- 17 Sep, 2010 12 commits
-
-
Christof Schmitt authored
Instead of calling the same handler for both, I/O and task management commands, use different handlers that call a function for the common part. Reviewed-by:
Swen Schillig <swen@vnet.ibm.com> Signed-off-by:
Christof Schmitt <christof.schmitt@de.ibm.com> Signed-off-by:
James Bottomley <James.Bottomley@suse.de>
-
Christof Schmitt authored
With the change to use the data on the SCSI device, iterating through all LUNs/scsi_devices takes the SCSI host_lock. This triggers warnings from the lock dependency checker: ========================================================= [ INFO: possible irq lock inversion dependency detected ] 2.6.34.1 #97 --------------------------------------------------------- chchp/3224 just changed the state of lock: (&(shost->host_lock)->rlock){-.-...}, at: [<00000000003a73f4>] __scsi_iterate_devices+0x38/0xbc but this lock took another, HARDIRQ-unsafe lock in the past: (&(&qdio->req_q_lock)->rlock){+.-...} and interrupts could create inverse lock ordering between them. other info that might help us debug this: [ 24.972394] 2 locks held by chchp/3224: #0: (&(sch->lock)->rlock){-.-...}, at: [<0000000000401efa>] do_IRQ+0xb2/0x1e4 #1: (&adapter->port_list_lock){.-....}, at: [<0000000000490302>] zfcp_erp_modify_adapter_status+0x9e/0x16c [...] ========================================================= [ INFO: possible irq lock inversion dependency detected ] 2.6.34.1 #98 --------------------------------------------------------- chchp/3235 just changed the state of lock: (&(shost->host_lock)->rlock){-.-...}, at: [<00000000003a73f4>] __scsi_iterate_devices+0x38/0xbc but this lock took another, HARDIRQ-unsafe lock in the past: (&(&qdio->stat_lock)->rlock){+.-...} and interrupts could create inverse lock ordering between them. other info that might help us debug this: 2 locks held by chchp/3235: #0: (&(sch->lock)->rlock){-.-...}, at: [<0000000000401efa>] do_IRQ+0xb2/0x1e4 #1: (&adapter->port_list_lock){.-.-..}, at: [<00000000004902f6>] zfcp_erp_modify_adapter_status+0x9e/0x16c [...] To stop this warning, change the request queue lock to disable irqs, not only softirq. The changes are required only outside of the critical "send fcp command" path. Reviewed-by: -
Christof Schmitt authored
Enable the LUN scanning mechanism in the SCSI midlayer: - Do not set the disable_target_scan bit in the FC transport class. - Set max_lun to 0xFFFFFFFF to allow the midlayer scan to include the two-level hierachical LUNs (like 0x40XX40XX00000000, but in SCSI midlayer LUN format). - Set max_id to a high value to allow triggering the SCSI device rescan from sysfs. When running in NPIV mode, zfcp accepts all LUNs in slave_attach. When running in non-NPIV mode, the list of zfcp_unit structs determines which SCSI devices are allowed on the current system. Reviewed-by:
-
Christof Schmitt authored
This is the large change to switch from using the data in zfcp_unit to zfcp_scsi_dev. Keeping everything working requires doing the switch in one piece. To ensure that no code keeps using the data in zfcp_unit, this patch also removes the data from zfcp_unit that is now being replaced with zfcp_scsi_dev. For zfcp, the scsi_device together with zfcp_scsi_dev exist from the call of slave_alloc to the call of slave_destroy. The data in zfcp_scsi_dev is initialized in zfcp_scsi_slave_alloc and the LUN is opened; the final shutdown for the LUN is run from slave_destroy. Where the scsi_device or zfcp_scsi_dev is needed, the pointer to the scsi_device is passed as function argument and inside the function converted to the pointer to zfcp_scsi_dev; this avoids back and forth conversion betweeen scsi_device and zfcp_scsi_dev. While changing the function arguments from zfcp_unit to scsi_device, the functions names are renamed form "unit" to "lun". This is to have a seperation between zfcp_scsi_dev/LUN and the zfcp_unit; only code referring to the remaining configuration information in zfcp_unit struct uses "unit". Reviewed-by:
-
Christof Schmitt authored
With the change for the LUN data to be part of the scsi_device, the slave_destroy callback will be the final call to the zfcp_erp_unit_shutdown function. The erp tries to acquire a reference to run the action asynchronously and fail, if it cannot get the reference. But calling scsi_device_get from slave_destroy will fail, because the scsi_device is already in the state SDEV_DEL. Introduce a new call into the zfcp erp to solve this: The function zfcp_erp_unit_shutdown_wait will close the LUN and wait for the erp to finish without acquiring an additional reference. The wait allows to omit the reference; the caller waiting for the erp to finish already has a reference that holds the struct in place. Reviewed-by:
-
Christof Schmitt authored
These sysfs attributes will require different functions. Implement them without using macros, so that the open coded functions can be changed later. Reviewed-by:
-
Christof Schmitt authored
Move the code for managing zfcp_unit devices to the new file zfcp_unit.c. This is in preparation for the change that zfcp_unit will only track the LUNs configured via unit_add, other data will be moved from zfcp_unit to the new struct zfcp_scsi_dev. Reviewed-by:
-
Christof Schmitt authored
Add a new data structure zfcp_scsi_dev that holds zfcp private data for each SCSI device. Use scsi_transport_reserve_device to let the SCSI midlayer automatically allocate this with each SCSI device. Reviewed-by:
-
Brian King authored
By default, ibmvfc does not log any async events in order to avoid flooding the log with them. Improve on this by logging by default events that are not likely to flood the log, such as link up/down. Having these events in the log will improve the ability to debug issues with ibmvfc. Signed-off-by:
Brian King <brking@linux.vnet.ibm.com> Signed-off-by:
-
Shyam Sundar authored
Signed-off-by:
Giridhar Malavali <giridhar.malavali@qlogic.com> Signed-off-by:
-
Thomas Gleixner authored
Get rid of init_MUTEX[_LOCKED]() and use sema_init() instead. Signed-off-by:
Thomas Gleixner <tglx@linutronix.de> Cc: aacraid@adaptec.com Signed-off-by:
-
Nicholas Bellinger authored
This patch updates the scsi_debug virtual LLD to use root_device_register() and root_device_unregister() from include/linux/device.h instead of device_register() and device_unregister() respectively within scsi_debug_init() and scsi_debug_exit() This simply involved converting the static struct device pseudo_primary into a pointer that is setup by the call to root_device_register(). This patch also contains the correct IS_ERR() conditional check of root_device_register() from within scsi_debug_init(). Thanks to Richard Sharpe and Dmitry Torokhov for their help with this item. Signed-off-by:
Nicholas A. Bellinger <nab@linux-iscsi.org> Acked-by:
Douglas Gilbert <dgilbert@interlog.com> Signed-off-by:
-
- 09 Sep, 2010 7 commits
-
-
James Smart authored
When an rport is "blocked" and a bsg request is received, the bsg request gets placed on the queue but the queue stalls. If the fc object is then deleted - the bsg queue never restarts and keeps the reference on the object, and stops the overall teardown. This patch restarts the bsg queue on teardown and drains any pending requests, allowing the teardown to succeed. Signed-off-by:
Carl Lajeunesse <carl.lajeunesse@emulex.com> Signed-off-by:
James Smart <james.smart@emulex.com> Signed-off-by:
-
Christof Schmitt authored
Make sure that the rport registration did complete and then register SCSI device directly. Otherwise the unit_enqueue would race with the call to zfcp_scsi_queue_unit_register. Reviewed-by:
-
Wayne Boyer authored
The isr optimization patch that was submitted a few months ago exposed problems with receiving and handling spurious HRRQ interrutps. commit 64ffdb76 Author: Wayne Boyer <wayneb@linux.vnet.ibm.com> Date: Wed May 19 11:56:13 2010 -0700 [SCSI] ipr: improve interrupt service routine performance That patch is reverted with this one. A new patch will be submitted once the issue is better understood and properly handled in the driver. Signed-off-by:
Wayne Boyer <wayneb@linux.vnet.ibm.com> Acked-by:
-
Martin K. Petersen authored
The previous thin provisioning support was not very user friendly because it depended on all the relevant options being set on the command line. Implement support for the Thin Provisioning VPD page from SBC3 r24 and add module options for TPU (UNMAP) and TPWS (WRITE SAME (16) with UNMAP bit). This allows us to have sane default and to enable thin provisioning with a simple tpu=1 or tpws=1 on the command line depending on whether we want UNMAP or WRITE SAME behavior. Signed-off-by:
Martin K. Petersen <martin.petersen@oracle.com> Acked-by:
-
Randy Dunlap authored
Fix kconfig dependency warnings in scsi/Kconfig: warning: (SCSI_SAS_LIBSAS && SCSI || SCSI_MPT2SAS && SCSI_LOWLEVEL && PCI && SCSI || FUSION_SAS && FUSION && PCI && SCSI) selects SCSI_SAS_ATTRS which has unmet direct dependencies (SCSI && BLK_DEV_BSG) Signed-off-by:
Randy Dunlap <randy.dunlap@oracle.com> Signed-off-by:
-
Dan Carpenter authored
Return -ENOMEM if the allocations fail. Signed-off-by:
Dan Carpenter <error27@gmail.com> Acked-by:
Jack Wang <jack_wang@usish.com> Signed-off-by:
-
Randy Dunlap authored
Fix typo in scsi_transport_iscsi.c kernel-doc notation: Warning(drivers/scsi/scsi_transport_iscsi.c:548): No description found for parameter 'cmd' Signed-off-by:
Mike Christie <michaelc@cs.wisc.edu> Signed-off-by:
-
- 05 Sep, 2010 21 commits
-
-
Saurav Kashyap authored
Return recovered from pci_mmio_enabled function if it is 82XX. Signed-off-by:
-
Giridhar Malavali authored
The sg table size is increased from 128 to 1024 to support multiple sg lists and number of sg elements per scsi command. Signed-off-by:
-
Giridhar Malavali authored
[SCSI] qla2xxx: Clear local references of rport on device loss timeout notification from FC transport. The clearing of local references to rport was blocked when abort is active. This code clears the local references unconditionaly upon device loss timeout callback from FC transport. Signed-off-by:
-
Madhuranath Iyengar authored
In case the MPI times out, the FW issues an async event AE8002 to indicate this to every FCoE function. The FC/FCoE driver is required to handle this, by doing a soft reset and issuing a Write MPI register mailbox command to reset the MPI. Signed-off-by:
-
Lalit Chandivade authored
Currently the IOs are returned back in slot reset, this could be late and can cause error handler to invoke. If error handler invokes, eh_abort fails and escalate to device/bus/host resets causing issues. The commands need to be returned back to upper layer in io_error_detected. Signed-off-by:
-
Giridhar Malavali authored
Signed-off-by:
-
Giridhar Malavali authored
Signed-off-by:
-
Lalit Chandivade authored
Signed-off-by:
-
Giridhar Malavali authored
Signed-off-by:
-
Giridhar Malavali authored
Signed-off-by:
-
Mike Christie authored
We have been seeing the flush request timeout with a wide range of hardware from tgt+iser to FC targets from a major vendor. After discussions about if the value should be configurable and what the best value should be, this patch just increases the flush/sync cache timeout to 1 minute. 2 minutes was determined to be too long, and making it configurable was troublesome for users. This patch was made over Linus's tree. It is not made over scsi-misc or scsi-rc-fixes, because Linus's had block layer changes that my patch was built over. Signed-off-by:
Jens Axboe <jaxboe@fusionio.com> Signed-off-by:
-
Julia Lawall authored
In this code, 0 is returned on memory allocation failure, even though other failures return -ENOMEM or other similar values. A simplified version of the semantic match that finds this problem is as follows: (http://coccinelle.lip6.fr/) // <smpl> @@ expression ret; expression x,e1,e2,e3; @@ ret = 0 ... when != ret = e1 *x = \(kmalloc\|kcalloc\|kzalloc\)(...) ... when != ret = e2 if (x == NULL) { ... when != ret = e3 return ret; } // </smpl> Signed-off-by:
Julia Lawall <julia@diku.dk> Acked-by:
"Desai, Kashyap" <Kashyap.Desai@lsi.com> Signed-off-by:
-
Eddie Wai authored
Also updated maintainer info. Signed-off-by:
Eddie Wai <eddie.wai@broadcom.com> Signed-off-by:
-
Eddie Wai authored
In the case when an ep_connect request is interrupted due to route request stall, if the iSCSI daemon is terminated by the user, the chip will be left in a state which will not get cleaned up upon module removal. Upon module reload, when the same context id is used for a new connection, chip panic would occur. This patch adds chip cleanup in the module removal path. Signed-off-by:
Benjamin Li <benli@broadcom.com> Reviewed-by:
-
Eddie Wai authored
Specific to the Broadcom 10g chipset, the CFC delete operation must be coupled with the cm_abort/close with does the SRC delete/terminate offload operation prior. Signed-off-by:
Michael Chan <mchan@broadcom.com> Acked-by:
-
Eddie Wai authored
Expanded the TMF request routine to support other TMFs such as LUN RESET, etc. Signed-off-by:
Anil Veerabhadrappa <anilgv@broadcom.com> Reviewed-by:
-
Eddie Wai authored
According to RFC3720, nopout packet sent in response to unsolicited nopin packet requesting a response must retain the TTT of the requester. Signed-off-by:
-
Wayne Boyer authored
Bump the driver version. Signed-off-by:
-
Wayne Boyer authored
This patch fixes a resource address formatting problem where the first byte was being zeroed out. Also, the device ID is now made available as a sysfs attribute. Signed-off-by:
-
kxie@chelsio.com authored
Signed-off-by:
Karen Xie <kxie@chelsio.com> Reviewed-by:
-
kxie@chelsio.com authored
Added cxgb4i iSCSI driver. This patch implements the cxgb4i iscsi connection acceleration for the open-iscsi initiator. The cxgb4i driver offers the iscsi PDU based offload: - digest insertion and verification - payload direct-placement into host memory buffer. Signed-off-by:
-
