From mboxrd@z Thu Jan 1 00:00:00 1970 From: Sumit Saxena Subject: RE: [PATCH 4/7] megaraid_sas: Send SYNCHRONIZE_CACHE command to firmware Date: Mon, 17 Oct 2016 17:43:10 +0530 Message-ID: References: <1476699850-25083-1-git-send-email-sumit.saxena@broadcom.com> <1476699850-25083-5-git-send-email-sumit.saxena@broadcom.com> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: quoted-printable Return-path: Received: from mail-oi0-f47.google.com ([209.85.218.47]:34388 "EHLO mail-oi0-f47.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1757570AbcJQMNN (ORCPT ); Mon, 17 Oct 2016 08:13:13 -0400 Received: by mail-oi0-f47.google.com with SMTP id t73so209644947oie.1 for ; Mon, 17 Oct 2016 05:13:12 -0700 (PDT) In-Reply-To: Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: Hannes Reinecke , linux-scsi@vger.kernel.org Cc: martin.petersen@oracle.com, thenzl@redhat.com, jejb@linux.vnet.ibm.com, Kashyap Desai >-----Original Message----- >From: Hannes Reinecke [mailto:hare@suse.de] >Sent: Monday, October 17, 2016 5:04 PM >To: Sumit Saxena; linux-scsi@vger.kernel.org >Cc: martin.petersen@oracle.com; thenzl@redhat.com; jejb@linux.vnet.ibm.com; >kashyap.desai@broadcom.com >Subject: Re: [PATCH 4/7] megaraid_sas: Send SYNCHRONIZE_CACHE command to >firmware > >On 10/17/2016 12:24 PM, Sumit Saxena wrote: >> megaraid_sas driver returns SYNCHRONIZE_CACHE command back to SCSI >> layer without sending it to firmware as firmware takes care of flushing cache. >> This patch will change the driver behavior wrt SYNCHRONIZE_CACHE handling. >> If underlying firmware has support to handle the SYNCHORNIZE_CACHE, >> driver will send it for firmware otherwise complete it back to SCSI >> layer with SUCCESS immediately. >> If Firmware handle SYNCHORNIZE_CACHE for both VD and JBOD >> "canHandleSyncCache" bit in scratch pad register(offset 0x00B4) will >> be set. >> >> This behavior can be controlled via module parameter and user can >> fallback to old behavior of returning SYNCHRONIZE_CACHE by driver only >> without sending it to firmware. >> >> Signed-off-by: Sumit Saxena >> Signed-off-by: Kashyap Desai >> --- >> drivers/scsi/megaraid/megaraid_sas.h | 3 +++ >> drivers/scsi/megaraid/megaraid_sas_base.c | 14 ++++++-------- >> drivers/scsi/megaraid/megaraid_sas_fusion.c | 3 +++ >> drivers/scsi/megaraid/megaraid_sas_fusion.h | 2 ++ >> 4 files changed, 14 insertions(+), 8 deletions(-) >> >> diff --git a/drivers/scsi/megaraid/megaraid_sas.h >> b/drivers/scsi/megaraid/megaraid_sas.h >> index ca86c88..43fd14f 100644 >> --- a/drivers/scsi/megaraid/megaraid_sas.h >> +++ b/drivers/scsi/megaraid/megaraid_sas.h >> @@ -1429,6 +1429,8 @@ enum FW_BOOT_CONTEXT { >> #define MR_MAX_REPLY_QUEUES_EXT_OFFSET_SHIFT 14 >> #define MR_MAX_MSIX_REG_ARRAY 16 >> #define MR_RDPQ_MODE_OFFSET 0X00800000 >> +#define MR_CAN_HANDLE_SYNC_CACHE_OFFSET 0X01000000 >> + >> /* >> * register set for both 1068 and 1078 controllers >> * structure extended for 1078 registers @@ -2140,6 +2142,7 @@ struct >> megasas_instance { >> u8 is_imr; >> u8 is_rdpq; >> bool dev_handle; >> + bool fw_sync_cache_support; >> }; >> struct MR_LD_VF_MAP { >> u32 size; >> diff --git a/drivers/scsi/megaraid/megaraid_sas_base.c >> b/drivers/scsi/megaraid/megaraid_sas_base.c >> index 092387f..a4a8e2f 100644 >> --- a/drivers/scsi/megaraid/megaraid_sas_base.c >> +++ b/drivers/scsi/megaraid/megaraid_sas_base.c >> @@ -104,6 +104,10 @@ unsigned int scmd_timeout =3D >> MEGASAS_DEFAULT_CMD_TIMEOUT; module_param(scmd_timeout, int, >> S_IRUGO); MODULE_PARM_DESC(scmd_timeout, "scsi command timeout >> (10-90s), default 90s. See megasas_reset_timer."); >> >> +bool block_sync_cache; >> +module_param(block_sync_cache, bool, S_IRUGO); >> +MODULE_PARM_DESC(block_sync_cache, "Block SYNC CACHE by driver >> +Default: 0(Send it to firmware)"); >> + >> MODULE_LICENSE("GPL"); >> MODULE_VERSION(MEGASAS_VERSION); >> MODULE_AUTHOR("megaraidlinux.pdl@avagotech.com"); >> @@ -1700,16 +1704,10 @@ megasas_queue_command(struct Scsi_Host >*shost, struct scsi_cmnd *scmd) >> goto out_done; >> } >> >> - switch (scmd->cmnd[0]) { >> - case SYNCHRONIZE_CACHE: >> - /* >> - * FW takes care of flush cache on its own >> - * No need to send it down >> - */ >> + if ((scmd->cmnd[0] =3D=3D SYNCHRONIZE_CACHE) && >> + (!instance->fw_sync_cache_support)) { >> scmd->result =3D DID_OK << 16; >> goto out_done; >> - default: >> - break; >> } >> >> return instance->instancet->build_and_issue_cmd(instance, scmd); >> diff --git a/drivers/scsi/megaraid/megaraid_sas_fusion.c >> b/drivers/scsi/megaraid/megaraid_sas_fusion.c >> index 2159f6a..8237580 100644 >> --- a/drivers/scsi/megaraid/megaraid_sas_fusion.c >> +++ b/drivers/scsi/megaraid/megaraid_sas_fusion.c >> @@ -747,6 +747,9 @@ megasas_ioc_init_fusion(struct megasas_instance >*instance) >> ret =3D 1; >> goto fail_fw_init; >> } >> + if (!block_sync_cache) >> + instance->fw_sync_cache_support =3D (scratch_pad_2 & >> + MR_CAN_HANDLE_SYNC_CACHE_OFFSET) ? 1 : 0; >> >> IOCInitMessage =3D >> dma_alloc_coherent(&instance->pdev->dev, >> diff --git a/drivers/scsi/megaraid/megaraid_sas_fusion.h >> b/drivers/scsi/megaraid/megaraid_sas_fusion.h >> index e3bee04..2857154 100644 >> --- a/drivers/scsi/megaraid/megaraid_sas_fusion.h >> +++ b/drivers/scsi/megaraid/megaraid_sas_fusion.h >> @@ -72,6 +72,8 @@ >> #define MPI2_SUP_REPLY_POST_HOST_INDEX_OFFSET (0x0000030C) >> #define MPI2_REPLY_POST_HOST_INDEX_OFFSET (0x0000006C) >> >> +extern bool block_sync_cache; >> + >> /* >> * Raid context flags >> */ >> >Be extra careful with that. > >SYNCHRONIZE_CACHE has (potentially) a global scope, as there might be an >array-wide cache, and a cache flush would affect the entire cache. >Linux is using SYNCHRONIZE_CACHE as a per device setting, ie it assumes that the >effects of a cache flush is restricted to the device in question. Yes, cache flush will be done per device only. > >If this is _not_ the case I'd rather not enable it. >Have you checked that enabling this functionality doesn't have negative >performance impact? Yes, there will be negative performance impact on wrokloads where SYNCHRONIZE_CACHE is fired very frequently. We have provided a module parameter to fallback to older behavior, if anyone sees negative performance impact because of this and there is external power to drive. This change is done for deployment where there is no external power to drive and abrupt shutdown may cause data loss as cache may not be flushed to disk. Thanks, Sumit > >Cheers, > >Hannes >-- >Dr. Hannes Reinecke Teamlead Storage & Networking >hare@suse.de +49 911 74053 688 >SUSE LINUX GmbH, Maxfeldstr. 5, 90409 N=C3=BCrnberg >GF: F. Imend=C3=B6rffer, J. Smithard, J. Guild, D. Upmanyu, G. Norton HRB 21284 (AG >N=C3=BCrnberg)