[PATCH] IB/ipath: reduce overhead on receive interrupts
authorBryan O'Sullivan <bos@pathscale.com>
Sat, 1 Jul 2006 11:36:04 +0000 (04:36 -0700)
committerLinus Torvalds <torvalds@g5.osdl.org>
Sat, 1 Jul 2006 16:56:00 +0000 (09:56 -0700)
commit13aef4942c291742064c1d5ac71df6493c4a00a9
treeb52657fa68485a00b5c94d3f37ed0a9f0cb7179d
parentf37bda92461313ad3bbfbf5660adc849c69718bf
[PATCH] IB/ipath: reduce overhead on receive interrupts

Also count the number of interrupts where that works (fastrcvint).  On any
interrupt where the port0 head and tail registers are not equal, just call the
ipath_kreceive code without reading the interrupt status, thus saving the
approximately 0.25usec processor stall waiting for the read to return.  If any
other interrupt bits are set, or head==tail, take the normal path, but that
has been reordered to handle read ahead of pioavail.  Also no longer call
ipath_kreceive() from ipath_qcheck(), because that just seems to make things
worse, and isn't really buying us anything, these days.

Also no longer loop in ipath_kreceive(); better to not hold things off too
long (I saw many cases where we would loop 4-8 times, and handle thousands (up
to 3500) in a single call).

Signed-off-by: Dave Olson <dave.olson@qlogic.com>
Signed-off-by: Bryan O'Sullivan <bryan.osullivan@qlogic.com>
Cc: "Michael S. Tsirkin" <mst@mellanox.co.il>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
drivers/infiniband/hw/ipath/ipath_common.h
drivers/infiniband/hw/ipath/ipath_driver.c
drivers/infiniband/hw/ipath/ipath_intr.c
drivers/infiniband/hw/ipath/ipath_stats.c