RE: [PATCH 00/12] LLD host_lock-less conversion status for .38

From: Desai, Kashyap
Date: Mon Dec 20 2010 - 10:08:33 EST



> -----Original Message-----
> From: Nicholas A. Bellinger [mailto:nab@xxxxxxxxxxxxxxx]
> Sent: Monday, December 20, 2010 2:52 AM
> To: linux-scsi; linux-kernel; James Bottomley; Jeff Garzik; Christoph
> Hellwig; FUJITA Tomonori; Hannes Reinecke; Mike Christie
> Cc: Mike Anderson; Tejun Heo; Vasu Dev; Tim Chen; Andi Kleen; Ravi
> Anand; Andrew Vasquez; Joe Eykholt; James Smart; Douglas Gilbert; adam
> radford; Desai, Kashyap; DL-MPT Fusion Linux; Nicholas Bellinger
> Subject: [PATCH 00/12] LLD host_lock-less conversion status for .38
>
> From: Nicholas Bellinger <nab@xxxxxxxxxxxxxxx>
>
> Greetings James and Co,
>
> Attached are the current set of LLD host_lock-less conversions in the
> queue
> for .38, that have now been pushed into lio-core-2.6.git/lock_less-
> LLDs-for-38-v3
>
> New in this v3 series are the megaraid_sas LLD changes necessary to run
> in
> host_lock-less mode, which have been tested thus far in legacy x86
> interrupt
> pin mode using 8708EM2 QEMU HBA emulation into KVM guest using TCM_Loop
> virtual SCSI LUNs.
>
> This series also drops the drivers/scsi patches that have been resolved
> with
> the following (Thanks!)
>
> commit 459dbf72e4d2b4aa13620e6b70d54f098547bf13
> Author: James Bottomley <James.Bottomley@xxxxxxx>
> Date: Wed Nov 17 10:10:57 2010 -0600
>
> [SCSI] Eliminate error handler overload of the SCSI serial number
>
> At this point the remaining code TODO items for the LLD conversion
> patches
> include
>
> *) libiscsi: NULL sc->scsi_done() callback in exception path in
> iscsi_queuecommand().
> The last status on this from Mike Christie was:
>
> "This will NULL pointer. See a couple lines above where we NULL it.
> iscsi_free_task checks if the scsi_done pointer is set and if it is it
> will call scsi_done.
>
> It is a hack to prevent the normal completion path from calling
> scsi_done. For the case where we return SCSI_MLQUEUE_TARGET_BUSY (the
> prepd_reject case) we need something to prevent scsi_done from getting
> called.
>
> For the return 0/prepd_fault case we can just call sc->scsi_done, but
> we
> have to move some code around."
>
> mnc, have you been able to take another look at this..?
>
> *) mpt2sas: Locking considerations for firmware specific data
> structures
>
> A patch to mptsas was included originally to convert to host_lock less
> w/ interrupts
> disabled externally here:
>
> http://git.kernel.org/?p=linux/kernel/git/nab/lio-core-
> 2.6.git;a=commitdiff;h=87ba9071b46f37f1adab1732d41b87bf195d191f
>
> but due to concerns from jgarzik this had been dropped from the
> lock_less-LLDs-for-38-v2
> during the last round. The feedback from Kashyap @ LSI is that mpt2sas
> technically should be
> able to run in host_lock-less, but there needs to be some explict
> clarification in the code
> for the conversion to happen. More details are here:
>
> http://marc.info/?l=linux-scsi&m=129022827729386&w=2
>
> Kashyap, can you take another look at this and make sure this is
> acceptable..?

Nicholas, I have replied my input on the same thread.

>
> *) megaraid_sas: Running in host_lock-less mode with MSI-X interrupts
>
> This has not been tested yet, but I would like to get some review from
> the LSI
> megaraid_sas folks now that their patches to enable MSI-X interrupts
> has been sent
> to linux-scsi.
>
> AdamR, any thoughts or comments to add here..?
>
> -----------------------------------------------------------------------
> -----------
>
> So at this point the next steps will be to shaking out breakage in
> corner cases for
> struct scsi_device hotplug removal, more testing+validation, and more
> LLD code review.
>
> For the CC'ed LLD maintainers, please have another look at the changes
> to enable
> host_lock-less operation within your code and send the necessary ACKs
> to linux-scsi
> at your earliest convience. Many thanks to everyone who has
> contributed!
>
> If anyone has any other patches, comments or concerns please let me
> know and they
> will be queued up for the next v4 for-38 round.
>
> Best Regards,
>
> Signed-off-by: Nicholas A. Bellinger <nab@xxxxxxxxxxxxxxx>
>
> Nicholas Bellinger (12):
> libiscsi: Convert to host_lock less w/ interrupts disabled internally
> scsi: Add IRQ_DISABLE_SCSI_QCMD wrapper
> libsas: Convert to host_lock less w/ interrupts disabled externally
> message: Convert to host_lock less w/ interrupts disabled externally
> fnic: Convert to host_lock less w/ interrupts disabled externally
> lpfc: Convert to host_lock less w/ interrupts disabled externally
> qla2xxx: Convert to host_lock less w/ interrupts disabled externally
> qla4xxx: Convert to host_lock less w/ interrupts disabled externally
> scsi_debug: Convert to host_lock less
> megaraid_sas: Add smp_mb__after_atomic_*() for
> instance->fw_outstanding
> megaraid_sas: Convert instance->issuepend_done to atomic_t
> megaraid_sas: Convert SHT->queuecommand() to run host_lock-less
>
> drivers/message/fusion/mptfc.c | 2 +-
> drivers/message/fusion/mptsas.c | 6 +-
> drivers/message/fusion/mptscsih.c | 8 ++-
> drivers/message/fusion/mptscsih.h | 2 +-
> drivers/message/fusion/mptspi.c | 6 +-
> drivers/scsi/fnic/fnic_scsi.c | 14 +----
> drivers/scsi/libiscsi.c | 10 +---
> drivers/scsi/libsas/sas_scsi_host.c | 12 +----
> drivers/scsi/lpfc/lpfc_scsi.c | 9 ++--
> drivers/scsi/megaraid/megaraid_sas.c | 85 ++++++++++++++++++++++----
> --------
> drivers/scsi/megaraid/megaraid_sas.h | 2 +-
> drivers/scsi/qla2xxx/qla_os.c | 11 +---
> drivers/scsi/qla4xxx/ql4_os.c | 10 +---
> drivers/scsi/scsi_debug.c | 16 +++----
> include/scsi/scsi_host.h | 14 ++++++
> 15 files changed, 107 insertions(+), 100 deletions(-)
>
> --
> 1.7.3.4

--
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/