Errors from SIGA instructions are stored in the per queue qdio_error
and reported back when the queue handler is called. That opens a race
when multiple error conditions occur simultanously.

Report SIGA errors immediately in the return value of do_QDIO so the
upper layer can react and SIGA errors no longer interfere with other
errors.

Move the SIGA error handling in qeth from the outbound handler to
qeth_flush_buffers.
Signed-off-by: NJan Glauber <jang@linux.vnet.ibm.com>

If the qdio module is unloaded the tiqdio tasklet must be terminated
by tasklet_kill. Move the tasklet_kill after the unregistration of
the adapter interrupt so the tiqdio tasklet will not be scheduled
anymore before calling tasklet_kill.
Signed-off-by: NJan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

The index value that indicated that the input queue moved was also used to
store the index of the first acknowledged buffer. For non-qebsm only the
newest buffer is acknowledged which may be different from the last move index
so two seperate values are needed to track the input queue.
Signed-off-by: NJan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

The ACKnowledgement state should be set on the newest SBAL so an
adapter interrupt surpression check needs to scan fewer SBALs.
Signed-off-by: NJan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

qdio_cleanup is a wrapper function that should call qdio_shutdown and
qdio_free. qdio_free was not called if an error occured in qdio_shutdown
resulting in a missing free of allocated resources.

Call qdio_free regardless of the return value of qdio_shutdown.
Signed-off-by: NJan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

The queue tasklets were stopped with tasklet_disable. Although tasklet_disable
prevents the tasklet from beeing executed it is still possible that a tasklet
is scheduled on a CPU at that point. A following qdio_establish calls
tasklet_init which clears the tasklet count and the tasklet state leading to
the following Oops:

    <2>kernel BUG at kernel/softirq.c:392!
    <4>illegal operation: 0001 [#1] SMP
    <4>Modules linked in: iptable_filter ip_tables x_tables dm_round_robin dm_multipath scsi_dh sg sd_mod crc_t10dif nfs lockd nfs
_acl sunrpc fuse loop dm_mod qeth_l3 ipv6 zfcp qeth scsi_transport_fc qdio scsi_tgt scsi_mod chsc_sch ccwgroup dasd_eckd_mod dasdm
od ext3 mbcache jbd
    <4>Supported: Yes
    <4>CPU: 0 Not tainted 2.6.27.13-1.1.mz13-default #1
    <4>Process blast.LzS_64 (pid: 16445, task: 000000006cc02538, ksp: 000000006cb67998)
    <4>Krnl PSW : 0704c00180000000 00000000001399f4 (tasklet_action+0xc8/0x1d4)
    <4>           R:0 T:1 IO:1 EX:1 Key:0 M:1 W:0 P:0 AS:3 CC:0 PM:0 EA:3
    <4>Krnl GPRS: ffffffff00000030 0000000000000002 0000000000000002 fffffffffffffffe
    <4>           000000000013aabe 00000000003b6a18 fffffffffffffffd 0000000000000000
    <4>           00000000006705a8 000000007d0914a8 000000007d0914b0 000000007fecfd30
    <4>           0000000000000000 00000000003b63e8 000000007fecfd90 000000007fecfd30
    <4>Krnl Code: 00000000001399e8: b9200021            cgr     %r2,%r1
    <4>           00000000001399ec: a7740004            brc     7,1399f4
    <4>           00000000001399f0: a7f40001            brc     15,1399f2
    <4>          >00000000001399f4: c0100027e8ee        larl    %r1,636bd0
    <4>           00000000001399fa: bf1f1008            icm     %r1,15,8(%r1)
    <4>           00000000001399fe: a7840019            brc     8,139a30
    <4>           0000000000139a02: c0300027e8ef        larl    %r3,636be0
    <4>           0000000000139a08: e3c030000004        lg      %r12,0(%r3)
    <4>Call Trace:
    <4>([<0000000000139c12>] tasklet_hi_action+0x112/0x1d4)
    <4> [<000000000013aabe>] __do_softirq+0xde/0x1c4
    <4> [<000000000010fa2e>] do_softirq+0x96/0xb0
    <4> [<000000000013a8d8>] irq_exit+0x70/0xcc
    <4> [<000000000010d1d8>] do_extint+0xf0/0x110
    <4> [<0000000000113b10>] ext_no_vtime+0x16/0x1a
    <4> [<000003e0000a3662>] ext3_dirty_inode+0xe6/0xe8 [ext3]
    <4>([<00000000001f6cf2>] __mark_inode_dirty+0x52/0x1d4)
    <4> [<000003e0000a44f0>] ext3_ordered_write_end+0x138/0x190 [ext3]
    <4> [<000000000018d5ec>] generic_perform_write+0x174/0x230
    <4> [<0000000000190144>] generic_file_buffered_write+0xb4/0x194
    <4> [<0000000000190864>] __generic_file_aio_write_nolock+0x418/0x454
    <4> [<0000000000190ee2>] generic_file_aio_write+0x76/0xe4
    <4> [<000003e0000a05c2>] ext3_file_write+0x3e/0xc8 [ext3]
    <4> [<00000000001cc2fe>] do_sync_write+0xd6/0x120
    <4> [<00000000001ccfc8>] vfs_write+0xac/0x184
    <4> [<00000000001cd218>] SyS_write+0x68/0xe0
    <4> [<0000000000113402>] sysc_noemu+0x10/0x16
    <4> [<0000020000043188>] 0x20000043188
    <4>Last Breaking-Event-Address:
    <4> [<00000000001399f0>] tasklet_action+0xc4/0x1d4
    <6>qdio: 0.0.c61b ZFCP on SC f67 using AI:1 QEBSM:0 PCI:1 TDD:1 SIGA: W AOP
    <4> <0>Kernel panic - not syncing: Fatal exception in interrupt

Use tasklet_kill instead of tasklet_disbale. Since tasklet_schedule must not be
called after tasklet_kill use the QDIO_IRQ_STATE_STOPPED to inidicate that a
queue is going down and prevent further tasklet schedules in that case.

Remove superflous tasklet_schedule from input queue setup, at that time
the queues are not ready so the schedule results in a NOP.
Signed-off-by: NJan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Remove the call to qdio_shutdown from qdio_activate since the upper-layer
drivers are responsible to call qdio_shutdown when qdio_activate returns
with an error.
Signed-off-by: NJan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Add a mutex to protect the tiq_list. Although reading the list is done
using RCU adding and removing elements from the list must still
happen locked since multiple qdio devices may change the list in parallel
otherwise.
Signed-off-by: NJan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Subchannel reprobing can block the kslowcrw workqueue indefinitely
while waiting for device recognition to finish which is also scheduled
to run on kslowcrw. Prevent this deadlock by moving the waiting
portion of subchannel reprobing to the cio workqueue.
Signed-off-by: NPeter Oberparleiter <peter.oberparleiter@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Remove unused subchannel pointer in io_subchannel_recog_done.
Signed-off-by: NSebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Fix incorrect check for active I/O in interrogate function.
Signed-off-by: NPeter Oberparleiter <peter.oberparleiter@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

In some situations a rc in __chsc_do_secm will be overwritten
by another one. This shouldn't do harm since todays callers
don't check for _specific_ errors but fix it for the sake of
correctness.
Signed-off-by: NSebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Writing only spaces to /proc/cio_ignore will cause a buffer overflow
since the size_t value i will not become negative and so buf[-1UL] is
accessed. Change the value of i to ssize_t.
Signed-off-by: NSebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

For a ccw group device unbinding it from its driver should do the
same as a call to ungroup, since this virtual device can not exist
without a driver.
Signed-off-by: NSebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Some sanity checks in the ccw group driver test the output of
container_of macros to be !NULL. Test the input parameters instead.
Signed-off-by: NSebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

In case the ccw group driver refuses to set a device [on|off]line,
we should transmit the return code to the caller.
Signed-off-by: NSebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

MAX_ISC is a valid isc number, so arrays with an index of isc
need to have a length of MAX_ISC+1
Signed-off-by: NSebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Since some callers rely on for_each_subchannel_staged to not fail,
fall back to brute force scanning using get_subchannel_by_schid in
case of a oom situation.
Signed-off-by: NSebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Add barrier to prevent compiler from reloading pointer to irq handler.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

The crw_unregister_handler uses xchg + synchronize_sched when
unregistering a crw_handler.
This doesn't protect crw_collect_info to potentially jump to NULL since
it has unlocked code like this:

if (crw_handlers[i])
        crw_handlers[i](NULL, NULL, 1);

So add a mutex which protects the crw handler array for changes.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

In case the ccw driver refuses to set a device offline, we should
transmit the return code to the caller.
Signed-off-by: NSebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Use ccw_device_set_notoper() (which also deletes the device
timer and disables the subchannel) instead of simply setting
the state to DEV_STATE_NOT_OPER in the generic not operational
handling code. This prevents unexpected interrupts popping up
for devices that are deemed not operational.
Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Acting upon the assumption that cio_disable_subchannel()
is only called when we really want to disable the subchannel
(a) remove the check for activity (it is already done in
    ccw_device_offline(), which is the place where it matters)
(b) collect pending status via tsch() and ignore it (it
    can't matter anymore since the subchannel will be disabled).
Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

The common I/O layer may encounter a situation where the
device number of a ccw device has changed or a device
driver doesn't want to keep a formerly disconnected device
becoming operational again. Instead of using device_del()/
device_add() as now, we can just unbind the driver from the
device and rebind it to get the desired effect (rebinding)
with less overhead.
Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Split machine check handler code and move it to cio and kernel code
where it belongs to. No functional change.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

If CONFIG_DEBUG_FS is disabled debugfs_create_file returns -ENODEV instead
of zero. Since the return value is stored and used at shutdown to remove
existing entries an OOPS may be triggered.

Add a check of the debugfs_create_file return value and in case of an error
set the entry to NULL so it will be ignored at shutdown.
Signed-off-by: NJan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

When I review ocfs2 code, find there are 2 typos to "successfull".  After
doing grep "successfull " in kernel tree, 22 typos found totally -- great
minds always think alike :)

This patch fixes all the similar typos. Thanks for Randy's ack and comments.
Signed-off-by: NColy Li <coyli@suse.de>
Acked-by: NRandy Dunlap <randy.dunlap@oracle.com>
Acked-by: NRoland Dreier <rolandd@cisco.com>
Cc: Jeremy Kerr <jk@ozlabs.org>
Cc: Jeff Garzik <jeff@garzik.org>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Vlad Yasevich <vladislav.yasevich@hp.com>
Cc: Sridhar Samudrala <sri@us.ibm.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Distinguish the cputime of the idle process where idle is actually using
cpu cycles from the cputime where idle is sleeping on an enabled wait psw.
The former is accounted as system time, the later as idle time.
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Signed-off-by: NMichael Ernst <mernst@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

To check if multipath is available we count the bits set in lpm,
which could change over time (via configure [on|off] of a path).

The following patch uses the pim (which is persistent) for this
decision.
Signed-off-by: NSebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Sometimes we change the pmcw configuration but don't call msch
to transmit these changes to the channel subsystem.

The patch fixes this by calling cio_commit_config in such cases.
Signed-off-by: NSebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

To change the configuration of a subchannel we alter the modifiable
bits of the subchannel's schib field and issue a modify subchannel.
There can be the case that not all changes were applied -or worse-
quietly overwritten by the hardware. With the next store subchannel
we obtain the current state of the hardware but lose our target
configuration.

With this patch we introduce a subchannel_config structure which
contains the target subchannel configuration. Additionally the msch
wrapper cio_modify is replaced with cio_commit_config which
copies the desired changes to a temporary schib. msch is then
called with the temporary schib. This schib is only written back
to the subchannel if all changes were applied.
Signed-off-by: NSebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

There is the chance that we get condition code 0 for a stsch but
the resulting schib is not vaild. In the current code there are
2 cases:
* we do a check for validity of the schib after stsch, but at this
  time we have already stored the invaild schib in the subchannel
  structure. This may lead to problems.
* we don't do a check for validity, which is not that good either.

The patch addresses both issues by introducing the stsch wrapper
cio_update_schib which performs stsch on a local schib. This schib
is only written back to the subchannel if it's valid.

side note: For some functions (chp_events) the return codes are
different now (-ENXIO vs -ENODEV) but this shouldn't do harm
since the caller doesn't check for _specific_ errors.
Signed-off-by: NSebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Check if a ccw device is registered via device_is_registered()
and not via the old kludge of checking the membership in driver
core internal klists.
Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Just put the cdev's reference count to give up our reference.
Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

If we fail the probe for an I/O subchannel, we won't be able
to unregister it again since there are no sch_event()
callbacks for unbound subchannels. Just succeed the probe in
any case and schedule unregistering the subchannel.
Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

There is a race between io_subchannel_register() and
io_subchannel_sch_event() which may cause a subchannel to be
unregistered because it is no longer operational before
io_subchannel_register() had run. We need to check whether the
subchannel is still registered before the ccw device can be
registered and just bail out if it is not.
Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Subchannel refcounting was incorrect in some places, especially
a refcount was missing when ccw_device_call_sch_unregister()
was called and the refcount was not correctly switched after
moving devices.

Fix this by establishing the following rules:
- The ccw_device obtains a reference on its parent subchannel
  when dev.parent is set and gives it up in its release
  function. This is needed because we need a parent reference
  for correct refcounting even before the ccw device is (if at
  all) registered.
- When calling device_move(), obtain a reference on the new
  subchannel before moving the ccw device and give up the
  reference on the old parent after moving. This brings the
  refcount in line with the first rule.
Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

The current code attempts to get an extra reference count
for online devices by doing a get_device() in ccw_device_online()
and a put_device() in ccw_device_done(). However, this
- incorrectly obtains an extra reference for disconnected
  devices becoming available again (since they are already
  online)
- needs special checks for css_init_done in order to handle
  the console device
- is not obvious and
- may incorretly drop a reference count in ccw_device_done() if
  that function is called after path verification for a device
  that just became not operational.

So let's just get the reference in ccw_device_set_online() and
drop it in ccw_device_set_offline(). (Unfortunately, we still
need the special case in io_subchannel_probe().)
Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>