[PATCH linux-2.6-block:master] overview of soon-to-be-posted patches

[PATCH linux-2.6-block:master] overview of soon-to-be-posted patches

[Tejun Heo]

  Hello, Jens.

  As requested, I've regenerated dispatch queue and ordered patchset and 
will soon post 18 patches.  All patches are against the current master 
of linux-2.6-block:master (3d80636a0d5f056ffc26472d05b6027a7a9f6e1c).

patch #1	: fix-elevator_find.  remove try_module_get race in
		  elevator_find.
patch #2-6	: generic dispatch queue patchset.
patch #7	: reimplement-elevator-switch.  reimplements elevator
		  switch using generic dispatch queue.
patch #8-17	: ordered reimplementation patchset

  Patches should be applied in above order.  I proof-read changes and 
tested things to make sure changes made to the tree since the last 
posting doesn't break anything.  Well, it seems okay and works for me.

  Thanks.

-- 
tejun

[PATCH linux-2.6-block:master] overview of soon-to-be-posted patches

[Tejun Heo]

 Hello, Jens.

 I hope you had fun on your vacation and at OLS.  I'm posting 18
welcome-back patches today. :-p This mail is to show the overview of
the patches.  All patches are against master head of linux-2.6-block
tree.

patch #1	: fix-elevator_find.  remove try_module_get race in
		  elevator_find.
patch #2	: fix-cfq_find_next_crq.  fix cfq_find_next_crq bug
		  in cfq.
patch #3-7	: generic dispatch queue patchset.  implements generic
		  dispatch queue.
patch #8	: reimplement-elevator-switch.  reimplements elevator
		  switch using generic dispatch queue.  draining isn't
		  needed anymore.
patch #9-#18	: ordered reimplementation patchset.  reimplements
		  I/O barrier handling.

 Both generic dispatch queue patchset and ordered reimplementation
patchset were previously posted.  They are reordered (as you asked)
and I've added missing bits (all elevators are updated, docs are
updated).  Also, there were a few changes and fixes.  I'll mention
them when I post those patches.

 I've tested these changes by running parallelly...

* random raw read (concurrency 8)
* repeatedly mounting a ext3 fs with -o barrier=1, copying, syncing &
  checksumming a 128M file, and unmounting the fs.
* periodic scheduler switch (each iosched runs for 3 minutes and then
  switched to the next one)

 This test has been running for several hours without a problem and I
will keep it running for today and maybe tomorrow unless I have to use
the test machine for some other purpose.

 Thanks.

--
tejun

Re: [PATCH linux-2.6-block:master] overview of soon-to-be-posted patches

[Jens Axboe]

On Tue, Jul 26 2005, Tejun Heo wrote:
>  Hello, Jens.
> 
>  I hope you had fun on your vacation and at OLS.  I'm posting 18
> welcome-back patches today. :-p This mail is to show the overview of
> the patches.  All patches are against master head of linux-2.6-block
> tree.
> 
> patch #1	: fix-elevator_find.  remove try_module_get race in
> 		  elevator_find.
> patch #2	: fix-cfq_find_next_crq.  fix cfq_find_next_crq bug
> 		  in cfq.
> patch #3-7	: generic dispatch queue patchset.  implements generic
> 		  dispatch queue.
> patch #8	: reimplement-elevator-switch.  reimplements elevator
> 		  switch using generic dispatch queue.  draining isn't
> 		  needed anymore.
> patch #9-#18	: ordered reimplementation patchset.  reimplements
> 		  I/O barrier handling.

Nice work! I'll review this, added to list...

-- 
Jens Axboe