openwrt/staging/blogic.git
15 years agoperf_counter: x86: Fix PMU resource leak
Peter Zijlstra [Wed, 9 Sep 2009 08:04:47 +0000 (10:04 +0200)]
perf_counter: x86: Fix PMU resource leak

Dave noticed that we leak the PMU resource reservations when we
fail the hardware counter init.

Reported-by: David Miller <davem@davemloft.net>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Acked-by: David Miller <davem@davemloft.net>
LKML-Reference: <1252483487.7746.164.camel@twins>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoperf util: SVG performance improvements
Arjan van de Ven [Sun, 20 Sep 2009 16:14:38 +0000 (18:14 +0200)]
perf util: SVG performance improvements

Tweak the output SVG to increase performance in SVG viewers by
limiting the different types of font sizes and by smarter
transformations on the text.

At least with Inkscape this gives a notable performance improvement
during zoom and scrolling.

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
LKML-Reference: <20090920181438.3a49cb93@linux.intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoperf util: Make the timechart SVG width dynamic
Arjan van de Ven [Sun, 20 Sep 2009 16:14:16 +0000 (18:14 +0200)]
perf util: Make the timechart SVG width dynamic

This patch adds a command line option for timechart that allows the
user to specify the width of the SVG file.

This patch also makes sure that each second of recording has at
least 200 units (pixels at 96 DPI) of width.  This impacts
recordings longer than 5 seconds; recordings shorter than 5 second
will scale up to have a width of 1000 units for the whole recording
(as before).

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
LKML-Reference: <20090920181416.69570c5d@linux.intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoperf timechart: Show the duration of scheduler delays in the SVG
Arjan van de Ven [Sun, 20 Sep 2009 16:13:53 +0000 (18:13 +0200)]
perf timechart: Show the duration of scheduler delays in the SVG

Given that scheduler latencies are the hot thing nowadays, show the
duration of said latencies in the SVG in text form.

In addition, if the latency is more than 10 msec, pick a brighter
yellow color as a way to point these long delays out.

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
LKML-Reference: <20090920181353.796f4509@linux.intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoperf timechart: Show the name of the waker/wakee in timechart
Arjan van de Ven [Sun, 20 Sep 2009 16:13:28 +0000 (18:13 +0200)]
perf timechart: Show the name of the waker/wakee in timechart

Timechart currently shows thin green lines for sending or receiving
wakeups. This patch also prints (in a very small font) the name of
the process that is being woken/wakes up this process.

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
LKML-Reference: <20090920181328.68baa978@linux.intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoperf_counter: Fix perf_copy_attr() pointer arithmetic
Ian Schram [Fri, 18 Sep 2009 19:26:26 +0000 (21:26 +0200)]
perf_counter: Fix perf_copy_attr() pointer arithmetic

There is still some weird code in per_copy_attr(). Which supposedly
checks that all bytes trailing a struct are zero.

It doesn't seem to get pointer arithmetic right. Since it
increments an iterating pointer by sizeof(unsigned long) rather
than 1.

Signed-off-by: Ian Schram <ischram@telenet.be>
[ v2: clean up the messy PTR_ALIGN logic as well. ]
Signed-off-by: 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>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: <stable@kernel.org> # for v2.6.31.x
LKML-Reference: <4AB3DEE2.3030600@telenet.be>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoperf utils: Use a define for the maximum length of a trace event
Arjan van de Ven [Sat, 19 Sep 2009 11:36:30 +0000 (13:36 +0200)]
perf utils: Use a define for the maximum length of a trace event

As per Ingo's review: use a #define rather than an open coded constant
for the maximum length of a trace event for storing in the perf.data file.

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Cc: fweisbec@gmail.com
Cc: peterz@infradead.org
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <20090919133630.10533d3e@infradead.org>
[ add a few comments to nearby functions ]
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoperf: Add timechart help text and add timechart to "perf help"
Arjan van de Ven [Sat, 19 Sep 2009 11:36:04 +0000 (13:36 +0200)]
perf: Add timechart help text and add timechart to "perf help"

As suggested by Ingo, add a timechart man page help text, as well
as add it to the "perf help" overview.

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Cc: fweisbec@gmail.com
Cc: peterz@infradead.org
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <20090919133604.3767fa35@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agotracing, x86, cpuidle: Move the end point of a C state in the power tracer
Arjan van de Ven [Sat, 19 Sep 2009 11:35:33 +0000 (13:35 +0200)]
tracing, x86, cpuidle: Move the end point of a C state in the power tracer

The "end of a C state" trace point currently happens before
the code runs that corrects the TSC for having stopped during idle.

The result of this is that the timestamp of the end-of-C-state event
is garbage on cpus where the TSC stops during idle.

This patch moves the end point of the C state to after the timekeeping
engine of the kernel has been corrected.

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Cc: Len Brown <len.brown@intel.com>
Cc: fweisbec@gmail.com
Cc: peterz@infradead.org
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <20090919133533.139c2a46@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoperf utils: Be consistent about minimum text size in the svghelper
Arjan van de Ven [Sat, 19 Sep 2009 11:35:07 +0000 (13:35 +0200)]
perf utils: Be consistent about minimum text size in the svghelper

Be more consistent in the svghelper about the minimum text size
by having a global #define for this.

There needs to be a minimum text size in order to keep the size
of the SVG file within the reach of what current SVG viewers can
cope with.

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Cc: fweisbec@gmail.com
Cc: peterz@infradead.org
Cc: Paul Mackerras <paulus@samba.org>
Cc: Arjan van de Ven <arjan@infradead.org>
LKML-Reference: <20090919133507.7374ef8b@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoperf timechart: Add "perf timechart record"
Arjan van de Ven [Sat, 19 Sep 2009 11:34:42 +0000 (13:34 +0200)]
perf timechart: Add "perf timechart record"

Add a command line option to record a trace, similar to "perf sched record".

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Cc: fweisbec@gmail.com
Cc: peterz@infradead.org
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <20090919133442.0dc2c7f5@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoperf: Add the timechart tool
Arjan van de Ven [Sat, 12 Sep 2009 05:53:05 +0000 (07:53 +0200)]
perf: Add the timechart tool

timechart is a tool to visualize what is going on in the system.

The user makes a trace of what is going on with

 > perf record --timechart /usr/bin/some_command

and then can turn the output of this into an svg file

 > perf timechart

which then can be viewed with any SVG view; inkscape works well
enough for me.

The idea behind timechart is to create a "infinitely zoomable"
picture; something that has high level information on a 1:1 zoom
level, but which exposes more details every time you zoom into a
specific area.

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
LKML-Reference: <20090912130713.6a77bbc0@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoperf: Add a SVG helper library file
Arjan van de Ven [Sat, 12 Sep 2009 10:52:11 +0000 (12:52 +0200)]
perf: Add a SVG helper library file

The timechart tool writes out SVG format output; this patch adds a
set of helper functions to abstract dealing with SVG from the core
timechart code.

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
LKML-Reference: <20090912130613.677f0516@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agotracing, perf: Convert the power tracer into an event tracer
Arjan van de Ven [Thu, 17 Sep 2009 14:11:28 +0000 (16:11 +0200)]
tracing, perf: Convert the power tracer into an event tracer

This patch converts the existing power tracer into an event tracer,
so that power events (C states and frequency changes) can be
tracked via "perf".

This also removes the perl script that was used to demo the tracer;
its functionality is being replaced entirely with timechart.

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
LKML-Reference: <20090912130542.6d314860@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoperf: Add a sample_event type to the event_union
Arjan van de Ven [Sat, 12 Sep 2009 05:53:00 +0000 (07:53 +0200)]
perf: Add a sample_event type to the event_union

Add a sample_event type to the event_union so that raw samples can
be processed easily.

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
LKML-Reference: <20090912130511.411434b5@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoperf: Allow perf utilities to have "callback" options without arguments
Arjan van de Ven [Sat, 12 Sep 2009 05:52:54 +0000 (07:52 +0200)]
perf: Allow perf utilities to have "callback" options without arguments

timechart needs to add a "callback" type command line argument that
does not take arguments.

This patch adds the parse-options.h infrastructure to make this
possible.

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
LKML-Reference: <20090912130440.548666c1@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoperf: Store trace event name/id pairs in perf.data
Arjan van de Ven [Sat, 12 Sep 2009 05:52:51 +0000 (07:52 +0200)]
perf: Store trace event name/id pairs in perf.data

The trace event name<->id mapping is dynamic for each kernel
compile. In order for perf.data to be useable outside the actual
system, we thus need to store a table of this mapping for later
use.

This patch adds this table to perf.data, and provides helper
functions for lookup up fields from this table.

To avoid mistakes, lookup-from-table is kept completely seprate
from lookup-from-local-debugfs.

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
LKML-Reference: <20090912130405.6960d099@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoperf: Add a timestamp to fork events
Arjan van de Ven [Sat, 12 Sep 2009 05:52:47 +0000 (07:52 +0200)]
perf: Add a timestamp to fork events

perf timechart needs to know when a process forked, in order to be
able to visualize properly when tasks start.

This patch adds a time field to the event structure, and fills it
in appropriately.

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
LKML-Reference: <20090912130341.51ad2de2@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoMerge branch 'linus' into perfcounters/core
Ingo Molnar [Sat, 19 Sep 2009 09:27:32 +0000 (11:27 +0200)]
Merge branch 'linus' into perfcounters/core

Merge reason: Bring in tracing changes we depend on.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agosched_clock: Make it NMI safe
Peter Zijlstra [Fri, 18 Sep 2009 18:14:01 +0000 (20:14 +0200)]
sched_clock: Make it NMI safe

Arjan complained about the suckyness of TSC on modern machines, and
asked if we could do something about that for PERF_SAMPLE_TIME.

Make cpu_clock() NMI safe by removing the spinlock and using
cmpxchg. This also makes it smaller and more robust.

Affects architectures that use HAVE_UNSTABLE_SCHED_CLOCK, i.e. IA64
and x86.

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoperf_counter: Fix up swcounter throttling
Peter Zijlstra [Fri, 18 Sep 2009 10:18:14 +0000 (12:18 +0200)]
perf_counter: Fix up swcounter throttling

/me dons the brown paper bag.

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86, perf_counter, bts: Optimize BTS overflow handling
Markus Metzger [Tue, 15 Sep 2009 11:00:23 +0000 (13:00 +0200)]
x86, perf_counter, bts: Optimize BTS overflow handling

Draining the BTS buffer on a buffer overflow interrupt takes too
long resulting in a kernel lockup when tracing the kernel.

Restructure perf_counter sampling into sample creation and sample
output.

Prepare a single reference sample for BTS sampling and update the
from and to address fields when draining the BTS buffer. Drain the
entire BTS buffer between a single perf_output_begin() /
perf_output_end() pair.

Signed-off-by: Markus Metzger <markus.t.metzger@intel.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <20090915130023.A16204@sedona.ch.intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agopty_write: don't do a tty_wakeup() when the buffers are full
Linus Torvalds [Fri, 18 Sep 2009 14:05:58 +0000 (07:05 -0700)]
pty_write: don't do a tty_wakeup() when the buffers are full

Commit ac89a9174 ("pty: don't limit the writes to 'pty_space()' inside
'pty_write()'") removed the pty_space() checking, in order to let the
regular tty buffer code limit the buffering itself.

That was all good, but as a subtle side effect it meant that we'd be
doing a tty_wakeup() even in the case where the buffers were all filled
up, and didn't actually make any progress on the write.

Which sounds innocuous, but it interacts very badly with the ppp_async
code, which has an infinite loop in ppp_async_push() that tries to push
out data to the tty.  When we call tty_wakeup(), that loop ends up
thinking that progress was made (see the subtle interactions between
XMIT_WAKEUP and 'tty_stuffed' for details).  End result: one unhappy ppp
user.

Fixed by noticing when tty_insert_flip_string() didn't actually do
anything, and then not doing any more processing (including, very much
not calling tty_wakeup()).

Bisected-and-tested-by: Peter Volkov <pva@gentoo.org>
Cc: stable@kernel.org (2.6.31)
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agoperf sched: Add --input=file option to builtin-sched.c
Mike Galbraith [Fri, 18 Sep 2009 06:22:24 +0000 (08:22 +0200)]
perf sched: Add --input=file option to builtin-sched.c

perf sched record passes unparsed args on to perf record, so
specifying an output file via perf sched record -o FILE (cmd) just
works.  Ergo, provide an option to specify input file as well.

Also add the missing 'map' command to help.

Signed-off-by: Mike Galbraith <efault@gmx.de>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <1253254944.20589.11.camel@marge.simson.net>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoMerge branch 'x86-mce-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Fri, 18 Sep 2009 04:07:08 +0000 (21:07 -0700)]
Merge branch 'x86-mce-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'x86-mce-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip: (21 commits)
  x86, mce: Fix compilation with !CONFIG_DEBUG_FS in mce-severity.c
  x86, mce: CE in last bank prevents panic by unknown MCE
  x86, mce: Fake panic support for MCE testing
  x86, mce: Move debugfs mce dir creating to mce.c
  x86, mce: Support specifying raise mode for software MCE injection
  x86, mce: Support specifying context for software mce injection
  x86, mce: fix reporting of Thermal Monitoring mechanism enabled
  x86, mce: remove never executed code
  x86, mce: add missing __cpuinit tags
  x86, mce: fix "mce" boot option handling for CONFIG_X86_NEW_MCE
  x86, mce: don't log boot MCEs on Pentium M (model == 13) CPUs
  x86: mce: Lower maximum number of banks to architecture limit
  x86: mce: macros to compute banks MSRs
  x86: mce: Move per bank data in a single datastructure
  x86: mce: Move code in mce.c
  x86: mce: Rename CONFIG_X86_NEW_MCE to CONFIG_X86_MCE
  x86: mce: Remove old i386 machine check code
  x86: mce: Update X86_MCE description in x86/Kconfig
  x86: mce: Make CONFIG_X86_ANCIENT_MCE dependent on CONFIG_X86_MCE
  x86, mce: use atomic_inc_return() instead of add by 1
  ...

Manually fixed up trivial conflicts:
Documentation/feature-removal-schedule.txt
arch/x86/kernel/cpu/mcheck/mce.c

15 years agoMerge branch 'sched-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 18 Sep 2009 04:00:02 +0000 (21:00 -0700)]
Merge branch 'sched-core-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'sched-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip: (37 commits)
  sched: Fix SD_POWERSAVING_BALANCE|SD_PREFER_LOCAL vs SD_WAKE_AFFINE
  sched: Stop buddies from hogging the system
  sched: Add new wakeup preemption mode: WAKEUP_RUNNING
  sched: Fix TASK_WAKING & loadaverage breakage
  sched: Disable wakeup balancing
  sched: Rename flags to wake_flags
  sched: Clean up the load_idx selection in select_task_rq_fair
  sched: Optimize cgroup vs wakeup a bit
  sched: x86: Name old_perf in a unique way
  sched: Implement a gentler fair-sleepers feature
  sched: Add SD_PREFER_LOCAL
  sched: Add a few SYNC hint knobs to play with
  sched: Fix sync wakeups again
  sched: Add WF_FORK
  sched: Rename sync arguments
  sched: Rename select_task_rq() argument
  sched: Feature to disable APERF/MPERF cpu_power
  x86: sched: Provide arch implementations using aperf/mperf
  x86: Add generic aperf/mperf code
  x86: Move APERF/MPERF into a X86_FEATURE
  ...

Fix up trivial conflict in arch/x86/include/asm/processor.h due to
nearby addition of amd_get_nb_id() declaration from the EDAC merge.

15 years agoMerge branch 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Fri, 18 Sep 2009 03:58:11 +0000 (20:58 -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, pat: don't use rb-tree based lookup in reserve_memtype()
  x86: Increase MIN_GAP to include randomized stack

15 years agoMerge branch 'tracing-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 18 Sep 2009 03:56:37 +0000 (20:56 -0700)]
Merge branch 'tracing-core-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'tracing-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip: (44 commits)
  vsnprintf: remove duplicate comment of vsnprintf
  softirq: add BLOCK_IOPOLL to softirq_to_name
  oprofile: fix oprofile regression: select RING_BUFFER_ALLOW_SWAP
  tracing: switch function prints from %pf to %ps
  vsprintf: add %ps that is the same as %pS but is like %pf
  tracing: Fix minor bugs for __unregister_ftrace_function_probe
  tracing: remove notrace from __kprobes annotation
  tracing: optimize global_trace_clock cachelines
  MAINTAINERS: Update tracing tree details
  ftrace: document function and function graph implementation
  tracing: make testing syscall events a separate configuration
  tracing: remove some unused macros
  ftrace: add compile-time check on F_printk()
  tracing: fix F_printk() typos
  tracing: have TRACE_EVENT macro use __flags to not shadow parameter
  tracing: add static to generated TRACE_EVENT functions
  ring-buffer: typecast cmpxchg to fix PowerPC warning
  tracing: add filter event logic to special, mmiotrace and boot tracers
  tracing: remove trace_event_types.h
  tracing: use the new trace_entries.h to create format files
  ...

15 years agoMerge branch 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzi...
Linus Torvalds [Fri, 18 Sep 2009 03:54:49 +0000 (20:54 -0700)]
Merge branch 'upstream-linus' of git://git./linux/kernel/git/jgarzik/libata-dev

* 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzik/libata-dev:
  libata: Add pata_atp867x driver for Artop/Acard ATP867X controllers
  pata_amd: do not filter out valid modes in nv_mode_filter
  sata_promise: update reset code
  sata_promise: disable hotplug on 1st gen chips
  libata: fix spurious WARN_ON_ONCE() on port freeze
  ahci: restore pci_intx() handling

15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6
Linus Torvalds [Fri, 18 Sep 2009 03:54:12 +0000 (20:54 -0700)]
Merge git://git./linux/kernel/git/davem/sparc-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6:
  sparc: Update defconfigs.
  sparc: Kill PROM console driver.

15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Fri, 18 Sep 2009 03:53:52 +0000 (20:53 -0700)]
Merge git://git./linux/kernel/git/davem/net-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6: (66 commits)
  be2net: fix some cmds to use mccq instead of mbox
  atl1e: fix 2.6.31-git4 -- ATL1E 0000:03:00.0: DMA-API: device driver frees DMA
  pkt_sched: Fix qstats.qlen updating in dump_stats
  ipv6: Log the affected address when DAD failure occurs
  wl12xx: Fix print_mac() conversion.
  af_iucv: fix race when queueing skbs on the backlog queue
  af_iucv: do not call iucv_sock_kill() twice
  af_iucv: handle non-accepted sockets after resuming from suspend
  af_iucv: fix race in __iucv_sock_wait()
  iucv: use correct output register in iucv_query_maxconn()
  iucv: fix iucv_buffer_cpumask check when calling IUCV functions
  iucv: suspend/resume error msg for left over pathes
  wl12xx: switch to %pM to print the mac address
  b44: the poll handler b44_poll must not enable IRQ unconditionally
  ipv6: Ignore route option with ROUTER_PREF_INVALID
  bonding: make ab_arp select active slaves as other modes
  cfg80211: fix SME connect
  rc80211_minstrel: fix contention window calculation
  ssb/sdio: fix printk format warnings
  p54usb: add Zcomax XG-705A usbid
  ...

15 years agoMerge branch 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus
Linus Torvalds [Fri, 18 Sep 2009 03:52:32 +0000 (20:52 -0700)]
Merge branch 'upstream' of git://ftp.linux-mips.org/upstream-linus

* 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus: (51 commits)
  MIPS: BCM63xx: Add integrated ethernet mac support.
  MIPS: BCM63xx: Add support for the Broadcom BCM63xx family of SOCs.
  MIPS: BCM63xx: Add Broadcom 63xx CPU definitions.
  MIPS: Octeon:  Move some platform device registration to its own file.
  MIPS: Don't corrupt page tables on vmalloc fault.
  MIPS: Shrink the size of tlb handler
  MIPS: Alchemy: override loops_per_jiffy detection
  MIPS: hw_random: Add hardware RNG for Octeon SOCs.
  MIPS: Octeon:  Add hardware RNG platform device.
  MIPS: Remove useless zero initializations.
  MIPS: Alchemy: get rid of allow_au1k_wait
  MIPS: Octeon: Set kernel_uses_llsc to false on non-SMP builds.
  MIPS: Allow kernel use of LL/SC to be separate from the presence of LL/SC.
  MIPS: Get rid of CONFIG_CPU_HAS_LLSC
  MIPS: Malta: Remove pointless use use of CONFIG_CPU_HAS_LLSC
  MIPS: Rewrite clearing of ll_bit on context switch in C
  MIPS: Rewrite sysmips(MIPS_ATOMIC_SET, ...) in C with inline assembler
  MIPS: Consolidate all CONFIG_CPU_HAS_LLSC use in a single C file.
  MIPS: Clean up linker script using new linker script macros.
  MIPS: Use PAGE_SIZE in assembly instead of _PAGE_SIZE.
  ...

15 years agoperf trace: Sample timestamp and cpu when using record flag
Li Zefan [Thu, 17 Sep 2009 08:49:43 +0000 (16:49 +0800)]
perf trace: Sample timestamp and cpu when using record flag

Sample timestamp and cpu just like the -R option.

Before:
            init-0     [-01] 1266874889.17179869184709551615: irq_handler_entry: irq=18 handler=eth0
            init-0     [-01] 1266874889.17179869184709551615: irq_handler_entry: irq=18 handler=eth0
            init-0     [-01] 1266874889.17179869184709551615: irq_handler_entry: irq=1 handler=i8042
            init-0     [-01] 1266874889.17179869184709551615: irq_handler_entry: irq=18 handler=eth0
            init-0     [-01] 1266874889.17179869184709551615: irq_handler_entry: irq=1 handler=i8042

After:
            init-0     [001]  7364.568965353: irq_handler_entry: irq=18 handler=eth0
            init-0     [001]  7365.530226877: irq_handler_entry: irq=1 handler=i8042
            init-0     [001]  7365.542831563: irq_handler_entry: irq=18 handler=eth0
            init-0     [001]  7365.644156299: irq_handler_entry: irq=18 handler=eth0
            init-0     [001]  7365.694556201: irq_handler_entry: irq=18 handler=eth0

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Acked-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
LKML-Reference: <4AB1F827.8040905@cn.fujitsu.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoperf tools: Increase MAX_EVENT_LENGTH
Li Zefan [Thu, 17 Sep 2009 08:34:51 +0000 (16:34 +0800)]
perf tools: Increase MAX_EVENT_LENGTH

The name length of some trace events is longer than 30, like
sys_enter_sched_get_priority_max and
ext4_mb_discard_preallocations.

Passing those events to perf-record will fail, try:

  # ./perf record -f -e syscalls:sys_enter_sched_get_priority_max -F 1 -a

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Acked-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
LKML-Reference: <4AB1F4AB.7050205@cn.fujitsu.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoperf tools: Fix memory leak in read_ftrace_printk()
Li Zefan [Thu, 17 Sep 2009 08:34:23 +0000 (16:34 +0800)]
perf tools: Fix memory leak in read_ftrace_printk()

get_tracing_file() should be paired with put_tracing_file().

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Acked-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
LKML-Reference: <4AB1F48F.4070807@cn.fujitsu.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoMerge branch 'x86/pat' into x86/urgent
H. Peter Anvin [Thu, 17 Sep 2009 21:40:19 +0000 (14:40 -0700)]
Merge branch 'x86/pat' into x86/urgent

Merge reason:

Suresh Siddha (1):
      x86, pat: don't use rb-tree based lookup in reserve_memtype()

... requires previous x86/pat commits already pushed to Linus.

Signed-off-by: H. Peter Anvin <hpa@zytor.com>
15 years agox86, pat: don't use rb-tree based lookup in reserve_memtype()
Suresh Siddha [Wed, 16 Sep 2009 21:28:03 +0000 (14:28 -0700)]
x86, pat: don't use rb-tree based lookup in reserve_memtype()

Recent enhancement of rb-tree based lookup exposed a  bug with the lookup
mechanism in the reserve_memtype() which ensures that there are no conflicting
memtype requests for the memory range.

memtype_rb_search() returns an entry which has a start address <= new start
address. And from here we traverse the linear linked list to check if there
any conflicts with the existing mappings. As the rbtree is based on the
start address of the memory range, it is quite possible that we have several
overlapped mappings whose start address is much less than new requested start
but the end is >= new requested end. This results in conflicting memtype
mappings.

Same bug exists with the old code which uses cached_entry from where
we traverse the linear linked list. But the new rb-tree code exposes this
bug fairly easily.

For now, don't use the memtype_rb_search() and always start the search from
the head of linear linked list in reserve_memtype(). Linear linked list
for most of the systems grow's to few 10's of entries(as we track memory type
of RAM pages using struct page). So we should be ok for now.

We still retain the rbtree and use it to speed up free_memtype() which
doesn't have the same bug(as we know what exactly we are searching for
in free_memtype).

Also use list_for_each_entry_from() in free_memtype() so that we start
the search from rb-tree lookup result.

Reported-by: Markus Trippelsdorf <markus@trippelsdorf.de>
Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Cc: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
LKML-Reference: <1253136483.4119.12.camel@sbs-t61.sc.intel.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
15 years agolibata: Add pata_atp867x driver for Artop/Acard ATP867X controllers
John(Jung-Ik) Lee [Tue, 15 Sep 2009 04:32:33 +0000 (21:32 -0700)]
libata: Add pata_atp867x driver for Artop/Acard ATP867X controllers

This is a new pata driver for ARTOP 867X 64bit 4-channel UDMA133 ATA ctrls.
Based on the Atp867 data sheet rev 1.2, Acard, and in part on early ide codes
from Eric Uhrhane <ericu@google.com>.

Signed-off-by: John(Jung-Ik) Lee <jilee@google.com>
Reviewed-by: Grant Grundler <grundler@google.com>
Reviewed-by: Gwendal Gringo <gwendal@google.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agopata_amd: do not filter out valid modes in nv_mode_filter
Robert Hancock [Sun, 13 Sep 2009 05:54:47 +0000 (23:54 -0600)]
pata_amd: do not filter out valid modes in nv_mode_filter

On a Compaq Presario V3000 laptop (NVIDIA MCP51 chipset), pata_amd selects
PIO0 mode for the PATA DVD-RAM drive instead of MWDMA2 which it supports:

ata4.00: ATAPI: HL-DT-ST DVDRAM GSA-4084N, KQ09, max MWDMA2
ata4: nv_mode_filter: 0x39f&0x7001->0x1, BIOS=0x0 (0x0) ACPI=0x7001 (60:600:0x11)
ata4.00: configured for PIO0

For some reason, the BIOS-set UDMA configuration returns 0 and the ACPI _GTM
reports that UDMA2 and PIO0 are enabled. This causes nv_mode_filter to end up
allowing only PIO0 and UDMA0-2. Since the drive doesn't support UDMA we end up
using PIO0.

Since the controllers should always support PIO4, MWDMA2 and UDMA2 regardless
of what cable type is used, let's make sure we don't filter out these modes
regardless of what wacky settings the BIOS is using.

Signed-off-by: Robert Hancock <hancockrwd@gmail.com>
Cc: stable@kernel.org
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agosata_promise: update reset code
Mikael Pettersson [Tue, 15 Sep 2009 13:08:47 +0000 (15:08 +0200)]
sata_promise: update reset code

sata_promise's reset code has deviated quite a bit from
the Promise reference driver's, and it has been observed
to fail to recover from errors in some cases.

This patch thus updates the reset code to more closely
match the reference driver:

- soft reset (pdc_reset_port):
  * wait for ATA engine to not be in packet command mode
    (2nd gen only)
  * write reset bit in PDC_CTLSTAT before the first read
    in the loop
  * for 2nd gen SATA follow up with FPDMA reset and clearing
    error status registers
- hard reset (pdc_sata_hardreset):
  * wait for ATA engine to not be in packet command mode
    (2nd gen only)
  * reset ATA engine via the PCI control register
  * Tejun's change to use non-waiting hardreset + follow-up SRST

I'm not changing the hotplug mask bits since they are taken care
of by sata_promise's ->freeze() and ->thaw() operations. And I'm
not writing the PMP port # because that's always zero (for now).

Tested here on various controllers. In particular, one disk
which used to timeout and fail to recover from certain hdparm
and smartmonctl commands now works nicely.

Signed-off-by: Mikael Pettersson <mikpe@it.uu.se>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agosata_promise: disable hotplug on 1st gen chips
Mikael Pettersson [Tue, 15 Sep 2009 13:07:32 +0000 (15:07 +0200)]
sata_promise: disable hotplug on 1st gen chips

1st generation Promise SATA chips are prone to generating spurious
hotplug events which can disrupt normal operation. This has been
observed on 20376 and 20378 chips. This patch thus disables hotplug
support on 1st gen chips while leaving it enabled for 2nd gen chips.

The pdc_sata_hotplug_offset() function becomes redundant so it is
removed.

Tested on 1st gen 20376 and 20378 mainboard chips and on a 2nd gen
SATA300 PCI card.

Signed-off-by: Mikael Pettersson <mikpe@it.uu.se>
Tested-by: Kurt Roeckx <kurt@roeckx.be>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agolibata: fix spurious WARN_ON_ONCE() on port freeze
Tejun Heo [Thu, 17 Sep 2009 09:45:27 +0000 (18:45 +0900)]
libata: fix spurious WARN_ON_ONCE() on port freeze

Commit 54c38444fad6a99b4b19512f8f0055d69115e69e makes libata abort qcs
after the port is frozen.  This is necessary to guarantee that TF
registers are accessed after the DMA engine is shutdown after an
error.  However, this triggers WARN_ON_ONCE() check in
ata_qc_complete() spuriously.  Move WARN_ON_ONCE() downwards such that
failing commands while frozen doesn't trigger it.

Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: stable <stable@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agoahci: restore pci_intx() handling
Tejun Heo [Wed, 16 Sep 2009 15:34:39 +0000 (00:34 +0900)]
ahci: restore pci_intx() handling

Commit a5bfc4714b3f01365aef89a92673f2ceb1ccf246 dropped explicit
pci_intx() manipulation from ahci because it seemed unnecessary and
ahci doesn't seem to be the right place to be tweaking it if it were.
This was largely okay but there are exceptions.  There was one on an
embedded platform which was fixed via firmware and now bko#14124
reports it on a HP DL320.

  http://bugzilla.kernel.org/show_bug.cgi?id=14124

I still think this isn't something libata drivers should be caring
about (the only ones which are calling pci_intx() explicitly are
libata ones and one other driver) but for now reverting the change
seems to be the right thing to do.

Signed-off-by: Tejun Heo <tj@kernel.org>
Reported-by: Thomas Jarosch <thomas.jarosch@intra2net.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
15 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6
Linus Torvalds [Thu, 17 Sep 2009 20:21:52 +0000 (13:21 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/tiwai/sound-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6:
  ALSA: hda - Fix MSI GX620 mixer
  ASoC: remove unused #include <linux/version.h>
  ASoC: S3C lrsync function made to work with IRQs disabled.
  ALSA: hda - Fix Dell S14 pin setup
  ALSA: hda - Fix IDT92HD83* codec setup
  ASoC: Fix display of stream name in DAPM debugfs
  ALSA: hda - Add support for HP dv6
  ALSA: hda - Fix HP/line-out initialization with IDT/STAC codecs
  ALSA: hda - Set default GPIO for IDT92HD71bxx
  ALSA: hda - Set default GPIO for STAC/IDT codecs
  ASoC: Clean up error handling in MPC5200 DMA setup
  ALSA: hda - Add missing model=auto entry for ALC269

15 years agoperf sched: Determine the number of CPUs automatically
Ingo Molnar [Thu, 17 Sep 2009 16:24:55 +0000 (18:24 +0200)]
perf sched: Determine the number of CPUs automatically

For 'perf sched map' output, determine max_cpu automatically,
instead of the static default of 15.

[ v2: use sysconf() pointed out by Arjan van de Ven <arjan@infradead.org> ]

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>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoperf record: Disable profiling before draining the buffer
Peter Zijlstra [Thu, 17 Sep 2009 17:59:05 +0000 (19:59 +0200)]
perf record: Disable profiling before draining the buffer

I noticed that perf-record continues profiling itself after the
child terminated and we're draining the buffer.

This can cause a _lot_ of overhead with --all recording - we keep
and keep recording, which produces new and new events.

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoperf_counter: Allow for a wakeup watermark
Peter Zijlstra [Thu, 17 Sep 2009 17:01:10 +0000 (19:01 +0200)]
perf_counter: Allow for a wakeup watermark

Currently we wake the mmap() consumer once every PAGE_SIZE of data
and/or once event wakeup_events when specified.

For high speed sampling this results in too many wakeups wrt. the
buffer size, hence change this.

We move the default wakeup limit to 1/4-th the buffer size, and
provide for means to manually specify this limit.

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoperf_counter: Do not throttle single swcounter events
Peter Zijlstra [Thu, 17 Sep 2009 16:47:11 +0000 (18:47 +0200)]
perf_counter: Do not throttle single swcounter events

We can have swcounter events that contribute more than a single
count per event, when used with a non-zero period, those can
generate multiple events, which is when we need throttling.

However, swcounter that contribute only a single count per event
can only come as fast as we can run code, hence don't throttle
them.

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agovsnprintf: remove duplicate comment of vsnprintf
Steven Rostedt [Thu, 17 Sep 2009 13:27:29 +0000 (09:27 -0400)]
vsnprintf: remove duplicate comment of vsnprintf

Remove the duplicate comment of bstr_printf that is the same as the
vsnprintf.

Add the 's' option to the comment for the pointer function. This is
more of an internal function so the little duplication of the comment
here is OK.

Reported-by: Zhaolei <zhaolei@cn.fujitsu.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
15 years agosoftirq: add BLOCK_IOPOLL to softirq_to_name
Li Zefan [Thu, 17 Sep 2009 09:38:32 +0000 (17:38 +0800)]
softirq: add BLOCK_IOPOLL to softirq_to_name

With BLOCK_IOPOLL_SOFTIRQ added, softirq_to_name[] and
show_softirq_name() needs to be updated.

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
LKML-Reference: <4AB20398.8070209@cn.fujitsu.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
15 years agooprofile: fix oprofile regression: select RING_BUFFER_ALLOW_SWAP
Christian Borntraeger [Wed, 16 Sep 2009 19:56:49 +0000 (21:56 +0200)]
oprofile: fix oprofile regression: select RING_BUFFER_ALLOW_SWAP

commit 85bac32c4a52c592b857f2c360cc5ec93a097d70
    ring-buffer: only enable ring_buffer_swap_cpu when needed
broke oprofile (at least on s390, but likely on all platforms).

this patch lets oprofile select RING_BUFER_ALLOW_SWAP to make
ring_buffer_swap_cpu usable for oprofile.

Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
LKML-Reference: <200909162156.49239.borntraeger@de.ibm.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Robert Richter <robert.richter@amd.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
15 years agotracing: switch function prints from %pf to %ps
Steven Rostedt [Thu, 17 Sep 2009 04:05:58 +0000 (00:05 -0400)]
tracing: switch function prints from %pf to %ps

For direct function pointers (like what mcount provides) PowerPC64
requires the use of %ps, otherwise nothing is printed.

This patch converts all prints of functions retrieved through mcount
to use the %ps format from the %pf.

Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
15 years agovsprintf: add %ps that is the same as %pS but is like %pf
Steven Rostedt [Thu, 17 Sep 2009 00:03:06 +0000 (20:03 -0400)]
vsprintf: add %ps that is the same as %pS but is like %pf

On PowerPC64 function pointers do not point directly at the functions,
but instead point to pointers to the functions. The output of %pF expects
to point to a pointer to the function, whereas %pS will show the function
itself.

mcount returns the direct pointer to the function and not the pointer to
the pointer. Thus %pS must be used to show this. The function tracer
requires printing of the functions without offsets and uses the %pf
instead.

 %pF produces run_local_timers+0x4/0x1f
 %pf produces just run_local_timers

For PowerPC64, we need to use the direct pointer, and we only have
%pS which will produce .run_local_timers+0x4/0x1f

This patch creates a %ps that matches the %pf as %pS matches %pF.

Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Zhao Lei <zhaolei@cn.fujitsu.com>
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
15 years agoMerge branch 'fix/hda' into for-linus
Takashi Iwai [Thu, 17 Sep 2009 19:08:56 +0000 (21:08 +0200)]
Merge branch 'fix/hda' into for-linus

* fix/hda:
  ALSA: hda - Fix MSI GX620 mixer
  ALSA: hda - Fix Dell S14 pin setup
  ALSA: hda - Fix IDT92HD83* codec setup
  ALSA: hda - Add support for HP dv6
  ALSA: hda - Fix HP/line-out initialization with IDT/STAC codecs
  ALSA: hda - Set default GPIO for IDT92HD71bxx
  ALSA: hda - Set default GPIO for STAC/IDT codecs
  ALSA: hda - Add missing model=auto entry for ALC269

15 years agoMerge branch 'fix/asoc' into for-linus
Takashi Iwai [Thu, 17 Sep 2009 19:08:53 +0000 (21:08 +0200)]
Merge branch 'fix/asoc' into for-linus

* fix/asoc:
  ASoC: remove unused #include <linux/version.h>
  ASoC: S3C lrsync function made to work with IRQs disabled.
  ASoC: Fix display of stream name in DAPM debugfs
  ASoC: Clean up error handling in MPC5200 DMA setup

15 years agoMerge branch 'linus' into tracing/core
Ingo Molnar [Thu, 17 Sep 2009 18:52:23 +0000 (20:52 +0200)]
Merge branch 'linus' into tracing/core

Merge reason: Pick up kernel/softirq.c update for dependent fix.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoMIPS: BCM63xx: Add integrated ethernet mac support.
Maxime Bizon [Tue, 18 Aug 2009 12:23:40 +0000 (13:23 +0100)]
MIPS: BCM63xx: Add integrated ethernet mac support.

Signed-off-by: Maxime Bizon <mbizon@freebox.fr>
Signed-off-by: Florian Fainelli <florian@openwrt.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: BCM63xx: Add support for the Broadcom BCM63xx family of SOCs.
Maxime Bizon [Tue, 18 Aug 2009 12:23:37 +0000 (13:23 +0100)]
MIPS: BCM63xx: Add support for the Broadcom BCM63xx family of SOCs.

Signed-off-by: Maxime Bizon <mbizon@freebox.fr>
Signed-off-by: Florian Fainelli <florian@openwrt.org>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: BCM63xx: Add Broadcom 63xx CPU definitions.
Maxime Bizon [Tue, 18 Aug 2009 12:23:37 +0000 (13:23 +0100)]
MIPS: BCM63xx: Add Broadcom 63xx CPU definitions.

Todo: Nothing ever detects CPU_BCM6338 but the code tests for it anyway.

Signed-off-by: Maxime Bizon <mbizon@freebox.fr>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Octeon: Move some platform device registration to its own file.
David Daney [Wed, 16 Sep 2009 21:54:18 +0000 (14:54 -0700)]
MIPS: Octeon:  Move some platform device registration to its own file.

There is a bunch of platform device registration in
arch/mips/cavium-octeon/setup.c.  We move it to its own file in
preparation for adding more platform devices.

Signed-off-by: David Daney <ddaney@caviumnetworks.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Don't corrupt page tables on vmalloc fault.
David Daney [Wed, 2 Sep 2009 22:47:34 +0000 (15:47 -0700)]
MIPS: Don't corrupt page tables on vmalloc fault.

The code after the vmalloc_fault: label in do_page_fault() modifies
user page tables, this is not correct for 64-bit kernels.

For 64-bit kernels we should go straight to the no_context handler
skipping vmalloc_fault.

Signed-off-by: David Daney <ddaney@caviumnetworks.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Shrink the size of tlb handler
Wu Fei [Thu, 3 Sep 2009 14:29:53 +0000 (22:29 +0800)]
MIPS: Shrink the size of tlb handler

By combining swapper_pg_dir and module_pg_dir, several if conditions
can be eliminated from the tlb exception handler. The reason they
can be combined is that, the effective virtual address of vmalloc
returned is at the bottom, and of module_alloc returned is at the
top. It also fixes the bug in vmalloc(), which happens when its
return address is not covered by the first pgd.

Signed-off-by: Wu Fei <at.wufei@gmail.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Alchemy: override loops_per_jiffy detection
Manuel Lauss [Thu, 27 Aug 2009 17:21:18 +0000 (19:21 +0200)]
MIPS: Alchemy: override loops_per_jiffy detection

loops_per_jiffy depends on coreclk speed;  preset it instead of
letting the kernel waste precious microseconds trying to approximate it.

Signed-off-by: Manuel Lauss <manuel.lauss@gmail.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: hw_random: Add hardware RNG for Octeon SOCs.
David Daney [Thu, 20 Aug 2009 21:10:23 +0000 (14:10 -0700)]
MIPS: hw_random: Add hardware RNG for Octeon SOCs.

Signed-off-by: David Daney <ddaney@caviumnetworks.com>
Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Octeon: Add hardware RNG platform device.
David Daney [Thu, 20 Aug 2009 21:10:22 +0000 (14:10 -0700)]
MIPS: Octeon:  Add hardware RNG platform device.

Add a platform device for the Octeon Random Number Generator (RNG).

Signed-off-by: David Daney <ddaney@caviumnetworks.com>
Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Remove useless zero initializations.
Ralf Baechle [Thu, 17 Sep 2009 00:25:07 +0000 (02:25 +0200)]
MIPS: Remove useless zero initializations.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Alchemy: get rid of allow_au1k_wait
Manuel Lauss [Sat, 22 Aug 2009 16:09:27 +0000 (18:09 +0200)]
MIPS: Alchemy: get rid of allow_au1k_wait

Eliminate the 'allow_au1k_wait' variable.  MIPS kernel installs the
Alchemy-specific wait code before timer initialization;  if the C0
timer must be used for timekeeping the wait function is set to NULL
which means no wait implementation is available.

As a sideeffect, the 'wait instruction available' output in
/proc/cpuinfo now correctly indicates whether 'wait' is usable.

Run-tested on DB1200.

Signed-off-by: Manuel Lauss <manuel.lauss@gmail.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Octeon: Set kernel_uses_llsc to false on non-SMP builds.
David Daney [Mon, 13 Jul 2009 18:15:20 +0000 (11:15 -0700)]
MIPS: Octeon: Set kernel_uses_llsc to false on non-SMP builds.

Signed-off-by: David Daney <ddaney@caviumnetworks.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Allow kernel use of LL/SC to be separate from the presence of LL/SC.
David Daney [Mon, 13 Jul 2009 18:15:19 +0000 (11:15 -0700)]
MIPS: Allow kernel use of LL/SC to be separate from the presence of LL/SC.

On some CPUs, it is more efficient to disable and enable interrupts in the
kernel rather than use ll/sc for atomic operations.  But if we were to set
cpu_has_llsc to false, we would break the userspace futex interface (in
asm/futex.h).

We separate the two concepts, with a new predicate kernel_uses_llsc, that
lets us disable the kernel's use of ll/sc while still allowing the futex
code to use it.

Also there were a couple of cases in bitops.h where we were using ll/sc
unconditionally even if cpu_has_llsc were false.

Signed-off-by: David Daney <ddaney@caviumnetworks.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Get rid of CONFIG_CPU_HAS_LLSC
Ralf Baechle [Thu, 17 Sep 2009 00:25:06 +0000 (02:25 +0200)]
MIPS: Get rid of CONFIG_CPU_HAS_LLSC

CONFIG_CPU_HAS_LLSC duplicated the function of cpu_has_llsc for no good
reason and and the results if the one was enabled and the other disabled
was very unobvious.  Remove it now that there are no more remaining users.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Malta: Remove pointless use use of CONFIG_CPU_HAS_LLSC
Ralf Baechle [Thu, 17 Sep 2009 00:25:06 +0000 (02:25 +0200)]
MIPS: Malta: Remove pointless use use of CONFIG_CPU_HAS_LLSC

All CPUs for Malta support LL/SC.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Rewrite clearing of ll_bit on context switch in C
Ralf Baechle [Thu, 17 Sep 2009 00:25:05 +0000 (02:25 +0200)]
MIPS: Rewrite clearing of ll_bit on context switch in C

This also means there is now only one implementation not 3 left.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Rewrite sysmips(MIPS_ATOMIC_SET, ...) in C with inline assembler
Ralf Baechle [Thu, 17 Sep 2009 00:25:05 +0000 (02:25 +0200)]
MIPS: Rewrite sysmips(MIPS_ATOMIC_SET, ...) in C with inline assembler

This way it doesn't have to use CONFIG_CPU_HAS_LLSC anymore.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Consolidate all CONFIG_CPU_HAS_LLSC use in a single C file.
Ralf Baechle [Thu, 17 Sep 2009 00:25:05 +0000 (02:25 +0200)]
MIPS: Consolidate all CONFIG_CPU_HAS_LLSC use in a single C file.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Clean up linker script using new linker script macros.
Nelson Elhage [Fri, 31 Jul 2009 20:58:19 +0000 (16:58 -0400)]
MIPS: Clean up linker script using new linker script macros.

This patch results in fewer output sections and in some data being
reordered, but should have no functional impact.

Signed-off-by: Nelson Elhage <nelhage@ksplice.com>
Signed-off-by: Tim Abbott <tabbott@ksplice.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Use PAGE_SIZE in assembly instead of _PAGE_SIZE.
Nelson Elhage [Fri, 31 Jul 2009 20:58:18 +0000 (16:58 -0400)]
MIPS: Use PAGE_SIZE in assembly instead of _PAGE_SIZE.

Now that PAGE_SIZE is available to assembly directly, there is no need
to separately expose it as _PAGE_SIZE through asm-offsets.

In addition, remove _PAGE_SHIFT from asm-offsets, since it was never
needed, and is not used anywhere.

Signed-off-by: Nelson Elhage <nelhage@ksplice.com>
Signed-off-by: Tim Abbott <tabbott@ksplice.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Convert to asm-generic/hardirq.h
Christoph Hellwig [Tue, 4 Aug 2009 14:54:04 +0000 (16:54 +0200)]
MIPS: Convert to asm-generic/hardirq.h

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: AR7: Make board code register ar7_wdt as a platform device
Florian Fainelli [Wed, 15 Jul 2009 10:09:34 +0000 (12:09 +0200)]
MIPS: AR7: Make board code register ar7_wdt as a platform device

This patch makes the board code register the ar7_wdt driver as a platform
device. We move the dynamic resource calculation here since the driver
should not be aware of the AR7 SoC version it is running on.

Signed-off-by: Florian Fainelli <florian@openwrt.org>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Loongson: Update the default config file for fuloong2e
Wu Zhangjin [Thu, 2 Jul 2009 15:28:04 +0000 (23:28 +0800)]
MIPS: Loongson: Update the default config file for fuloong2e

Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Loongson: Add GCC 4.4 support for Loongson2E
Wu Zhangjin [Thu, 2 Jul 2009 15:27:41 +0000 (23:27 +0800)]
MIPS: Loongson: Add GCC 4.4 support for Loongson2E

Because only gcc >=4.4 have loongson-specific support, we need to choose
the suitable -march argument for gcc <= 4.3 and gcc >= 4.4, and here, we
use -march=loongson2e for loongson2e.

Thanks goes to Arnaud Patard <apatard@mandriva.com> for suggestion of
using cc-options(Documentation/kbuild/makefiles.txt). and thanks Zhang
Le for introducing the new CPU_LOONGSON2E kernel option.

NOTE: -mtune option is not need if -march and -mtune use the same value.

Signed-off-by: Zhang Le <r0bertz@gentoo.org>
Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Loongson: Add a machtype kernel command line argument
Wu Zhangjin [Thu, 2 Jul 2009 15:27:12 +0000 (23:27 +0800)]
MIPS: Loongson: Add a machtype kernel command line argument

The difference between some loongson-based machines is very small, so, if
there is no necessary to add new kernel config options to cope with this
difference, it will be better to share the same kernel image file between
them, benefit from this, the linux distribution developers only have a need
to compile the kernel one time.

This machtype kernel command line argument will be used later to share the
same kernel image file between two different machines(menglong & yeeloong)
made by lemote.

Thanks very much to Zhang Le for cleaning up the machtype implementation.

Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Loongson: Split common loongson source code out
Wu Zhangjin [Thu, 2 Jul 2009 15:26:45 +0000 (23:26 +0800)]
MIPS: Loongson: Split common loongson source code out

To share common loongson source code between all of the loongson-based
machines. there is a need to split it out of the fuloong-2e/ directory.
at the same time, other according tuning is needed. the machine-specific
parts are defined as macros in relative header file, pci.h, mem.h,
machine.h.

Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Loongson: Change naming methods
Wu Zhangjin [Thu, 2 Jul 2009 15:26:08 +0000 (23:26 +0800)]
MIPS: Loongson: Change naming methods

To make source code of loongson sharable to the machines(such as gdium)
made by the other companies, we rename arch/mips/lemote to
arch/mips/loongson, asm/mach-lemote to asm/mach-loongson, and rename lm2e
to the name of the machine: fuloong-2e. accordingly, FULONG are renamed to
FULOONG2E to make it distinguishable to the future FULOONG2F.  and also,
some other relative tuning is needed.

Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Loongson: Add oprofile support
Wu Zhangjin [Thu, 2 Jul 2009 15:25:46 +0000 (23:25 +0800)]
MIPS: Loongson: Add oprofile support

This kernel support is needed by the user-space tool:oprofile to profile
linux kernel or applications via loongson2 performance counters. you can
enable this driver via CONFIG_OPROFILE = y or m.

On Loongson2 there are two performance counters, each one can count 16
events respectively. when anyone of the performance counter overflows, an
interrupt will be generated and is routed to the IRQ MIPS_CPU_IRQ_BASE + 6.

Signed-off-by: Yanhua <yanh@lemote.com>
Signed-off-by: Wu Zhangjin <wuzj@lemote.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Loongson: RTC: Enable legacy RTC driver on fulong
Wu Zhangjin [Thu, 2 Jul 2009 15:24:01 +0000 (23:24 +0800)]
MIPS: Loongson: RTC: Enable legacy RTC driver on fulong

RTC_LIB is selected by MIPS by default, and therefore, the legacy RTC
driver is disabled. but fortunately, RTC_LIB not works on fulong, so,
enabling the legcy RTC driver is needed, otherwise, the tools like
hwclock will not work.

Signed-off-by: Wu Zhangjin <wuzj@lemote.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Loongson: PCI: Clean up pcimap setup
Wu Zhangjin [Thu, 2 Jul 2009 15:23:30 +0000 (23:23 +0800)]
MIPS: Loongson: PCI: Clean up pcimap setup

Fixup the wrong original comment of pcimap, and make the source code more
understandable. and also, some new extra consideration is added in.

Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Loongson: clean up the coding style
Wu Zhangjin [Thu, 2 Jul 2009 15:23:03 +0000 (23:23 +0800)]
MIPS: Loongson: clean up the coding style

With the help of script/checkpatch.pl, i have cleaned up the coding
style.

1. remove un-needed header files and tune some comments.
2. remove some un-needed { }

add a new header file loongson.h:

3. move some common header files to loongson.h
4. move some common extern declartions to loongson.h

and this new header file is needed for future loongson2f support.

Signed-off-by: Wu Zhangjin <wuzj@lemote.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Loongson: Split the implementation of prom and setup parts
Wu Zhangjin [Thu, 2 Jul 2009 15:22:36 +0000 (23:22 +0800)]
MIPS: Loongson: Split the implementation of prom and setup parts

This patch split the old initilization and setup implementation to
several file, one file one logic function.

the other main changes include:

1. as the script/checkpatch.pl suggests, use strict_strtol instead
of simple_strtol in arch/mips/lemote/lm2e/cmdline.c

2. use the existed macros in asm/mips-boards/bonito64.h as the
arguments of set_io_port_base() and remove the un-needed ones in
asm/mach-lemote/pci.h

Signed-off-by: Wu Zhangjin <wuzj@lemote.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Loongson: PCI: use existing mips_io_port_base
Wu Zhangjin [Thu, 2 Jul 2009 15:22:11 +0000 (23:22 +0800)]
MIPS: Loongson: PCI: use existing mips_io_port_base

mips_io_port_base is initialized via set_io_port_base() in
arch/mips/lemote/lm2e/setup.c, we can use it directly here.

Signed-off-by: Wu Zhangjin <wuzj@lemote.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Loongson: pm: clean up the reboot support
Wu Zhangjin [Thu, 2 Jul 2009 15:21:27 +0000 (23:21 +0800)]
MIPS: Loongson: pm: clean up the reboot support

Several magic numbers have been replaced by relative macros, which will be
more readable and understandable.

Signed-off-by: Wu Zhangjin <wuzj@lemote.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Loongson: pm: Remove redundant source code
Wu Zhangjin [Thu, 2 Jul 2009 15:20:56 +0000 (23:20 +0800)]
MIPS: Loongson: pm: Remove redundant source code

The implmentation of loongson2e_power_off and loongson2e_halt is almostly
the same, just preserve one of them.

Signed-off-by: Wu Zhangjin <wuzj@lemote.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Loongson: Add new early_printk implmentation
Wu Zhangjin [Thu, 2 Jul 2009 15:20:20 +0000 (23:20 +0800)]
MIPS: Loongson: Add new early_printk implmentation

This patch is based on the implementation in the lm2e-fixes branch of
Philippe's git://git.linux-cisco.org/linux-mips.git and the
malta-specific early_printk implementation.

Signed-off-by: Wu Zhangjin <wuzj@lemote.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Loongson: Remove out-of-date board-specific kgdb source code
Wu Zhangjin [Thu, 2 Jul 2009 15:19:33 +0000 (23:19 +0800)]
MIPS: Loongson: Remove out-of-date board-specific kgdb source code

Since the re-implementation of kgdb by commit
8854700115ecf8aa6f087aa915b7b6cf18090d39 the platform-specific version
has become superfluous, remove it.

Signed-off-by: Wu Zhangjin <wuzj@lemote.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Loongson: Remove existing early_printk implementation
Wu Zhangjin [Thu, 2 Jul 2009 15:18:09 +0000 (23:18 +0800)]
MIPS: Loongson: Remove existing early_printk implementation

Removes the existing implementation of early_printk for fulong.  The old
implementation was based on the outdated board-specific dbg_io.c.

Since commit 8d60a903d986ffa26c41f0092320a3b9da20bfaf dbg_io.c is not longer
needed; it will be removed by the next patch, and a new implementation of
early_printk will be added later.

Signed-off-by: Wu Zhangjin <wuzj@lemote.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: IP22, IP28: Build with -Werror
Ralf Baechle [Tue, 15 Sep 2009 21:04:59 +0000 (23:04 +0200)]
MIPS: IP22, IP28: Build with -Werror

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: FW: Build with -Werror
Ralf Baechle [Tue, 15 Sep 2009 20:58:23 +0000 (22:58 +0200)]
MIPS: FW: Build with -Werror

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Fulong: Convert reset initialization to initcall.
Ralf Baechle [Tue, 15 Sep 2009 13:00:02 +0000 (15:00 +0200)]
MIPS: Fulong: Convert reset initialization to initcall.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Malta: Convert reset initialization to initcall.
Ralf Baechle [Tue, 15 Sep 2009 12:50:39 +0000 (14:50 +0200)]
MIPS: Malta: Convert reset initialization to initcall.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: Get rid of duplicate cpu_idle() prototype.
Ralf Baechle [Tue, 15 Sep 2009 11:36:13 +0000 (13:36 +0200)]
MIPS: Get rid of duplicate cpu_idle() prototype.

Since 2.6.11-rc1 there is a prototype in <linux/smp.h>.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
15 years agoMIPS: TXx9: Fix error handling.
Julia Lawall [Sun, 13 Sep 2009 19:15:18 +0000 (21:15 +0200)]
MIPS: TXx9: Fix error handling.

Error handling code following a kzalloc should free the allocated data.
Error handling code following an ioremap should iounmap the allocated data.

The semantic match that finds the first problem is as follows:
(http://www.emn.fr/x-info/coccinelle/)

// <smpl>
@r exists@
local idexpression x;
statement S;
expression E;
identifier f,f1,l;
position p1,p2;
expression *ptr != NULL;
@@

x@p1 = \(kmalloc\|kzalloc\|kcalloc\)(...);
...
if (x == NULL) S
<... when != x
     when != if (...) { <+...x...+> }
(
x->f1 = E
|
 (x->f1 == NULL || ...)
|
 f(...,x->f1,...)
)
...>
(
 return \(0\|<+...x...+>\|ptr\);
|
 return@p2 ...;
)

@script:python@
p1 << r.p1;
p2 << r.p2;
@@

print "* file: %s kmalloc %s return %s" % (p1[0].file,p1[0].line,p2[0].line)
// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>