27 Jul, 2008
40 commits
-
The last_sector_bug flag was added to work around a bug in certain usb
cardreaders, where they would crash if a multiple sector read included the
last sector. The original implementation avoids this by e.g. splitting an 8
sector read which includes the last sector into a 7 sector read, and a single
sector read for the last sector. The flag is enabled for all USB devices.This revealed a second bug in other usb cardreaders, which crash when they
get a multiple sector read which stops 1 sector short of the last sector.
Affected hardware includes the Kingston "MobileLite" external USB cardreader
and the internal USB cardreader on the Asus EeePC.Extend the last_sector_bug workaround to ensure that any access which touches
the last 8 hardware sectors of the device is a single sector long. Requests
are shrunk as necessary to meet this constraint.This gives us a safety margin against potential unknown or future bugs
affecting multi-sector access to the end of the device. The two known bugs
only affect the last 2 sectors. However, they suggest that these devices
are prone to fencepost errors and that multi-sector access to the end of the
device is not well tested. Popular OS's use multi-sector accesses, but they
rarely read the last few sectors. Linux (with udev & vol_id) automatically
reads sectors from the end of the device on insertion. It is assumed that
single sector accesses are more thoroughly tested during development.Signed-off-by: Alan Jenkins
Tested-by: Alan Jenkins
Signed-off-by: James Bottomley -
Signed-off-by: Andrew Vasquez
Signed-off-by: James Bottomley -
Minor fixes addressing:
- rport managements during vport deletion.
- acquire proper physical-ha during qla24xx_abort_command() and
qla24xx_queuecommand()
- do not needlessly acquire the pha for non-NPIV capable ISPs.Signed-off-by: Andrew Vasquez
Signed-off-by: James Bottomley -
drivers/scsi/qla2xxx/qla_os.c: In function 'qla2x00_post_work':
drivers/scsi/qla2xxx/qla_os.c:2158: warning: 'flags' may be used uninitialized in this functionSigned-off-by: Andrew Morton
Signed-off-by: Andrew Vasquez
Signed-off-by: James Bottomley -
Signed-off-by: Akinobu Mita
Signed-off-by: Andrew Morton
Signed-off-by: Andrew Vasquez
Signed-off-by: James Bottomley -
As the original code would incorrectly call the non-ISP24xx/25xx
callbacks during recovery, a stop-firmware failure could result
in improper bit-banging of the RISC and in some cases manifest in
a NMI-watchdog trigger due to the RISC not coming out of its
reset state.Signed-off-by: Andrew Vasquez
Signed-off-by: James Bottomley -
The following commit causes ch_remove oops:
commit 24b42566c3fcbb5a9011d1446783d0f5844ccd45
Author: Greg Kroah-Hartman
Date: Fri May 16 17:55:12 2008 -0700SCSI: fix race in device_create
There is a race from when a device is created with device_create() and
then the drvdata is set with a call to dev_set_drvdata() in which a
sysfs file could be open, yet the drvdata will be NULL, causing all
sorts of bad things to happen.This patch fixes the problem by using the new function,
device_create_drvdata(). It fixes the problem in all of the scsi
drivers that need it.Cc: Kay Sievers
Cc: Doug Gilbert
Cc: James E.J. Bottomley
Signed-off-by: Greg Kroah-HartmanThe problem is ch_probe stores ch's private data at a wrong place.
We need to store it at scsi_device->sdev_gendev but the above patch
stores it at device struct that device_create_drvdata returns. So we
hit an oops when ch_remove accesses
scsi_device->sdev_gendev->driver_data, which is NULL.Actually, there wasn't a race because ch doesn't create sysfs files
with device struct that device_create returns. This patch puts back
dev_set_drvdata() to set ch's private data properly.Signed-off-by: FUJITA Tomonori
Signed-off-by: James Bottomley -
This patch for the 3w-9xxx scsi driver applies on top of the
BKL-pushdown changes in -git9.This patch does the following:
- Increase max AENs drained to 256.
- Add MSI support and "use_msi" module parameter.
- Fix bug in twa_get_param() on 4GB+.
- Use pci_resource_len() for ioremap().Signed-off-by: James Bottomley
-
I goofed and did not see the macro for checking if a request is tagged.
This patch has us use blk_rq_tagged instead of digging into the req->tag.Patch was made over scsi-misc.
Signed-off-by: Mike Christie
Signed-off-by: James Bottomley -
Update driver version to 1.0.1.
Signed-off-by: Brian King
Signed-off-by: James Bottomley -
Add an ADISC to the target discovery job in order to sanity check whether or
not we need to re-login to the target.Signed-off-by: Brian King
Signed-off-by: James Bottomley -
Properly setup the size of the async event queue. This fixes a bug where async events
were not getting processed by the driver.Setup target_id field in the driver's target struct so that target sysfs attributes
work for multiple targets.Signed-off-by: Brian King
Signed-off-by: James Bottomley -
If certain ELS events are received during module removal, after the kthread
is stopped, the rmmod can hang. This fixes the ibmvfc driver so that ELS
events during rmmod are ignored by stopping all device activity prior to
killing the kthread and also changes reinitialization to not attempt a reinit
if the adapter has been taken offline.Signed-off-by: Brian King
Signed-off-by: James Bottomley -
Fix up some refcounting on the ibmvfc drivers internal target struct
when accessed through some sysfs attributes.Signed-off-by: Brian King
Signed-off-by: James Bottomley -
Reduces some unnecessary log noise by removing a printk during
host port state query and increasing the log level required to
log received async events.Signed-off-by: Brian King
Signed-off-by: James Bottomley -
This patch frees the luntbl dma area in sym_hcb_free if allocated.
Since the luntbl is part of a larger dma coherent area not freeing the
luntbl kept a 64k dma coherent area previous allocated through
dma_alloc_coherent allocated. This prevented a DLPAR remove IO
operation from completing successfully.Signed-off-by: Mike Anderson
Cc: Matthew Wilcox
Signed-off-by: James Bottomley -
The mutex is released on a successful return, so it would seem that it
should be released on an error return as well.The semantic patch that makes this change is as follows:
(http://www.emn.fr/x-info/coccinelle/)//
@@
expression l;
@@mutex_lock(l);
... when != mutex_unlock(l)
when any
when strict
(
if (...) { ... when != mutex_unlock(l)
+ mutex_unlock(l);
return ...;
}
|
mutex_unlock(l);
)
//Signed-off-by: Julia Lawall
Signed-off-by: James Bottomley -
This patch (as1116) fixes a bug in scsi_eh_prep_cmnd() and
scsi_eh_restore_cmnd(). These routines are supposed to save any
values they change and restore them later, but someone forgot to
save & restore scmd->underflow.This fixes part of the problem reported in Bugzilla #9638.
[jejb: fix up rejections around DIF/DIX]
Signed-off-by: Alan Stern
Signed-off-by: James Bottomley -
Support for controllers and disks that implement DIF protection
information:- During command preparation the RDPROTECT/WRPROTECT must be set
correctly if the target has DIF enabled.- READ(6) and WRITE(6) are not supported when DIF is on.
- The controller must be told how to handle the I/O via the
protection operation field in scsi_cmnd.- Refactor the I/O completion code that extracts failed LBA from the
returned sense data and handle DIF failures correctly.- sd_dif.c implements the functions required to prepare and complete
requests with protection information attached.Signed-off-by: Martin K. Petersen
Signed-off-by: James Bottomley -
If a disk is formatted with protection information (Inquiry bit
PROTECT=1) it is required to support Read Capacity(16). Force use of
the 16-bit command in this case and extract the P_TYPE field which
indicates whether the disk is formatted using DIF Type 1, 2 or 3.The ATO (App Tag Own) bit in the Control Mode Page indicates whether
the storage device or the initiator own the contents of the
DIF application tag.Signed-off-by: Martin K. Petersen
Signed-off-by: James Bottomley -
If initiator or target reject the I/O due to DIF errors there is no
point in retrying.Signed-off-by: Martin K. Petersen
Signed-off-by: James Bottomley -
Implement support for DMA of protection information for devices that
are data integrity capable.- Add support for mapping an extra scatter-gather list containing
the protection information.- Allocate protection scsi_data_buffer if host is DIX (integrity DMA)
capable.- Accessor function for checking whether a device has protection
enabled.Signed-off-by: Martin K. Petersen
Signed-off-by: James Bottomley -
Controllers that support DMA of protection information must be told
explicitly how to handle the I/O. The controller has no knowledge of
the protection capabilities of the target device so this information
must be passed in the scsi_cmnd.- The protection operation tells the HBA whether to generate, strip or
verify protection information.- The protection type tells the HBA which layout the target is
formatted with. This is necessary because the controller must be
able to correctly interpret the included protection information in
order to verify it.- When a scsi_cmnd is reused for error handling the protection
operation must be cleared and saved while error handling is in
progress.- prot_op and prot_type are placed in an existing hole in scsi_cmnd
and don't cause the structure to grow.Signed-off-by: Martin K. Petersen
Signed-off-by: James Bottomley -
Controllers that support protection information must indicate this to
the SCSI midlayer so that the ULD can prepare scsi_cmnds accordingly.This patch implements a host mask and various types of protection:
- DIF Type 1-3 (between HBA and disk)
- DIX Type 0-3 (between OS and HBA)The patch also allows the HBA to set the guard type to something
different than the T10-mandated CRC.Signed-off-by: Martin K. Petersen
Signed-off-by: James Bottomley -
Create a cache of devices that are seen in a system. This will avoid
the unnecessary traversal of the device list in the scsi_dh when there
are multiple luns of a same type.Signed-off-by: Chandra Seetharaman
Signed-off-by: Hannes Reinecke
Signed-off-by: James Bottomley -
multipath keeps a separate device table which may be
more current than the built-in one.
So we should make sure to always call ->attach whenever
a multipath map with hardware handler is instantiated.
And we should call ->detach on removal, too.[sekharan: update as per comments from agk]
Signed-off-by: Hannes Reinecke
Signed-off-by: Chandra Seetharaman
Signed-off-by: James Bottomley -
Signed-off-by: Hannes Reinecke
Signed-off-by: Chandra Seetharaman
Signed-off-by: James Bottomley -
This patch updates the RDAC device handler to
refuse to attach to devices not supporting the
RDAC vpd pages.Signed-off-by: Hannes Reinecke
Signed-off-by: Chandra Seetharaman
Signed-off-by: James Bottomley -
This patch updates the hp_sw device handler to properly
check the return codes etc.
And adds the 'correct' machine definitions.Signed-off-by: Hannes Reinecke
Signed-off-by: Chandra Seetharaman
Signed-off-by: James Bottomley -
This patch converts the EMC device handler to use a proper
state machine. We now also parse the extended INQUIRY
information to determine if long trespass commands are
supported. And we're now using the long trespass command
correctly. And finally there's now an check at init time
to refuse to attach to devices not supporting EMC-specific
VPD pages.Signed-off-by: Hannes Reinecke
Signed-off-by: Chandra Seetharaman
Signed-off-by: James Bottomley -
Implement a 'dh_state' sdev attribute for dynamic device handler
manipulation. A read on the attribute will return the name of
the currently attached device handler or 'detached' if no handler
is attached.
The attribute allows the following strings to be written:
- The name of the device handler to be attached if the state is
'detached'.
- 'activate' to trigger path activation if a device handler
is attached.
- 'detach' to detach the currently attached device handler.Signed-off-by: Hannes Reinecke
Signed-off-by: Chandra Seetharaman
Signed-off-by: James Bottomley -
Instead of having each and every driver implement its own
device table scanning code we should rather implement a common
routine and scan the device tables there.
This allows us also to implement a general notifier chain
callback for all device handler instead for one per handler.[sekharan: Fix rejections caused by conflicting bug fix]
Signed-off-by: Hannes Reinecke
Signed-off-by: Chandra Seetharaman
Signed-off-by: James Bottomley -
Daniel Debonzi reports that he has managed to wrap host_no. Increasing
the number of host numbers available to 32-bit from 16-bit allows the
problem to be evaded for another hundred years.Signed-off-by: Matthew Wilcox
Signed-off-by: James Bottomley -
Update index allocation as follows.
* sd_index_idr is used only for ID allocation and mapping
functionality is not used. Use more memory efficient ida instead.* idr and ida have their own locks inside them and don't need them for
operation. Drop it.* index wasn't freed if probing failed after index allocation. fix
it.* ida allocation should be repeated if it fails with -EAGAIN.
Signed-off-by: Tejun Heo
Signed-off-by: James Bottomley -
When drivers use a shared tag map we can end up with more requests
than tags, because the tag map is shost->can_queue tags and there
can be sdevs * sdev->queue_depth requests. In scsi_request_fn
if tag allocation fails we just drop down to just dequeueing the
tag without a tag. The problem is that drivers using the shared tag
map rely on a valid tag always being set, because it will use the
tag number to lookup commands later.This patch has us check if we got a valid tag when the host lock
is held right before we check if the host queue is ready. We do the
check here because to allocate the tag we need the q lock, but
if the tag is bad we want to add the device/q onto the starved list
which requires the host lock.Signed-off-by: Mike Christie
Signed-off-by: James Bottomley -
We want to set the queue depth to something reasonable - not
the can_queue.Signed-off-by: Mike Christie
Cc: Ed Lin
Signed-off-by: James Bottomley -
We want to set the queue depth to something reasonable - not
the can_queue.Signed-off-by: Mike Christie
Cc: David Somayajulu
Signed-off-by: James Bottomley -
Currently qla4xxx and stex pass in their can_queue values into
scsi_activate_tcq because they wanted the tag map that large.
The problem with this is that it ends up also setting the queue
depth to that large value. All we want to do this in this case
is set the device queue depth and the other device settings.
We do not need to touch the tag map sizing because the drivers
had setup that map according to their can_queue limits when the
shared map was created.The scsi mid layer in request_fn will then handle the case where we
have more requests than available tags when it checks the host
queue ready function.Signed-off-by: Mike Christie
Signed-off-by: James Bottomley -
Mike Christie noticed a bogus memset. It can be removed as dead code
since the number of bytes in the driver buffer in fixed block mode is
always a multiple of the tape block size.Signed-off-by: Kai Mäkisara
Signed-off-by: James Bottomley -
Move buffer pointer back when data could not be written. Bug found by
Mike Christie.Signed-off-by: Kai Mäkisara
Signed-off-by: James Bottomley
