this locally.
Regards,
Dan
ioat: fix 'ack' handling, driver must ensure that 'ack' is zero
From: Dan Williams [EMAIL PROTECTED]
Initialize 'ack' to zero in case the descriptor has been recycled.
Signed-off-by: Dan Williams [EMAIL PROTECTED]
---
drivers/dma/ioat_dma.c |2 ++
1 files changed, 2
heheh.
it's really easy to reproduce the hang without the patch -- i could
hang the box in under 20 min on 2.6.22+ w/XFS and raid5 on 7x750GB.
i'll try with ext3... Dan's experiences suggest it won't happen with ext3
(or is even more rare), which would explain why this has is overall a
rare
do not see this change making the situation
any worse. In fact, it may make it a bit better since there is a
higher chance for the thread submitting i/o to MD to do its own i/o to
the backing disks.
Reviewed-by: Dan Williams [EMAIL PROTECTED]
-
To unsubscribe from this list: send the line
On Sun, 2007-12-30 at 10:58 -0700, dean gaudet wrote:
On Sat, 29 Dec 2007, Dan Williams wrote:
On Dec 29, 2007 1:58 PM, dean gaudet [EMAIL PROTECTED] wrote:
On Sat, 29 Dec 2007, Dan Williams wrote:
On Dec 29, 2007 9:48 AM, dean gaudet [EMAIL PROTECTED] wrote:
hmm bummer
On Jan 9, 2008 5:09 PM, Neil Brown [EMAIL PROTECTED] wrote:
On Wednesday January 9, [EMAIL PROTECTED] wrote:
On Sun, 2007-12-30 at 10:58 -0700, dean gaudet wrote:
i have evidence pointing to d89d87965dcbe6fe4f96a2a7e8421b3a75f634d1
On Wed, 2008-01-09 at 20:57 -0700, Neil Brown wrote:
So I'm incline to leave it as do as much work as is available to be
done as that is simplest. But I can probably be talked out of it
with a convincing argument
Well, in an age of CFS and CFQ it smacks of 'unfairness'. But does that
On Jan 7, 2008 6:44 AM, Radu Rendec [EMAIL PROTECTED] wrote:
I'm experiencing trouble when trying to add a new disk to a raid 1 array
after having replaced a faulty disk.
[..]
# mdadm --version
mdadm - v2.6.2 - 21st May 2007
[..]
However, this happens with both mdadm 2.6.2 and 2.6.4. I
: [EMAIL PROTECTED]
Cc: Dan Williams [EMAIL PROTECTED]
Signed-off-by: Neil Brown [EMAIL PROTECTED]
Acked-by: Dan Williams [EMAIL PROTECTED]
-
To unsubscribe from this list: send the line unsubscribe linux-raid in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http
now is safer long-term.
This bug exists in 2.6.23 and 2.6.24-rc
Cc: [EMAIL PROTECTED]
Cc: Dan Williams [EMAIL PROTECTED]
Signed-off-by: Neil Brown [EMAIL PROTECTED]
Acked-by: Dan Williams [EMAIL PROTECTED]
On closer look the safer test is:
!test_bit(STRIPE_OP_COMPUTE_BLK, sh
In case someone else happens upon this I have found that mdadm =
v2.6.2 cannot add a disk to a degraded raid1 array created with mdadm
2.6.2.
I bisected the problem down to mdadm git commit
2fb749d1b7588985b1834e43de4ec5685d0b8d26 which appears to make an
incompatible change to the super block's
On Dec 29, 2007 9:48 AM, dean gaudet [EMAIL PROTECTED] wrote:
hmm bummer, i'm doing another test (rsync 3.5M inodes from another box) on
the same 64k chunk array and had raised the stripe_cache_size to 1024...
and got a hang. this time i grabbed stripe_cache_active before bumping
the size
On Dec 29, 2007 1:58 PM, dean gaudet [EMAIL PROTECTED] wrote:
On Sat, 29 Dec 2007, Dan Williams wrote:
On Dec 29, 2007 9:48 AM, dean gaudet [EMAIL PROTECTED] wrote:
hmm bummer, i'm doing another test (rsync 3.5M inodes from another box) on
the same 64k chunk array and had raised
On Nov 23, 2007 11:19 AM, Joshua Johnson [EMAIL PROTECTED] wrote:
Greetings, long time listener, first time caller.
I recently replaced a disk in my existing 8 disk RAID 6 array.
Previously, all disks were PATA drives connected to the motherboard
IDE and 3 promise Ultra 100/133 controllers.
On Nov 14, 2007 5:05 PM, Justin Piszcz [EMAIL PROTECTED] wrote:
On Wed, 14 Nov 2007, Bill Davidsen wrote:
Justin Piszcz wrote:
This is a known bug in 2.6.23 and should be fixed in 2.6.23.2 if the RAID5
bio* patches are applied.
Note below he's running 2.6.22.3 which doesn't have the bug
[ Adding Neil, stable@, DaveJ, and GregKH to the cc ]
On Nov 13, 2007 11:20 AM, Peter [EMAIL PROTECTED] wrote:
Hi
I had a 3 disc raid5 array running fine with Fedora 7 (32bit) kernel
2.6.23.1-fc7 on an old Athlon XP using a two sata_sil cards.
I replaced the hardware with an Athlon64 X2
On Nov 13, 2007 5:23 PM, Greg KH [EMAIL PROTECTED] wrote:
On Tue, Nov 13, 2007 at 04:22:14PM -0800, Greg KH wrote:
On Mon, Oct 22, 2007 at 05:15:27PM +1000, NeilBrown wrote:
It appears that a couple of bugs slipped in to md for 2.6.23.
These two patches fix them and are appropriate for
void handle_stripe5(struct stripe_head *sh)
raid5-fix-unending-write-sequence.patch is in -mm and I believe is
waiting on an Acked-by from Neil?
thanks,
greg k-h
Thanks,
Dan
raid5: fix clearing of biofill operations
From: Dan Williams [EMAIL PROTECTED]
ops_complete_biofill() runs outside
From: Dan Williams [EMAIL PROTECTED]
debug output from Joël's system
handling stripe 7629696, state=0x14 cnt=1, pd_idx=2 ops=0:0:0
check 5: state 0x6 toread read write
f800ffcffcc0 written
check 4: state 0x6 toread read
, pending))
- ops_run_postxor(sh, tx);
+ ops_run_postxor(sh, tx, pending);
if (test_bit(STRIPE_OP_CHECK, pending))
ops_run_check(sh);
raid5: fix unending write sequence
From: Dan Williams [EMAIL PROTECTED]
---
drivers/md/raid5.c | 16
On 11/4/07, Justin Piszcz [EMAIL PROTECTED] wrote:
On Mon, 5 Nov 2007, Neil Brown wrote:
On Sunday November 4, [EMAIL PROTECTED] wrote:
# ps auxww | grep D
USER PID %CPU %MEMVSZ RSS TTY STAT START TIME COMMAND
root 273 0.0 0.0 0 0 ?DOct21
On 11/5/07, Justin Piszcz [EMAIL PROTECTED] wrote:
[..]
Are you seeing the same md thread takes 100% of the CPU that Joël is
reporting?
Yes, in another e-mail I posted the top output with md3_raid5 at 100%.
This seems too similar to Joël's situation for them not to be
correlated, and it
On 11/1/07, Yuri Tikhonov [EMAIL PROTECTED] wrote:
Hi Dan,
Honestly I tried to fix this quickly using the approach similar to proposed
by you, with one addition though (in fact, deletion of BUG_ON(chan ==
tx-chan) in async_tx_run_dependencies()). And this led to Kernel stack
overflow.
On 10/27/07, BERTRAND Joël [EMAIL PROTECTED] wrote:
Dan Williams wrote:
Can you collect some oprofile data, as Ming suggested, so we can maybe
see what md_d0_raid5 and istd1 are fighting about? Hopefully it is as
painless to run on sparc as it is on IA:
opcontrol --start --vmlinux
On 10/24/07, tirumalareddy marri [EMAIL PROTECTED] wrote:
Hi,
I am looking for best way of understanding MD
driver(including raid5/6) architecture. I am
developing driver for one of the PPC based SOC. I have
done some code reading and tried to use HW debugger to
walk through the code.
On 10/24/07, BERTRAND Joël [EMAIL PROTECTED] wrote:
Hello,
Any news about this trouble ? Any idea ? I'm trying to fix it, but I
don't see any specific interaction between raid5 and istd. Does anyone
try to reproduce this bug on another arch than sparc64 ? I only use
sparc32
On Fri, 2007-10-19 at 05:23 -0700, Yuri Tikhonov wrote:
Hello Dan,
Hi Yuri, sorry it has taken me so long to get back to you...
I have a suggestion regarding the async_tx_find_channel() procedure.
First, a little introduction. Some processors (e.g. ppc440spe) have several
DMA
On Fri, 2007-10-19 at 14:04 -0700, BERTRAND Joël wrote:
Sorry for this last mail. I have found another mistake, but I
don't
know if this bug comes from iscsi-target or raid5 itself. iSCSI target
is disconnected because istd1 and md_d0_raid5 kernel threads use 100%
of
CPU each !
On Fri, 2007-10-19 at 01:04 -0700, BERTRAND Joël wrote:
I never see any oops with this patch. But I cannot create a
RAID1 array
with a local RAID5 volume and a foreign RAID5 array exported by iSCSI.
iSCSI seems to works fine, but RAID1 creation randomly aborts due to a
unknown SCSI
On 10/17/07, Dan Williams [EMAIL PROTECTED] wrote:
On 10/17/07, BERTRAND Joël [EMAIL PROTECTED] wrote:
BERTRAND Joël wrote:
Hello,
I run 2.6.23 linux kernel on two T1000 (sparc64) servers. Each
server has a partitionable raid5 array (/dev/md/d0) and I have to
synchronize
into handle_stripe5 and adds some debug
information.
--
Dan
raid5: fix clearing of biofill operations (try2)
From: Dan Williams [EMAIL PROTECTED]
ops_complete_biofill() runs outside of spin_lock(sh-lock) and clears the
'pending' and 'ack' bits. Since the test_and_ack_op() macro only checks
against 'complete
On 10/13/07, Hod Greeley [EMAIL PROTECTED] wrote:
Hello,
I tried to create a raid device starting with
foo:~ 1032% mdadm --create -l1 -n2 /dev/md1 /dev/sda1 missing
mdadm: /dev/sda1 is too small: 0K
mdadm: create aborted
Quick sanity check, is /dev/sda1 still a block device node with
On Mon, 2007-10-08 at 23:21 -0700, Neil Brown wrote:
On Saturday October 6, [EMAIL PROTECTED] wrote:
Neil,
Here is the latest spin of the 'stripe_queue' implementation.
Thanks to
raid6+bitmap testing done by Mr. James W. Laferriere there have been
several cleanups and fixes since the
On 10/6/07, Justin Piszcz [EMAIL PROTECTED] wrote:
On Sat, 6 Oct 2007, Dan Williams wrote:
Neil,
Here is the latest spin of the 'stripe_queue' implementation. Thanks to
raid6+bitmap testing done by Mr. James W. Laferriere there have been
several cleanups and fixes since the last
.
Andrew,
These are updated in the git-md-accel tree, but I will work the
finalized versions through Neil's 'Signed-off-by' path.
Dan Williams (4):
raid5: add the stripe_queue object for tracking raid io requests (rev3)
raid5: split allocation of stripe_heads and stripe_queues
-off-by: Dan Williams [EMAIL PROTECTED]
---
drivers/md/raid5.c | 564 +++-
include/linux/raid/raid5.h | 28 +-
2 files changed, 364 insertions(+), 228 deletions(-)
diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index f96dea9..a13de7d 100644
. Laferriere [EMAIL PROTECTED]
Signed-off-by: Dan Williams [EMAIL PROTECTED]
---
drivers/md/raid5.c | 316
include/linux/raid/raid5.h | 11 +-
2 files changed, 239 insertions(+), 88 deletions(-)
diff --git a/drivers/md/raid5.c b/drivers/md
a stripe_head is attached.
Tested-by: Mr. James W. Laferriere [EMAIL PROTECTED]
Signed-off-by: Dan Williams [EMAIL PROTECTED]
---
drivers/md/raid5.c | 53
include/linux/raid/raid5.h |6 +
2 files changed, 40 insertions(+), 19 deletions
and release_queue to remove STRIPE_QUEUE_HANDLE and
sq-sh back references
* kill init_sh and allocate init_sq on the stack
Tested-by: Mr. James W. Laferriere [EMAIL PROTECTED]
Signed-off-by: Dan Williams [EMAIL PROTECTED]
---
drivers/md/raid5.c | 843
Linus, please pull from:
git://lost.foo-projects.org/~dwillia2/git/iop async-tx-fixes-for-linus
to receive:
Dan Williams (3):
async_tx: usage documentation and developer notes (v2)
async_tx: fix dma_wait_for_async_tx
raid5: fix 2 bugs in ops_complete_biofill
The raid5
Fix a couple bugs and provide documentation for the async_tx api.
Neil, please 'ack' patch #3.
git://lost.foo-projects.org/~dwillia2/git/iop async-tx-fixes-for-linus
Dan Williams (3):
async_tx: usage documentation and developer notes
async_tx: fix dma_wait_for_async_tx
raid5
Signed-off-by: Dan Williams [EMAIL PROTECTED]
---
Documentation/crypto/async-tx-api.txt | 217 +
1 files changed, 217 insertions(+), 0 deletions(-)
diff --git a/Documentation/crypto/async-tx-api.txt
b/Documentation/crypto/async-tx-api.txt
new file mode 100644
Fix dma_wait_for_async_tx to not loop forever in the case where a
dependency chain is longer than two entries. This condition will not
happen with current in-kernel drivers, but fix it for future drivers.
Found-by: Saeed Bishara [EMAIL PROTECTED]
Signed-off-by: Dan Williams [EMAIL PROTECTED
).
ops_complete_biofill can run in tasklet context, so rather than upgrading
all the stripe locks from spin_lock to spin_lock_bh this patch just moves
read completion handling back into handle_stripe.
Found-by: Yuri Tikhonov [EMAIL PROTECTED]
Signed-off-by: Dan Williams [EMAIL PROTECTED]
---
drivers/md/raid5.c
On 9/13/07, Yuri Tikhonov [EMAIL PROTECTED] wrote:
Hi Dan,
On Friday 07 September 2007 20:02, you wrote:
You need to fetch from the 'md-for-linus' tree. But I have attached
them as well.
git fetch git://lost.foo-projects.org/~dwillia2/git/iop
md-for-linus:md-for-linus
Thanks.
-for-linus:refs/heads/md-for-linus
raid5: fix the 'more_to_read' case in ops_complete_biofill
From: Dan Williams [EMAIL PROTECTED]
Prevent ops_complete_biofill from running concurrently with add_queue_bio
---
drivers/md/raid5.c | 33 +++--
1 files changed, 19
On 8/30/07, saeed bishara [EMAIL PROTECTED] wrote:
you are right, I've another question regarding the function
dma_wait_for_async_tx from async_tx.c, here is the body of the code:
/* poll through the dependency chain, return when tx is complete */
1.do {
2. iter
On 8/25/07, Mr. James W. Laferriere [EMAIL PROTECTED] wrote:
Hello Dan ,
On Mon, 20 Aug 2007, Dan Williams wrote:
On 8/18/07, Mr. James W. Laferriere [EMAIL PROTECTED] wrote:
Hello All , Here we go again . Again attempting to do bonnie++
testing
on a small array
On 8/26/07, Justin Piszcz [EMAIL PROTECTED] wrote:
On Sun, 26 Aug 2007, Abe Skolnik wrote:
Dear Mr./Dr./Prof. Brown et al,
I recently had the unpleasant experience of creating an MD array for
the purpose of booting off it and then not being able to do so. Since
I had already made
On 8/26/07, Abe Skolnik [EMAIL PROTECTED] wrote:
Dear Mr./Dr. Williams,
Just Dan is fine :-)
Because you can rely on the configuration file to be certain about
which disks to pull in and which to ignore. Without the config file
the auto-detect routine may not always do the right thing
(stripe-queue) , Dan Williams [EMAIL PROTECTED]
Hello James,
Thanks for the report.
I tried to reproduce this on my system, no luck. However it looks
like their is a potential race between 'handle_queue' and
'add_queue_bio'. The attached patch moves these critical sections
under spin_lock(sq-lock
[trimmed all but linux-raid from the cc]
On 7/30/07, Justin Piszcz [EMAIL PROTECTED] wrote:
CONFIG:
Software RAID 5 (400GB x 6): Default mkfs parameters for all filesystems.
Kernel was 2.6.21 or 2.6.22, did these awhile ago.
Can you give 2.6.22.1-iop1 a try to see what affect it has on
insertions(+), 407 deletions(-)
Dan Williams (2):
raid5: add the stripe_queue object for tracking raid io requests (take2)
raid5: use stripe_queues to prioritize the most deserving requests
(take4)
I initially considered them 2.6.24 material but after fixing the sync+io
data corruption
. Pre-patch
throughput hovers at ~85MB/s for this dd command.
Changes in take2:
* leave the flags with the buffers, prevents a data corruption issue
whereby stale buffer state flags are attached to newly initialized
buffers
Signed-off-by: Dan Williams [EMAIL PROTECTED]
---
drivers/md/raid5.c
Per Bill Davidsen's request I have made available a 2.6.22.1 based
kernel with the current raid5 performance changes I have been working
on:
1/ Offload engine acceleration (recently merged for the 2.6.23
development cycle)
2/ Stripe-queue, an evolutionary change to the raid5 queuing model (take4)
offload copies for TCP when there will be a context switch
Dan Aloni (1):
I/OAT: fix I/OAT for kexec
Dan Williams (20):
dmaengine: refactor dmaengine around dma_async_tx_descriptor
dmaengine: make clients responsible for managing channels
xor: make 'xor_blocks' a library
Neil, Andrew,
The following patches replace the stripe-queue patches currently in -mm.
Following your suggestion, Neil, I gathered blktrace data on the number
of reads generated by sequential write stimulus. It turns out that
reduced pre-reading is not the cause of the performance increase, but
compared to ~120MB/s without. Pre-patch
throughput hovers at ~85MB/s for this dd command.
Changes in take2:
* leave the flags with the buffers, prevents a data corruption issue
whereby stale buffer state flags are attached to newly initialized buffers
Signed-off-by: Dan Williams [EMAIL PROTECTED
The first take of the stripe-queue implementation[1] had a performance
limiting bug in __wait_for_inactive_queue. Fixing that issue
drastically changed the performance characteristics. The following data
from tiobench shows the relative performance difference of the
stripe-queue patchset.
Unit
[ trimmed the cc ]
On 6/26/07, Satyam Sharma [EMAIL PROTECTED] wrote:
Hi Dan,
[ Minor thing ... ]
Not a problem, thanks for taking a look...
On 6/27/07, Dan Williams [EMAIL PROTECTED] wrote:
The async_tx api tries to use a dma engine for an operation, but will fall
back to an optimized
Raz's stripe-deadline patch illuminated the fact that the current
queuing model leaves write performance on the table in some cases. The
following patches introduce a new queuing model which attempts to
recover this performance.
On an ARM based iop13xx platform I see an averaged %14.7 increase
to
pull for 2.6.23.
git://lost.foo-projects.org/~dwillia2/git/iop md-accel-linus
Dan Williams (19):
dmaengine: refactor dmaengine around dma_async_tx_descriptor
dmaengine: make clients responsible for managing channels
xor: make 'xor_blocks' a library routine for use
, set_dest, and tx_submit descriptor specific methods
Cc: Jeff Garzik [EMAIL PROTECTED]
Cc: Chris Leech [EMAIL PROTECTED]
Cc: Shannon Nelson [EMAIL PROTECTED]
Signed-off-by: Dan Williams [EMAIL PROTECTED]
---
drivers/dma/dmaengine.c | 182 ++
drivers/dma/ioatdma.c
xor_block = xor_blocks, suggested by Adrian Bunk
* ensure that xor.o initializes before md.o in the built-in case
* checkpatch.pl fixes
* mark calibrate_xor_blocks __init, Adrian Bunk
Cc: Adrian Bunk [EMAIL PROTECTED]
Cc: NeilBrown [EMAIL PROTECTED]
Cc: Herbert Xu [EMAIL PROTECTED]
Signed-off-by: Dan
share algorithms in the future
* move large inline functions into c files
* checkpatch.pl fixes
* gpl v2 only correction
Cc: Herbert Xu [EMAIL PROTECTED]
Signed-off-by: Dan Williams [EMAIL PROTECTED]
Acked-By: NeilBrown [EMAIL PROTECTED]
---
crypto/Kconfig |6
crypto/Makefile
Replaces PRINTK with pr_debug, and kills the RAID5_DEBUG definition in
favor of the global DEBUG definition. To get local debug messages just add
'#define DEBUG' to the top of the file.
Signed-off-by: Dan Williams [EMAIL PROTECTED]
---
drivers/md/raid5.c | 116
ops_complete_biofill
* remove test_and_set/test_and_clear BUG_ONs, Neil Brown
* remove explicit interrupt handling for channel switching, this feature
was absorbed (i.e. it is now implicit) by the async_tx api
Signed-off-by: Dan Williams [EMAIL PROTECTED]
Acked-By: NeilBrown [EMAIL PROTECTED]
---
drivers
and completion of operations.
Signed-off-by: Dan Williams [EMAIL PROTECTED]
Acked-By: NeilBrown [EMAIL PROTECTED]
---
drivers/md/raid5.c | 67 +---
1 files changed, 58 insertions(+), 9 deletions(-)
diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
the R5_Wantcompute flag there is no facility to pass the
async_tx dependency chain across successive calls to raid5_run_ops. The
req_compute variable protects against this case.
Changelog:
* remove the req_compute BUG_ON
Signed-off-by: Dan Williams [EMAIL PROTECTED]
Acked-By: NeilBrown [EMAIL PROTECTED
.
Changelog:
* remove test_and_set/test_and_clear BUG_ONs, Neil Brown
Signed-off-by: Dan Williams [EMAIL PROTECTED]
Acked-By: NeilBrown [EMAIL PROTECTED]
---
drivers/md/raid5.c | 84
1 files changed, 65 insertions(+), 19 deletions(-)
diff --git
arrive while raid5_run_ops is running they will not be handled until
handle_stripe is scheduled to run again.
Changelog:
* cleanup to_read and to_fill accounting
* do not fail reads that have reached the cache
Signed-off-by: Dan Williams [EMAIL PROTECTED]
Acked-By: NeilBrown [EMAIL PROTECTED
differentiate expand operations from normal write operations.
Signed-off-by: Dan Williams [EMAIL PROTECTED]
Acked-By: NeilBrown [EMAIL PROTECTED]
---
drivers/md/raid5.c | 50 ++
1 files changed, 38 insertions(+), 12 deletions(-)
diff --git a/drivers/md
replaced by raid5_run_ops
Signed-off-by: Dan Williams [EMAIL PROTECTED]
Acked-By: NeilBrown [EMAIL PROTECTED]
---
drivers/md/raid5.c | 124
1 files changed, 0 insertions(+), 124 deletions(-)
diff --git a/drivers/md/raid5.c b/drivers/md
.
The following new routines are shared between raid5 and raid6:
handle_completed_write_requests
handle_requests_to_failed_array
handle_stripe_expansion
Changes in v2:
* fixed 'conf-raid_disk-1' for the raid6 'handle_stripe_expansion' path
Signed-off-by: Dan Williams [EMAIL
error fix from Kirill A. Shutemov
* rebase for async_tx changes
* add interrupt support
* do not call platform register macros in driver code
* remove unnecessary ARM assembly statement
* checkpatch.pl fixes
* gpl v2 only correction
Cc: Russell King [EMAIL PROTECTED]
Signed-off-by: Dan Williams
]
Signed-off-by: Dan Williams [EMAIL PROTECTED]
---
arch/arm/mach-iop32x/glantank.c|2
arch/arm/mach-iop32x/iq31244.c |5
arch/arm/mach-iop32x/iq80321.c |3
arch/arm/mach-iop32x/n2100.c |2
arch/arm/mach-iop33x/iq80331.c |3
arch/arm
Cc: Russell King [EMAIL PROTECTED]
Signed-off-by: Dan Williams [EMAIL PROTECTED]
---
arch/arm/Kconfig |2 ++
1 files changed, 2 insertions(+), 0 deletions(-)
diff --git a/arch/arm/Kconfig b/arch/arm/Kconfig
index 50d9f3e..0cb2d4f 100644
--- a/arch/arm/Kconfig
+++ b/arch/arm/Kconfig
On 6/26/07, Mr. James W. Laferriere [EMAIL PROTECTED] wrote:
Hello Dan ,
On Tue, 26 Jun 2007, Dan Williams wrote:
Greetings,
Per Andrew's suggestion this is the md raid5 acceleration patch set
updated with more thorough changelogs to lower the barrier to entry for
reviewers
7. And now, the question: the best absolute 'write' performance comes
with a stripe_cache_size value of 4096 (for my setup). However, any
value of stripe_cache_size above 384 really, really hurts 'check' (and
rebuild, one can assume) performance. Why?
Question:
After performance goes bad does
Andrew's concerns about the commit messages.
Dan Williams (2):
raid5: refactor handle_stripe5 and handle_stripe6
raid5: replace custom debug print with standard pr_debug
-
To unsubscribe from this list: send the line unsubscribe linux-raid in
the body of a message to [EMAIL PROTECTED]
More
.
The following new routines are shared between raid5 and raid6:
handle_completed_write_requests
handle_requests_to_failed_array
handle_stripe_expansion
Signed-off-by: Dan Williams [EMAIL PROTECTED]
---
drivers/md/raid5.c | 1484
Replaces PRINTK with pr_debug, and kills the RAID5_DEBUG definition in
favor of the global DEBUG definition. To get local debug messages just add
'#define DEBUG' to the top of the file.
Signed-off-by: Dan Williams [EMAIL PROTECTED]
---
drivers/md/raid5.c | 116
From: Dan Williams [EMAIL PROTECTED]
Document the overloading of struct bio fields.
Signed-off-by: Dan Williams [EMAIL PROTECTED]
---
[ drop this if you think it is too much commenting/unnecessary, but I figured I
would leave some
breadcrumbs for the next guy. ]
drivers/md/raid5.c | 26
I am pleased to release this latest spin of the raid acceleration
patches for merge consideration. This release aims to address all
pending review items including MD bug fixes and async_tx api changes
from Neil, and concerns on channel management from Chris and others.
Data integrity tests using
to interoperate with async_tx
calls
* hookup ioat_tx_submit
* convert channel capabilities to a 'cpumask_t like' bitmap
Cc: Chris Leech [EMAIL PROTECTED]
Signed-off-by: Dan Williams [EMAIL PROTECTED]
---
drivers/dma/dmaengine.c | 182 +
drivers/dma/ioatdma.c
to ignore
a channel if it does not meet extra client specific constraints beyond
simple base capabilities.
This patch also fixes up the NET_DMA client to use the new mechanism.
Cc: Chris Leech [EMAIL PROTECTED]
Signed-off-by: Dan Williams [EMAIL PROTECTED]
---
drivers/dma/dmaengine.c | 206
Cc: Russell King [EMAIL PROTECTED]
Signed-off-by: Dan Williams [EMAIL PROTECTED]
---
arch/arm/Kconfig |2 ++
1 files changed, 2 insertions(+), 0 deletions(-)
diff --git a/arch/arm/Kconfig b/arch/arm/Kconfig
index e7baca2..74077e3 100644
--- a/arch/arm/Kconfig
+++ b/arch/arm/Kconfig
counts = 1
* implicitly handle hardware concerns like channel switching and
interrupts, Neil Brown
* remove the per operation type list, and distribute operation capabilities
evenly amongst the available channels
* simplify async_tx_find_channel to optimize the fast path
Signed-off-by: Dan Williams
* remove test_and_set/test_and_clear BUG_ONs, Neil Brown
* remove explicit interrupt handling
Signed-off-by: Dan Williams [EMAIL PROTECTED]
---
drivers/md/raid5.c | 539
include/linux/raid/raid5.h | 63 +
2 files changed, 599 insertions
to clear 'pending' and 'ack'.
Signed-off-by: Dan Williams [EMAIL PROTECTED]
---
drivers/md/raid5.c | 65 +---
1 files changed, 56 insertions(+), 9 deletions(-)
diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index 0251bca..14e9f6a 100644
* remove test_and_set/test_and_clear BUG_ONs, Neil Brown
Signed-off-by: Dan Williams [EMAIL PROTECTED]
---
drivers/md/raid5.c | 151 +---
1 files changed, 130 insertions(+), 21 deletions(-)
diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index
.
Changelog:
* remove the req_compute BUG_ON
Signed-off-by: Dan Williams [EMAIL PROTECTED]
---
drivers/md/raid5.c | 126 +++-
1 files changed, 94 insertions(+), 32 deletions(-)
diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index 03a435d
-by: Dan Williams [EMAIL PROTECTED]
---
drivers/md/raid5.c | 80
1 files changed, 61 insertions(+), 19 deletions(-)
diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index 844bd9b..f8a4522 100644
--- a/drivers/md/raid5.c
+++ b/drivers/md/raid5.c
Use raid5_run_ops to carry out the memory copies for a raid5 read request.
Changelog:
* cleanup to_read and to_fill accounting
* do not fail reads that have reached the cache
Signed-off-by: Dan Williams [EMAIL PROTECTED]
---
drivers/md/raid5.c | 61
.
The bulk copy operation to the new stripe is handled inline by async_tx.
Signed-off-by: Dan Williams [EMAIL PROTECTED]
---
drivers/md/raid5.c | 48
1 files changed, 36 insertions(+), 12 deletions(-)
diff --git a/drivers/md/raid5.c b/drivers/md
handle_stripe now only updates the state of stripes. All execution of
operations is moved to raid5_run_ops.
Signed-off-by: Dan Williams [EMAIL PROTECTED]
---
drivers/md/raid5.c | 68
1 files changed, 10 insertions(+), 58 deletions(-)
diff
replaced by raid5_run_ops
Signed-off-by: Dan Williams [EMAIL PROTECTED]
---
drivers/md/raid5.c | 124
1 files changed, 0 insertions(+), 124 deletions(-)
diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index c9b91e3..74ce354 100644
* convert capabilities over to dma_cap_mask_t
Signed-off-by: Dan Williams [EMAIL PROTECTED]
---
drivers/dma/Kconfig |8
drivers/dma/Makefile|1
drivers/dma/iop-adma.c | 1464 +++
include/asm-arm/hardware
error fix from Kirill A. Shutemov
* rebase for async_tx changes
* add interrupt support
* do not call platform register macros in driver code
Cc: Russell King [EMAIL PROTECTED]
Signed-off-by: Dan Williams [EMAIL PROTECTED]
---
arch/arm/mach-iop13xx/setup.c | 208
include/asm
boards
* do not call platform register macros in driver code
* remove switch() statements for compatible register offsets/layouts
* change over to bitmap based capabilities
Cc: Russell King [EMAIL PROTECTED]
Signed-off-by: Dan Williams [EMAIL PROTECTED]
---
arch/arm/mach-iop32x/glantank.c
On 4/30/07, Jan Engelhardt [EMAIL PROTECTED] wrote:
Hi list,
when a user does `mdadm -C /dev/md0 -l any -n whatever fits
devices`, the array gets rebuilt for at least RAID1 and RAID5, even if
the disk contents are most likely not of importance (otherwise we would
not be creating a raid array
1 - 100 of 226 matches
Mail list logo
