Various PCI bus errors can be signaled by newer PCI controllers. This
patch adds the PCI error recovery callbacks to the IPR SCSI device driver.
The patch has been tested, and appears to work well.
Signed-off-by: Linas Vepstas <linas@linas.org>
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Since scsi core is always sending scatterlists now, remove
some code which was written with the bad assumption that
a small transfer would not be sent down in a scatterlist.
Without this fix, the ipr driver ends up sending garbage
data to the adapter following a reset, causing it to
fail the reset and take the adapter offline.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Leave the overloaded "hotplug" word to susbsystems which are handling
real devices. The driver core does not "plug" anything, it just exports
the state to userspace and generates events.
Signed-off-by: Kay Sievers <kay.sievers@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The distinction between hotplug and uevent does not make sense these
days, netlink events are the default.
udev depends entirely on netlink uevents. Only during early boot and
in initramfs, /sbin/hotplug is needed. So merge the two functions and
provide only one interface without all the options.
The netlink layer got a nice generic interface with named slots
recently, which is probably a better facility to plug events for
subsystem specific events.
Also the new poll() interface to /proc/mounts is a nicer way to
notify about changes than sending events through the core.
The uevents should only be used for driver core related requests to
userspace now.
Signed-off-by: Kay Sievers <kay.sievers@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
When kexec booting a kernel when the previous kernel did not
call ipr's shutdown method, the ipr adapter does not get
properly initialized, which can result in the ipr adapter
completing commands issued by the previous kernel. Fix ipr
to detect this scenario by reading the adapter's interrupt
mask register and the microprocessor interrupt register.
If the interrupt mask register indicates that interrupts
are enabled or the reset alert bit is set when the card is
probed, this means the card is in an unknown state and we
hard reset the card.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Some new ipr adapters do not support some of the initialization
commands currently sent to it from the driver. Handle these
commands failing and continue on with the adapter initialization.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Support now exists in some ipr adapters to issue a device reset
to an Advanced Function disk.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
New ipr adapters support a new device queueing model in the
adapter firmware. The queueing model is the NACA queueing model,
but it does not mean use of NACA is required. The new model removes
some of the adapter firmware queue state that made handling QERR=0
almost impossible. The queueing model on older adapters included the
concept of a queue frozen state, which would freeze the response
queue in the adapter when a check condition occurred, requiring a
a primitive to resume the queue. The new queueing model removes this
complexity.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Handle some new types of ipr errors that can be returned by the adapter.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Some ipr adapters will automatically create single device
RAID 0 arrays for all unconfigured RAID capable devices found
at adapter initialization time. This patch adds a module parameter
to disable this behavior.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Some IPR RAID adapter will automatically create single device RAID arrays
for all attached devices when the card is initialized. Setting the
RUNTIME_RESET doorbell bit will prevent this from occurring, since we
only want this behavior the first time the card is initialized and not
each time the card happens to get reset.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Add support for handling some new errors that may be returned
by ipr adapters.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
If an ipr adapter repeatedly fails its initialization
the ipr driver will take the adapter offline and never talk
to it again. This provides a method for the user to manually
try the initialization again through sysfs.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Make some compile time debugging options runtime module options.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
If the write buffer command that is issued to the ipr adapter
to update its microcode fails for some reason, the DMA buffer
will never get unmapped. Move the pci_map/unmap out of the
IOA reset job so that the buffer is always clearly mapped
and unmapped.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Convert appropriate kmalloc/memset calls to use kzalloc.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Adds a scsi_host sysfs attribute and module parm to enable/disable
the write cache on an ipr adapter.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Optimize ipr's slave_alloc to return -ENXIO for devices that
do not exist.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Set the no_uld_attach for devices ipr does not want
upper layer drivers to attach to. These devices are
only reported for RAID management and only sg should
be used to talk to them.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Fix ipr to include all disks in the supported device list,
not just disks formatted to advanced function format.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Simplify error logging path, sanitize error length returned
by the adapter.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Better handle errors received which are not known to the device driver.
Just dump the hex data so that we have a hope of figuring out what
went wrong.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
The generic ipr adapter error log currently logs 2 lines of useless
data. Delete these lines.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Adds a macro in the ipr driver for logging a physical device location.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Simplify the ipr error structures a bit by removing some duplication.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
IPR RAID arrays show up on a virtual scsi bus, with a scsi bus number
of 255, which is generated by the adapter microcode. For the initial
scan of the host, we manually scan this bus since it does not obey
SAM in regards to sparse LUNs and the disk array devices do not have
a consistent product id to use scsi core's blacklist. If /proc/scsi/scsi
or sysfs is used to delete one of these devices, the device will not
be able to get added back by rescanning the host since scsi core
will see ipr's max_channel as 4, rather than 255. Update max_channel
after the initial scan so that ipr raid arrays can get re-added
if they get deleted.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
IPR scsi adapter have an exposure today in that they issue BIST to the adapter
to reset the card. If, during the time it takes to complete BIST, userspace
attempts to access PCI config space, the host bus bridge will master abort the
access since the ipr adapter does not respond on the PCI bus for a brief
period of time when running BIST. On PPC64 hardware, this master abort
results in the host PCI bridge isolating that PCI device from the rest of the
system, making the device unusable until Linux is rebooted. This patch makes
use of some newly added PCI layer APIs that allow for protection from
userspace accessing config space of a device in scenarios such as this.
Signed-off-by: Brian King <brking@us.ibm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: James Bottomley <James.Bottomley@steeleye.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
drivers/scsi/ipr.c | 2 ++
1 file changed, 2 insertions(+)
Now we can change the pci core to always set this pointer, as pci drivers
should use it, not the driver core callback.
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Currently, during PCI hotplug remove, if the upper layer
drivers of the attached devices send commands down as part
of the remove action, like a CDROM, the hotplug action
will hang forever due to the ipr driver returning
SCSI_MLQUEUE_HOST_BUSY. Patch fixes this.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Bugme 4547. The following patch fixes a bug in ipr's error logging.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Initial git repository build. I'm not bothering with the full history,
even though we have it. We can create a separate "historical" git
archive of that later if we want to, and in the meantime it's about
3.2GB when imported into git - space that would just make the early
git days unnecessarily complicated, when we don't have a lot of good
infrastructure for it.
Let it rip!
Linas Vepstas