Re: [PATCH v2] use dynamically allocated sense buffer

2008-01-17 Thread FUJITA Tomonori
On Thu, 17 Jan 2008 09:58:11 -0600 James Bottomley <[EMAIL PROTECTED]> wrote: > > On Thu, 2008-01-17 at 18:13 +0900, FUJITA Tomonori wrote: > > On Wed, 16 Jan 2008 14:35:50 +0200 > > Benny Halevy <[EMAIL PROTECTED]> wrote: > > > > > On Jan. 15, 2008, 17:20 +0200, James Bottomley <[EMAIL PROTECTE

Re: [PATCH v2] use dynamically allocated sense buffer

2008-01-17 Thread James Bottomley
On Thu, 2008-01-17 at 18:13 +0900, FUJITA Tomonori wrote: > On Wed, 16 Jan 2008 14:35:50 +0200 > Benny Halevy <[EMAIL PROTECTED]> wrote: > > > On Jan. 15, 2008, 17:20 +0200, James Bottomley <[EMAIL PROTECTED]> wrote: > > > On Tue, 2008-01-15 at 18:23 +0900, FUJITA Tomonori wrote: > > >> This is t

Re: [PATCH v2] use dynamically allocated sense buffer

2008-01-17 Thread FUJITA Tomonori
On Wed, 16 Jan 2008 14:35:50 +0200 Benny Halevy <[EMAIL PROTECTED]> wrote: > On Jan. 15, 2008, 17:20 +0200, James Bottomley <[EMAIL PROTECTED]> wrote: > > On Tue, 2008-01-15 at 18:23 +0900, FUJITA Tomonori wrote: > >> This is the second version of > >> > >> http://marc.info/?l=linux-scsi&m=1199336

Re: [PATCH v2] use dynamically allocated sense buffer

2008-01-16 Thread Benny Halevy
On Jan. 15, 2008, 17:20 +0200, James Bottomley <[EMAIL PROTECTED]> wrote: > On Tue, 2008-01-15 at 18:23 +0900, FUJITA Tomonori wrote: >> This is the second version of >> >> http://marc.info/?l=linux-scsi&m=119933628210006&w=2 >> >> I gave up once, but I found that the performance loss is negligible

Re: [PATCH v2] use dynamically allocated sense buffer

2008-01-15 Thread FUJITA Tomonori
On Tue, 15 Jan 2008 17:44:14 +0200 Boaz Harrosh <[EMAIL PROTECTED]> wrote: > > If __scsi_put_command puts a command to shost->free_list, it doesn't > > free scmd->sense_buffer since it's the sense_buffer for the backup > > sense_buffer. If __scsi_put_command puts a command to > > shost->cmd_pool->

Re: [PATCH v2] use dynamically allocated sense buffer

2008-01-15 Thread Boaz Harrosh
On Tue, Jan 15 2008 at 17:20 +0200, James Bottomley <[EMAIL PROTECTED]> wrote: > On Tue, 2008-01-15 at 18:23 +0900, FUJITA Tomonori wrote: >> This is the second version of >> >> http://marc.info/?l=linux-scsi&m=119933628210006&w=2 >> >> I gave up once, but I found that the performance loss is negli

Re: [PATCH v2] use dynamically allocated sense buffer

2008-01-15 Thread Boaz Harrosh
On Tue, Jan 15 2008 at 17:08 +0200, FUJITA Tomonori <[EMAIL PROTECTED]> wrote: > On Tue, 15 Jan 2008 15:56:56 +0200 > Boaz Harrosh <[EMAIL PROTECTED]> wrote: > >> On Tue, Jan 15 2008 at 11:23 +0200, FUJITA Tomonori <[EMAIL PROTECTED]> >> wrote: >>> This is the second version of >>> >>> http://mar

Re: [PATCH v2] use dynamically allocated sense buffer

2008-01-15 Thread James Bottomley
On Tue, 2008-01-15 at 18:23 +0900, FUJITA Tomonori wrote: > This is the second version of > > http://marc.info/?l=linux-scsi&m=119933628210006&w=2 > > I gave up once, but I found that the performance loss is negligible > (within 1%) by using kmem_cache_alloc instead of mempool. > > I use scsi_d

Re: [PATCH v2] use dynamically allocated sense buffer

2008-01-15 Thread FUJITA Tomonori
On Tue, 15 Jan 2008 15:56:56 +0200 Boaz Harrosh <[EMAIL PROTECTED]> wrote: > On Tue, Jan 15 2008 at 11:23 +0200, FUJITA Tomonori <[EMAIL PROTECTED]> wrote: > > This is the second version of > > > > http://marc.info/?l=linux-scsi&m=119933628210006&w=2 > > > > I gave up once, but I found that the

Re: [PATCH v2] use dynamically allocated sense buffer

2008-01-15 Thread Boaz Harrosh
On Tue, Jan 15 2008 at 11:23 +0200, FUJITA Tomonori <[EMAIL PROTECTED]> wrote: > This is the second version of > > http://marc.info/?l=linux-scsi&m=119933628210006&w=2 > > I gave up once, but I found that the performance loss is negligible > (within 1%) by using kmem_cache_alloc instead of mempoo

[PATCH v2] use dynamically allocated sense buffer

2008-01-15 Thread FUJITA Tomonori
This is the second version of http://marc.info/?l=linux-scsi&m=119933628210006&w=2 I gave up once, but I found that the performance loss is negligible (within 1%) by using kmem_cache_alloc instead of mempool. I use scsi_debug with fake_rw=1 and disktest (DIO reads with 8 threads) again: scsi-mi