Do not flush queues then block session. This will cause commands
to needlessly swing around on us and remove goofy
recovery_failed field and replace with state value.
And do not start recovery from within the host reset function.
This causeis too many problems becuase open-iscsi was desinged to
call out to userspace then have userpscae decide if we should
go into recovery or kill the session.
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Patch from david.somayajulu@qlogic.com and cleaned up by Tomo.
qla4xxx is going to have a different daemon so this patch
just routes the events to the right daemon.
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Discovered by steven@hayter.me.uk and patch by michaelc@cs.wisc.edu
The dtask mempool is reserving 261120 items per session! Since we are now
sending headers with sendmsg there is no reason for the mempool and that
was causing us to us carzy amounts of mem. We can preallicate a header in
the r2t and task struct and reuse them
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
We only use the mtask data buffer for login tasks so we do not
need to preallocate a buffer for every mtask. This saves
8 * 31 KB.
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
From Zhen and ported by Mike:
Don't use sendpage for the headers. sendpage for the pdu headers
does not seem to have a performance impact, makes life harder
for mutiple data pdus to be in flight and still trips up some
network cards when it is from slab mem.
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Received From Mark Salyzyn
The queue tracking is just not being used, not even for debugging. Information
about outstanding commands can be acquired from the scsi structures.
Signed-off-by: Mark Haverkamp <markh@osdl.org>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Received From Mark Salyzyn
A race condition existed that could result in a lost completion of a
command to the ppc based cards.
Signed-off-by: Mark Haverkamp <markh@osdl.org>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Received From Mark Salyzyn
Add the ability to adjust for unusual corner case failures. Both of
these additional module parameters deal with embedded, non-intel or
complicated system scenarios.
Aif_timeout can be increased past the default 2 minute timeout to drop
application registrations when a system has an unusually high event load
resulting from continuing management requests, or simultaneous builds,
or sluggish user space as a result of system load.
Startup_timeout can be increased past the default 3 minute timeout to
drop an adapter initialization for systems that have a very large number
of targets, or slow to spin-up targets, or a complicated set of array
configurations that extend the time for the firmware to declare that it
is operational. This timeout would only have an affect on non-intel
based systems, as the (more patient) BIOS would generally be where the
startup delay would be dealt with.
Signed-off-by: Mark Haverkamp <markh@osdl.org>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Received From Mark Salyzyn
Slight space and speed efficiency improvement.
Signed-off-by: Mark Haverkamp <markh@osdl.org>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Received From Mark Salyzyn
Since new commands to the card are quiesced, respect the changes in
the SCSI error path which dropped locking around the hba reset handler
and similarly drop the lock requirement in the driver's path.
Signed-off-by: Mark Haverkamp <markh@osdl.org>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Problem spotted by: Suzuki K P <suzuki@in.ibm.com>
A zero return on success isn't correct for filesystem write functions.
They should either return negative error or the length of bytes
consumed. Add code to convert our zero on success error return to
return the length of bytes passed in.
This fixes the following:
$ echo "scsi remove-single-device 0 0 3 0" > /proc/scsi/scsi
bash: echo: write error: No such device or address"
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
debugged by wrwhitehead@novell.com
patch and analysis by fujita.tomonori@lab.ntt.co.jp
Only tcp_read_sock and recv_actor (iscsi_tcp_data_recv for us) see
desc.count. It is is used just for permitting tcp_read_sock to read
the portion of data in the socket.
When iscsi_tcp_data_recv sees a partial header, it sets
desc.count. However, it is possible that the next skb (containing the
rest of the header) still does not come. So I'm not sure that this
scheme is completely correct.
Ideally, we should use the exact length of the data in the socket for
desc.count. However, it is not so simple (see SIOCINQ in
tcp_ioctl). So I think that iscsi_tcp_data_recv can just stop playing
with desc.count and tell tcp_read_sock to read the all skbs. As
proposed already, if iscsi_tcp_data_ready sets desc.count to
non-zero, tcp_read_sock does that.
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
debugged by Ming and Rohan:
The problem Ming and Rohan debugged was that during a normal session
login, open-iscsi is not incrementing the exp_statsn counter. It was
stuck at zero. From the RFC, it looks like if the login response PDU has
a successful status then we should be incrementing that value. Also from
the RFC, it looks like if when we drop a connection then reconnect, we
should be using the exp_statsn from the old connection in the next
relogin attempt.
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
align printk output
Signed-off-by: Or Gerlitz <ogerlitz@voltaire.com>
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
from patmans@us.ibm.com and michaelc@cs.wisc.edu
Fix bugs when forcing a mgmt task to fail and allow
session recovery to cleanup the session/connection
of any running mgmt tasks. When called during
the in login state.
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
add transport end point callbacks so iscsi drivers that cannot connect
from userspace, like iscsi tcp, using sockets do not have to
implement their own socket infrastructure.
Signed-off-by: Or Gerlitz <ogerlitz@voltaire.com>
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
drivers/scsi/scsi_lib.c: In function `scsi_kmap_atomic_sg':
drivers/scsi/scsi_lib.c:2394: warning: unsigned int format, different type arg (arg 3)
drivers/scsi/scsi_lib.c:2394: warning: unsigned int format, different type arg (arg 4)
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
drivers/scsi/scsi_scan.c: In function `scsi_probe_and_add_lun':
drivers/scsi/scsi_scan.c:926: warning: unused variable `vend'
drivers/scsi/scsi_scan.c:926: warning: unused variable `mod'
drivers/scsi/scsi_scan.c: At top level:
drivers/scsi/scsi_scan.c:829: warning: `scsi_inq_str' defined but not used
Fix those, tighten up the (somewhat poorly-designed) logging macro and fix
some coding-style warts.
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
The current dc395x driver uses PIO to transfer up to 4 bytes which do not
get transferred by DMA (under unclear circumstances). For this the driver
uses page_address() which is broken on highmem. Apart from this the
actual calculation of the virtual address is wrong (even without highmem).
So, e.g., for reading it reads bytes from the driver to a wrong address
and returns wrong data, I guess, for writing it would just output random
data to the device.
The proper fix, as suggested by many, is to dynamically map data using
kmap_atomic(page, KM_BIO_SRC_IRQ) / kunmap_atomic(virt). The reason why it
has not been done until now, although I've done some preliminary patches
more than a year ago was that nobody interested in fixing this problem was
able to reliably reproduce it. Now it changed - with the help from
Sebastian Frei (CC'ed) I was able to trigger the PIO path. Thus, I was
also able to test and debug it.
There are 4 cases when PIO is used in dc395x - data-in / -out with and
without scatter-gather. I was able to reproduce and test only data-in with
and without SG. So, the data-out path is still untested, but it is also
somewhat simpler than the data-in. Fredrik Roubert (also CC'ed) also had
PIO triggering on his system, and in his case it was data-out without SG.
It would be great if he could test the attached patch on his system, but
even if he cannot, I would still request to apply the patch and just wait
if anybody cries...
Implementation: I put 2 new functions in scsi_lib.c and their declarations
in scsi_cmnd.h. I exported them without _GPL, although, I don't feel
strongly about that - not many drivers are likely to use them. But there
is at least one more - I want to use them in tmscsim.c. Whether these are
the right files for the functions and their declarations - not sure
either. Actually, they are not scsi-specific, so, might go somewhere
around other scattergather magic? They are not platform specific either,
and most SG functions are defined under arch/*/... As these issues were
discussed previously there were some more routines suggested to manipulate
scattergather buffers, I think, some of them were needed around
crypto code... So, might be a common place reasonable, like
lib/scattergather.c? I am open here.
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Convert kmalloc + memset to kcalloc in ibmvscsi
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Acked-by: Dave Boutcher <sleddog@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Conflicts:
include/scsi/scsi_devinfo.h
Same number for two BLIST flags: BLIST_MAX_512 and BLIST_ATTACH_PQ3
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Any end device that can't support any of the scanning protocols
shouldn't be scanned, so set its id to -1 to prevent
scsi_scan_target() being called for it.
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
This just converts iscsi_tcp to the lib
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
There is a lot of code duplcited between iscsi_tcp
and the upcoming iscsi_iser driver. This patch puts
the duplicated code in a lib. There is more code
to move around but this takes care of the
basics. For iscsi_offload if they use the lib we will
probably move some things around. For example in the
queuecommand we will not assume that the LLD wants
to do queue_work, but it is better to handle that
later when we know for sure what iscsi_offload looks
like (we could probably do this for iscsi_iser though to).
Ideally I would like to get the iscsi_transports modules
to a place where all they really have to do is put data
on the wire, but how to do that will hopefully be more clear
when we see other modules like iscsi_offload. Or maybe
iscsi_offload will not use the lib and it will just be
iscsi_iser and iscsi_tcp and maybe the iscsi_tcp_tgt if that
is allowed in mainline.
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
The current iscsi_tcp eh is not nicely setup for dm-multipath
and performs some extra task management functions when they
are not needed.
The attached patch:
- Fixes the TMF issues. If a session is rebuilt
then we do not send aborts.
- Fixes the problem where if the host reset fired, we would
return SUCCESS even though we had not really done anything
yet. This ends up causing problem with scsi_error.c's TUR.
- If someone has turned on the userspace nop daemon code to try
and detect network problems before the scsi command timeout
we can now drop and clean up the session before the scsi command
timesout and fires the eh speeding up the time it takes for a
command to go from one patch to another. For network problems
we fail the command with DID_BUS_BUSY so if failfast is set
scsi_decide_disposition fails the command up to dm for it to
try on another path.
- And we had to add some basic iscsi session block code. Previously
if we were trying to repair a session we would retrun a MLQUEUE code
in the queuecommand. This worked but it was not the most efficient
or pretty thing to do since it would take a while to relogin
to the target. For iscsi_tcp/open-iscsi a lot of the iscsi error handler
is in userspace the block code is pretty bare. We will be
adding to that for qla4xxx.
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
For iscsi boot when going from initramfs to the real root we
need to stop the userpsace iscsi daemon. To later restart it
iscsid needs to be able to rebuild itself and part of that
process is matching a session running the kernel with the
iscsid representation. To do this the attached patch
adds several required iscsi values. If the LLD does not provide
them becuase, login is done in userspace, then the transport
class and userspace set ths up for the LLD.
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
from hare@suse.de and michaelc@cs.wisc.edu
hw iscsi like qla4xxx does not allocate a host per session and
for userspace it is difficult to restart iscsid using the
"iscsi handles" for the session and connection, so this
patch just has the class or userspace allocate the id for
the session and connection.
Note: this breaks userspace and requires users to upgrade to the newest
open-iscsi tools. Sorry about his but open-iscsi is still too new to
say we have a stable user-kernel api and we were not good nough
designers to know that other hw iscsi drivers and iscsid itself would
need such changes. Actually we sorta did but at the time we did not
have the HW available to us so we could only guess.
Luckily, the only tools hooking into the class are the open-iscsi ones
or other tools like iscsitart hook into the open-iscsi engine from
userspace or prgroams like anaconda call our tools so they are not affected.
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Some devices report a peripheral qualifier of 3 for LUN 0; with the original
code, we would still try a REPORT_LUNS scan (if SCSI level is >= 3 or if we
have the BLIST_REPORTLUNS2 passed in), but NOT any sequential scan.
Also, the device at LUN 0 (which is not connected according to the PQ) is not
registered with the OS.
Unfortunately, SANs exist that are SCSI-2 and do NOT support REPORT_LUNS, but
report a unknown device with PQ 3 on LUN 0. We still need to scan them, and
most probably we even need BLIST_SPARSELUN (and BLIST_LARGELUN). See the bug
reference for an infamous example.
This is patch 3/3:
3. Implement the blacklist flag BLIST_ATTACH_PQ3 that makes the scsi
scanning code register PQ3 devices and continues scanning; only sg
will attach thanks to scsi_bus_match().
Signed-off-by: Kurt Garloff <garloff@suse.de>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Some devices report a peripheral qualifier of 3 for LUN 0; with the original
code, we would still try a REPORT_LUNS scan (if SCSI level is >= 3 or if we
have the BLIST_REPORTLUNS2 passed in), but NOT any sequential scan.
Also, the device at LUN 0 (which is not connected according to the PQ) is not
registered with the OS.
Unfortunately, SANs exist that are SCSI-2 and do NOT support REPORT_LUNS, but
report a unknown device with PQ 3 on LUN 0. We still need to scan them, and
most probably we even need BLIST_SPARSELUN (and BLIST_LARGELUN). See the bug
reference for an infamous example.
This patch 2/3:
If a PQ3 device is found, log a message that describes the device
(INQUIRY DATA and C:B:T:U tuple) and make a suggestion for blacklisting
it.
Signed-off-by: Kurt Garloff <garloff@suse.de>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Some devices report a peripheral qualifier of 3 for LUN 0; with the original
code, we would still try a REPORT_LUNS scan (if SCSI level is >= 3 or if we
have the BLIST_REPORTLUNS2 passed in), but NOT any sequential scan.
Also, the device at LUN 0 (which is not connected according to the PQ) is not
registered with the OS.
Unfortunately, SANs exist that are SCSI-2 and do NOT support REPORT_LUNS, but
report a unknown device with PQ 3 on LUN 0. We still need to scan them, and
most probably we even need BLIST_SPARSELUN (and BLIST_LARGELUN). See the bug
reference for an infamous example.
This is patch 1/3:
If we end up in sequential scan, at least try LUN 1 for devices
that reported a PQ of 3 for LUN 0.
Also return blacklist flags, even for PQ3 devices.
Signed-off-by: Kurt Garloff <garloff@suse.de>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Equivalent of the same patch for the 3w-xxxx driver.
Signed-off-by: Adam Radford <linuxraid@amcc.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
scsi_kill_request() completes requests via normal SCSI completion path
which decrements busy counts; however, requests which get passed to
scsi_kill_request() aren't holding busy counts and scsi_kill_request()
don't increment them before invoking completion path resulting in
incorrect busy counts. Bump up busy counts before invoking completion
path.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
As previously reported via Michael Reed, the FC transport took a hit
in 2.6.15 (perhaps a little earlier) when we solved a recursion error.
There are 2 deadlocks occurring:
- With scan and the delete items sharing the same workq, flushing the
workq for the delete code was getting it stalled behind a very long
running scan code path.
- There's a deadlock where scsi_remove_target() has to sit behind
scsi_scan_target() due to contention over the scan_lock().
This patch resolves the 1st deadlock and significantly reduces the
odds of the second. So far, we have only replicated the 2nd deadlock
on a highly-parallel SMP system. More on the 2nd deadlock in a following
email.
This patch reworks the transport to:
- Only use the scsi host workq for scanning
- Use 2 other workq's internally. One for deletions, the other for
scheduled deletions. Originally, we tried this with a single workq,
but the occassional flushes of the scheduled queues was hitting the
second deadlock with a slightly higher frequency. In the future, we'll
look at the LLDD's and the transport to see if we can get rid of this
extra overhead.
- When moving to the other workq's we tightened up some object states
and some lock handling.
- Properly syncs adds/deletes
- minor code cleanups
- directly reference fc_host_attrs, rather than through attribute
macros
- flush the right workq on delayed work cancel failures.
Large kudos to Michael Reed who has been working this issue for the last
month.
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
When a target is added aic79xx tries to be overly clever: it changes
the command on the fly to TEST UNIT READY and tries to requeue the
original command. Sadly this breaks SCSI compability and of course
the midlayer is getting a bit confused by it.
So we're just removing that bit of code and let the midlayer deal with
it. It's clever enough by now. And the driver code is getting simpler.
Signed-off-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
It's no longer needed after the convrsion to use the linux srp.h file.
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
As James B. correctly noted, ahd_reset_channel() in
ahd_linux_bus_reset() should be protected by ahd_lock(). However, the
main reason for not doing so was a deadlock with the interesting
polling mechanism to detect the end a bus reset.
This patch replaces the polling mechanism with a saner signalling via
flags; it also gives us the benefit of detecting any multiple calls to
ahd_reset_channel().
Signed-off-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Original From: Ingo Flaschberger <if@xip.at>
To support the RA4100 array from Compaq.
This patch now correctly handles SCSI_UNKNOWN types with regard to
BLIST_REPORTLUNS2 (allow it) and cdb[1] LUN inclusion (don't).
It also allows a BLIST_MAX_512 flag to restrict the maximum transfer
length to 512 blocks (apparently this is an RA4100 problem).
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
When spinlock debugging is turned on, a struct completion grows beyond the
size allowed for the scsi_pointer. So move the struct completion back onto
the stack. The additional memory barriers are to keep us from completing
a random piece of kernel stack if the command happens to complete after
the error handling has finished.
Signed-off-by: Matthew Wilcox <matthew@wil.cx>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Encapsulate some more of the device reset processing in
preparation for SATA support.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Remove some unused printk macros, make some more robust, and
convert some to use standard printk macros when possible.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Simplify the dumping of the command status area by
removing some device specific information that has proven
to not be worthwhile.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Fixup a check used by the ipr driver to determine if a given
device is a SCSI disk. Due to the addition of support for
attaching SATA devices, this check needs to be more robust.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
Instead of NULLing the resource entry pointer when a disk
goes away to prevent any new commands being sent to it,
set the adapter resource handle to an invalid value so
new ops getting sent to it will fail with a selection timeout
response. This patch is needed for future SATA patches.
Signed-off-by: Brian King <brking@us.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
when the sg driver is unable to setup direct IO, free that scatter
gather list prior to falling back to indirect IO
Further to this thread started by Bryan Holty:
http://marc.theaimsgroup.com/?l=linux-scsi&m=114306885116728&w=2
Here is the reworked patch again. This time it has been
tested with a program provided by Bryan.
Signed-off-by: Douglas Gilbert <dougg@torque.net>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
for_each_cpu() actually iterates across all possible CPUs. We've had mistakes
in the past where people were using for_each_cpu() where they should have been
iterating across only online or present CPUs. This is inefficient and
possibly buggy.
We're renaming for_each_cpu() to for_each_possible_cpu() to avoid this in the
future.
This patch replaces for_each_cpu with for_each_possible_cpu.
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
