aoe: become I/O request queue handler for increased user control
authorEd Cashin <ecashin@coraid.com>
Fri, 5 Oct 2012 00:16:23 +0000 (17:16 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Fri, 5 Oct 2012 18:05:25 +0000 (03:05 +0900)
commit69cf2d85de773d998798e47e3335b85e5645d157
tree765eb2be45726e7e098fe73b7f368239c0461342
parent896831f5909e2733c13c9cb13a1a215f10c3eaa8
aoe: become I/O request queue handler for increased user control

To allow users to choose an elevator algorithm for their particular
workloads, change from a make_request-style driver to an
I/O-request-queue-handler-style driver.

We have to do a couple of things that might be surprising.  We manipulate
the page _count directly on the assumption that we still have no guarantee
that users of the block layer are prohibited from submitting bios
containing pages with zero reference counts.[1] If such a prohibition now
exists, I can get rid of the _count manipulation.

Just as before this patch, we still keep track of the sk_buffs that the
network layer still hasn't finished yet and cap the resources we use with
a "pool" of skbs.[2]

Now that the block layer maintains the disk stats, the aoe driver's
diskstats function can go away.

1. https://lkml.org/lkml/2007/3/1/374
2. https://lkml.org/lkml/2007/7/6/241

Signed-off-by: Ed Cashin <ecashin@coraid.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
drivers/block/aoe/aoe.h
drivers/block/aoe/aoeblk.c
drivers/block/aoe/aoechr.c
drivers/block/aoe/aoecmd.c
drivers/block/aoe/aoedev.c