[RFC v4 00/19] scsi: Add optional host_lock less scsi_dispatch_cmd()

From: Nicholas A. Bellinger
Date: Mon Sep 27 2010 - 22:05:22 EST


From: Nicholas Bellinger <nab@xxxxxxxxxxxxxxx>

Greetings all,

This is the forth RFC for allowing an optional struct Scsi_Host->host_lock
less bit for high performance LLDs who where originaly dropping and reaquiring
host_lock from within their own SHT->queuecommand() callers in order to increase
performance. This series includes the wholesale conversion of existing libraries
and all other SCSI LLD ->queuecommand() callers that use the now extinct host_lock
optimization of:

unlock() ; do_some_lld_work() ; lock() -> return from ->mylld_queuecommand()

This series includes patches to drop this historical optimization (and prevent a
sure-fire deadlock w/ host_lock here) in libiscsi, libsas, libfc, libata and in
lpfc, qla4xxx, qla2xxx, fnic, and buslogic SCSI LLDs.

This patch also converts the scsi_cmd_get_serial() call to a counting by 2 odd
counter that is no longer used directly and by default for LLDs in scsi_dispatch_cmd(),
but is still required for mpt-fusion, mpt2sas, and a handful of old LLDs.

For this case we add explict scsi_cmd_get_serial() calls into SHT->queuecommand()
callers for mpt2sas, mpt/fusion, dpt_i2o, eata and u14-34f. There are other
SCSI LLDs that use struct scsi_cmnd->serial_number informationally, but no other
ones (AFAICT) actually depend on having ->serial_number set to function properly.

So if anyone else knows if any SCSI LLDs that either:

*) drop and request struct Scsi_Host->host_lock in their SHT->queuecommand()
to prevent dead-lock

*) or depend upon struct scsi_cmnd->serial_number for some reason to prevent
any other old/obsecure LLDs from having problems with a zero ->serial_number.

please let me know and I will queue a follow up patches to address any other issues.

Please note that this patch has a bug wrt to scsi_error.c usage mentioned by
Mike Anderson, and will be addressed in a RFC v5 series.

Changes from RFC v3 -> v4 include:

*) Use scsi_dispatch_cmd_locked() by default when LLD is not registering
it's struct scsi_host_template with SHT->unlocked_qcmd=1
*) Update aic94xx, mvsas, pm8001 and other LLDs included in RFCv3 to use
new SHT->unlocked_qcmd=1 who where already using the ->queuecommand()
'optimization' discussed above.

Many thanks go out to the linux.intel.com team (Vasu Dev, Tim Chen, Andi Kleen, Matthew Wilcox)
Joe Eykholt (Cisco), Mike Christie (RHAT), James Bottomley (Suse.de), Mike Anderson + Brian King
(linux.vnet.ibm.com) and Christof Schmitt (de.ibm.com) for their invaluable input towards
the whole-sale conversion of drivers/scsi to an optional host_lock less SHT->queuecommand()
dispatcher!

Thank you!

Signed-off-by: Nicholas A. Bellinger <nab@xxxxxxxxxxxxxxx>

Nicholas Bellinger (19):
scsi: Add SHT->unlocked_qcmd for host_lock less SHT->queuecommand()
dispatch
libiscsi: Remove host_lock unlock() + lock() from
iscsi_queuecommand()
libsas: Remove host_lock unlock() + lock() from sas_queuecommand()
aic94xx: Set SHT->unlocked_qcmd=1 for libsas queuecommand()
mvsas: Set SHT->unlocked_qcmd=1 for libsas queuecommand()
pm8001: Set SHT->unlocked_qcmd=1 for libsas queuecommand()
libfc: Remove host_lock unlock() + lock() from fc_queuecommand()
fcoe: Set SHT->unlocked_qcmd=1 for libfc fc_queuecommand()
libata: Remove host_lock unlock() + lock() from ata_scsi_queuecmd()
lpfc: Remove host_lock unlock() + lock() from lpfc_queuecommand()
qla4xxx: Remove host_lock unlock() + lock() from
qla4xxx_queuecommand()
qla2xxx: Remove host_lock unlock() + lock() from
qla2xxx_queuecommand()
fnic: Remove host_lock unlock() + lock() from fnic_queuecommand()
buslogic: Remove host_lock unlock() + lock() from
BusLogic_QueueCommand()
mpt2sas: Add scsi_cmd_get_serial() call
mpt/fusion: Add scsi_cmd_get_serial() call
dpt_i2o: Add scsi_cmd_get_serial() call
eata: Add scsi_cmd_get_serial() call
u14-34f: Add scsi_cmd_get_serial() call

block/blk.h | 4 ++
drivers/ata/libata-scsi.c | 4 +-
drivers/message/fusion/mptfc.c | 1 +
drivers/message/fusion/mptsas.c | 1 +
drivers/message/fusion/mptscsih.c | 5 ++
drivers/message/fusion/mptspi.c | 1 +
drivers/scsi/BusLogic.c | 4 --
drivers/scsi/aic94xx/aic94xx_init.c | 1 +
drivers/scsi/dpt_i2o.c | 5 ++
drivers/scsi/eata.c | 5 ++
drivers/scsi/fcoe/fcoe.c | 1 +
drivers/scsi/fnic/fnic_main.c | 1 +
drivers/scsi/fnic/fnic_scsi.c | 9 ----
drivers/scsi/hosts.c | 5 ++
drivers/scsi/iscsi_tcp.c | 1 +
drivers/scsi/libfc/fc_fcp.c | 5 +--
drivers/scsi/libiscsi.c | 4 --
drivers/scsi/libsas/sas_scsi_host.c | 5 --
drivers/scsi/lpfc/lpfc_scsi.c | 4 +-
drivers/scsi/mpt2sas/mpt2sas_scsih.c | 6 ++
drivers/scsi/mvsas/mv_init.c | 1 +
drivers/scsi/pm8001/pm8001_init.c | 1 +
drivers/scsi/qla2xxx/qla_os.c | 8 +--
drivers/scsi/qla4xxx/ql4_os.c | 9 +---
drivers/scsi/scsi.c | 89 ++++++++++++++++++++++++++--------
drivers/scsi/scsi_error.c | 10 +++-
drivers/scsi/u14-34f.c | 6 ++
include/linux/libata.h | 1 +
include/scsi/scsi_cmnd.h | 1 +
include/scsi/scsi_host.h | 16 +++++-
30 files changed, 144 insertions(+), 70 deletions(-)

--
1.7.3

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/