM.patch
swap 04/13 page_convert_anon -ENOMEM
hugh-05-page_convert_anon-unlocking.patch
swap 05/13 page_convert_anon unlocking
hugh-06-wrap-below-vm_start.patch
swap 06/13 wrap below vm_start
hugh-07-objrmap-page_table_lock.patch
swap 07/13 objrmap page_table_lock
hugh-08-rmap-comments.patch
swap 08/13 rmap comments
hugh-11-fix-unuse_pmd-fixme.patch
swap 11/13 fix unuse_pmd fixme
tmpfs-blk_congestion_wait-fix.patch
tmpfs blk_congestion_wait fix
page_convert_anon-locking-fix.patch
page_convert_anon locking fix
monotonic-clock-hangcheck.patch
monotonic clock source for hangcheck timer
module_load_notification.patch
module load notification
put_task_struct-debug.patch
remove-kdev_name.patch
remove kdevname() before someone starts using it again
percpu_counter.patch
percpu_counters: approximate but scalabel counters
blockgroup_lock.patch
blockgroup_lock: hashed spinlocks for ext2 and ext3 blockgroup locking
ext2-no-lock_super-ng.patch
ext2-ialloc-no-lock_super-ng.patch
stat_t-larger-dev_t.patch
struct stat - support larger dev_t
kill-TIOCTTYGSTRUCT.patch
kill TIOCTTYGSTRUCT
dev_t-32-bit.patch
[for playing only] change type of dev_t
dev_t-remove-B_FREE.patch
dev_t: eliminate B_FREE
sg-dev_t-fix.patch
32-bit dev_t fix for sg
misc.patch
misc fixes
sony-apm-fix.patch
fix ec_read using wrong #define's in sonypi driver.
init-sections-in-kallsyms.patch
Put all functions in kallsyms
3c59x-980-support.patch
Additional 3c980 device support
fadvise-flush-data.patch
oops-dump-preceding-code.patch
i386 oops output: dump preceding code
console-scrollback.patch
add vt console scrollback ioctl
usb-disconnect-crash-fix.patch
Subject: Re: [linux-usb-devel] timer hang with current 2.5 BK
devfs-rescan_partitions-fix.patch
Fix devfs' partition handling
umsdos-fixes.patch
umsdos fixes
exp_parent-locking-fix.patch
exp_parent locking fixes
real_lookup-race-fix.patch
real_lookup race fix
remove-dparent_lock.patch
remove dparent_lock
conntrack-use-after-free-fix.patch
fix use-after-free in ip_conntrack
PCI-aliases-fix.patch
Fix PCI aliases.
jbd_expect.patch
Add less-severe assert-failure form for ext3.
jbd-assert-io-failure-fix.patch
Fix jbd assert failure on IO error.
acpi-build-fix.patch
ACPI build fix
ext3_mark_inode_dirty-speedup.patch
ext3_mark_inode_dirty() speedup
ext3_mark_inode_dirty-less-calls.patch
ext3_commit_write speedup
ext3-handle-cache.patch
ext3: create a slab cache for transaction handles
jbd-handle-journal-io-errors.patch
ext3 journal commit I/O error fix
jbd-handle-journal-io-errors-fix.patch
ext3-no-bkl.patch
journal_dirty_metadata-speedup.patch
journal_get_write_access-speedup.patch
ext3-concurrent-block-inode-allocation.patch
Subject: [PATCH] concurrent block/inode allocation for EXT3
ext3-concurrent-block-allocation-fix-1.patch
ext3-concurrent-block-allocation-hashed.patch
Subject: Re: [PATCH] concurrent block/inode allocation for EXT3
--
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/ .
Don't email: <a href=mailto:"aart@kvack.org">aart@kvack.org</a>
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: 2.5.66-mm2
2003-04-01 8:01 2.5.66-mm2 Andrew Morton
@ 2003-04-01 8:48 ` Andrew Morton
2003-04-01 9:15 ` 2.5.66-mm2 Andrew Morton
2003-04-02 1:10 ` 2.5.66-mm2 Martin J. Bligh
1 sibling, 1 reply; 6+ messages in thread
From: Andrew Morton @ 2003-04-01 8:48 UTC (permalink / raw)
To: linux-kernel, linux-mm
Andrew Morton <akpm@digeo.com> wrote:
>
>
> ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.5/2.5.66/2.5.66-mm2/
oops. 2.5.66-mm2.gz is missing lots of patches.
Please use 2.5.66-mm2-1.gz from that directory.
--
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/ .
Don't email: <a href=mailto:"aart@kvack.org">aart@kvack.org</a>
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: 2.5.66-mm2
2003-04-01 8:48 ` 2.5.66-mm2 Andrew Morton
@ 2003-04-01 9:15 ` Andrew Morton
0 siblings, 0 replies; 6+ messages in thread
From: Andrew Morton @ 2003-04-01 9:15 UTC (permalink / raw)
To: linux-kernel, linux-mm
Andrew Morton <akpm@digeo.com> wrote:
>
> Andrew Morton <akpm@digeo.com> wrote:
> >
> >
> > ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.5/2.5.66/2.5.66-mm2/
>
> oops. 2.5.66-mm2.gz is missing lots of patches.
>
> Please use 2.5.66-mm2-1.gz from that directory.
And now kernel.org mirroring is having another sulk.
There's a copy at
http://www.zip.com.au/~akpm/linux/patches/2.5/2.5.66-mm2-1.gz
--
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/ .
Don't email: <a href=mailto:"aart@kvack.org">aart@kvack.org</a>
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: 2.5.66-mm2
2003-04-01 8:01 2.5.66-mm2 Andrew Morton
2003-04-01 8:48 ` 2.5.66-mm2 Andrew Morton
@ 2003-04-02 1:10 ` Martin J. Bligh
2003-04-02 1:34 ` 2.5.66-mm2 Andrew Morton
1 sibling, 1 reply; 6+ messages in thread
From: Martin J. Bligh @ 2003-04-02 1:10 UTC (permalink / raw)
To: Andrew Morton, Alex Tomas; +Cc: linux-kernel, linux-mm
> +ext3-concurrent-block-allocation-hashed.patch
>
> Change the ext3 BKL-removal code to use the hashed locking and
> percpu_counters from ext2.
Ho hum. All very strange. Kernbench seems to be really behaving itself
quite well now, but SDET sucks worse than ever. The usual 16x NUMA-Q
machine ....
Kernbench: (make -j N vmlinux, where N = 2 x num_cpus)
Elapsed System User CPU
2.5.66-mm2 44.04 81.12 569.40 1476.75
2.5.66-mm2-ext3 44.43 84.10 568.82 1469.00
Kernbench: (make -j N vmlinux, where N = 16 x num_cpus)
Elapsed System User CPU
2.5.66-mm2 44.36 89.64 575.45 1499.25
2.5.66-mm2-ext3 44.79 91.15 575.48 1488.25
Kernbench: (make -j vmlinux, maximal tasks)
Elapsed System User CPU
2.5.66-mm2 44.06 87.03 574.31 1501.00
2.5.66-mm2-ext3 44.35 89.33 573.47 1495.75
DISCLAIMER: SPEC(tm) and the benchmark name SDET(tm) are registered
trademarks of the Standard Performance Evaluation Corporation. This
benchmarking was performed for research purposes only, and the run results
are non-compliant and not-comparable with any published results.
Results are shown as percentages of the first set displayed
SDET 1 (see disclaimer)
Throughput Std. Dev
2.5.66-mm2 100.0% 1.9%
2.5.66-mm2-ext3 92.6% 1.8%
SDET 2 (see disclaimer)
Throughput Std. Dev
2.5.66-mm2 100.0% 0.0%
2.5.66-mm2-ext3 88.4% 5.1%
SDET 4 (see disclaimer)
Throughput Std. Dev
2.5.66-mm2 100.0% 1.9%
2.5.66-mm2-ext3 26.3% 7.8%
SDET 8 (see disclaimer)
Throughput Std. Dev
2.5.66-mm2 100.0% 1.0%
2.5.66-mm2-ext3 8.0% 3.1%
SDET 16 (see disclaimer)
Throughput Std. Dev
2.5.66-mm2 100.0% 1.0%
2.5.66-mm2-ext3 5.6% 1.6%
SDET 32 (see disclaimer)
Throughput Std. Dev
2.5.66-mm2 100.0% 0.7%
2.5.66-mm2-ext3 4.7% 1.5%
SDET 64 (see disclaimer)
Throughput Std. Dev
2.5.66-mm2 100.0% 0.7%
2.5.66-mm2-ext3 3.6% 1.9%
SDET 128 (see disclaimer)
Throughput Std. Dev
2.5.66-mm2 100.0% 0.6%
2.5.66-mm2-ext3 3.9% 0.4%
diffprofile for SDET 128 (+ worse with ext3, - better)
10410044 1594.1% total
8799596 2479.9% default_idle
649656 15845.3% __down
347530 0.0% .text.lock.sched
218233 8660.0% schedule
183618 9984.7% __wake_up
117605 0.0% .text.lock.transaction
30896 0.0% do_get_write_access
18162 1319.9% .text.lock.attr
14815 0.0% journal_dirty_metadata
13655 0.0% journal_get_write_access
12762 79762.5% cpu_idle
11023 7252.0% __blk_queue_bounce
9404 0.0% start_this_handle
9310 0.0% journal_add_journal_head
5566 0.0% block_write_full_page
5478 0.0% journal_stop
4210 0.0% ext3_get_inode_loc
4127 668.9% __find_get_block_slow
3852 1965.3% .text.lock.sem
3660 0.0% ext3_mark_iloc_dirty
3630 184.5% __find_get_block
3360 239.8% __brelse
2911 0.0% ext3_do_update_inode
2672 16700.0% scsi_request_fn
2565 0.0% journal_start
2432 0.0% journal_commit_transaction
1945 0.0% journal_unlock_journal_head
1937 0.0% ext3_new_inode
1820 0.0% ext3_orphan_del
1731 13315.4% __make_request
1606 1784.4% .text.lock.ioctl
1591 103.9% .text.lock.base
1575 0.0% find_next_usable_block
1563 142.2% __block_prepare_write
1483 0.0% ext3_orphan_add
1428 0.0% __journal_unfile_buffer
1367 0.0% journal_get_undo_access
1365 156.7% kmap_atomic
1360 1766.2% default_wake_function
1344 40.5% do_anonymous_page
1232 0.0% ext3_reserve_inode_write
1180 64.7% current_kernel_time
1172 0.0% journal_invalidatepage
1139 136.4% kmalloc
1121 37366.7% mempool_alloc
1017 0.0% __journal_file_buffer
...
-1080 -55.4% d_alloc
-1105 -67.4% filemap_nopage
-1153 -100.0% ext2_new_inode
-1182 -37.3% pte_alloc_one
-1186 -46.4% copy_process
-1470 -98.4% .text.lock.highmem
-1496 -95.2% .text.lock.file_table
-1555 -64.2% file_move
-1596 -94.4% __read_lock_failed
-1601 -61.8% path_release
-1602 -100.0% grab_block
-1649 -51.0% copy_mm
-2413 -57.4% remove_shared_vm_struct
-2540 -78.9% free_pages_and_swap_cache
-3236 -93.1% .text.lock.namei
-3600 -47.5% page_add_rmap
-4671 -72.9% path_lookup
-4720 -95.8% .text.lock.dcache
-4921 -42.9% __d_lookup
-5140 -86.9% follow_mount
-6005 -39.7% copy_page_range
-6534 -72.0% release_pages
-6905 -94.2% .text.lock.dec_and_lock
-7158 -74.2% atomic_dec_and_lock
-7307 -47.3% page_remove_rmap
-9399 -48.2% zap_pte_range
Looks just like horrific semaphore contention to me.
I dumped the full results to:
ftp://ftp.kernel.org/pub/linux/kernel/people/mbligh/benchmarks/2.5.66-mm2/
ftp://ftp.kernel.org/pub/linux/kernel/people/mbligh/benchmarks/2.5.66-mm2-ext3/
as you probably want the gory detail, knowing you ;-)
M.
--
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/ .
Don't email: <a href=mailto:"aart@kvack.org">aart@kvack.org</a>
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: 2.5.66-mm2
2003-04-02 1:10 ` 2.5.66-mm2 Martin J. Bligh
@ 2003-04-02 1:34 ` Andrew Morton
2003-04-02 15:34 ` 2.5.66-mm2 Martin J. Bligh
0 siblings, 1 reply; 6+ messages in thread
From: Andrew Morton @ 2003-04-02 1:34 UTC (permalink / raw)
To: Martin J. Bligh; +Cc: bzzz, linux-kernel, linux-mm
"Martin J. Bligh" <mbligh@aracnet.com> wrote:
>
> Ho hum. All very strange. Kernbench seems to be really behaving itself
> quite well now, but SDET sucks worse than ever. The usual 16x NUMA-Q
> machine ....
>
> Kernbench: (make -j N vmlinux, where N = 2 x num_cpus)
> Elapsed System User CPU
> 2.5.66-mm2 44.04 81.12 569.40 1476.75
> 2.5.66-mm2-ext3 44.43 84.10 568.82 1469.00
Is this ext2 versus ext3? If so, that's a pretty good result isn't it? I
forget what kernbench looked like for stock ext3.
> SDET 32 (see disclaimer)
> Throughput Std. Dev
> 2.5.66-mm2 100.0% 0.7%
> 2.5.66-mm2-ext3 4.7% 1.5%
Yes, this is presumably a lot more metadata-intensive, so we're just
hammering the journal semaphore to death. We're working on it.
> ftp://ftp.kernel.org/pub/linux/kernel/people/mbligh/benchmarks/2.5.66-mm2-ext3/
Offtopic, a raw sdet64 profile says:
5392317 total
4478683 default_idle
307163 __down
169770 .text.lock.sched
106769 schedule
88092 __wake_up
57280 .text.lock.transaction
I'm slightly surprised that the high context switch rate is showing up so
much contention in sched.c. I'm assuming that it's on the sleep/wakeup path
and not in the context switch path. It would be interesting to inline the
spinlock code and reprofile.
We really should be using the waker-removes-wakee facility in the semaphore
code, but that's not completely trivial.
--
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/ .
Don't email: <a href=mailto:"aart@kvack.org">aart@kvack.org</a>
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: 2.5.66-mm2
2003-04-02 1:34 ` 2.5.66-mm2 Andrew Morton
@ 2003-04-02 15:34 ` Martin J. Bligh
0 siblings, 0 replies; 6+ messages in thread
From: Martin J. Bligh @ 2003-04-02 15:34 UTC (permalink / raw)
To: Andrew Morton; +Cc: bzzz, linux-kernel, linux-mm
>> Ho hum. All very strange. Kernbench seems to be really behaving itself
>> quite well now, but SDET sucks worse than ever. The usual 16x NUMA-Q
>> machine ....
>>
>> Kernbench: (make -j N vmlinux, where N = 2 x num_cpus)
>> Elapsed System User CPU
>> 2.5.66-mm2 44.04 81.12 569.40 1476.75
>> 2.5.66-mm2-ext3 44.43 84.10 568.82 1469.00
>
> Is this ext2 versus ext3? If so, that's a pretty good result isn't it? I
> forget what kernbench looked like for stock ext3.
Yes, it's splendid. Used to look more like this:
Kernbench: (make -j N vmlinux, where N = 2 x num_cpus)
Elapsed System User CPU
2.5.61-mjb0.1 46.04 115.46 563.07 1472.25
2.5.61-mjb0.1-ext3 48.45 143.79 564.14 1459.00
That was before I had noatime though (I think) ...
Kernbench: (make -j N vmlinux, where N = 2 x num_cpus)
Elapsed System User CPU
2.5.65-mjb1 43.73 81.69 563.54 1475.00
2.5.65-mjb1-ext3 44.13 79.77 564.56 1460.25
So I think after noatime, SDET was really the big problem.
>> SDET 32 (see disclaimer)
>> Throughput Std. Dev
>> 2.5.66-mm2 100.0% 0.7%
>> 2.5.66-mm2-ext3 4.7% 1.5%
>
> Yes, this is presumably a lot more metadata-intensive, so we're just
> hammering the journal semaphore to death. We're working on it.
Ah, that makes sense, thanks.
>> ftp://ftp.kernel.org/pub/linux/kernel/people/mbligh/benchmarks/2.5.66-mm2-ext3/
>
> Offtopic, a raw sdet64 profile says:
>
> 5392317 total
> 4478683 default_idle
> 307163 __down
> 169770 .text.lock.sched
> 106769 schedule
> 88092 __wake_up
> 57280 .text.lock.transaction
>
> I'm slightly surprised that the high context switch rate is showing up so
> much contention in sched.c. I'm assuming that it's on the sleep/wakeup path
> and not in the context switch path. It would be interesting to inline the
> spinlock code and reprofile.
OK, done. diffprofile with an without spinlines below:
350487 158.8% schedule
67691 0.6% total
49184 0.5% default_idle
43095 1680.1% journal_start
42048 136.1% do_get_write_access
34988 638.7% journal_stop
16813 1527.1% inode_change_ok
6752 3.6% __wake_up
5960 0.9% __down
3942 3718.9% sem_exit
3323 725.5% inode_setattr
3191 1470.5% proc_pid_readlink
1646 2743.3% sys_ioctl
418 16.8% atomic_dec_and_lock
408 4.4% journal_add_journal_head
303 841.7% proc_root_lookup
258 4.6% __find_get_block
256 33.0% follow_mount
245 1.7% journal_dirty_metadata
223 4.7% __find_get_block_slow
213 2366.7% chrdev_open
209 1492.9% proc_root_readdir
202 2.6% find_get_page
171 8.8% journal_unlock_journal_head
146 6.2% kmem_cache_free
126 2.7% do_anonymous_page
120 1.3% copy_page_range
111 1387.5% sys_sysctl
107 93.9% journal_get_create_access
106 0.8% cpu_idle
106 963.6% __posix_lock_file
101 280.6% put_filp
100 1666.7% de_put
...
-102 -13.1% fget
-118 -100.0% .text.lock.char_dev
-120 -2.2% block_write_full_page
-123 -20.0% block_prepare_write
-127 -93.4% remove_from_page_cache
-142 -100.0% .text.lock.sysctl
-167 -1.5% __blk_queue_bounce
-183 -27.8% do_generic_mapping_read
-203 -11.7% free_hot_cold_page
-205 -100.0% .text.lock.dcache
-211 -13.0% buffered_rmqueue
-237 -98.8% .text.lock.namei
-426 -100.0% .text.lock.dec_and_lock
-458 -100.0% .text.lock.root
-516 -8.8% __copy_to_user_ll
-781 -100.0% .text.lock.journal
-1696 -100.0% .text.lock.ioctl
-3123 -100.0% .text.lock.base
-4048 -100.0% .text.lock.sem
-19538 -100.0% .text.lock.attr
-117523 -99.9% .text.lock.transaction
-347530 -100.0% .text.lock.sched
Thanks,
M.
--
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/ .
Don't email: <a href=mailto:"aart@kvack.org">aart@kvack.org</a>
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2003-04-02 15:34 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2003-04-01 8:01 2.5.66-mm2 Andrew Morton
2003-04-01 8:48 ` 2.5.66-mm2 Andrew Morton
2003-04-01 9:15 ` 2.5.66-mm2 Andrew Morton
2003-04-02 1:10 ` 2.5.66-mm2 Martin J. Bligh
2003-04-02 1:34 ` 2.5.66-mm2 Andrew Morton
2003-04-02 15:34 ` 2.5.66-mm2 Martin J. Bligh
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox