Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753276Ab3GSXij (ORCPT ); Fri, 19 Jul 2013 19:38:39 -0400 Received: from longford.logfs.org ([213.229.74.203]:59838 "EHLO longford.logfs.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752949Ab3GSXii (ORCPT ); Fri, 19 Jul 2013 19:38:38 -0400 Date: Fri, 19 Jul 2013 18:07:00 -0400 From: =?utf-8?B?SsO2cm4=?= Engel To: Nagalakshmi Nandigama , Sreekanth Reddy , support@lsi.com, "James E.J. Bottomley" , DL-MPTFusionLinux@lsi.com, linux-scsi@vger.kernel.org, linux-kernel@vger.kernel.org, mit@purestorage.com Subject: [PATCH] mpt2sas: don't handle broadcast primitives Message-ID: <20130719220659.GF29404@logfs.org> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline User-Agent: Mutt/1.5.20 (2009-06-14) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 10280 Lines: 326 The handling of broadcast primitives involves _scsih_block_io_all_device(), which does what the name implies. I have observed cases with >60s of blocking io on all devices, caused by a single bad device. The downsides of this code are obvious, while the upsides are more elusive. Signed-off-by: Joern Engel --- drivers/scsi/mpt2sas/mpt2sas_base.c | 4 - drivers/scsi/mpt2sas/mpt2sas_scsih.c | 238 ---------------------------------- 2 files changed, 242 deletions(-) diff --git a/drivers/scsi/mpt2sas/mpt2sas_base.c b/drivers/scsi/mpt2sas/mpt2sas_base.c index 21c0a27..bba2209 100644 --- a/drivers/scsi/mpt2sas/mpt2sas_base.c +++ b/drivers/scsi/mpt2sas/mpt2sas_base.c @@ -586,9 +586,6 @@ _base_display_event_data(struct MPT2SAS_ADAPTER *ioc, printk("\n"); return; } - case MPI2_EVENT_SAS_BROADCAST_PRIMITIVE: - desc = "SAS Broadcast Primitive"; - break; case MPI2_EVENT_SAS_INIT_DEVICE_STATUS_CHANGE: desc = "SAS Init Device Status Change"; break; @@ -4370,7 +4367,6 @@ mpt2sas_base_attach(struct MPT2SAS_ADAPTER *ioc) /* here we enable the events we care about */ _base_unmask_events(ioc, MPI2_EVENT_SAS_DISCOVERY); - _base_unmask_events(ioc, MPI2_EVENT_SAS_BROADCAST_PRIMITIVE); _base_unmask_events(ioc, MPI2_EVENT_SAS_TOPOLOGY_CHANGE_LIST); _base_unmask_events(ioc, MPI2_EVENT_SAS_DEVICE_STATUS_CHANGE); _base_unmask_events(ioc, MPI2_EVENT_SAS_ENCL_DEVICE_STATUS_CHANGE); diff --git a/drivers/scsi/mpt2sas/mpt2sas_scsih.c b/drivers/scsi/mpt2sas/mpt2sas_scsih.c index 3fffb95..856a502 100644 --- a/drivers/scsi/mpt2sas/mpt2sas_scsih.c +++ b/drivers/scsi/mpt2sas/mpt2sas_scsih.c @@ -2915,31 +2915,6 @@ _scsih_fw_event_cleanup_queue(struct MPT2SAS_ADAPTER *ioc) } /** - * _scsih_ublock_io_all_device - unblock every device - * @ioc: per adapter object - * - * change the device state from block to running - */ -static void -_scsih_ublock_io_all_device(struct MPT2SAS_ADAPTER *ioc) -{ - struct MPT2SAS_DEVICE *sas_device_priv_data; - struct scsi_device *sdev; - - shost_for_each_device(sdev, ioc->shost) { - sas_device_priv_data = sdev->hostdata; - if (!sas_device_priv_data) - continue; - if (!sas_device_priv_data->block) - continue; - sas_device_priv_data->block = 0; - dewtprintk(ioc, sdev_printk(KERN_INFO, sdev, "device_running, " - "handle(0x%04x)\n", - sas_device_priv_data->sas_target->handle)); - scsi_internal_device_unblock(sdev, SDEV_RUNNING); - } -} -/** * _scsih_ublock_io_device - set the device state to SDEV_RUNNING * @ioc: per adapter object * @handle: device handle @@ -2971,34 +2946,6 @@ _scsih_ublock_io_device(struct MPT2SAS_ADAPTER *ioc, u64 sas_address) } /** - * _scsih_block_io_all_device - set the device state to SDEV_BLOCK - * @ioc: per adapter object - * @handle: device handle - * - * During device pull we need to appropiately set the sdev state. - */ -static void -_scsih_block_io_all_device(struct MPT2SAS_ADAPTER *ioc) -{ - struct MPT2SAS_DEVICE *sas_device_priv_data; - struct scsi_device *sdev; - - shost_for_each_device(sdev, ioc->shost) { - sas_device_priv_data = sdev->hostdata; - if (!sas_device_priv_data) - continue; - if (sas_device_priv_data->block) - continue; - sas_device_priv_data->block = 1; - dewtprintk(ioc, sdev_printk(KERN_INFO, sdev, "device_blocked, " - "handle(0x%04x)\n", - sas_device_priv_data->sas_target->handle)); - scsi_internal_device_block(sdev); - } -} - - -/** * _scsih_block_io_device - set the device state to SDEV_BLOCK * @ioc: per adapter object * @handle: device handle @@ -5829,166 +5776,6 @@ _scsih_sas_enclosure_dev_status_change_event(struct MPT2SAS_ADAPTER *ioc, } /** - * _scsih_sas_broadcast_primitive_event - handle broadcast events - * @ioc: per adapter object - * @fw_event: The fw_event_work object - * Context: user. - * - * Return nothing. - */ -static void -_scsih_sas_broadcast_primitive_event(struct MPT2SAS_ADAPTER *ioc, - struct fw_event_work *fw_event) -{ - struct scsi_cmnd *scmd; - unsigned long ser_no; - struct scsi_device *sdev; - u16 smid, handle; - u32 lun; - struct MPT2SAS_DEVICE *sas_device_priv_data; - u32 termination_count; - u32 query_count; - Mpi2SCSITaskManagementReply_t *mpi_reply; - Mpi2EventDataSasBroadcastPrimitive_t *event_data = fw_event->event_data; - u16 ioc_status; - unsigned long flags; - int r; - u8 max_retries = 0; - u8 task_abort_retries; - - mutex_lock(&ioc->tm_cmds.mutex); - dewtprintk(ioc, printk(MPT2SAS_INFO_FMT "%s: enter: phy number(%d), " - "width(%d)\n", ioc->name, __func__, event_data->PhyNum, - event_data->PortWidth)); - - _scsih_block_io_all_device(ioc); - - spin_lock_irqsave(&ioc->scsi_lookup_lock, flags); - mpi_reply = ioc->tm_cmds.reply; -broadcast_aen_retry: - - /* sanity checks for retrying this loop */ - if (max_retries++ == 5) { - dewtprintk(ioc, printk(MPT2SAS_INFO_FMT "%s: giving up\n", - ioc->name, __func__)); - goto out; - } else if (max_retries > 1) - dewtprintk(ioc, printk(MPT2SAS_INFO_FMT "%s: %d retry\n", - ioc->name, __func__, max_retries - 1)); - - termination_count = 0; - query_count = 0; - for (smid = 1; smid <= ioc->scsiio_depth; smid++) { - if (ioc->shost_recovery) - goto out; - scmd = _scsih_scsi_lookup_get(ioc, smid); - if (!scmd) - continue; - sdev = scmd->device; - ser_no = scmd->serial_number; - sas_device_priv_data = sdev->hostdata; - if (!sas_device_priv_data || !sas_device_priv_data->sas_target) - continue; - /* skip hidden raid components */ - if (sas_device_priv_data->sas_target->flags & - MPT_TARGET_FLAGS_RAID_COMPONENT) - continue; - /* skip volumes */ - if (sas_device_priv_data->sas_target->flags & - MPT_TARGET_FLAGS_VOLUME) - continue; - - handle = sas_device_priv_data->sas_target->handle; - lun = sas_device_priv_data->lun; - query_count++; - - if (ioc->shost_recovery) - goto out; - - spin_unlock_irqrestore(&ioc->scsi_lookup_lock, flags); - r = mpt2sas_scsih_issue_tm(ioc, handle, 0, 0, lun, - MPI2_SCSITASKMGMT_TASKTYPE_QUERY_TASK, smid, 30, 0, - TM_MUTEX_OFF); - if (r == FAILED) { - sdev_printk(KERN_WARNING, sdev, - "mpt2sas_scsih_issue_tm: FAILED when sending " - "QUERY_TASK: scmd(%p)\n", scmd); - spin_lock_irqsave(&ioc->scsi_lookup_lock, flags); - goto broadcast_aen_retry; - } - ioc_status = le16_to_cpu(mpi_reply->IOCStatus) - & MPI2_IOCSTATUS_MASK; - if (ioc_status != MPI2_IOCSTATUS_SUCCESS) { - sdev_printk(KERN_WARNING, sdev, "query task: FAILED " - "with IOCSTATUS(0x%04x), scmd(%p)\n", ioc_status, - scmd); - spin_lock_irqsave(&ioc->scsi_lookup_lock, flags); - goto broadcast_aen_retry; - } - - /* see if IO is still owned by IOC and target */ - if (mpi_reply->ResponseCode == - MPI2_SCSITASKMGMT_RSP_TM_SUCCEEDED || - mpi_reply->ResponseCode == - MPI2_SCSITASKMGMT_RSP_IO_QUEUED_ON_IOC) { - spin_lock_irqsave(&ioc->scsi_lookup_lock, flags); - continue; - } - task_abort_retries = 0; - tm_retry: - if (task_abort_retries++ == 60) { - dewtprintk(ioc, printk(MPT2SAS_INFO_FMT - "%s: ABORT_TASK: giving up\n", ioc->name, - __func__)); - spin_lock_irqsave(&ioc->scsi_lookup_lock, flags); - goto broadcast_aen_retry; - } - - if (ioc->shost_recovery) - goto out_no_lock; - - r = mpt2sas_scsih_issue_tm(ioc, handle, sdev->channel, sdev->id, - sdev->lun, MPI2_SCSITASKMGMT_TASKTYPE_ABORT_TASK, smid, 30, - ser_no, TM_MUTEX_OFF); - if (r == FAILED) { - sdev_printk(KERN_WARNING, sdev, - "mpt2sas_scsih_issue_tm: ABORT_TASK: FAILED : " - "scmd(%p)\n", scmd); - goto tm_retry; - } - - if (task_abort_retries > 1) - sdev_printk(KERN_WARNING, sdev, - "mpt2sas_scsih_issue_tm: ABORT_TASK: RETRIES (%d):" - " scmd(%p)\n", - task_abort_retries - 1, scmd); - - termination_count += le32_to_cpu(mpi_reply->TerminationCount); - spin_lock_irqsave(&ioc->scsi_lookup_lock, flags); - } - - if (ioc->broadcast_aen_pending) { - dewtprintk(ioc, printk(MPT2SAS_INFO_FMT "%s: loop back due to" - " pending AEN\n", ioc->name, __func__)); - ioc->broadcast_aen_pending = 0; - goto broadcast_aen_retry; - } - - out: - spin_unlock_irqrestore(&ioc->scsi_lookup_lock, flags); - out_no_lock: - - dewtprintk(ioc, printk(MPT2SAS_INFO_FMT - "%s - exit, query_count = %d termination_count = %d\n", - ioc->name, __func__, query_count, termination_count)); - - ioc->broadcast_aen_busy = 0; - if (!ioc->shost_recovery) - _scsih_ublock_io_all_device(ioc); - mutex_unlock(&ioc->tm_cmds.mutex); -} - -/** * _scsih_sas_discovery_event - handle discovery events * @ioc: per adapter object * @fw_event: The fw_event_work object @@ -7081,8 +6868,6 @@ _scsih_remove_unresponding_sas_devices(struct MPT2SAS_ADAPTER *ioc) } printk(MPT2SAS_INFO_FMT "removing unresponding devices: complete\n", ioc->name); - /* unblock devices */ - _scsih_ublock_io_all_device(ioc); } static void @@ -7358,10 +7143,6 @@ _firmware_event_work(struct work_struct *work) _scsih_sas_discovery_event(ioc, fw_event); break; - case MPI2_EVENT_SAS_BROADCAST_PRIMITIVE: - _scsih_sas_broadcast_primitive_event(ioc, - fw_event); - break; case MPI2_EVENT_SAS_ENCL_DEVICE_STATUS_CHANGE: _scsih_sas_enclosure_dev_status_change_event(ioc, fw_event); @@ -7419,25 +7200,6 @@ mpt2sas_scsih_event_callback(struct MPT2SAS_ADAPTER *ioc, u8 msix_index, event = le16_to_cpu(mpi_reply->Event); switch (event) { - /* handle these */ - case MPI2_EVENT_SAS_BROADCAST_PRIMITIVE: - { - Mpi2EventDataSasBroadcastPrimitive_t *baen_data = - (Mpi2EventDataSasBroadcastPrimitive_t *) - mpi_reply->EventData; - - if (baen_data->Primitive != - MPI2_EVENT_PRIMITIVE_ASYNCHRONOUS_EVENT) - return 1; - - if (ioc->broadcast_aen_busy) { - ioc->broadcast_aen_pending++; - return 1; - } else - ioc->broadcast_aen_busy = 1; - break; - } - case MPI2_EVENT_SAS_TOPOLOGY_CHANGE_LIST: _scsih_check_topo_delete_events(ioc, (Mpi2EventDataSasTopologyChangeList_t *) -- 1.7.10.4 -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/