summaryrefslogtreecommitdiff
path: root/Documentation
diff options
context:
space:
mode:
Diffstat (limited to 'Documentation')
-rw-r--r--Documentation/scsi/ChangeLog.megaraid_sas23
-rw-r--r--Documentation/scsi/hpsa.txt23
-rw-r--r--Documentation/scsi/scsi_mid_low_api.txt14
3 files changed, 59 insertions, 1 deletions
diff --git a/Documentation/scsi/ChangeLog.megaraid_sas b/Documentation/scsi/ChangeLog.megaraid_sas
index b64d10d221ec..4d9ce73ff730 100644
--- a/Documentation/scsi/ChangeLog.megaraid_sas
+++ b/Documentation/scsi/ChangeLog.megaraid_sas
@@ -1,3 +1,26 @@
+Release Date : Thu. Feb 24, 2011 17:00:00 PST 2010 -
+ (emaild-id:megaraidlinux@lsi.com)
+ Adam Radford
+Current Version : 00.00.05.34-rc1
+Old Version : 00.00.05.29-rc1
+ 1. Fix some failure gotos from megasas_probe_one(), etc.
+ 2. Add missing check_and_restore_queue_depth() call in
+ complete_cmd_fusion().
+ 3. Enable MSI-X before calling megasas_init_fw().
+ 4. Call tasklet_schedule() even if outbound_intr_status == 0 for MFI based
+ boards in MSI-X mode.
+ 5. Fix megasas_probe_one() to clear PCI_MSIX_FLAGS_ENABLE in msi control
+ register in kdump kernel.
+ 6. Fix megasas_get_cmd() to only print "Command pool empty" if
+ megasas_dbg_lvl is set.
+ 7. Fix megasas_build_dcdb_fusion() to not filter by TYPE_DISK.
+ 8. Fix megasas_build_dcdb_fusion() to use io_request->LUN[1] field.
+ 9. Add MR_EVT_CFG_CLEARED to megasas_aen_polling().
+ 10. Fix tasklet_init() in megasas_init_fw() to use instancet->tasklet.
+ 11. Fix fault state handling in megasas_transition_to_ready().
+ 12. Fix max_sectors setting for IEEE SGL's.
+ 13. Fix iMR OCR support to work correctly.
+-------------------------------------------------------------------------------
Release Date : Tues. Dec 14, 2010 17:00:00 PST 2010 -
(emaild-id:megaraidlinux@lsi.com)
Adam Radford
diff --git a/Documentation/scsi/hpsa.txt b/Documentation/scsi/hpsa.txt
index dca658362cbf..891435a72fce 100644
--- a/Documentation/scsi/hpsa.txt
+++ b/Documentation/scsi/hpsa.txt
@@ -28,6 +28,12 @@ boot parameter "hpsa_allow_any=1" is specified, however these are not tested
nor supported by HP with this driver. For older Smart Arrays, the cciss
driver should still be used.
+The "hpsa_simple_mode=1" boot parameter may be used to prevent the driver from
+putting the controller into "performant" mode. The difference is that with simple
+mode, each command completion requires an interrupt, while with "performant mode"
+(the default, and ordinarily better performing) it is possible to have multiple
+command completions indicated by a single interrupt.
+
HPSA specific entries in /sys
-----------------------------
@@ -39,6 +45,8 @@ HPSA specific entries in /sys
/sys/class/scsi_host/host*/rescan
/sys/class/scsi_host/host*/firmware_revision
+ /sys/class/scsi_host/host*/resettable
+ /sys/class/scsi_host/host*/transport_mode
the host "rescan" attribute is a write only attribute. Writing to this
attribute will cause the driver to scan for new, changed, or removed devices
@@ -55,6 +63,21 @@ HPSA specific entries in /sys
root@host:/sys/class/scsi_host/host4# cat firmware_revision
7.14
+ The transport_mode indicates whether the controller is in "performant"
+ or "simple" mode. This is controlled by the "hpsa_simple_mode" module
+ parameter.
+
+ The "resettable" read-only attribute indicates whether a particular
+ controller is able to honor the "reset_devices" kernel parameter. If the
+ device is resettable, this file will contain a "1", otherwise, a "0". This
+ parameter is used by kdump, for example, to reset the controller at driver
+ load time to eliminate any outstanding commands on the controller and get the
+ controller into a known state so that the kdump initiated i/o will work right
+ and not be disrupted in any way by stale commands or other stale state
+ remaining on the controller from the previous kernel. This attribute enables
+ kexec tools to warn the user if they attempt to designate a device which is
+ unable to honor the reset_devices kernel parameter as a dump device.
+
HPSA specific disk attributes:
------------------------------
diff --git a/Documentation/scsi/scsi_mid_low_api.txt b/Documentation/scsi/scsi_mid_low_api.txt
index df322c103466..5f17d29c59b5 100644
--- a/Documentation/scsi/scsi_mid_low_api.txt
+++ b/Documentation/scsi/scsi_mid_low_api.txt
@@ -1343,7 +1343,7 @@ Members of interest:
underruns (overruns should be rare). If possible an LLD
should set 'resid' prior to invoking 'done'. The most
interesting case is data transfers from a SCSI target
- device device (i.e. READs) that underrun.
+ device (e.g. READs) that underrun.
underflow - LLD should place (DID_ERROR << 16) in 'result' if
actual number of bytes transferred is less than this
figure. Not many LLDs implement this check and some that
@@ -1351,6 +1351,18 @@ Members of interest:
report a DID_ERROR. Better for an LLD to implement
'resid'.
+It is recommended that a LLD set 'resid' on data transfers from a SCSI
+target device (e.g. READs). It is especially important that 'resid' is set
+when such data transfers have sense keys of MEDIUM ERROR and HARDWARE ERROR
+(and possibly RECOVERED ERROR). In these cases if a LLD is in doubt how much
+data has been received then the safest approach is to indicate no bytes have
+been received. For example: to indicate that no valid data has been received
+a LLD might use these helpers:
+ scsi_set_resid(SCpnt, scsi_bufflen(SCpnt));
+where 'SCpnt' is a pointer to a scsi_cmnd object. To indicate only three 512
+bytes blocks has been received 'resid' could be set like this:
+ scsi_set_resid(SCpnt, scsi_bufflen(SCpnt) - (3 * 512));
+
The scsi_cmnd structure is defined in include/scsi/scsi_cmnd.h