- Retry PLOGI up to 48 times when LS_RJT reason is
"Unable to supply requested data."
- When dev loss timeout occures do not change state if there
is an outstanding REG_LOGIN.
- Add logic to ignore REG_LOGIN completion if discovery is
restarted while waiting for REG_LOGIN.
- Only change state on REG_LOGIN completion if still in
state waiting for REG_LOGIN completion.
- Only send ADISCs to FCP-2 Targets (not Initiators).
Signed-off-by: Alex Iannicelli <alex.iannicelli@emulex.com>
Signed-off-by: James Smart <james.smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
A change in the hardware design of the chip for the new adapters changes the
default endianness of MMIO operations. This patch adds a register definition
which when written to with a predefined value will change the endianness
back to what the driver expects.
This patch also fixes two problems found during testing.
First, the first reserved field in the ipr_hostrcb64_fabirc_desc structure only
reserved one byte. The correct amount to reserve is 2 bytes.
Second, the reserved field of the ipr_hostrcb64_error structure only reserved
2 bytes. The correct amount to reserve is 16 bytes.
Signed-off-by: Wayne Boyer <wayneb@linux.vnet.ibm.com>
Acked-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
The function adpt_i2o_post_wait is called from several places, in some of
which, such as adpt_abort, a lock may be held.
The functions adpt_i2o_reparse_lct and adpt_i2o_lct_get are called from
several places, including adpt_rescan where a lock may be held.
The semantic patch that makes this change is as follows:
(http://coccinelle.lip6.fr/)
// <smpl>
@gfp exists@
identifier fn;
position p;
@@
fn(...) {
... when != spin_unlock_irqrestore
when any
GFP_KERNEL@p
... when any
}
@locked@
identifier gfp.fn;
@@
spin_lock_irqsave(...)
... when != spin_unlock_irqrestore
fn(...)
@depends on locked@
position gfp.p;
@@
- GFP_KERNEL@p
+ GFP_ATOMIC
// </smpl>
Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Driver was not unprotecting correctly, use correct bits
to unprotect the flash on ISP 82xx.
Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Outstanding mailbox commands, have no way to recover on f/w hung, and we
timeout on waiting for mbx response. This in turn affects the recovery process
as follows:
- We might already be in dpc while waiting for mbx to complete, so recovery for
that pci function will never get invoked. Reset Timeout (10 sec) is far less
than mbx timeout (30 sec).
- Other mbx cmds will get stuck due to serial mbx access.
Solution is to identify fw-hung scenario and handle outstanding mbx commands to
have an early-exit instead of waiting for response.
Other mbx commands waiting for access will also do an early-exit if fw-hung is
still applicable.
Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
After topology change ISP is reset and VPs are re-enabled.
If the topology is not fabric, VPs could falsely get enabled.
Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Used bootloder address from FLT while loading FW from flash as well.
Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
In case of both target and initiator capabilities reported by fc port,
the fc port port capability is made mutualy exclusive with priority given
for target capabilities.
Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
* hold the hardware_lock throughout the duration of ctx-sp
timeout handling -- could result in use-after-free oops.
* retry a timed-out login-request.
* done() routines are called with the hardware-lock held, issue
qla2x00_mark_device_lost() with proper 'defer' flag.
* FCP2 capabilities are only relevant to target devices.
Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Hold a reference to the srb (sp) while aborting an I/O -- as the
I/O can/will complete from within the interrupt-context.
Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
The TMFs used for pre-24xx ISPs incorrectly assumed 'cpu' tag
data could be valid. These chips have no multi-q/cpu-affinity
support. This corrects an oops seen on ISP23xx parts.
Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
This helps to correlate submission/completion messages during
triaging.
Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
The CRB drive active register is cleared when driver is unloaded
or when driver enters failed state.
Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
We have 32 (MAXSGENTRIES) scatter gather elements embedded
in the command. With all these, the total command size is
about 576 bytes. However, the last entry in the block fetch table
is 35. (the block fetch table contains the number of 16-byte chunks
the firmware needs to fetch for a given number of scatter gather
elements.) 35 * 16 = 560 bytes, which isn't enough. It needs to be
36. (36 * 16 == 576) or, MAXSGENTRIES + 4. (plus 4 because there's a
bunch of stuff at the front of the command before the first scatter
gather element that takes up 4 * 16 bytes.) Without this fix, the
controller may have to perform two DMA operations to fetch the
command since the first one may not get the whole thing.
Signed-off-by: Don Brace <brace@beardog.cce.hp.com>
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
before trying to enter simple mode transport method.
Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
The end of the function is reachable both when host is and is not NULL.
A simplified version of the semantic match that finds this problem is as
follows: (http://coccinelle.lip6.fr/)
// <smpl>
@r exists@
expression E,E1;
identifier f;
statement S1,S2,S3;
@@
if ((E == NULL && ...) || ...)
{
... when != if (...) S1 else S2
when != E = E1
* E->f
... when any
return ...;
}
else S3
// </smpl>
Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
I may have an explanation for the LSI 1068 HBA hangs provoked by ATA
pass-through commands, in particular by smartctl.
First, my version of the symptoms. On an LSI SAS1068E B3 HBA running
01.29.00.00 firmware, with SATA disks, and with smartd running, I'm seeing
occasional task, bus, and host resets, some of which lead to hard faults of
the HBA requiring a reboot. Abusively looping the smartctl command,
# while true; do smartctl -a /dev/sdb > /dev/null; done
dramatically increases the frequency of these failures to nearly one per
minute. A high IO load through the HBA while looping smartctl seems to
improve the chance of a full scsi host reset or a non-recoverable hang.
I reduced what smartctl was doing down to a simple test case which
causes the hang with a single IO when pointed at the sd interface. See
the code at the bottom of this e-mail. It uses an SG_IO ioctl to issue
a single pass-through ATA identify device command. If the buffer
userspace gives for the read data has certain alignments, the task is
issued to the HBA but the HBA fails to respond. If run against the sg
interface, neither the test code nor smartctl causes a hang.
sd and sg handle the SG_IO ioctl slightly differently. Unless you
specifically set a flag to do direct IO, sg passes a buffer of its own,
which is page-aligned, to the block layer and later copies the result
into the userspace buffer regardless of its alignment. sd, on the other
hand, always does direct IO unless the userspace buffer fails an
alignment test at block/blk-map.c line 57, in which case a page-aligned
buffer is created and used for the transfer.
The alignment test currently checks for word-alignment, the default
setup by scsi_lib.c; therefore, userspace buffers of almost any
alignment are given directly to the HBA as DMA targets. The LSI 1068
hardware doesn't seem to like at least a couple of the alignments which
cross a page boundary (see the test code below). Curiously, many
page-boundary-crossing alignments do work just fine.
So, either the hardware has an bug handling certain alignments or the
hardware has a stricter alignment requirement than the driver is
advertising. If stricter alignment is required, then in no case should
misaligned buffers from userspace be allowed through without being
bounced or at least causing an error to be returned.
It seems the mptsas driver could use blk_queue_dma_alignment() to advertise
a stricter alignment requirement. If it does, sd does the right thing and
bounces misaligned buffers (see block/blk-map.c line 57). The following
patch to 2.6.34-rc5 makes my symptoms go away. I'm sure this is the wrong
place for this code, but it gets my idea across.
Acked-by: "Desai, Kashyap" <Kashyap.Desai@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Adding DIF Type 2 protection support, as well as turning on 32 byte cdb's,
and setting the cdb length for > 16 byte in the SCSI_IO->control parameter.
Signed-off-by: Martin Petersen <martin.petersen@oracle.com>
Signed-off-by: Eric Moore <eric.moore@lsi.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
A driver needs to be ready to take an interrupt as soon as it registers
an interrupt handler. I noticed the following oops when testing kdump:
ipr: IBM Power RAID SCSI Device Driver version: 2.5.0 (February 11, 2010)
ibmvscsi 30000002: SRP_VERSION: 16.a
ibmvscsi 30000002: SRP_VERSION: 16.a
Unable to handle kernel paging request for data at address 0x00000000
...
pc: c000000004085e34: .tasklet_action+0xf4/0x1dc
...
c000000004086fe4 .__do_softirq+0x16c/0x2c0
c00000000403138c .call_do_softirq+0x14/0x24
c00000000400ee14 .do_softirq+0xa0/0x104
c00000000408690c .irq_exit+0x70/0xd0
c00000000400f190 .do_IRQ+0x214/0x2a8
c000000004004804 hardware_interrupt_entry+0x1c/0x98
--- Exception: 501 (Hardware Interrupt) at c00000000400c544 .raw_local_irq_restore+0x48/0x54
c00000000465d2a8 ._raw_spin_unlock_irqrestore+0x74/0xa0
c0000000040e7f00 .__setup_irq+0x2ec/0x3f0
c0000000040e8198 .request_threaded_irq+0x194/0x22c
c00000000446d854 .rpavscsi_init_crq_queue+0x284/0x3f0
c00000000446c764 .ibmvscsi_probe+0x688/0x710
c00000000402903c .vio_bus_probe+0x37c/0x3e4
c000000004403f10 .driver_probe_device+0xec/0x1b8
c000000004404088 .__driver_attach+0xac/0xf4
c000000004403184 .bus_for_each_dev+0x98/0x104
c000000004403c98 .driver_attach+0x40/0x60
c0000000044026f0 .bus_add_driver+0x154/0x324
c0000000044045d0 .driver_register+0xe8/0x1ac
c00000000402b2a8 .vio_register_driver+0x54/0x74
c000000004933ea4 .ibmvscsi_module_init+0x80/0xc0
c000000004009834 .do_one_initcall+0x98/0x1d8
c0000000049005b4 .kernel_init+0x27c/0x33c
c000000004031550 .kernel_thread+0x54/0x70
srp_task needs to be setup before request_irq. The patch below fixes the oops.
Signed-off-by: Anton Blanchard <anton@samba.org>
Acked-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Commit 64deb6efdc changed the way status
read buffers are handled but forgot to adjust the mempool to the new
size. Add the call to resize the mempool after the exchange config
data. Also use the define instead of the hard coded number in the fsf
callback for consistency.
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>
Trying to read the FC host statistics on an offline adapter results in
a 5 seconds wait. Reading the statistics tries to issue an exchange
port data request which first waits up to 5 seconds for an entry in
the request queue.
Change the strategy for getting a free SBAL to exit when the queue is
stopped. Reading the statistics will then fail without the wait.
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>
A false check was performed whether an unchained ct_els
is possible or not.
Signed-off-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>
It was possible to overflow the buffer used to print out the formatted
version of the resource path. The fix is to limit the number of
bytes that get formatted.
This patch also updates the ipr_show_resource_path function to display the
resource address for devices that are attached to adapters that don't
support resource paths.
Signed-off-by: Wayne Boyer <wayneb@linux.vnet.ibm.com>
Acked-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
