md: support for asynchronous execution of RAID6 operations
v3 cover md: support for asynchronous execution of RAID6 operations
Hello,
This is the next attempt on asynchronous RAID-6 support. This patch-set
has the Dan Williams' comments (Dec, 17) addressed with the following exception:
- I still think that using 'enum dma_ctrl_flags' for PQ-specific operations is better than introducing another group of flags and enhance the device_prep_dma_pq()/device_prep_dma_pqzero_sum() with one more parameter. If unchanged 'enum dma_ctrl_flags' will be a criteria of acceptance, please let me know - I'll re-implement this exactly as Dan suggested.
Fearing to look like a spammer, I post only those patches, which had
been affected by the changes intended to address Dan's comments. These are the following five:
0002-async_tx-add-support-for-asynchronous-GF-multiplica.patch 0003-async_tx-add-support-for-asynchronous-RAID6-recover.patch 0004-md-run-RAID-6-stripe-operations-outside-the-lock.patch 0008-md-asynchronous-handle_parity_check6.patch 0011-ppc440spe-adma-ADMA-driver-for-the-PPC440SP-e-syst.patch
As regarding the other six patches of asynchronous RAID-6 support
patchset:
0001-async_tx-don-t-use-src_list-argument-of-async_xor.patch 0005-md-common-schedule_reconstruction-for-raid5-6.patch 0006-md-change-handle_stripe_fill6-to-work-in-asynchrono.patch 0007-md-rewrite-handle_stripe_dirtying6-in-asynchronous.patch 0009-md-change-handle_stripe6-to-work-asynchronously.patch 0010-md-remove-unused-functions.patch
they are the same as in my 09 Dec 2008 post:
https://kerneltrap.org/mailarchive/linux-raid/2008/12/8/4367574
and are waiting for your comments.
Regards, Yuri
Quoted 6 lines · Yuri Tikhonov
Hello,
This is the next attempt on asynchronous RAID-6 support. This patch-set
has the Dan Williams' comments (Dec, 17) addressed with the following exception:
Quoted 6 lines
- I still think that using 'enum
dma_ctrl_flags' for PQ-specific operations is better than introducing another group of flags and enhance the device_prep_dma_pq()/device_prep_dma_pqzero_sum() with one more parameter. If unchanged 'enumdma_ctrl_flags' will be a criteria of acceptance, please let me know - I'll re-implement this exactly as Dan suggested.
Ignore that Dan character :-), dma_ctrl_flags is the proper place for this functionality.
However, I do not believe we need/want ASYNC_TX_ZERO_{P,Q}. I will reply to 0002 and 0003 with details.
Quoted 15 lines
Fearing to look like a spammer, I post only those patches, which had
been affected by the changes intended to address Dan's comments. These are the following five:0002-async_tx-add-support-for-asynchronous-GF-multiplica.patch 0003-async_tx-add-support-for-asynchronous-RAID6-recover.patch 0004-md-run-RAID-6-stripe-operations-outside-the-lock.patch 0008-md-asynchronous-handle_parity_check6.patch 0011-ppc440spe-adma-ADMA-driver-for-the-PPC440SP-e-syst.patch
As regarding the other six patches of asynchronous RAID-6 support
patchset:
Quoted 8 lines
0001-async_tx-don-t-use-src_list-argument-of-async_xor.patch 0005-md-common-schedule_reconstruction-for-raid5-6.patch 0006-md-change-handle_stripe_fill6-to-work-in-asynchrono.patch 0007-md-rewrite-handle_stripe_dirtying6-in-asynchronous.patch 0009-md-change-handle_stripe6-to-work-asynchronously.patch 0010-md-remove-unused-functions.patch
they are the same as in my 09 Dec 2008 post:
Do you have a git tree where you can post this series? That would make it easier for me to track/review.
Thanks,
Dan
Hello Dan,
On Wednesday, January 14, 2009 you wrote:
[..]
Quoted 2 lines
Do you have a git tree where you can post this series? That would make it easier for me to track/review.
Yes.
Please see the "raidstuff" branch in the linux-2.6-denx repository:
http://git.denx.de/?p=linux-2.6-denx.git;a=shortlog;h=refs/heads/raidstuff
Regards, Yuri
--
Yuri Tikhonov, Senior Software Engineer
Emcraft Systems, www.emcraft.com