12 years ago[S390] Add hwcap flag for the etf3 enhancement facility
Andreas Krebbel [Thu, 26 Mar 2009 14:24:56 +0000]
[S390] Add hwcap flag for the etf3 enhancement facility

The Extended Translation Facility 3 (ETF3) added instructions which
allow conversions between different unicode character maps (UTF-8,
UTF-32 ...).  These instructions got enhanced with a later version of
the ETF3 allowing malformed multibyte chars to be recognized and
reported correctly.  The attached patch reserves bit 8 in the elf
hwcaps vector for the enhanced version of ETF3. The bit corresponds to
the stfle bits 22 and 30 and will only be set if both of the stfle
bits are set.

Signed-off-by: Andreas Krebbel <krebbel@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] Ensure that ipl panic notifier is called late.
Michael Holzheu [Thu, 26 Mar 2009 14:24:55 +0000]
[S390] Ensure that ipl panic notifier is called late.

The s390 ipl panic notifier will stop the system or trigger a system dump.
This should be done as final action on the panic path. All other panic
notifiers should be executed before. Currently we use priority 0 for the ipl
notifier. In order to be called late, this patch changes the priority to
INT_MIN which is the lowest possible priority.

Signed-off-by: Michael Holzheu <holzheu@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] fix dfp elf hwcap/facility bit detection
Christian Ehrhardt [Thu, 26 Mar 2009 14:24:54 +0000]
[S390] fix dfp elf hwcap/facility bit detection

The old dfp detection wanted to check bit 43 (dfp high performance), but due
to a wrong calculation always used to check bit 42. Additionally the
"userspace expectation" is, that the dfp capability bit is set is if facility
bit 42 (decimal floating point facility available) and bit 44 (perform floating
point operation facility avail).
The patch fixes the bit calculation and extends the check to work like:
  elf hw cap dfp bit = facility bits 42 (dfp) & 44 (pfpo) available

Signed-off-by: Christian Ehrhardt <ehrhardt@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] smp: perform initial cpu reset before starting a cpu
Heiko Carstens [Thu, 26 Mar 2009 14:24:53 +0000]
[S390] smp: perform initial cpu reset before starting a cpu

Performing an initial cpu reset makes sure all registers and tlbs of
the targeted cpu are initialized and flushed.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] smp: fix memory leak on __cpu_up
Heiko Carstens [Thu, 26 Mar 2009 14:24:52 +0000]
[S390] smp: fix memory leak on __cpu_up

If sigp_set_prefix fails on __cpu_up we leak the lowcore structures
and async+panic stacks for the failed cpu.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] ipl: Improve checking logic and remove switch defaults.
Michael Holzheu [Thu, 26 Mar 2009 14:24:51 +0000]
[S390] ipl: Improve checking logic and remove switch defaults.

A code analysis tool reported two warnings:
"The expression `ipl_info.type == IPL_TYPE_FCP' is true whenever evaluated."
and "Default is not possible". This patch improves the corresponding if
statement logic and removes the unnecessary switch defaults.

Signed-off-by: Michael Holzheu <holzheu@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] s390dbf: Remove needless check for NULL pointer.
Michael Holzheu [Thu, 26 Mar 2009 14:24:50 +0000]
[S390] s390dbf: Remove needless check for NULL pointer.

The check for NULL pointer has already be done before. Therefore we can remove
the second check.

Signed-off-by: Michael Holzheu <holzheu@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] s390dbf: Remove redundant initilizations.
Michael Holzheu [Thu, 26 Mar 2009 14:24:49 +0000]
[S390] s390dbf: Remove redundant initilizations.

Signed-off-by: Michael Holzheu <holzheu@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] use kzfree()
Johannes Weiner [Thu, 26 Mar 2009 14:24:48 +0000]
[S390] use kzfree()

Use kzfree() instead of memset() + kfree().

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] BUG to BUG_ON changes
Stoyan Gaydarov [Thu, 26 Mar 2009 14:24:47 +0000]
[S390] BUG to BUG_ON changes

Signed-off-by: Stoyan Gaydarov <stoyboyker@gmail.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] zfcpdump: Prevent zcore from beeing built as a kernel module.
Michael Holzheu [Thu, 26 Mar 2009 14:24:46 +0000]
[S390] zfcpdump: Prevent zcore from beeing built as a kernel module.

The zcore code switches to real addressing mode when creating a kernel dump.
This is not possible, if it is built as a kernel module. With this patch
zcore (zfcpdump) can't be built as a kernel module any more.

Signed-off-by: Michael Holzheu <holzheu@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] Use csum_partial in checksum.h
Frank Munzert [Thu, 26 Mar 2009 14:24:45 +0000]
[S390] Use csum_partial in checksum.h

The cksm function in system.h is duplicate to csum_partial in checksum.h.
Remove cksm and use csum_partial instead.

Signed-off-by: Frank Munzert <munzert@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] cleanup lowcore.h
Martin Schwidefsky [Thu, 26 Mar 2009 14:24:44 +0000]
[S390] cleanup lowcore.h

The lowcore.h header has quite a lot of whitespace damage and a rather
wild collection of entries. Remove all that whitespace and tidy up the
order of the lowcore fields.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] eliminate ipl_device from lowcore
Martin Schwidefsky [Thu, 26 Mar 2009 14:24:43 +0000]
[S390] eliminate ipl_device from lowcore

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] eliminate cpuinfo_S390 structure
Martin Schwidefsky [Thu, 26 Mar 2009 14:24:42 +0000]
[S390] eliminate cpuinfo_S390 structure

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] Fix comments in lowcore structure
Christian Borntraeger [Thu, 26 Mar 2009 14:24:41 +0000]
[S390] Fix comments in lowcore structure

This patch fixes two addresses in the comments for the
lowcore structure. Looks like an copy-paste bug.

Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] Expose a constant for the number of words representing the CRs
Dan Smith [Thu, 26 Mar 2009 14:24:40 +0000]
[S390] Expose a constant for the number of words representing the CRs

We need to use this value in the checkpoint/restart code and would like to
have a constant instead of a magic '3'.

Cc: linux-s390@vger.kernel.org
Signed-off-by: Dan Smith <danms@us.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] list usage cleanup in s390
Eric Sesterhenn [Thu, 26 Mar 2009 14:24:39 +0000]
[S390] list usage cleanup in s390

Trivial cleanup, list_del(); list_add{,_tail}() is equivalent
to list_move{,_tail}(). Semantic patch for coccinelle can be
found at www.cccmz.de/~snakebyte/list_move_tail.spatch

Signed-off-by: Eric Sesterhenn <snakebyte@gmx.de>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] tape message cleanup
Carsten Otte [Thu, 26 Mar 2009 14:24:38 +0000]
[S390] tape message cleanup

This is a cleanup of all the messages this driver prints. It uses the
dev_message macros now.

Signed-off-by: Carsten Otte <cotte@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] use compiler builtin versions of strlen/strcpy/strcat
Heiko Carstens [Thu, 26 Mar 2009 14:24:37 +0000]
[S390] use compiler builtin versions of strlen/strcpy/strcat

Use builtin variants if gcc 4 or newer is used to compile the kernel.
Generates better code than the asm variants.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] module function call optimization
Martin Schwidefsky [Thu, 26 Mar 2009 14:24:36 +0000]
[S390] module function call optimization

Avoid the detour over the PLT if the branch target of a function call
in a module is in the range of the bras (16-bit) or brasl (32-bit)
instruction. The PLT is still generated but it is unused.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] bitops: remove likely annotations
Heiko Carstens [Thu, 26 Mar 2009 14:24:35 +0000]
[S390] bitops: remove likely annotations

likely/unlikely profiling revealed that none of the branches in bitops
is taken likely or unlikely. So remove the annotations.
In addition the generated code is shorter.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] remove duplicate nul-termination of string
Hendrik Brueckner [Thu, 26 Mar 2009 14:24:34 +0000]
[S390] remove duplicate nul-termination of string

strlcpy() does already NUL-terminate the destination string.

Signed-off-by: Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] allow usage of string functions in linux/string.h
Rusty Russell [Thu, 26 Mar 2009 14:24:33 +0000]
[S390] allow usage of string functions in linux/string.h

In introducing a trivial "strstarts()" function in linux/string.h, we
hit the following error on s390:

In file included from include/linux/bitmap.h:8,
                 from include/linux/cpumask.h:142,
                 from include/linux/smp.h:12,
                 from /home/rusty/devel/kernel/patches/linux-2.6/arch/s390/include/asm/spinlock.h:14,
                 from include/linux/spinlock.h:88,
                 from include/linux/seqlock.h:29,
                 from include/linux/time.h:8,
                 from include/linux/stat.h:60,
                 from include/linux/module.h:10,
                 from arch/s390/lib/string.c:13:
include/linux/string.h: In function 'strstarts':
include/linux/string.h:124: error: implicit declaration of function 'strlen'
include/linux/string.h:124: warning: incompatible implicit declaration of built-in function 'strlen'

Because when including asm/string.h from arch/s390/lib/string.c we
don't declare the string ops we are about to define, and
linux/string.h barfs.

The fix is to declare them in this IN_ARCH_STRING_C case, but in
general I wonder if there's a neater fix.

Reported-by: linux-next
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] qdio: merge inbound and outbound handler functions
Jan Glauber [Thu, 26 Mar 2009 14:24:32 +0000]
[S390] qdio: merge inbound and outbound handler functions

The inbound and outbound handlers are nearly identical if the outbound
handler uses first_to_check as end index instead of last_move. Since both
values are identical at that point the handlers can be merged.

Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] qdio: report SIGA errors directly
Jan Glauber [Thu, 26 Mar 2009 14:24:31 +0000]
[S390] qdio: report SIGA errors directly

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: Jan Glauber <jang@linux.vnet.ibm.com>

12 years ago[S390] qdio: tasklet termination in case of module unload
Jan Glauber [Thu, 26 Mar 2009 14:24:30 +0000]
[S390] qdio: tasklet termination in case of module unload

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: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] qdio: seperate last move index and polling index
Jan Glauber [Thu, 26 Mar 2009 14:24:29 +0000]
[S390] qdio: seperate last move index and polling index

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: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] qdio: move ACK to newest buffer for devices without QEBSM
Jan Glauber [Thu, 26 Mar 2009 14:24:28 +0000]
[S390] qdio: move ACK to newest buffer for devices without QEBSM

The ACKnowledgement state should be set on the newest SBAL so an
adapter interrupt surpression check needs to scan fewer SBALs.

Signed-off-by: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] qdio: call qdio_free also if qdio_shutdown fails
Jan Glauber [Thu, 26 Mar 2009 14:24:27 +0000]
[S390] qdio: call qdio_free also if qdio_shutdown fails

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: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] qdio: proper kill of qdio tasklets
Jan Glauber [Thu, 26 Mar 2009 14:24:26 +0000]
[S390] qdio: proper kill of qdio tasklets

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: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] qdio: Dont call qdio_shutdown in case qdio_activate fails
Jan Glauber [Thu, 26 Mar 2009 14:24:25 +0000]
[S390] qdio: Dont call qdio_shutdown in case qdio_activate fails

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: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] qdio: add missing tiq_list locking
Jan Glauber [Thu, 26 Mar 2009 14:24:24 +0000]
[S390] qdio: add missing tiq_list locking

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: Jan Glauber <jang@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] Add zcrypt section in MAINTAINERS
Felix Beck [Thu, 26 Mar 2009 14:24:23 +0000]
[S390] Add zcrypt section in MAINTAINERS

Add zcrypt section in S390 part of MAINTAINERS file.

Signed-off-by: Felix Beck <felix.beck@de.ibm.com>
Signed-off-by: Ralph Wuerthner <ralph.wuerthner@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] kernel: Disable switch_amode by default
Gerald Schaefer [Thu, 26 Mar 2009 14:24:22 +0000]
[S390] kernel: Disable switch_amode by default

Disable switch_amode by default because pagetable walk on pre z9
hardware has negative performance impact.

Signed-off-by: Gerald Schaefer <gerald.schaefer@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] clock sync mode flags
Martin Schwidefsky [Thu, 26 Mar 2009 14:24:21 +0000]
[S390] clock sync mode flags

The clock sync mode flag CLOCK_SYNC_STP is not cleared when stp
is set offline. In this case the get_sync_clock() function returns
-EACCESS and the dasd driver will block all i/o until stp is enabled
again. In addition get_sync_clock can return -EACCESS if the clock is
not in sync instead of -EAGAIN.

Rework the stp/etr online handling to fix these problems.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] cio: prevent workqueue deadlock
Peter Oberparleiter [Thu, 26 Mar 2009 14:24:20 +0000]
[S390] cio: prevent workqueue deadlock

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: Peter Oberparleiter <peter.oberparleiter@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] cio: remove unused local variable
Sebastian Ott [Thu, 26 Mar 2009 14:24:19 +0000]
[S390] cio: remove unused local variable

Remove unused subchannel pointer in io_subchannel_recog_done.

Signed-off-by: Sebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] cio: incorrect status check in interrogate function
Peter Oberparleiter [Thu, 26 Mar 2009 14:24:18 +0000]
[S390] cio: incorrect status check in interrogate function

Fix incorrect check for active I/O in interrogate function.

Signed-off-by: Peter Oberparleiter <peter.oberparleiter@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] cio: fix rc generation after chsc call
Sebastian Ott [Thu, 26 Mar 2009 14:24:17 +0000]
[S390] cio: fix rc generation after chsc call

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: Sebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] cio: fix wrong buffer access in cio_ignore_write
Sebastian Ott [Thu, 26 Mar 2009 14:24:16 +0000]
[S390] cio: fix wrong buffer access in cio_ignore_write

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: Sebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] cio: ccw group fix unbind behaviour.
Sebastian Ott [Thu, 26 Mar 2009 14:24:15 +0000]
[S390] cio: ccw group fix unbind behaviour.

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: Sebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] cio: fix sanity checks in ccwgroup driver.
Sebastian Ott [Thu, 26 Mar 2009 14:24:14 +0000]
[S390] cio: fix sanity checks in ccwgroup driver.

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: Sebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] cio: ccw group online store - report rcs to the caller.
Sebastian Ott [Thu, 26 Mar 2009 14:24:13 +0000]
[S390] cio: ccw group online store - report rcs to the caller.

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: Sebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] cio: airq - fix array boundary
Sebastian Ott [Thu, 26 Mar 2009 14:24:12 +0000]
[S390] cio: airq - fix array boundary

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: Sebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] cio: device scan oom fallback.
Sebastian Ott [Thu, 26 Mar 2009 14:24:11 +0000]
[S390] cio: device scan oom fallback.

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: Sebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] cio: ensure single load of irq handler pointer
Heiko Carstens [Thu, 26 Mar 2009 14:24:10 +0000]
[S390] cio: ensure single load of irq handler pointer

Add barrier to prevent compiler from reloading pointer to irq handler.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] cio/crw: add/fix locking
Heiko Carstens [Thu, 26 Mar 2009 14:24:09 +0000]
[S390] cio/crw: add/fix locking

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: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] cio: ccw device online store - report rc from ccw driver.
Sebastian Ott [Thu, 26 Mar 2009 14:24:08 +0000]
[S390] cio: ccw device online store - report rc from ccw driver.

In case the ccw driver refuses to set a device offline, we should
transmit the return code to the caller.

Signed-off-by: Sebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] cio: Use ccw_device_set_notoper().
Cornelia Huck [Thu, 26 Mar 2009 14:24:07 +0000]
[S390] cio: Use ccw_device_set_notoper().

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

12 years ago[S390] cio: Try harder to disable subchannel.
Cornelia Huck [Thu, 26 Mar 2009 14:24:06 +0000]
[S390] cio: Try harder to disable subchannel.

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

12 years ago[S390] cio: Use unbind/bind instead of unregister/register.
Cornelia Huck [Thu, 26 Mar 2009 14:24:05 +0000]
[S390] cio: Use unbind/bind instead of unregister/register.

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

12 years ago[S390] move EXPORT_SYMBOLs to definitions
Heiko Carstens [Thu, 26 Mar 2009 14:24:04 +0000]
[S390] move EXPORT_SYMBOLs to definitions

Move all EXPORT_SYMBOLs to their corresponding definitions.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] convert bitmap definitions to C
Heiko Carstens [Thu, 26 Mar 2009 14:24:03 +0000]
[S390] convert bitmap definitions to C

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] Remove CONFIG_MACHCHK_WARNING.
Heiko Carstens [Thu, 26 Mar 2009 14:24:02 +0000]
[S390] Remove CONFIG_MACHCHK_WARNING.

Everybody enables it so there is no point for an extra config option.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] split/move machine check handler code
Heiko Carstens [Thu, 26 Mar 2009 14:24:01 +0000]
[S390] split/move machine check handler code

Split machine check handler code and move it to cio and kernel code
where it belongs to. No functional change.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] Fix appldata build break with !NET
Sachin Sant [Thu, 26 Mar 2009 14:24:00 +0000]
[S390] Fix appldata build break with !NET

With CONFIG_NET not set appldata build breaks on s390.

arch/s390/appldata/built-in.o: In function appldata_get_net_sum_data:
appldata_net_sum.c:(.text+0x2684): undefined reference to dev_get_stats
appldata_net_sum.c:(.text+0x2688): undefined reference to init_net
appldata_net_sum.c:(.text+0x268c): undefined reference to init_net
appldata_net_sum.c:(.text+0x2694): undefined reference to dev_base_lock

The following patch fixes the issue for me.

Signed-off-by: Sachin Sant <sachinp@in.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] ftrace: dont trace machine check handler
Heiko Carstens [Thu, 26 Mar 2009 14:23:59 +0000]
[S390] ftrace: dont trace machine check handler

The ftrace code is currently not reentrant, so we better don't trace
our machine check handler. Machine checks are handled like NMIs on s390.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] Fix hypervisor detection for KVM
Christian Borntraeger [Thu, 26 Mar 2009 14:23:58 +0000]
[S390] Fix hypervisor detection for KVM

Currently we use the cpuid (via STIDP instruction) to recognize LPAR,
z/VM and KVM.
The architecture states, that bit 0-7 of STIDP returns all zero, and
if STIDP is executed in a virtual machine, the VM operating system
will replace bits 0-7 with FF.

KVM should not use FE to distinguish z/VM from KVM for interested
guests. The proper way to detect the hypervisor is the STSI (Store
System Information) instruction, which return information about the
hypervisors via function code 3, selector1=2, selector2=2.

This patch changes the detection routine of Linux to use STSI instead
of STIDP. This detection is earlier than bootmem, we have to use a
static buffer. Since STSI expects a 4kb block (4kb aligned) this
patch also changes the init.data alignment for s390. As this section
will be freed during boot, this should be no problem.

Patch is tested with LPAR, z/VM, KVM on LPAR, and KVM under z/VM.

Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] check addressing mode in s390_enable_sie
Carsten Otte [Thu, 26 Mar 2009 14:23:57 +0000]
[S390] check addressing mode in s390_enable_sie

The sie instruction requires address spaces to be switched
to run proper. This patch verifies that this is the case
in s390_enable_sie, otherwise the kernel would crash badly
as soon as the process runs into sie.

Signed-off-by: Carsten Otte <cotte@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] lockdep: trace hardirq off in smp_send_stop
Christian Borntraeger [Thu, 26 Mar 2009 14:23:56 +0000]
[S390] lockdep: trace hardirq off in smp_send_stop

With lockdep we got the following trace after a panic:

Badness at /home/autobuild/BUILD/linux-2.6.28-20090204/kernel/lockdep.c:2878
[...]
Call Trace:
 [<0000000000176334>] lock_acquire+0x54/0xbc
 [<000000000050b4fe>] __atomic_notifier_call_chain+0x6e/0xdc
 [<000000000050b59c>] atomic_notifier_call_chain+0x30/0x44
 [<0000000000504274>] panic+0xd0/0x1e8
[...]
INFO: lockdep is turned off.
Last Breaking-Event-Address:
 [<0000000000170e62>] check_flags+0xae/0x15c
possible reason: unannotated irqs-off.

lockdep is right. We missed a trace_hardirq_off in our smp_send_stop
function and smp_send_stop is called before the panic call chain.

Reported-by: Mijo <Safradin mijo@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] hvc_iucv: Provide IUCV z/VM user ID filtering
Hendrik Brueckner [Thu, 26 Mar 2009 14:23:55 +0000]
[S390] hvc_iucv: Provide IUCV z/VM user ID filtering

This patch introduces the kernel parameter hvc_iucv_allow= that specifies
a comma-separated list of z/VM user IDs.
If specified, the z/VM IUCV hypervisor console device driver accepts IUCV
connections from listed z/VM user IDs only.

Signed-off-by: Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] hvc_iucv: Update and add missing kernel messages
Hendrik Brueckner [Thu, 26 Mar 2009 14:23:54 +0000]
[S390] hvc_iucv: Update and add missing kernel messages

If the hvc_iucv= kernel parameter specifies a value that is not
valid, display an error message.
Minor changes to existing kernel messages.

Signed-off-by: Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] cputime: initialize per thread timer values on fork
Heiko Carstens [Thu, 26 Mar 2009 14:23:53 +0000]
[S390] cputime: initialize per thread timer values on fork

Initialize per thread timer values instead of just copying them from
the parent. That way it is easily possible to tell how much time a
thread spent in user/system context.
Doesn't fix a bug, this is just for debugging purposes.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] arch/s390/kernel/process.c: fix whitespace damage
Heiko Carstens [Thu, 26 Mar 2009 14:23:52 +0000]
[S390] arch/s390/kernel/process.c: fix whitespace damage

Fix all the whitespace damage in process.c, especially copy_thread().
Next patch will add code to copy_thread() which needs to 'fixed' first.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] delete drivers/s390/ebcdic.c
Heiko Carstens [Thu, 26 Mar 2009 14:23:51 +0000]
[S390] delete drivers/s390/ebcdic.c

Dead file. Seems to be a leftover from the 2.4->2.5 conversion.
The used and uptodate version of this file is in arch/s390/kernel.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] move sysinfo.c from drivers/s390 to arch/s390/kernel
Heiko Carstens [Thu, 26 Mar 2009 14:23:50 +0000]
[S390] move sysinfo.c from drivers/s390 to arch/s390/kernel

All in sysinfo.c is core kernel code and not driver code. So move it
to arch/s390/kernel. Also includes some small cleanups.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] dasd: message cleanup
Stefan Haberland [Thu, 26 Mar 2009 14:23:49 +0000]
[S390] dasd: message cleanup

Moved some Messages into s390 debug feature and changed remaining
messages to use the dev_xxx and pr_xxx macros.

Signed-off-by: Stefan Haberland <stefan.haberland@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] dasd: add High Performance FICON support
Stefan Weinhuber [Thu, 26 Mar 2009 14:23:48 +0000]
[S390] dasd: add High Performance FICON support

To support High Performance FICON, the DASD device driver has to
translate I/O requests into the new transport mode control words (TCW)
instead of the traditional (command mode) CCW requests.

Signed-off-by: Stefan Weinhuber <wein@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] dasd: add large volume support
Stefan Weinhuber [Thu, 26 Mar 2009 14:23:47 +0000]
[S390] dasd: add large volume support

The dasd device driver will now support ECKD devices with more then
65520 cylinders.
In the traditional ECKD adressing scheme each track is addressed
by a 16-bit cylinder and 16-bit head number. The new addressing
scheme makes use of the fact that the actual number of heads is
never larger then 15, so 12 bits of the head number can be redefined
to be part of the cylinder address.

Signed-off-by: Stefan Weinhuber <wein@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] dasd_eckd / Write format R0 is now allowed BB
Jean-Baptiste Joret [Thu, 26 Mar 2009 14:23:46 +0000]
[S390] dasd_eckd / Write format R0 is now allowed BB

Permission is now granted to the subsystem to format write R0 with:
* an ID = CCHHR, where CC = physical cylinder number,
  HH = physical head number, and R = 0
* a key length of zero
* a data length of eight
* a data field containing all zeros

Signed-off-by: Jean-Baptiste Joret <joret@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] dasd: enable compat ioctls
Heiko Carstens [Thu, 26 Mar 2009 14:23:45 +0000]
[S390] dasd: enable compat ioctls

All of the ioctls are compatible. Just enable them.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] page fault: invoke oom-killer
Heiko Carstens [Thu, 26 Mar 2009 14:23:44 +0000]
[S390] page fault: invoke oom-killer

s390 arch backend for 1c0fe6e3bda0464728c23c8d84aa47567e8b716c
"mm: invoke oom-killer from page fault".

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years ago[S390] Automatic IPL after dump
Frank Munzert [Thu, 26 Mar 2009 14:23:43 +0000]
[S390] Automatic IPL after dump

Provide new shutdown action "dump_reipl" for automatic ipl after dump.

Signed-off-by: Frank Munzert <munzert@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>

12 years ago[S390] fix dump_stack vs. %p and (null)
Christian Borntraeger [Thu, 26 Mar 2009 14:23:42 +0000]
[S390] fix dump_stack vs. %p and (null)

The s390 implemenation of dump_stack uses %p to display stack content.
Since d97106ab53f812910a62d18afb9dbe882819c1ba (Make %p print '(null)'
for NULL pointers) this causes a strange output for dump_stack:

[...]
Process basename (pid: 8822, task: 00000000b2ece038, ksp: 00000000b24d7b38)
04000000b5685c00 00000000b24d7760 0000000000000002 (null)
00000000b24d7800 00000000b24d7778 00000000b24d7778 00000000001052fe
(null) 00000000b24d7b38 (null) 000000000000000a
000000000000000d (null) 00000000b24d7760 00000000b24d77d8
000000000051a7e8 00000000001052fe 00000000b24d7760 00000000b24d77b0
Call Trace:
[...]

This patch changes our dump_stack to use the appropriate %x format.

Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

12 years agoLinux 2.6.29
Linus Torvalds [Mon, 23 Mar 2009 23:12:14 +0000]
Linux 2.6.29

12 years agoBuild with -fno-dwarf2-cfi-asm
Kyle McMartin [Mon, 23 Mar 2009 19:25:49 +0000]
Build with -fno-dwarf2-cfi-asm

With a sufficiently new compiler and binutils, code which wasn't
previously generating .eh_frame sections has begun to.  Certain
architectures (powerpc, in this case) may generate unexpected relocation
formats in response to this, preventing modules from loading.

While the new relocation types should probably be handled, revert to the
previous behaviour with regards to generation of .eh_frame sections.

(This was reported against Fedora, which appears to be the only distro
doing any building against gcc-4.4 at present: RH bz#486545.)

Signed-off-by: Kyle McMartin <kyle@redhat.com>
Acked-by: Roland McGrath <roland@redhat.com>
Cc: Alexandre Oliva <aoliva@redhat.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years agotrivial: fix orphan dates in ext2 documentation
Jody McIntyre [Mon, 23 Mar 2009 17:13:34 +0000]
trivial: fix orphan dates in ext2 documentation

Revert the change to the orphan dates of Windows 95, DOS, compression.
Add a new orphan date for OS/2.

Signed-off-by: Jody McIntyre <scjody@sun.com>
Acked-by: Pavel Machek <pavel@ucw.cz>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Mon, 23 Mar 2009 16:25:58 +0000]
Merge git://git./linux/kernel/git/davem/net-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6: (32 commits)
  ucc_geth: Fix oops when using fixed-link support
  dm9000: locking bugfix
  net: update dnet.c for bus_id removal
  dnet: DNET should depend on HAS_IOMEM
  dca: add missing copyright/license headers
  nl80211: Check that function pointer != NULL before using it
  sungem: missing net_device_ops
  be2net: fix to restore vlan ids into BE2 during a IF DOWN->UP cycle
  be2net: replenish when posting to rx-queue is starved in out of mem conditions
  bas_gigaset: correctly allocate USB interrupt transfer buffer
  smsc911x: reset last known duplex and carrier on open
  sh_eth: Fix mistake of the address of SH7763
  sh_eth: Change handling of IRQ
  netns: oops in ip[6]_frag_reasm incrementing stats
  net: kfree(napi->skb) => kfree_skb
  net: fix sctp breakage
  ipv6: fix display of local and remote sit endpoints
  net: Document /proc/sys/net/core/netdev_budget
  tulip: fix crash on iface up with shirq debug
  virtio_net: Make virtio_net support carrier detection
  ...

12 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6
Linus Torvalds [Mon, 23 Mar 2009 16:25:24 +0000]
Merge git://git./linux/kernel/git/davem/sparc-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6:
  sparc64: Fix crash with /proc/iomem
  sparc64: Reschedule KGDB capture to a software interrupt.
  sbus: Auto-load openprom module when device opened.

12 years agofix ptrace slowness
Miklos Szeredi [Mon, 23 Mar 2009 15:07:24 +0000]
fix ptrace slowness

This patch fixes bug #12208:

  Bug-Entry       : http://bugzilla.kernel.org/show_bug.cgi?id=12208
  Subject         : uml is very slow on 2.6.28 host

This turned out to be not a scheduler regression, but an already
existing problem in ptrace being triggered by subtle scheduler
changes.

The problem is this:

 - task A is ptracing task B
 - task B stops on a trace event
 - task A is woken up and preempts task B
 - task A calls ptrace on task B, which does ptrace_check_attach()
 - this calls wait_task_inactive(), which sees that task B is still on the runq
 - task A goes to sleep for a jiffy
 - ...

Since UML does lots of the above sequences, those jiffies quickly add
up to make it slow as hell.

This patch solves this by not rescheduling in read_unlock() after
ptrace_stop() has woken up the tracer.

Thanks to Oleg Nesterov and Ingo Molnar for the feedback.

Signed-off-by: Miklos Szeredi <mszeredi@suse.cz>
CC: stable@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years agoMerge branch 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/galak/powerpc
Linus Torvalds [Mon, 23 Mar 2009 15:03:09 +0000]
Merge branch 'merge' of git://git./linux/kernel/git/galak/powerpc

* 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/galak/powerpc:
  powerpc/mm: Fix Respect _PAGE_COHERENT on classic ppc32 SW TLB load machines

12 years agopowerpc/mm: Fix Respect _PAGE_COHERENT on classic ppc32 SW TLB load machines
Kumar Gala [Sat, 14 Mar 2009 14:23:03 +0000]
powerpc/mm: Fix Respect _PAGE_COHERENT on classic ppc32 SW TLB load machines

Grant picked up the wrong version of "Respect _PAGE_COHERENT on classic
ppc32 SW" (commit a4bd6a93c3f14691c8a29e53eb04dc734b27f0db)

It was missing the code to actually deal with the fixup of
_PAGE_COHERENT based on the CPU feature.

Signed-off-by: Kumar Gala <galak@kernel.crashing.org>

12 years agoucc_geth: Fix oops when using fixed-link support
Anton Vorontsov [Mon, 23 Mar 2009 04:30:52 +0000]
ucc_geth: Fix oops when using fixed-link support

commit b1c4a9dddf09fe99b8f88252718ac5b357363dc4 ("ucc_geth: Change
uec phy id to the same format as gianfar's") introduced a regression
in the ucc_geth driver that causes this oops when fixed-link is used:

Unable to handle kernel paging request for data at address 0x00000000
Faulting instruction address: 0xc0151270
Oops: Kernel access of bad area, sig: 11 [#1]
TMCUTU
NIP: c0151270 LR: c0151270 CTR: c0017760
REGS: cf81fa60 TRAP: 0300   Not tainted  (2.6.29-rc8)
MSR: 00009032 <EE,ME,IR,DR>  CR: 24024042  XER: 20000000
DAR: 00000000, DSISR: 20000000
TASK = cf81cba0[1] 'swapper' THREAD: cf81e000
GPR00: c0151270 cf81fb10 cf81cba0 00000000 c0272e20 c025f354 00001e80
cf86b08c
GPR08: d1068200 cffffb74 06000000 d106c200 42024042 10085148 0fffd000
0ffc81a0
GPR16: 00000001 00000001 00000000 007ffeb0 00000000 0000c000 cf83f36c
cf83f000
GPR24: 00000030 cf83f360 cf81fb20 00000000 d106c200 20000000 00001e80
cf83f360
NIP [c0151270] ucc_geth_open+0x330/0x1efc
LR [c0151270] ucc_geth_open+0x330/0x1efc
Call Trace:
[cf81fb10] [c0151270] ucc_geth_open+0x330/0x1efc (unreliable)
[cf81fba0] [c0187638] dev_open+0xbc/0x12c
[cf81fbc0] [c0187e38] dev_change_flags+0x8c/0x1b0

This patch fixes the issue by removing offending (and somewhat
duplicate) code from init_phy() routine, and changes _probe()
function to use uec_mdio_bus_name().

Also, since we fully construct phy_bus_id in the _probe() routine,
we no longer need ->phy_address and ->mdio_bus fields in
ucc_geth_info structure.

I wish the patch would be a bit shorter, but it seems like the only
way to fix the issue in a sane way. Luckily, the patch has been
tested with real PHYs and fixed-link, so no further regressions
expected.

Reported-by: Joakim Tjernlund <Joakim.Tjernlund@transmode.se>
Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Tested-by: Joakim Tjernlund <Joakim.Tjernlund@transmode.se>
Signed-off-by: David S. Miller <davem@davemloft.net>

12 years agodm9000: locking bugfix
David Brownell [Mon, 23 Mar 2009 04:28:39 +0000]
dm9000: locking bugfix

This fixes a locking bug in the dm9000 driver.  It calls
request_irq() without setting IRQF_DISABLED ... which is
correct for handlers that support IRQ sharing, since that
behavior is not guaranteed for shared IRQs.  However, its
IRQ handler then wrongly assumes that IRQs are blocked.
So the fix just uses the right spinlock primitives in the
IRQ handler.

NOTE:  this is a classic example of the type of bug which
lockdep currently masks by forcibly setting IRQF_DISABLED
on IRQ handlers that did not request that flag.

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

12 years agonet: update dnet.c for bus_id removal
Stephen Rothwell [Mon, 23 Mar 2009 04:22:48 +0000]
net: update dnet.c for bus_id removal

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

12 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/sam/kbuild-fixes
Linus Torvalds [Sun, 22 Mar 2009 18:41:02 +0000]
Merge git://git./linux/kernel/git/sam/kbuild-fixes

* git://git.kernel.org/pub/scm/linux/kernel/git/sam/kbuild-fixes:
  kconfig: improve seed in randconfig
  kconfig: fix randconfig for choice blocks

12 years agoMerge branch 'fix-includes' of git://git.kernel.org/pub/scm/linux/kernel/git/gerg...
Linus Torvalds [Sun, 22 Mar 2009 18:38:57 +0000]
Merge branch 'fix-includes' of git://git./linux/kernel/git/gerg/m68knommu

* 'fix-includes' of git://git.kernel.org/pub/scm/linux/kernel/git/gerg/m68knommu:
  m68k: merge the non-MMU and MMU versions of siginfo.h
  m68k: use the MMU version of unistd.h for all m68k platforms
  m68k: merge the non-MMU and MMU versions of signal.h
  m68k: merge the non-MMU and MMU versions of ptrace.h
  m68k: use MMU version of setup.h for both MMU and non-MMU
  m68k: merge the non-MMU and MMU versions of sigcontext.h
  m68k: merge the non-MMU and MMU versions of swab.h
  m68k: merge the non-MMU and MMU versions of param.h

12 years agoUpdate my email address
Gertjan van Wingerde [Sat, 21 Mar 2009 22:18:57 +0000]
Update my email address

Update all previous incarnations of my email address to the correct one.

Signed-off-by: Gertjan van Wingerde <gwingerde@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years agoeCryptfs: NULL crypt_stat dereference during lookup
Tyler Hicks [Fri, 20 Mar 2009 07:23:57 +0000]
eCryptfs: NULL crypt_stat dereference during lookup

If ecryptfs_encrypted_view or ecryptfs_xattr_metadata were being
specified as mount options, a NULL pointer dereference of crypt_stat
was possible during lookup.

This patch moves the crypt_stat assignment into
ecryptfs_lookup_and_interpose_lower(), ensuring that crypt_stat
will not be NULL before we attempt to dereference it.

Thanks to Dan Carpenter and his static analysis tool, smatch, for
finding this bug.

Signed-off-by: Tyler Hicks <tyhicks@linux.vnet.ibm.com>
Acked-by: Dustin Kirkland <kirkland@canonical.com>
Cc: Dan Carpenter <error27@gmail.com>
Cc: Serge Hallyn <serue@us.ibm.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years agoeCryptfs: Allocate a variable number of pages for file headers
Tyler Hicks [Fri, 20 Mar 2009 06:25:09 +0000]
eCryptfs: Allocate a variable number of pages for file headers

When allocating the memory used to store the eCryptfs header contents, a
single, zeroed page was being allocated with get_zeroed_page().
However, the size of an eCryptfs header is either PAGE_CACHE_SIZE or
ECRYPTFS_MINIMUM_HEADER_EXTENT_SIZE (8192), whichever is larger, and is
stored in the file's private_data->crypt_stat->num_header_bytes_at_front
field.

ecryptfs_write_metadata_to_contents() was using
num_header_bytes_at_front to decide how many bytes should be written to
the lower filesystem for the file header.  Unfortunately, at least 8K
was being written from the page, despite the chance of the single,
zeroed page being smaller than 8K.  This resulted in random areas of
kernel memory being written between the 0x1000 and 0x1FFF bytes offsets
in the eCryptfs file headers if PAGE_SIZE was 4K.

This patch allocates a variable number of pages, calculated with
num_header_bytes_at_front, and passes the number of allocated pages
along to ecryptfs_write_metadata_to_contents().

Thanks to Florian Streibelt for reporting the data leak and working with
me to find the problem.  2.6.28 is the only kernel release with this
vulnerability.  Corresponds to CVE-2009-0787

Signed-off-by: Tyler Hicks <tyhicks@linux.vnet.ibm.com>
Acked-by: Dustin Kirkland <kirkland@canonical.com>
Reviewed-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Eugene Teo <eugeneteo@kernel.sg>
Cc: Greg KH <greg@kroah.com>
Cc: dann frazier <dannf@dannf.org>
Cc: Serge E. Hallyn <serue@us.ibm.com>
Cc: Florian Streibelt <florian@f-streibelt.de>
Cc: stable@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years agoradeonfb: Whack the PCI PM register until it sticks
Benjamin Herrenschmidt [Thu, 19 Mar 2009 22:22:30 +0000]
radeonfb: Whack the PCI PM register until it sticks

This fixes a regression introduced when we switched to using the core
pci_set_power_state().  The chip seems to need the state to be written
over and over again until it sticks, so we do that.

Note that the code is a bit blunt, without timeout, etc...  but that's
pretty much because I put back in there the code exactly as it used to
be before the regression.  I still add a call to pci_set_power_state()
at the end so that ACPI gets called appropriately on x86.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Tested-by: Raymond Wooninck <tittiatcoke@gmail.com>
Acked-by: Rafael J. Wysocki <rjw@sisk.pl>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 years agodnet: DNET should depend on HAS_IOMEM
Ilya Yanok [Sat, 21 Mar 2009 23:58:47 +0000]
dnet: DNET should depend on HAS_IOMEM

Signed-off-by: Ilya Yanok <yanok@emcraft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

12 years agodca: add missing copyright/license headers
Maciej Sosnowski [Sat, 21 Mar 2009 20:31:23 +0000]
dca: add missing copyright/license headers

In two dca files copyright and license headers are missing.
This patch adds them there.

Signed-off-by: Maciej Sosnowski <maciej.sosnowski@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

12 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wirel...
David S. Miller [Fri, 20 Mar 2009 21:53:36 +0000]
Merge branch 'master' of git://git./linux/kernel/git/linville/wireless-2.6

12 years agonl80211: Check that function pointer != NULL before using it
Jouni Malinen [Fri, 20 Mar 2009 15:57:36 +0000]
nl80211: Check that function pointer != NULL before using it

NL80211_CMD_GET_MESH_PARAMS and NL80211_CMD_SET_MESH_PARAMS handlers
did not verify whether a function pointer is NULL (not supported by
the driver) before trying to call the function. The former nl80211
command is available for unprivileged users, too, so this can
potentially allow normal users to kill networking (or worse..) if
mac80211 is built without CONFIG_MAC80211_MESH=y.

Signed-off-by: Jouni Malinen <jouni.malinen@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

12 years agosungem: missing net_device_ops
Stephen Hemminger [Fri, 20 Mar 2009 06:58:01 +0000]
sungem: missing net_device_ops

Sungem driver only got partially converted to net_device_ops.
Since this could cause bugs, please push this to 2.6.29

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

12 years agobe2net: fix to restore vlan ids into BE2 during a IF DOWN->UP cycle
Sathya Perla [Fri, 20 Mar 2009 06:56:46 +0000]
be2net: fix to restore vlan ids into BE2 during a IF DOWN->UP cycle

This is a patch to reconfigure vlan-ids during an i/f down/up cycle

Signed-off-by: Sathya Perla <sathyap@serverengines.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

12 years agobe2net: replenish when posting to rx-queue is starved in out of mem conditions
Sathya Perla [Fri, 20 Mar 2009 06:56:20 +0000]
be2net: replenish when posting to rx-queue is starved in out of mem conditions

This is a patch to replenish the rx-queue when it is in a starved
state (due to out-of-mem conditions)

Signed-off-by: Sathya Perla <sathyap@serverengines.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

12 years agoaio: lookup_ioctx can return the wrong value when looking up a bogus context
Jeff Moyer [Thu, 19 Mar 2009 00:04:21 +0000]
aio: lookup_ioctx can return the wrong value when looking up a bogus context

The libaio test harness turned up a problem whereby lookup_ioctx on a
bogus io context was returning the 1 valid io context from the list
(harness/cases/3.p).

Because of that, an extra put_iocontext was done, and when the process
exited, it hit a BUG_ON in the put_iocontext macro called from exit_aio
(since we expect a users count of 1 and instead get 0).

The problem was introduced by "aio: make the lookup_ioctx() lockless"
(commit abf137dd7712132ee56d5b3143c2ff61a72a5faa).

Thanks to Zach for pointing out that hlist_for_each_entry_rcu will not
return with a NULL tpos at the end of the loop, even if the entry was
not found.

Signed-off-by: Jeff Moyer <jmoyer@redhat.com>
Acked-by: Zach Brown <zach.brown@oracle.com>
Acked-by: Jens Axboe <jens.axboe@oracle.com>
Cc: Benjamin LaHaise <bcrl@kvack.org>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>