linux-2.6
14 years agoMerge git://git.infradead.org/iommu-2.6
Linus Torvalds [Sun, 28 Jun 2009 18:06:11 +0000 (11:06 -0700)] 
Merge git://git.infradead.org/iommu-2.6

* git://git.infradead.org/iommu-2.6:
  intel-iommu: fix Identity Mapping to be arch independent

14 years agoMerge branch 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 28 Jun 2009 18:05:28 +0000 (11:05 -0700)] 
Merge branch 'x86-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  x86, delay: tsc based udelay should have rdtsc_barrier
  x86, setup: correct include file in <asm/boot.h>
  x86, setup: Fix typo "CONFIG_x86_64" in <asm/boot.h>
  x86, mce: percpu mcheck_timer should be pinned
  x86: Add sysctl to allow panic on IOCK NMI error
  x86: Fix uv bau sending buffer initialization
  x86, mce: Fix mce resume on 32bit
  x86: Move init_gbpages() to setup_arch()
  x86: ensure percpu lpage doesn't consume too much vmalloc space
  x86: implement percpu_alloc kernel parameter
  x86: fix pageattr handling for lpage percpu allocator and re-enable it
  x86: reorganize cpa_process_alias()
  x86: prepare setup_pcpu_lpage() for pageattr fix
  x86: rename remap percpu first chunk allocator to lpage
  x86: fix duplicate free in setup_pcpu_remap() failure path
  percpu: fix too lazy vunmap cache flushing
  x86: Set cpu_llc_id on AMD CPUs

14 years agoMerge branch 'timers-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 28 Jun 2009 18:05:16 +0000 (11:05 -0700)] 
Merge branch 'timers-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'timers-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  timer stats: Optimize by adding quick check to avoid function calls
  timers: Fix timer_migration interface which accepts any number as input

14 years agoMerge branch 'tracing-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 28 Jun 2009 18:05:04 +0000 (11:05 -0700)] 
Merge branch 'tracing-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'tracing-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  ftrace: Fix the output of profile
  ring-buffer: Make it generally available
  ftrace: Remove duplicate newline
  tracing: Fix trace_buf_size boot option
  ftrace: Fix t_hash_start()
  ftrace: Don't manipulate @pos in t_start()
  ftrace: Don't increment @pos in g_start()
  tracing: Reset iterator in t_start()
  trace_stat: Don't increment @pos in seq start()
  tracing_bprintk: Don't increment @pos in t_start()
  tracing/events: Don't increment @pos in s_start()

14 years agoMerge branch 'for-linus' of git://git.o-hand.com/linux-rpurdie-leds
Linus Torvalds [Sun, 28 Jun 2009 18:02:32 +0000 (11:02 -0700)] 
Merge branch 'for-linus' of git://git.o-hand.com/linux-rpurdie-leds

* 'for-linus' of git://git.o-hand.com/linux-rpurdie-leds:
  leds: Futher document blink_set
  leds: Add options to have GPIO LEDs start on or keep their state
  leds: LED driver for National Semiconductor LP3944 Funlight Chip
  leds: pca9532 - Indent using tabs, not spaces.
  leds: Remove an orphan Kconfig entry
  leds: Further document parameters for blink_set()
  leds: alix-leds2 fixed for Award BIOS
  leds: leds-gpio - fix a section mismatch
  leds: add the sysfs interface into the leds-bd2802 driver for changing wave pattern and led current.
  leds: change the license information
  leds: fix led-bd2802 errors while resuming

14 years agoperf stat: Micro-optimize the code: memcpy is only required if no event is selected...
Jaswinder Singh Rajput [Sat, 27 Jun 2009 18:19:09 +0000 (23:49 +0530)] 
perf stat: Micro-optimize the code: memcpy is only required if no event is selected and !null_run

Set attrs and nr_counters if no event is selected and !null_run.

Setting of attrs should depend on number of counters,
so we need to memcpy only for sizeof(default_attrs)

Also set nr_counters as ARRAY_SIZE(default_attrs) in place of
hardcoded value.

Signed-off-by: Jaswinder Singh Rajput <jaswinderrajput@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
LKML-Reference: <1246126749.32198.16.camel@hpdv5.satnam>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
14 years agoKVM: shut up uninit compiler warning in paging_tmpl.h
Jaswinder Singh Rajput [Wed, 20 May 2009 04:29:35 +0000 (09:59 +0530)] 
KVM: shut up uninit compiler warning in paging_tmpl.h

Dixes compilation warning:
  CC      arch/x86/kernel/io_delay.o
 arch/x86/kvm/paging_tmpl.h: In function ‘paging64_fetch’:
 arch/x86/kvm/paging_tmpl.h:279: warning: ‘sptep’ may be used uninitialized in this function
 arch/x86/kvm/paging_tmpl.h: In function ‘paging32_fetch’:
 arch/x86/kvm/paging_tmpl.h:279: warning: ‘sptep’ may be used uninitialized in this function

warning is bogus (always have a least one level), but need to shut the compiler
up.

Signed-off-by: Jaswinder Singh Rajput <jaswinderrajput@gmail.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
14 years agoKVM: Ignore reads to K7 EVNTSEL MSRs
Amit Shah [Mon, 15 Jun 2009 07:55:34 +0000 (13:25 +0530)] 
KVM: Ignore reads to K7 EVNTSEL MSRs

In commit 7fe29e0faacb650d31b9e9f538203a157bec821d we ignored the
reads to the P6 EVNTSEL MSRs. That fixed crashes on Intel machines.

Ignore the reads to K7 EVNTSEL MSRs as well to fix this on AMD
hosts.

This fixes Kaspersky antivirus crashing Windows guests on AMD hosts.

Signed-off-by: Amit Shah <amit.shah@redhat.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
14 years agoKVM: VMX: Handle vmx instruction vmexits
Avi Kivity [Tue, 16 Jun 2009 11:19:52 +0000 (14:19 +0300)] 
KVM: VMX: Handle vmx instruction vmexits

IF a guest tries to use vmx instructions, inject a #UD to let it know the
instruction is not implemented, rather than crashing.

This prevents guest userspace from crashing the guest kernel.

Cc: stable@kernel.org
Signed-off-by: Avi Kivity <avi@redhat.com>
14 years agoKVM: s390: Allow stfle instruction in the guest
Christian Borntraeger [Tue, 23 Jun 2009 15:24:07 +0000 (17:24 +0200)] 
KVM: s390: Allow stfle instruction in the guest

2.6.31-rc introduced an architecture level set checker based on facility
bits. e.g. if the kernel is compiled to run only on z9, several facility
bits are checked very early and the kernel refuses to boot if a z9 specific
facility is missing.
Until now kvm on s390 did not implement the store facility extended (STFLE)
instruction. A 2.6.31-rc kernel that was compiled for z9 or higher did not
boot in kvm. This patch implements stfle.

This patch should go in before 2.6.31.

Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
14 years agoKVM: kvm/x86_emulate.c toggle_interruptibility() should be static
Jaswinder Singh Rajput [Thu, 18 Jun 2009 11:23:25 +0000 (16:53 +0530)] 
KVM: kvm/x86_emulate.c toggle_interruptibility() should be static

toggle_interruptibility() is used only by same file, it should be static.

Fixed following sparse warning :

  arch/x86/kvm/x86_emulate.c:1364:6: warning: symbol 'toggle_interruptibility' was not declared. Should it be static?

Signed-off-by: Jaswinder Singh Rajput <jaswinderrajput@gmail.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
14 years agoKVM: ia64: fix ia64 build due to missing kallsyms_lookup() and double export
Jes Sorensen [Wed, 17 Jun 2009 09:08:08 +0000 (11:08 +0200)] 
KVM: ia64: fix ia64 build due to missing kallsyms_lookup() and double export

Fix problem with double export of certain symbols from vsprintf.c
which we do not wish to export from the kvm-intel.ko module.

In addition, we do not have access to kallsyms_lookup() from the
module, so make sure to #undef CONFIG_KALLSYMS

Signed-off-by: Jes Sorensen <jes@sgi.com>
Acked-by: Xiantao Zhang <xiantao.zhang@intel.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
14 years agoKVM: protect concurrent make_all_cpus_request
Marcelo Tosatti [Wed, 17 Jun 2009 13:53:47 +0000 (10:53 -0300)] 
KVM: protect concurrent make_all_cpus_request

make_all_cpus_request contains a race condition which can
trigger false request completed status, as follows:

CPU0                                              CPU1

if (test_and_set_bit(req,&vcpu->requests))
   ....                                            if (test_and_set_bit(req,&vcpu->requests))
   ..                                                  return
proceed to smp_call_function_many(wait=1)

Use a spinlock to serialize concurrent CPUs.

Cc: stable@kernel.org
Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
14 years agoKVM: MMU: Allow 4K ptes with bit 7 (PAT) set
Avi Kivity [Tue, 19 May 2009 10:29:27 +0000 (13:29 +0300)] 
KVM: MMU: Allow 4K ptes with bit 7 (PAT) set

Bit 7 is perfectly legal in the 4K page leve; it is used for the PAT.

Signed-off-by: Avi Kivity <avi@redhat.com>
14 years agoKVM: Fix dirty bit tracking for slots with large pages
Izik Eidus [Wed, 10 Jun 2009 16:23:24 +0000 (19:23 +0300)] 
KVM: Fix dirty bit tracking for slots with large pages

When slot is already allocated and being asked to be tracked we need
to break the large pages.

This code flush the mmu when someone ask a slot to start dirty bit
tracking.

Cc: stable@kernel.org
Signed-off-by: Izik Eidus <ieidus@redhat.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
14 years agomtd: nand: fix build failure and incorrect return from omap_wait()
vimal singh [Sat, 27 Jun 2009 05:37:06 +0000 (11:07 +0530)] 
mtd: nand: fix build failure and incorrect return from omap_wait()

We need to include jiffies.h manually in some cases, and the status
returned from omap_wait() was broken in two separate ways.

Also add cond_resched() to the loop.

Signed-off-by: Vimal Singh <vimalsingh@ti.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
14 years agoRevert "x86: cap iomem_resource to addressable physical memory"
H. Peter Anvin [Sat, 27 Jun 2009 19:22:27 +0000 (12:22 -0700)] 
Revert "x86: cap iomem_resource to addressable physical memory"

This reverts commit 95ee14e4379c5e19c0897c872350570402014742.
Mikael Petterson <mikepe@it.uu.se> reported that at least one of his
systems will not boot as a result.  We have ruled out the detection
algorithm malfunctioning, so it is not a matter of producing the
incorrect bitmasks; rather, something in the application of them
fails.

Revert the commit until we can root cause and correct this problem.

-stable team: this means the underlying commit should be rejected.

Reported-and-isolated-by: Mikael Petterson <mikpe@it.uu.se>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
LKML-Reference: <200906261559.n5QFxJH8027336@pilspetsen.it.uu.se>
Cc: stable@kernel.org
Cc: Grant Grundler <grundler@parisc-linux.org>
14 years agocifs: fix fh_mutex locking in cifs_reopen_file
Jeff Layton [Sat, 27 Jun 2009 11:04:55 +0000 (07:04 -0400)] 
cifs: fix fh_mutex locking in cifs_reopen_file

Fixes a regression caused by commit a6ce4932fbdbcd8f8e8c6df76812014351c32892

When this lock was converted to a mutex, the locks were turned into
unlocks and vice-versa.

Signed-off-by: Jeff Layton <jlayton@redhat.com>
Acked-by: Shirish Pargaonkar <shirishp@us.ibm.com>
Cc: Stable Tree <stable@kernel.org>
Signed-off-by: Steve French <sfrench@us.ibm.com>
14 years ago[SCSI] cxgb3i: fix connection error when vlan is enabled
Karen Xie [Fri, 26 Jun 2009 22:17:29 +0000 (15:17 -0700)] 
[SCSI] cxgb3i: fix connection error when vlan is enabled

There is a bug when VLAN is configured on the cxgb3 interface, the iscsi
conn. would be denied with message "cxgb3i: NOT going through cxgbi device."

This patch adds code to get the real egress net_device when vlan is configured.

Signed-off-by: Karen Xie <kxie@chelsio.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
14 years agoperf stat: Improve output
Jaswinder Singh Rajput [Fri, 26 Jun 2009 21:32:07 +0000 (03:02 +0530)] 
perf stat: Improve output

Increase size for event name to handle bigger names like
'L1-d$-prefetch-misses'

Changed scaled counters from percentage to a multiplicative
factor because the latter is more expressive.

Also aligned the scaling factor, otherwise sometimes it looks
like:

            384  iTLB-load-misses           (4.74x scaled)
         452029  branch-loads               (8.00x scaled)
           5892  branch-load-misses         (20.39x scaled)
         972315  iTLB-loads                 (3.24x scaled)

Before:
         150708  L1-d$-stores          (scaled from 23.57%)
         428804  L1-d$-prefetches      (scaled from 23.47%)
         314446  L1-d$-prefetch-misses  (scaled from 23.42%)
      252626137  L1-i$-loads           (scaled from 23.24%)
        5297550  dTLB-load-misses      (scaled from 23.96%)
      106992392  branch-loads          (scaled from 23.67%)
        5239561  branch-load-misses    (scaled from 23.43%)

After:
        1731713  L1-d$-loads               (  14.25x scaled)
          44241  L1-d$-prefetches          (   3.88x scaled)
          21076  L1-d$-prefetch-misses     (   3.40x scaled)
        5789421  L1-i$-loads               (   3.78x scaled)
          29645  dTLB-load-misses          (   2.95x scaled)
         461474  branch-loads              (   6.52x scaled)
           7493  branch-load-misses        (  26.57x scaled)

Reported-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Jaswinder Singh Rajput <jaswinderrajput@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
LKML-Reference: <1246051927.2988.10.camel@hpdv5.satnam>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
14 years ago[ARM] 5563/1: at91: at91sam9rlek lcd interface correction
Nicolas Ferre [Wed, 24 Jun 2009 16:13:48 +0000 (17:13 +0100)] 
[ARM] 5563/1: at91: at91sam9rlek lcd interface correction

Here is a little update to the at91sam9rlek lcd interface.
This will correct the power pin of the LCD.
It will also add precision to the struct atmel_lcdfb_info
scructure: backlight enabling  and wiring mode correction:
RGB wiring on the -EK board.

Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Acked-by: Andrew Victor <linux@maxim.org.za>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
14 years agomtd: Use BLOCK_NIL consistently in NFTL/INFTL
Julia Lawall [Sat, 27 Jun 2009 07:55:32 +0000 (09:55 +0200)] 
mtd: Use BLOCK_NIL consistently in NFTL/INFTL

Use BLOCK_NIL consistently rather than sometimes 0xffff and sometimes
BLOCK_NIL.

The semantic patch that finds this issue is below
(http://www.emn.fr/x-info/coccinelle/).  On the other hand, the changes
were made by hand, in part because drivers/mtd/inftlcore.c contains dead
code that causes spatch to ignore a relevant function.  Specifically, the
function INFTL_findwriteunit contains a do-while loop, but always takes a
return that leaves the loop on the first iteration.

// <smpl>
@r exists@
identifier f,C;
@@

f(...) { ... return C; }

@s@
identifier r.C;
expression E;
@@

@@
identifier r.f,r.C,I;
expression s.E;
@@

f(...) {
 <...
(
  I
|
- E
+ C
)
 ...>
}

// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
14 years agoperf stat: Fix multi-run stats
Ingo Molnar [Sat, 27 Jun 2009 04:24:32 +0000 (06:24 +0200)] 
perf stat: Fix multi-run stats

In multi-run (-r/--repeat) printouts, print out the noise of
the wall-clock average as well.

Also, fix a bug in printing out scaled counters: if it was not
scaled then we should not update the average with -1.

Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
14 years agoperf stat: Add -n/--null option to run without counters
Ingo Molnar [Sat, 27 Jun 2009 04:10:30 +0000 (06:10 +0200)] 
perf stat: Add -n/--null option to run without counters

Allow a no-counters run. This can be useful to measure just
elapsed wall-clock time - or to assess the raw overhead of perf
stat itself, without running any counters.

Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
14 years agoperf_counter tools: Remove dead code
Ingo Molnar [Sat, 27 Jun 2009 04:06:39 +0000 (06:06 +0200)] 
perf_counter tools: Remove dead code

Vince Weaver reported that there's a handful of #ifdef __MINGW32__
sections in the code.

Remove them as they are in essence dead code - as unlike upstream
Git, the perf tool is unlikely to be ported to Windows.

Reported-by: Vince Weaver <vince@deater.net>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
14 years agogro: Flush GRO packets in napi_disable_pending path
Herbert Xu [Sat, 27 Jun 2009 02:27:04 +0000 (19:27 -0700)] 
gro: Flush GRO packets in napi_disable_pending path

When NAPI is disabled while we're in net_rx_action, we end up
calling __napi_complete without flushing GRO packets.  This is
a bug as it would cause the GRO packets to linger, of course it
also literally BUGs to catch error like this :)

This patch changes it to napi_complete, with the obligatory IRQ
reenabling.  This should be safe because we've only just disabled
IRQs and it does not materially affect the test conditions in
between.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoinet: Call skb_orphan before tproxy activates
Herbert Xu [Sat, 27 Jun 2009 02:22:37 +0000 (19:22 -0700)] 
inet: Call skb_orphan before tproxy activates

As transparent proxying looks up the socket early and assigns
it to the skb for later processing, we must drop any existing
socket ownership prior to that in order to distinguish between
the case where tproxy is active and where it is not.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agomac80211: Use rcu_barrier() on unload.
Jesper Dangaard Brouer [Fri, 26 Jun 2009 10:45:53 +0000 (10:45 +0000)] 
mac80211: Use rcu_barrier() on unload.

The mac80211 module uses rcu_call() thus it should use rcu_barrier()
on module unload.

The rcu_barrier() is placed in mech.c ieee80211_stop_mesh() which is
invoked from ieee80211_stop() in case vif.type == NL80211_IFTYPE_MESH_POINT.

Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Acked-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: Jesper Dangaard Brouer <hawk@comx.dk>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosunrpc: Use rcu_barrier() on unload.
Jesper Dangaard Brouer [Fri, 26 Jun 2009 10:45:58 +0000 (10:45 +0000)] 
sunrpc: Use rcu_barrier() on unload.

The sunrpc module uses rcu_call() thus it should use rcu_barrier() on
module unload.

Have not verified that the possibility for new call_rcu() callbacks
has been disabled.  As a hint for checking, the functions calling
call_rcu() (unx_destroy_cred and generic_destroy_cred) are
registered as crdestroy function pointer in struct rpc_credops.

Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Acked-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Signed-off-by: Jesper Dangaard Brouer <hawk@comx.dk>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agobridge: Use rcu_barrier() instead of syncronize_net() on unload.
Jesper Dangaard Brouer [Fri, 26 Jun 2009 10:45:48 +0000 (10:45 +0000)] 
bridge: Use rcu_barrier() instead of syncronize_net() on unload.

When unloading modules that uses call_rcu() callbacks, then we must
use rcu_barrier().  This module uses syncronize_net() which is not
enough to be sure that all callback has been completed.

Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Jesper Dangaard Brouer <hawk@comx.dk>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoipv6: Use rcu_barrier() on module unload.
Jesper Dangaard Brouer [Fri, 26 Jun 2009 10:46:03 +0000 (10:46 +0000)] 
ipv6: Use rcu_barrier() on module unload.

The ipv6 module uses rcu_call() thus it should use rcu_barrier() on
module unload.

Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Jesper Dangaard Brouer <hawk@comx.dk>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agodecnet: Use rcu_barrier() on module unload.
Jesper Dangaard Brouer [Fri, 26 Jun 2009 10:46:08 +0000 (10:46 +0000)] 
decnet: Use rcu_barrier() on module unload.

The decnet module unloading as been disabled with a '#if 0' statement,
because it have had issues.

We add a rcu_barrier() anyhow for correctness.

The maintainer (Chrissie Caulfield) will look into the unload issue
when time permits.

Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Acked-by: Chrissie Caulfield <christine.caulfield@googlemail.com>
Signed-off-by: Jesper Dangaard Brouer <hawk@comx.dk>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoide-cd: handle fragmented packet commands gracefully
Borislav Petkov [Fri, 26 Jun 2009 18:22:37 +0000 (11:22 -0700)] 
ide-cd: handle fragmented packet commands gracefully

There are some devices in the wild that clear the DRQ bit during the
last word of a packet command and therefore could use a "second chance"
for that last word of data to be xferred instead of simply failing the
request. Do that by attempting to suck in those last bytes in PIO mode.

In addition, the ATA_ERR bit has to be cleared for we cannot be sure the
data is valid otherwise.

See http://bugzilla.kernel.org/show_bug.cgi?id=13399 for details.

Signed-off-by: Borislav Petkov <petkovbb@gmail.com>
Acked-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agomtd: m25p80 timeout too short for worst-case m25p16 devices
Steven A. Falco [Fri, 26 Jun 2009 16:42:47 +0000 (12:42 -0400)] 
mtd: m25p80 timeout too short for worst-case m25p16 devices

The m25p16 data sheet from numonyx lists the worst-case bulk erase time
(tBE) as 40 seconds.

Signed-off-by: Steven A. Falco <sfalco@harris.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/bp/bp
Linus Torvalds [Fri, 26 Jun 2009 16:39:40 +0000 (09:39 -0700)] 
Merge branch 'for-linus' of git://git./linux/kernel/git/bp/bp

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/bp/bp:
  amd64_edac: misc small cleanups
  amd64_edac: fix ecc_enable_override handling
  amd64_edac: check only ECC bit in amd64_determine_edac_cap

14 years agoMerge branch 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc
Linus Torvalds [Fri, 26 Jun 2009 16:39:02 +0000 (09:39 -0700)] 
Merge branch 'merge' of git://git./linux/kernel/git/benh/powerpc

* 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc: (29 commits)
  powerpc/rtas: Fix watchdog driver temperature read functionality
  powerpc/mm: Fix potential access to freed pages when using hugetlbfs
  powerpc/440: Fix warning early debug code
  powerpc/of: Fix usage of dev_set_name() in of_device_alloc()
  powerpc/pasemi: Use raw spinlock in SMP TB sync
  powerpc: Use one common impl. of RTAS timebase sync and use raw spinlock
  powerpc/rtas: Turn rtas lock into a raw spinlock
  powerpc: Add irqtrace support for 32-bit powerpc
  powerpc/BSR: Fix BSR to allow mmap of small BSR on 64k kernel
  powerpc/BSR: add 4096 byte BSR size
  powerpc: Map more memory early on 601 processors
  powerpc/pmac: Fix DMA ops for MacIO devices
  powerpc/mm: Make k(un)map_atomic out of line
  powerpc: Fix mpic alloc warning
  powerpc: Fix output from show_regs
  powerpc/pmac: Fix issues with PowerMac "PowerSurge" SMP
  powerpc/amigaone: Limit ISA I/O range to 4k in the device tree
  powerpc/warp: Platform fix for i2c change
  powerpc: Have git ignore generated files from dtc compile
  powerpc/mpic: Fix mapping of "DCR" based MPIC variants
  ...

14 years agokmemleak: Slightly change the policy on newly allocated objects
Catalin Marinas [Fri, 26 Jun 2009 16:38:29 +0000 (17:38 +0100)] 
kmemleak: Slightly change the policy on newly allocated objects

Newly allocated objects are more likely to be reported as false
positives. Kmemleak ignores the reporting of objects younger than 5
seconds. However, this age was calculated after the memory scanning
completed which usually takes longer than 5 seconds. This patch
make the minimum object age calculation in relation to the start of the
memory scanning.

Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
14 years agokmemleak: Do not trigger a scan when reading the debug/kmemleak file
Catalin Marinas [Fri, 26 Jun 2009 16:38:27 +0000 (17:38 +0100)] 
kmemleak: Do not trigger a scan when reading the debug/kmemleak file

Since there is a kernel thread for automatically scanning the memory, it
makes sense for the debug/kmemleak file to only show its findings. This
patch also adds support for "echo scan > debug/kmemleak" to trigger an
intermediate memory scan and eliminates the kmemleak_mutex (scan_mutex
covers all the cases now).

Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
14 years agokmemleak: Simplify the reports logged by the scanning thread
Catalin Marinas [Fri, 26 Jun 2009 16:38:26 +0000 (17:38 +0100)] 
kmemleak: Simplify the reports logged by the scanning thread

Because of false positives, the memory scanning thread may print too
much information. This patch changes the scanning thread to only print
the number of newly suspected leaks. Further information can be read
from the /sys/kernel/debug/kmemleak file.

Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
14 years agokmemleak: Enable task stacks scanning by default
Catalin Marinas [Fri, 26 Jun 2009 16:38:25 +0000 (17:38 +0100)] 
kmemleak: Enable task stacks scanning by default

This is to reduce the number of false positives reported.

Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
14 years agoMerge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux...
Linus Torvalds [Fri, 26 Jun 2009 16:37:52 +0000 (09:37 -0700)] 
Merge branch 'release' of git://git./linux/kernel/git/lenb/linux-acpi-2.6

* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-acpi-2.6:
  ACPI: video: prevent NULL deref in acpi_get_pci_dev()
  eeepc-laptop: add rfkill support for the 3G modem in Eee PC 901 Go
  eeepc-laptop: get the right value for CMSG
  eeepc-laptop: makes get_acpi() returns -ENODEV
  eeepc-laptop: right parent device
  eeepc-laptop: rfkill refactoring
  eeepc-laptop.c: use pr_fmt and pr_<level>
  eeepc-laptop: Register as a pci-hotplug device

14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6
Linus Torvalds [Fri, 26 Jun 2009 16:37:19 +0000 (09:37 -0700)] 
Merge git://git./linux/kernel/git/sfrench/cifs-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6:
  [CIFS] remove unknown mount option warning message
  [CIFS] remove bkl usage from umount begin
  cifs: Fix incorrect return code being printed in cFYI messages
  [CIFS] cleanup asn handling for ntlmssp
  [CIFS] Copy struct *after* setting the port, instead of before.
  cifs: remove rw/ro options
  cifs: fix problems with earlier patches
  cifs: have cifs parse scope_id out of IPv6 addresses and use it
  [CIFS] Do not send tree disconnect if session is already disconnected
  [CIFS] Fix build break
  cifs: display scopeid in /proc/mounts
  cifs: add new routine for converting AF_INET and AF_INET6 addrs
  cifs: have cifs_show_options show forceuid/forcegid options
  cifs: remove unneeded NULL checks from cifs_show_options

14 years agosky2: Fix checksum endianness
Anton Vorontsov [Fri, 26 Jun 2009 16:28:42 +0000 (09:28 -0700)] 
sky2: Fix checksum endianness

sky2 driver on PowerPC targets floods kernel log with following errors:

  eth1: hw csum failure.
  Call Trace:
  [ef84b8a0] [c00075e4] show_stack+0x50/0x160 (unreliable)
  [ef84b8d0] [c02fa178] netdev_rx_csum_fault+0x3c/0x5c
  [ef84b8f0] [c02f6920] __skb_checksum_complete_head+0x7c/0x84
  [ef84b900] [c02f693c] __skb_checksum_complete+0x14/0x24
  [ef84b910] [c0337e08] tcp_v4_rcv+0x4c8/0x6f8
  [ef84b940] [c031a9c8] ip_local_deliver+0x98/0x210
  [ef84b960] [c031a788] ip_rcv+0x38c/0x534
  [ef84b990] [c0300338] netif_receive_skb+0x260/0x36c
  [ef84b9c0] [c025de00] sky2_poll+0x5dc/0xcf8
  [ef84ba20] [c02fb7fc] net_rx_action+0xc0/0x144

The NIC is Yukon-2 EC chip revision 1.

Converting checksum field from le16 to CPU byte order fixes the issue.

Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoperf_counter: Complete counter swap
Peter Zijlstra [Fri, 26 Jun 2009 11:10:23 +0000 (13:10 +0200)] 
perf_counter: Complete counter swap

Complete the counter swap by indeed switching the times too and
updating the userpage after modifying the counter values.

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <1246014623.31755.195.camel@twins>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6
Linus Torvalds [Fri, 26 Jun 2009 15:48:42 +0000 (08:48 -0700)] 
Merge git://git./linux/kernel/git/davem/sparc-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6:
  sparc32: Fix makefile not generating required files
  sparc32: Fix tftpboot.img Makefile
  sparc: fix tftpboot.img build
  sparc32: Fix obvious build issues for tftpboot.img build.
  sparc64: Fix build warnings in piggyback_64.c
  sparc64: Don't use alloc_bootmem() in init_IRQ() code paths.

14 years ago[SCSI] FC transport: Locking fix for common-code FC pass-through patch
Christof Schmitt [Fri, 26 Jun 2009 14:30:43 +0000 (16:30 +0200)] 
[SCSI] FC transport: Locking fix for common-code FC pass-through patch

Fix this:
------------[ cut here ]------------
Badness at block/blk-core.c:244
CPU: 0 Tainted: G        W  2.6.31-rc1-00004-gd3a263a #3
Process zfcp_wq (pid: 901, task: 000000002fb7a038, ksp: 000000002f02bc78)
Krnl PSW : 0704300180000000 00000000002141ba (blk_remove_plug+0xb2/0xb8)
           R:0 T:1 IO:1 EX:1 Key:0 M:1 W:0 P:0 AS:0 CC:3 PM:0 EA:3
Krnl GPRS: 0000000000000001 0000000000000001 0000000022811440 0000000022811798
           000000000027ff4e 0000000000000000 0000000000000000 000000002f00f000
           070000000006a0f4 000000002af70000 000000002af2a800 00000000228d1c00
           0000000022811440 000000000050c708 000000002f02bca8 000000002f02bc80
Krnl Code: 00000000002141b0b9140022 lgfr %r2,%r2
           00000000002141b4: 07fe bcr 15,%r14
           00000000002141b6a7f40001 brc 15,2141b8
          >00000000002141baa7f4ffbe brc 15,214136
           00000000002141be: 0707 bcr 0,%r7
           00000000002141c0ebaff0680024 stmg %r10,%r15,104(%r15)
           00000000002141c6c0d00017c2a9 larl %r13,50c718
           00000000002141cca7f13fc0 tmll %r15,16320
Call Trace:
([<000000000050e7d8>] C.272.16122+0x88/0x110)
 [<00000000002141ec>] __blk_run_queue+0x2c/0x154
 [<000000000028013a>] fc_remote_port_add+0x85e/0x95c
 [<000000000037596e>] zfcp_scsi_rport_work+0xe6/0x148
 [<000000000006908c>] worker_thread+0x25c/0x318
 [<000000000006f10c>] kthread+0x94/0x9c
 [<000000000001c2b2>] kernel_thread_starter+0x6/0xc
 [<000000000001c2ac>] kernel_thread_starter+0x0/0xc
INFO: lockdep is turned off.
Last Breaking-Event-Address:
 [<00000000002141b6>] blk_remove_plug+0xae/0xb8

The FC pass-through support triggers the WARN_ON(!irqs_disabled()) in
blk_plug_device. Since blk_plug_device requires being called with
disabled interrupts, use spin_lock_irqsave in fc_bsg_goose_queue to
disable the interrupts before calling into the block layer.

Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Acked-by: James Smart <james.smart@emulex.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
14 years agoperf report: Print sorted callchains per histogram entries
Frederic Weisbecker [Fri, 26 Jun 2009 14:28:01 +0000 (16:28 +0200)] 
perf report: Print sorted callchains per histogram entries

Use the newly created callchains radix tree to gather the chains stats
from the recorded events and then print the callchains for all of them,
sorted by hits, using the "-c" parameter with perf report.

Example:

 66.15%  [k] atm_clip_exit
            63.08%
                0xffffffffffffff80
                0xffffffff810196a8
                0xffffffff810c14c8
                0xffffffff8101a79c
                0xffffffff810194f3
                0xffffffff8106ab7f
                0xffffffff8106abe5
                0xffffffff8106acde
                0xffffffff8100d94b
                0xffffffff8153e7ea
                [...]

             1.54%
                0xffffffffffffff80
                0xffffffff810196a8
                0xffffffff810c14c8
                0xffffffff8101a79c
[...]

Symbols are not yet resolved.

Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <1246026481-8314-3-git-send-email-fweisbec@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
14 years agoperf_counter tools: Prepare a small callchain framework
Frederic Weisbecker [Fri, 26 Jun 2009 14:28:00 +0000 (16:28 +0200)] 
perf_counter tools: Prepare a small callchain framework

We plan to display the callchains depending on some user-configurable
parameters.

To gather the callchains stats from the recorded stream in a fast way,
this patch introduces an ad hoc radix tree adapted for callchains and also
a rbtree to sort these callchains once we have gathered every events
from the stream.

Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <1246026481-8314-2-git-send-email-fweisbec@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
14 years agoamd64_edac: misc small cleanups
Borislav Petkov [Wed, 10 Jun 2009 15:36:57 +0000 (17:36 +0200)] 
amd64_edac: misc small cleanups

- cleanup debug calls
- shorten function names
- cleanup error exit paths

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
14 years agoamd64_edac: fix ecc_enable_override handling
Borislav Petkov [Mon, 22 Jun 2009 17:42:24 +0000 (19:42 +0200)] 
amd64_edac: fix ecc_enable_override handling

amd64_check_ecc_enabled() returns non-zero status when ECC
checking/correcting is disabled and this fails further loading of the
driver even when 'ecc_enable_override' boot param is used.

Fix that by clearing return status in that case.

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
14 years agoamd64_edac: check only ECC bit in amd64_determine_edac_cap
Borislav Petkov [Wed, 10 Jun 2009 16:29:54 +0000 (18:29 +0200)] 
amd64_edac: check only ECC bit in amd64_determine_edac_cap

Checking whether the machine is using ECC enabled DRAM is done through
testing the DimmEccEn bit in the DRAM Cfg Low register (F2x[1,0]90). Do
that instead of testing all bits from the DimmEccEn upwards.

Also, remove mci->edac_cap assignment and use value returned from
amd64_determine_edac_cap().

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
14 years agointel-iommu: fix Identity Mapping to be arch independent
Chris Wright [Fri, 26 Jun 2009 01:52:05 +0000 (18:52 -0700)] 
intel-iommu: fix Identity Mapping to be arch independent

Drop the e820 scanning and use existing function for finding valid
RAM regions to add to 1:1 mapping.

Signed-off-by: Chris Wright <chrisw@redhat.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
14 years agoftrace: Fix the output of profile
Li Zefan [Fri, 26 Jun 2009 03:15:37 +0000 (11:15 +0800)] 
ftrace: Fix the output of profile

The first entry of the ftrace profile was always skipped when
reading trace_stat/functionX.

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
LKML-Reference: <4A443D59.4080307@cn.fujitsu.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
14 years agosh: ms7724se: Add sh_eth support
Kuninori Morimoto [Fri, 26 Jun 2009 07:05:39 +0000 (07:05 +0000)] 
sh: ms7724se: Add sh_eth support

Signed-off-by: Kuninori Morimoto <morimoto.kuninori@renesas.com>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
14 years agoMerge commit 'kumar/next' into merge
Benjamin Herrenschmidt [Fri, 26 Jun 2009 06:58:01 +0000 (16:58 +1000)] 
Merge commit 'kumar/next' into merge

14 years agoide: always kill the whole request on error
Bartlomiej Zolnierkiewicz [Tue, 23 Jun 2009 11:27:27 +0000 (11:27 +0000)] 
ide: always kill the whole request on error

* Use blk_rq_bytes() instead of obsolete ide_rq_bytes() in ide_kill_rq()
  and ide_floppy_do_request() for failed requests.
  [ bugfix part ]

* Use blk_rq_bytes() instead of obsolete ide_rq_bytes() in ide_do_devset()
  and ide_complete_drive_reset().  Then remove ide_rq_bytes().
  [ cleanup part ]

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoide: fix ide_kill_rq() for special ide-{floppy,tape} driver requests
Bartlomiej Zolnierkiewicz [Tue, 23 Jun 2009 11:26:06 +0000 (11:26 +0000)] 
ide: fix ide_kill_rq() for special ide-{floppy,tape} driver requests

Such requests should be failed with -EIO (like all other requests
in this function) instead of being completed successfully.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agopowerpc/rtas: Fix watchdog driver temperature read functionality
Adrian Reber [Thu, 26 Mar 2009 02:05:42 +0000 (02:05 +0000)] 
powerpc/rtas: Fix watchdog driver temperature read functionality

Using the RTAS watchdog driver to read out the temperature crashes
on a PXCAB:

Unable to handle kernel paging request for data at address 0xfe347b50
Faulting instruction address: 0xc00000000001af64
Oops: Kernel access of bad area, sig: 11 [#1]

The wrong usage of "(void *)__pa(&temperature)" in rtas_call() is
removed by using the function rtas_get_sensor() which does the
right thing.

Signed-off-by: Adrian Reber <adrian@lisas.de>
Acked-by: Utz Bacher <utz.bacher@de.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
14 years agopowerpc/mm: Fix potential access to freed pages when using hugetlbfs
Benjamin Herrenschmidt [Mon, 15 Jun 2009 16:53:43 +0000 (16:53 +0000)] 
powerpc/mm: Fix potential access to freed pages when using hugetlbfs

When using 64k page sizes, our PTE pages are split in two halves,
the second half containing the "extension" used to keep track of
individual 4k pages when not using HW 64k pages.

However, our page tables used for hugetlb have a slightly different
format and don't carry that "second half".

Our code that batched PTEs to be invalidated unconditionally reads
the "second half" (to put it into the batch), which means that when
called to invalidate hugetlb PTEs, it will access unrelated memory.

It breaks when CONFIG_DEBUG_PAGEALLOC is enabled.

This fixes it by only accessing the second half when the _PAGE_COMBO
bit is set in the first half, which indicates that we are dealing with
a "combo" page which represents 16x4k subpages. Anything else shouldn't
have this bit set and thus not require loading from the second half.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
14 years agopowerpc/440: Fix warning early debug code
Benjamin Herrenschmidt [Tue, 16 Jun 2009 15:55:19 +0000 (15:55 +0000)] 
powerpc/440: Fix warning early debug code

The function udbg_44x_as1_flush() has the wrong prototype causing
a warning when enabling 440 early debug.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
14 years agopowerpc/of: Fix usage of dev_set_name() in of_device_alloc()
Benjamin Herrenschmidt [Tue, 16 Jun 2009 15:55:18 +0000 (15:55 +0000)] 
powerpc/of: Fix usage of dev_set_name() in of_device_alloc()

dev_set_name() takes a format string, so use it properly and avoid
a warning with recent gcc's

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
14 years agopowerpc/pasemi: Use raw spinlock in SMP TB sync
Benjamin Herrenschmidt [Tue, 16 Jun 2009 16:42:51 +0000 (16:42 +0000)] 
powerpc/pasemi: Use raw spinlock in SMP TB sync

spin_lock() can hang if called while the timebase is frozen,
so use a raw lock instead, also disable interrupts while
at it.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
14 years agopowerpc: Use one common impl. of RTAS timebase sync and use raw spinlock
Benjamin Herrenschmidt [Tue, 16 Jun 2009 16:42:50 +0000 (16:42 +0000)] 
powerpc: Use one common impl. of RTAS timebase sync and use raw spinlock

Several platforms use their own copy of what is essentially the same code,
using RTAS to synchronize the timebases when bringing up new CPUs. This
moves it all into a single common implementation and additionally
turns the spinlock into a raw spinlock since the former can rely on
the timebase not being frozen when spinlock debugging is enabled, and finally
masks interrupts while the timebase is disabled.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
14 years agopowerpc/rtas: Turn rtas lock into a raw spinlock
Benjamin Herrenschmidt [Tue, 16 Jun 2009 16:42:49 +0000 (16:42 +0000)] 
powerpc/rtas: Turn rtas lock into a raw spinlock

RTAS currently uses a normal spinlock. However it can be called from
contexts where this is not necessarily a good idea. For example, it
can be called while syncing timebases, with the core timebase being
frozen. Unfortunately, that will deadlock in case of lock contention
when spinlock debugging is enabled as the spin lock debugging code
will try to use __delay() which ... relies on the timebase being
enabled.

Also RTAS can be used in some low level IRQ handling code path so it
may as well be a raw spinlock for -rt sake.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
14 years agopowerpc: Add irqtrace support for 32-bit powerpc
Benjamin Herrenschmidt [Wed, 17 Jun 2009 17:43:59 +0000 (17:43 +0000)] 
powerpc: Add irqtrace support for 32-bit powerpc

Based on initial work from: Dale Farnsworth <dale@farnsworth.org>

Add the low level irq tracing hooks for 32-bit powerpc needed
to enable full lockdep functionality.

The approach taken to deal with the code in entry_32.S is that
we don't trace all the transitions of MSR:EE when we just turn
it off to peek at TI_FLAGS without races. Only when we are
calling into C code or returning from exceptions with a state
that have changed from what lockdep thinks.

There's a little bugger though: If we take an exception that
keeps interrupts enabled (such as an alignment exception) while
interrupts are enabled, we will call trace_hardirqs_on() on the
way back spurriously. Not a big deal, but to get rid of it would
require remembering in pt_regs that the exception was one of the
type that kept interrupts enabled which we don't know at this
stage. (Well, we could test all cases for regs->trap but that
sucks too much).

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Tested-by: Kumar Gala <galak@kernel.crashing.org>
14 years agopowerpc/BSR: Fix BSR to allow mmap of small BSR on 64k kernel
Sonny Rao [Thu, 18 Jun 2009 15:13:04 +0000 (15:13 +0000)] 
powerpc/BSR: Fix BSR to allow mmap of small BSR on 64k kernel

On Mon, Nov 17, 2008 at 01:26:13AM -0600, Sonny Rao wrote:
> On Fri, Nov 07, 2008 at 04:28:29PM +1100, Paul Mackerras wrote:
> > Sonny Rao writes:
> >
> > > Fix the BSR driver to allow small BSR devices, which are limited to a
> > > single 4k space, on a 64k page kernel.  Previously the driver would
> > > reject the mmap since the size was smaller than PAGESIZE (or because
> > > the size was greater than the size of the device).  Now, we check for
> > > this case use remap_4k_pfn(). Also, take out code to set vm_flags,
> > > as the remap_pfn functions will do this for us.
> >
> > Thanks.
> >
> > Do we know that the BSR size will always be 4k if it's not a multiple
> > of 64k?  Is it possible that we could get 8k, 16k or 32k or BSRs?
> > If it is possible, what does the user need to be able to do?  Do they
> > just want to map 4k, or might then want to map the whole thing?
>
>
> Hi Paul, I took a look at changing the driver to reject a request for
> mapping more than a single 4k page, however the only indication we get
> of the requested size in the mmap function is the vma size, and this
> is always one page at minimum.  So, it's not possible to determine if
> the user wants one 4k page or more.  As I noted in my first response,
> there is only one case where this is even possible and I don't think
> it is a significant concern.
>
> I did notice that I left out the check to see if the user is trying to
> map more than the device length, so I fixed that.  Here's the revised
> patch.

Alright, I've reworked this now so that if we get one of these cases
where there's a bsr that's > 4k and < 64k on a 64k kernel we'll only
advertise that it is a 4k BSR to userspace.  I think this is the best
solution since user programs are only supposed to look at sysfs to
determine how much can be mapped, and libbsr does this as well.

Please consider for 2.6.31 as a fix, thanks.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
14 years agopowerpc/BSR: add 4096 byte BSR size
Sonny Rao [Thu, 18 Jun 2009 15:14:36 +0000 (15:14 +0000)] 
powerpc/BSR: add 4096 byte BSR size

Add a 4096 byte BSR size which will be used on new machines.  Also, remove
the warning when we run into an unknown size, as this can spam the kernel
log excessively.

Signed-off-by: Sonny Rao <sonnyrao@us.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
14 years agopowerpc: Map more memory early on 601 processors
Benjamin Herrenschmidt [Thu, 18 Jun 2009 19:17:39 +0000 (19:17 +0000)] 
powerpc: Map more memory early on 601 processors

The 32-bit kernel relies on some memory being mapped covering
the kernel text,data and bss at least, early during boot before
the full MMU setup is done. On 32-bit "classic" processors, this
is done using BAT registers.

On 601, the size of BATs is limited to 8M and we use 2 of them
for that initial mapping. This can become quite tight when enabling
features like lockdep, so let's use a 3rd one to bump that mapping
from 16M to 24M. We keep the 4th BAT free as it can be useful for
debugging early boot code to map things like serial ports.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
14 years agopowerpc/pmac: Fix DMA ops for MacIO devices
Benjamin Herrenschmidt [Thu, 18 Jun 2009 19:20:51 +0000 (19:20 +0000)] 
powerpc/pmac: Fix DMA ops for MacIO devices

The macio_dev's created to map devices inside the MacIO ASICs
don't have proper dma_ops. This causes crashes on some machines
since the SCSI code calls dma_map_* on our behalf using the
device we hang from.

This fixes it by copying the parent PCI device dma_ops into
the macio_dev when creating it.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
14 years agopowerpc/mm: Make k(un)map_atomic out of line
Benjamin Herrenschmidt [Thu, 18 Jun 2009 19:25:00 +0000 (19:25 +0000)] 
powerpc/mm: Make k(un)map_atomic out of line

Those functions are way too big to be inline, besides, kmap_atomic()
wants to call debug_kmap_atomic() which isn't exported for modules
and causes module link failures.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
14 years agopowerpc: Fix mpic alloc warning
Kumar Gala [Thu, 18 Jun 2009 22:01:20 +0000 (22:01 +0000)] 
powerpc: Fix mpic alloc warning

Since we can use kmalloc earlier we are getting the following since the
mpic_alloc() code calls alloc_bootmem().  Move to using kzalloc() to
remove the warning.

------------[ cut here ]------------
Badness at c0583248 [verbose debug info unavailable]
NIP: c0583248 LR: c0583210 CTR: 00000004
REGS: c0741de0 TRAP: 0700   Not tainted  (2.6.30-06736-g12a31df)
MSR: 00021000 <ME,CE>  CR: 22024024  XER: 00000000
TASK = c070d3b8[0] 'swapper' THREAD: c0740000 CPU: 0
<6>GPR00: 00000001 c0741e90 c070d3b8 00000001 00000210 00000020 3fffffff 00000000
<6>GPR08: 00000000 c0c85700 c04f8c40 0000002d 22044022 1004a388 7ffd9400 00000000
<6>GPR16: 00000000 7ffcd100 7ffcd100 7ffcd100 c04f8c40 00000000 c059f62c c075a0c0
<6>GPR24: c059f648 00000000 0000000f 00000210 00000020 00000000 3fffffff 00000210
NIP [c0583248] alloc_arch_preferred_bootmem+0x50/0x80
LR [c0583210] alloc_arch_preferred_bootmem+0x18/0x80
Call Trace:
[c0741e90] [c07343b0] devtree_lock+0x0/0x24 (unreliable)
[c0741ea0] [c0583b14] ___alloc_bootmem_nopanic+0x54/0x108
[c0741ee0] [c0583e18] ___alloc_bootmem+0x18/0x50
[c0741ef0] [c057b9cc] mpic_alloc+0x48/0x710
[c0741f40] [c057ecf4] mpc85xx_ds_pic_init+0x190/0x1b8
[c0741f90] [c057633c] init_IRQ+0x24/0x34
[c0741fa0] [c05738b8] start_kernel+0x260/0x3dc
[c0741ff0] [c00003c8] skpinv+0x2e0/0x31c
Instruction dump:
409e001c 7c030378 80010014 83e1000c 38210010 7c0803a6 4e800020 3d20c0c8
39295700 80090004 7c000034 5400d97e <0f0000002f800000 409e001c 38800000

BenH: Changed to use GFP_KERNEL, the allocator will do the right thing

Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
14 years agopowerpc: Fix output from show_regs
Kumar Gala [Thu, 18 Jun 2009 22:29:55 +0000 (22:29 +0000)] 
powerpc: Fix output from show_regs

For some reason we've had an explicit KERN_INFO for GPR dumps.  With
recent changes we get output like:

<6>GPR00: 00000000 ef855eb0 ef858000 00000001 000000d0 f1000000 ffbc8000 ffffffff

The KERN_INFO is causing the <6>.  Don't see any reason to keep it
around.

Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
14 years agopowerpc/pmac: Fix issues with PowerMac "PowerSurge" SMP
Benjamin Herrenschmidt [Thu, 18 Jun 2009 23:30:07 +0000 (23:30 +0000)] 
powerpc/pmac: Fix issues with PowerMac "PowerSurge" SMP

The old PowerSurge SMP (ie, dual or quad 604 machines) code has
numerous issues in modern world.

One is cpu_possible_map is set too late (the device-tree is bogus)
so we fail to allocate the interrupt stacks and crash. Another
problem is the fact the timebase is frozen by the bringup of the
second CPU so the delays in the generic code will hang, we need
to move some of the calling procedure to inside the powermac code.

This makes it boot again for me

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
14 years agopowerpc/amigaone: Limit ISA I/O range to 4k in the device tree
Gerhard Pircher [Fri, 19 Jun 2009 11:42:36 +0000 (11:42 +0000)] 
powerpc/amigaone: Limit ISA I/O range to 4k in the device tree

The kernel reserves the I/O address space from 0x0 to 0xfff for legacy
ISA devices. Change the ranges property for the PCI2ISA bridge to match
the kernels behavior, even if the ranges property isn't used for now.

Signed-off-by: Gerhard Pircher <gerhard_pircher@gmx.net>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
14 years agopowerpc/warp: Platform fix for i2c change
Sean MacLennan [Fri, 19 Jun 2009 19:43:59 +0000 (19:43 +0000)] 
powerpc/warp: Platform fix for i2c change

A change to the i2c subsystem breaks the warp platform code. The patch
is cleaner anyway, the old way was a bit crufty.

For those with keen eyes, the gratuitous change in the string from
PIKA to Warp is just so the logs look a bit nicer. The following two
lines tend to be printed one after another.

  Warp POST OK
  Warp DTM thread running.

Yeah, this will be the third patch to warp.c submitted in this
release....

Cheers,
   Sean

The i2c_client struct changed, breaking the code that looked for the ad7414
chip. Use the new of_find_i2c_device_by_node function added in 2.6.29.

Signed-off-by: Sean MacLennan <smaclennan@pikatech.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
14 years agopowerpc: Have git ignore generated files from dtc compile
Jon Smirl [Sun, 21 Jun 2009 15:28:00 +0000 (15:28 +0000)] 
powerpc: Have git ignore generated files from dtc compile

Have git ignore generated files from dtc compile

Signed-off-by: Jon Smirl <jonsmirl@gmail.com>
Acked-by: David Gibson <david@gibson.dropbear.id.au>
Acked-by: Sean MacLennan <smaclennan@pikatech.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
14 years agopowerpc/mpic: Fix mapping of "DCR" based MPIC variants
Benjamin Herrenschmidt [Mon, 22 Jun 2009 16:47:59 +0000 (16:47 +0000)] 
powerpc/mpic: Fix mapping of "DCR" based MPIC variants

Commit 31207dab7d2e63795eb15823947bd2f7025b08e2
"Fix incorrect allocation of interrupt rev-map"
introduced a regression crashing on boot on machines using
a "DCR" based MPIC, such as the Cell blades.

The reason is that the irq host data structure is initialized
much later as a result of that patch, causing our calls to
mpic_map() do be done before we have a host setup.

Unfortunately, this breaks _mpic_map_dcr() which uses the
mpic->irqhost to get to the device node.

This fixes it by, instead, passing the device node explicitely
to mpic_map().

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Acked-by: Akira Tsukamoto <akirat@rd.scei.sony.co.jp>
14 years agopowerpc: Swiotlb breaks pseries
Michael Ellerman [Mon, 22 Jun 2009 23:13:48 +0000 (23:13 +0000)] 
powerpc: Swiotlb breaks pseries

Turning on SWIOTLB selects or enables PPC_NEED_DMA_SYNC_OPS, which means
we get the non empty versions of dma_sync_* in asm/dma-mapping.h

On my pseries machine the dma_ops have no such routines and we die with
a null pointer - this patch gets it booting, is there a more elegant way
to do it?

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
14 years agoACPI: video: prevent NULL deref in acpi_get_pci_dev()
Troy Moure [Thu, 25 Jun 2009 23:05:35 +0000 (17:05 -0600)] 
ACPI: video: prevent NULL deref in acpi_get_pci_dev()

ref: http://thread.gmane.org/gmane.linux.kernel/857228/focus=857468

When the ACPI video driver initializes, it does a namespace walk
looking for for supported devices. When we find an appropriate
handle, we walk up the ACPI tree looking for a PCI root bus, and
then walk back down the PCI bus, assuming that every device
inbetween is a P2P bridge.

This assumption is not correct, and is reported broken on at
least:

Dell Latitude E6400
ThinkPad X61
Dell XPS M1330

Add a NULL deref check to prevent boot panics.

Reported-by: Alessandro Suardi <alessandro.suardi@gmail.com>
Signed-off-by: Troy Moure <twmoure@szypr.net>
Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
14 years agoeeepc-laptop: add rfkill support for the 3G modem in Eee PC 901 Go
Corentin Chary [Thu, 25 Jun 2009 11:25:42 +0000 (13:25 +0200)] 
eeepc-laptop: add rfkill support for the 3G modem in Eee PC 901 Go

Signed-off-by: Janne Grunau <j@jannau.net>
Signed-off-by: Corentin Chary <corentincj@iksaif.net>
Signed-off-by: Len Brown <len.brown@intel.com>
14 years agoeeepc-laptop: get the right value for CMSG
Corentin Chary [Thu, 25 Jun 2009 11:25:41 +0000 (13:25 +0200)] 
eeepc-laptop: get the right value for CMSG

CMSG is an ACPI method used to find features available on
an Eee PC. But some features are never repported, even if present.

If the getter of a feature is present, this patch will set
the corresponding bit in cmsg.

Signed-off-by: Corentin Chary <corentincj@iksaif.net>
Signed-off-by: Len Brown <len.brown@intel.com>
14 years agoeeepc-laptop: makes get_acpi() returns -ENODEV
Corentin Chary [Thu, 25 Jun 2009 11:25:40 +0000 (13:25 +0200)] 
eeepc-laptop: makes get_acpi() returns -ENODEV

If there is there is no getter defined, get_acpi()
will return -ENODEV.

Signed-off-by: Corentin Chary <corentincj@iksaif.net>
Signed-off-by: Len Brown <len.brown@intel.com>
14 years agoeeepc-laptop: right parent device
Corentin Chary [Thu, 25 Jun 2009 11:25:39 +0000 (13:25 +0200)] 
eeepc-laptop: right parent device

Signed-off-by: Corentin Chary <corentincj@iksaif.net>
Signed-off-by: Len Brown <len.brown@intel.com>
14 years agoeeepc-laptop: rfkill refactoring
Corentin Chary [Thu, 25 Jun 2009 11:25:38 +0000 (13:25 +0200)] 
eeepc-laptop: rfkill refactoring

Refactor rfkill code, because we'll add another
rfkill for wwan3g later.

Signed-off-by: Corentin Chary <corentincj@iksaif.net>
Signed-off-by: Len Brown <len.brown@intel.com>
14 years agoeeepc-laptop.c: use pr_fmt and pr_<level>
Joe Perches [Thu, 25 Jun 2009 11:25:37 +0000 (13:25 +0200)] 
eeepc-laptop.c: use pr_fmt and pr_<level>

Convert the unusual printk(EEEPC_<level> uses to
the more standard pr_fmt and pr_<level>(.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Corentin Chary <corentincj@iksaif.net>
Signed-off-by: Len Brown <len.brown@intel.com>
14 years agoeeepc-laptop: Register as a pci-hotplug device
Corentin Chary [Thu, 25 Jun 2009 11:25:36 +0000 (13:25 +0200)] 
eeepc-laptop: Register as a pci-hotplug device

The eee contains a logically (but not physically) hotpluggable PCIe slot.
Currently this is handled by adding or removing the PCI device in response
to rfkill events, but if a user has forced pciehp to bind to it (with the
force=1 argument) then both drivers will try to handle the event and
hilarity (in the form of oopses) will ensue. This can be avoided by having
eee-laptop register the slot as a hotplug slot. Only one of pciehp and
eee-laptop will successfully register this, avoiding the problem.

Signed-off-by: Matthew Garrett <mjg@redhat.com>
Signed-off-by: Corentin Chary <corentincj@iksaif.net>
Tested-by: Darren Salt <linux@youmustbejoking.demon.co.uk>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Len Brown <len.brown@intel.com>
14 years ago[CIFS] remove unknown mount option warning message
Steve French [Fri, 26 Jun 2009 04:07:18 +0000 (04:07 +0000)] 
[CIFS] remove unknown mount option warning message

Jeff's previous patch which removed the unneeded rw/ro
parsing can cause a minor warning in dmesg (about the
unknown rw or ro mount option) at mount time. This
patch makes cifs ignore them in kernel to remove the warning
(they are already handled in the mount helper and VFS).

Signed-off-by: Steve French <sfrench@us.ibm.com>
14 years ago[CIFS] remove bkl usage from umount begin
Steve French [Fri, 26 Jun 2009 03:25:49 +0000 (03:25 +0000)] 
[CIFS] remove bkl usage from umount begin

The lock_kernel call moved into the fs for umount_begin
is not needed.  This adds a check to make sure we don't
call umount_begin twice on the same fs.

umount_begin for cifs is probably not needed and
may eventually be able to be removed, but in
the meantime this smaller patch is safe and
gets rid of the bkl from this path which provides
some benefit.

Acked-by: Jeff Layton <redhat.com>
Signed-off-by: Steve French <sfrench@us.ibm.com>
14 years agomdio add missing GPL flag
Nicolas Reinecke [Thu, 25 Jun 2009 02:55:31 +0000 (02:55 +0000)] 
mdio add missing GPL flag

Add missing GPL flag and description.

mdio: module license 'unspecified' taints kernel.
Disabling lock debugging due to kernel taint

Signed-off-by: Nicolas Reinecke <nr <at> das-labor.org>
Acked-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosh_eth: remove redundant test on unsigned
roel kluin [Mon, 22 Jun 2009 07:38:00 +0000 (07:38 +0000)] 
sh_eth: remove redundant test on unsigned

Unsigned boguscnt cannot be less than 0.

Signed-off-by: Roel Kluin <roel.kluin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agofsl_pq_mdio: Fix fsl_pq_mdio to work with modules
Ionut Nicu [Wed, 24 Jun 2009 22:23:39 +0000 (22:23 +0000)] 
fsl_pq_mdio: Fix fsl_pq_mdio to work with modules

This patch fixes the case when ucc_geth or gianfar are compiled
as modules. Without this patch the call to phy_connect() fails.

Signed-off-by: Ionut Nicu <ionut.nicu@freescale.com>
Acked-by: Andy Fleming <afleming@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoipv6: avoid wraparound for expired preferred lifetime
Jens Rosenboom [Thu, 25 Jun 2009 04:55:50 +0000 (04:55 +0000)] 
ipv6: avoid wraparound for expired preferred lifetime

Avoid showing wrong high values when the preferred lifetime of an address
is expired.

Signed-off-by: Jens Rosenboom <me@jayr.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agotcp: missing check ACK flag of received segment in FIN-WAIT-2 state
Wei Yongjun [Wed, 24 Jun 2009 22:29:31 +0000 (22:29 +0000)] 
tcp: missing check ACK flag of received segment in FIN-WAIT-2 state

RFC0793 defined that in FIN-WAIT-2 state if the ACK bit is off drop
the segment and return[Page 72]. But this check is missing in function
tcp_timewait_state_process(). This cause the segment with FIN flag but
no ACK has two diffent action:

Case 1:
    Node A                      Node B
              <-------------    FIN,ACK
                                (enter FIN-WAIT-1)
    ACK       ------------->
                                (enter FIN-WAIT-2)
    FIN       ------------->    discard
                                (move sk to tw list)

Case 2:
    Node A                      Node B
              <-------------    FIN,ACK
                                (enter FIN-WAIT-1)
    ACK       ------------->
                                (enter FIN-WAIT-2)
                                (move sk to tw list)
    FIN       ------------->

              <-------------    ACK

This patch fixed the problem.

Signed-off-by: Wei Yongjun <yjwei@cn.fujitsu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosparc32: Fix makefile not generating required files
Julian Calaby [Tue, 23 Jun 2009 01:45:46 +0000 (01:45 +0000)] 
sparc32: Fix makefile not generating required files

The tftpboot build was failing with missing file errors.

It turns out that $(obj)/image wasn't being generated which was causing the a.out conversion to be skipped and hence piggyback to be called with nonexistent files.

Signed-off-by: Julian Calaby <julian.calaby@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosparc32: Fix tftpboot.img Makefile
Julian Calaby [Sun, 21 Jun 2009 16:45:01 +0000 (16:45 +0000)] 
sparc32: Fix tftpboot.img Makefile

Signed-off-by: Julian Calaby <julian.calaby@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosparc: fix tftpboot.img build
Sam Ravnborg [Sun, 21 Jun 2009 16:46:10 +0000 (16:46 +0000)] 
sparc: fix tftpboot.img build

Kjetil Oftedal mentioned that piggyback_32 was failing
when building a sparc image.

I tracked this down to the fact that the kernel no longer
provided an absolute symbol named "end".

Commit 86ed40bd6fe511d26bb8f3fa65a84cb65c235366 ("sparc: unify sections.h")
renamed end to _end but failed to update piggyback_32.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Cc: Kjetil Oftedal <oftedal@gmail.com>
Cc: Robert Reif <reif@earthlink.net>
Signed-off-by: Julian Calaby <julian.calaby@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosparc32: Fix obvious build issues for tftpboot.img build.
Robert Reif [Sun, 21 Jun 2009 16:45:44 +0000 (16:45 +0000)] 
sparc32: Fix obvious build issues for tftpboot.img build.

Signed-off-by: Robert Reif <reif@earthlink.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosparc64: Fix build warnings in piggyback_64.c
Julian Calaby [Sun, 21 Jun 2009 16:44:13 +0000 (16:44 +0000)] 
sparc64: Fix build warnings in piggyback_64.c

This patch fixes the following build warnings:
arch/sparc/boot/piggyback_64.c: In function 'main':
arch/sparc/boot/piggyback_64.c:44: warning: 'end' may be used uninitialized in this function
arch/sparc/boot/piggyback_64.c:44: warning: 'start' may be used uninitialized in this function

Signed-off-by: Julian Calaby <julian.calaby@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosparc64: Don't use alloc_bootmem() in init_IRQ() code paths.
David S. Miller [Fri, 26 Jun 2009 02:00:47 +0000 (19:00 -0700)] 
sparc64: Don't use alloc_bootmem() in init_IRQ() code paths.

The page allocator and SLAB are available at this point now,
and if we still try to use bootmem allocations here the kernel
spits out warnings.

Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoMerge branch 'drm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied...
Linus Torvalds [Fri, 26 Jun 2009 00:04:37 +0000 (17:04 -0700)] 
Merge branch 'drm-fixes' of git://git./linux/kernel/git/airlied/drm-2.6

* 'drm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6: (28 commits)
  drm: remove unused #include <linux/version.h>'s
  drm/radeon: fix driver initialization order so radeon kms can be builtin
  drm: Fix shifts which were miscalculated when converting from bitfields.
  drm/radeon: Clear surface registers at initialization time.
  drm/radeon: Don't initialize acceleration related fields of struct fb_info.
  drm/radeon: fix radeon kms framebuffer device
  drm/i915: initialize fence registers to zero when loading GEM
  drm/i915: Fix HDMI regression introduced in new chipset support
  drm/i915: fix LFP data fetch
  drm/i915: set TV detection mode when tv is already connected
  drm/i915: Catch up to obj_priv->page_list rename in disabled debug code.
  drm/i915: Fix size_t handling in off-by-default debug printfs
  drm/i915: Don't change the blank/sync width when calculating scaled modes
  drm/i915: Add support for changing LVDS panel fitting using an output property.
  drm/i915: correct suspend/resume ordering
  drm/i915: Add missing dependency on Intel AGP support.
  drm/i915: Generate 2MHz clock for display port aux channel I/O. Retry I/O.
  drm/i915: Clarify error returns from display port aux channel I/O
  drm/i915: Add CLKCFG register definition
  drm/i915: Split array of DAC limits into separate structures.
  ...