Distros generally (I looked at Debian, RHEL5 and SLES11) seem to
enable CONFIG_HIGHPTE for any x86 configuration which has highmem
enabled. This means that the overhead applies even to machines which
have a fairly modest amount of high memory and which therefore do not
really benefit from allocating PTEs in high memory but still pay the
price of the additional mapping operations.
Running kernbench on a 4G box I found that with CONFIG_HIGHPTE=y but
no actual highptes being allocated there was a reduction in system
time used from 59.737s to 55.9s.
With CONFIG_HIGHPTE=y and highmem PTEs being allocated:
Average Optimal load -j 4 Run (std deviation):
Elapsed Time 175.396 (0.238914)
User Time 515.983 (5.85019)
System Time 59.737 (1.26727)
Percent CPU 263.8 (71.6796)
Context Switches 39989.7 (4672.64)
Sleeps 42617.7 (246.307)
With CONFIG_HIGHPTE=y but with no highmem PTEs being allocated:
Average Optimal load -j 4 Run (std deviation):
Elapsed Time 174.278 (0.831968)
User Time 515.659 (6.07012)
System Time 55.9 (1.07799)
Percent CPU 263.8 (71.266)
Context Switches 39929.6 (4485.13)
Sleeps 42583.7 (373.039)
This patch allows the user to control the allocation of PTEs in
highmem from the command line ("userpte=nohigh") but retains the
status-quo as the default.
It is possible that some simple heuristic could be developed which
allows auto-tuning of this option however I don't have a sufficiently
large machine available to me to perform any particularly meaningful
experiments. We could probably handwave up an argument for a threshold
at 16G of total RAM.
Assuming 768M of lowmem we have 196608 potential lowmem PTE
pages. Each page can map 2M of RAM in a PAE-enabled configuration,
meaning a maximum of 384G of RAM could potentially be mapped using
lowmem PTEs.
Even allowing generous factor of 10 to account for other required
lowmem allocations, generous slop to account for page sharing (which
reduces the total amount of RAM mappable by a given number of PT
pages) and other innacuracies in the estimations it would seem that
even a 32G machine would not have a particularly pressing need for
highmem PTEs. I think 32G could be considered to be at the upper bound
of what might be sensible on a 32 bit machine (although I think in
practice 64G is still supported).
It's seems questionable if HIGHPTE is even a win for any amount of RAM
you would sensibly run a 32 bit kernel on rather than going 64 bit.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
LKML-Reference: <1266403090-20162-1-git-send-email-ian.campbell@citrix.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
This will save 64K bytes from memory when loading linux if DMI is
disabled, which is good for embedded systems.
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@holoscopio.com>
LKML-Reference: <1265758732-19320-1-git-send-email-cascardo@holoscopio.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Update rcu_dereference() primitives to use new lockdep-based
checking. The rcu_dereference() in __in6_dev_get() may be
protected either by rcu_read_lock() or RTNL, per Eric Dumazet.
The rcu_dereference() in __sk_free() is protected by the fact
that it is never reached if an update could change it. Check
for this by using rcu_dereference_check() to verify that the
struct sock's ->sk_wmem_alloc counter is zero.
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Acked-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: laijs@cn.fujitsu.com
Cc: dipankar@in.ibm.com
Cc: mathieu.desnoyers@polymtl.ca
Cc: josh@joshtriplett.org
Cc: dvhltc@us.ibm.com
Cc: niv@us.ibm.com
Cc: peterz@infradead.org
Cc: rostedt@goodmis.org
Cc: Valdis.Kletnieks@vt.edu
Cc: dhowells@redhat.com
LKML-Reference: <1266887105-1528-5-git-send-email-paulmck@linux.vnet.ibm.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The theory is that use of bare rcu_dereference() is more prone
to error than use of the RCU list-traversal primitives.
Therefore, disable lockdep RCU read-side critical-section
checking in these primitives for the time being. Once all of
the rcu_dereference() uses have been dealt with, it may be time
to re-enable lockdep checking for the RCU list-traversal
primitives.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: laijs@cn.fujitsu.com
Cc: dipankar@in.ibm.com
Cc: mathieu.desnoyers@polymtl.ca
Cc: josh@joshtriplett.org
Cc: dvhltc@us.ibm.com
Cc: niv@us.ibm.com
Cc: peterz@infradead.org
Cc: rostedt@goodmis.org
Cc: Valdis.Kletnieks@vt.edu
Cc: dhowells@redhat.com
LKML-Reference: <1266887105-1528-4-git-send-email-paulmck@linux.vnet.ibm.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Inspection is proving insufficient to catch all RCU misuses,
which is understandable given that rcu_dereference() might be
protected by any of four different flavors of RCU (RCU, RCU-bh,
RCU-sched, and SRCU), and might also/instead be protected by any
of a number of locking primitives. It is therefore time to
enlist the aid of lockdep.
This set of patches is inspired by earlier work by Peter
Zijlstra and Thomas Gleixner, and takes the following approach:
o Set up separate lockdep classes for RCU, RCU-bh, and RCU-sched.
o Set up separate lockdep classes for each instance of SRCU.
o Create primitives that check for being in an RCU read-side
critical section. These return exact answers if lockdep is
fully enabled, but if unsure, report being in an RCU read-side
critical section. (We want to avoid false positives!)
The primitives are:
For RCU: rcu_read_lock_held(void)
For RCU-bh: rcu_read_lock_bh_held(void)
For RCU-sched: rcu_read_lock_sched_held(void)
For SRCU: srcu_read_lock_held(struct srcu_struct *sp)
o Add rcu_dereference_check(), which takes a second argument
in which one places a boolean expression based on the above
primitives and/or lockdep_is_held().
o A new kernel configuration parameter, CONFIG_PROVE_RCU, enables
rcu_dereference_check(). This depends on CONFIG_PROVE_LOCKING,
and should be quite helpful during the transition period while
CONFIG_PROVE_RCU-unaware patches are in flight.
The existing rcu_dereference() primitive does no checking, but
upcoming patches will change that.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: laijs@cn.fujitsu.com
Cc: dipankar@in.ibm.com
Cc: mathieu.desnoyers@polymtl.ca
Cc: josh@joshtriplett.org
Cc: dvhltc@us.ibm.com
Cc: niv@us.ibm.com
Cc: peterz@infradead.org
Cc: rostedt@goodmis.org
Cc: Valdis.Kletnieks@vt.edu
Cc: dhowells@redhat.com
LKML-Reference: <1266887105-1528-1-git-send-email-paulmck@linux.vnet.ibm.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Specify proper quirk models for FSC and Quanta machines with ALC269 codec.
Signed-off-by: Kailang Yang <kailang@realtek.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
- Fixed alc_subsystem_id( ) typo and add new function.
- !(ass & 0x100000)) ==> Delete this check. It is unnecessary check.
- Add porti
- ALC670 support
Signed-off-by: Kailang Yang <kailang@realtek.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
All of the SH clocksource drivers follow the scheme that the IRQ is setup
prior to registering the clockevent. The interrupt handler in the
clockevent cases looks to the event handler function pointer being filled
in by the registration code, permitting us to get in to situations where
asserted IRQs step in to the handler before registration has had a chance
to complete and hitting a NULL pointer deref.
In practice this is not an issue for most platforms, but some of them
with fairly special loaders (or that are chain-loading from another
kernel) may enter in to this situation. This fixes up the oops reported
by Rafael on hp6xx.
Reported-and-tested-by: Rafael Ignacio Zurita <rafaelignacio.zurita@gmail.com>
Cc: stable@kernel.org
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
KEYSC::SCN register of SH7724 is 3bit.
Thus, scan_timing should be 0 - 7 here.
Signed-off-by: Kuninori Morimoto <morimoto.kuninori@renesas.com>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
This fixes corrupted CIPSO packets when SELinux categories greater than 127
are used. The bug occured on the second (and later) loops through the
while; the inner for loop through the ebitmap->maps array used the same
index as the NetLabel catmap->bitmap array, even though the NetLabel bitmap
is twice as long as the SELinux bitmap.
Signed-off-by: Joshua Roys <joshua.roys@gtri.gatech.edu>
Acked-by: Paul Moore <paul.moore@hp.com>
Signed-off-by: James Morris <jmorris@namei.org>
convert it to a real mutex.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <sfrench@us.ibm.com>
Mike Cui reported that his system with an NVIDIA MCP79 (aka MCP7A)
chipset stopped working with 2.6.32. The problem appears to be that
2.6.32 now enables the FPDMA auto-activate optimization in the ahci
driver. The drive works fine with this enabled on an Intel AHCI so
this appears to be a chipset bug. Since MCP79 is a fairly recent
NVIDIA chipset and we don't have any info on whether any other NVIDIA
chipsets have this issue, disable FPDMA AA optimization on all NVIDIA
AHCI controllers for now.
Should address http://bugzilla.kernel.org/show_bug.cgi?id=14922
Signed-off-by: Robert Hancock <hancockrwd@gmail.com>
While-we-investigate-issue-this-patch-looks-good-to-me-by:
Prajakta Gudadhe <pgudadhe@nvidia.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Cc: stable@kernel.org
syscall_name() helper, which resolves a syscall arch number to
its name, is not yet available as we first need to implement
event injection for it to work.
Remove it from the documentation or tag its references as
unavailable yet. Once it's implemented, we can just revert
the current patch.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Keiichi KII <k-keiichi@bx.jp.nec.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Also small update to perf-trace-perl and perf-trace docs.
Signed-off-by: Tom Zanussi <tzanussi@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Keiichi KII <k-keiichi@bx.jp.nec.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
LKML-Reference: <1264580883-15324-13-git-send-email-tzanussi@gmail.com>
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
If we know the size of a tuple in advance, there's no need to resize
it - start out with the known size in the first place.
Signed-off-by: Tom Zanussi <tzanussi@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Keiichi KII <k-keiichi@bx.jp.nec.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
LKML-Reference: <1266822779.6426.4.camel@tropicana>
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Adds a set of scripts that aggregate system call totals and system
call errors. Most are Python scripts that also test basic
functionality of the new Python engine, but there's also one Perl
script added for comparison and for reference in some new
Documentation contained in a later patch.
Signed-off-by: Tom Zanussi <tzanussi@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Keiichi KII <k-keiichi@bx.jp.nec.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
LKML-Reference: <1264580883-15324-8-git-send-email-tzanussi@gmail.com>
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Add base support for Python scripting to perf trace.
Signed-off-by: Tom Zanussi <tzanussi@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Keiichi KII <k-keiichi@bx.jp.nec.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
LKML-Reference: <1264580883-15324-6-git-send-email-tzanussi@gmail.com>
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
The anon_inodes interface has been split to allow creating a bare
(non-installed) file pointer and also extended to allow specifying
O_RDONLY in the flags. This makes it a suitable replacement for the
private "infinibandeventfs" pseudo-filesystem used by uverbs, and this
replacement saves a small chunk of boilerplate code.
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Now the omap4 clock framework is in mainline and clk_get_rate()
is functional. Hence reomve the hardcoded clock hacks.
This patch also fixes
Division by zero in kernel.
Backtrace:
[<c0025fb8>] (dump_backtrace+0x0/0x110) from [<c017febc>] (dump_stack+0x18/0x1c)
r7:60000093 r6:c0641050 r5:c0223e78 r4:c02126b4
[<c017fea4>] (dump_stack+0x0/0x1c) from [<c00260fc>] (__div0+0x18/0x20)
[<c00260e4>] (__div0+0x0/0x20) from [<c01431fc>] (Ldiv0+0x8/0x10)
[<c00318d4>] (omap_dm_timer_stop+0x0/0xb0) from [<c002c148>] (omap2_gp_timer_set_mode+0x1c/0x68)
r5:c0223e78 r4:00000000
[<c002c12c>] (omap2_gp_timer_set_mode+0x0/0x68) from [<c0063270>] (clockevents_set_mode+0x30/0x64)
r5:c020cae0 r4:00000000
[<c0063240>] (clockevents_set_mode+0x0/0x64) from [<c00632fc>] (clockevents_exchange_device+0x30/0x9c)
r5:c020cae0 r4:c02146e0
[<c00632cc>] (clockevents_exchange_device+0x0/0x9c) from [<c00636e0>] (tick_notify+0x17c/0x404)
r7:00000000 r6:c0641050 r5:00000000 r4:c020cae0
[<c0063564>] (tick_notify+0x0/0x404) from [<c005d5fc>] (notifier_call_chain+0x34/0x78)
[<c005d5c8>] (notifier_call_chain+0x0/0x78) from [<c005d684>] (__raw_notifier_call_chain+0x1c/0x24)
[<c005d668>] (__raw_notifier_call_chain+0x0/0x24) from [<c005d6ac>] (raw_notifier_call_chain+0x20/0x28)
[<c005d68c>] (raw_notifier_call_chain+0x0/0x28) from [<c0062e78>] (clockevents_do_notify+0x1c/0x24)
[<c0062e5c>] (clockevents_do_notify+0x0/0x24) from [<c0062f18>] (clockevents_register_device+0x98/0xd0)
[<c0062e80>] (clockevents_register_device+0x0/0xd0) from [<c001a194>] (percpu_timer_setup+0x80/0x9c)
r7:00000000 r6:00000002 r5:00000002 r4:00000003
[<c001a114>] (percpu_timer_setup+0x0/0x9c) from [<c000e9f0>] (smp_prepare_cpus+0xb0/0xe8)
[<c000e940>] (smp_prepare_cpus+0x0/0xe8) from [<c00084e8>] (kernel_init+0x5c/0x1fc)
r7:00000000 r6:00000000 r5:00000000 r4:c001b8a4
[<c000848c>] (kernel_init+0x0/0x1fc) from [<c0046c50>] (do_exit+0x0/0x604)
r7:00000000 r6:00000000 r5:00000000 r4:00000000
Signed-off-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
On OMAP4 platform the iclk control is completly under hardware control
and no software control is available.
This difference w.r.t previous OMAP's needs all the common driver
accross OMAP's , cpu_is_xxxx() checks. To avoid poulluting the
drivers dummy clock nodes are created (The autogeneration
script has been updated accordingly).
Signed-off-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Signed-off-by: Rajendra Nayak <rnayak@ti.com>
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
[paul@pwsan.com: made OMAP1 dummy_ck common and edited patch to reuse that]
Signed-off-by: Paul Walmsley <paul@pwsan.com>
All leaf clock nodes are renamed for OMAP4 to have a clk name which
end with a _ick or a _fck. This is done so that the naming convention
is same as that followed on older OMAPs.
Signed-off-by: Rajendra Nayak <rnayak@ti.com>
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Simplify the code in the omap2_clk_disable() and omap2_clk_enable()
functions, reducing levels of indentation. This makes the code easier
to read. Add some additional debugging pr_debug()s here also to help
others understand what is going on.
Revise the omap2_clk_disable() logic so that it now attempts to
disable the clock's clockdomain before recursing up the clock tree.
Simultaneously, ensure that omap2_clk_enable() is called on parent
clocks first, before enabling the clockdomain. This ensures that a
parent clock's clockdomain is enabled before the child clock's
clockdomain. These sequences should be the inverse of each other.
Revise the omap2_clk_enable() logic so that it now cleans up after
itself upon encountering an error. Previously, an error enabling a
parent clock could have resulted in inconsistent usecounts on the
enclosing clockdomain.
Remove the trivial _omap2_clk_disable() and _omap2_clk_enable() static
functions, and replace it with the clkops calls that they were
executing.
For all this to work, the clockdomain omap2_clkdm_clk_enable() and
omap2_clkdm_clk_disable() code must not return an error on clockdomains
without CLKSTCTRL registers; so modify those functions to simply return 0
in that case.
While here, add some basic kerneldoc documentation on both functions,
and get rid of some old non-CodingStyle-compliant comments that have
existed since the dawn of time (at least, the OMAP clock framework's
time).
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Richard Woodruff <r-woodruff2@ti.com>
Cc: Rajendra Nayak <rnayak@ti.com>
The OMAP2 and OMAP3 boot-time MPU rate change code is almost
identical. Merge them into mach-omap2/clock.c, and add kerneldoc
documentation.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
_clkdm_add_autodeps() and _clkdm_del_autodeps() will attempt to dereference
a NULL pointer if no autodeps were supplied to clkdm_init().
Based on a patch from Roel Kluin <roel.kluin@gmail.com> - thanks Roel.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Roel Kluin <roel.kluin@gmail.com>
Add support for categorizing and iterating over hardware IP blocks by
the "class" of the IP block. The class is the type of the IP block:
e.g., "timer", "timer1ms", etc. Move the OCP_SYSCONFIG/SYSSTATUS data
from the struct omap_hwmod into the struct omap_hwmod_class, since
it's expected to stay consistent for each class. While here, fix some
comments.
The hwmod_class structures in this patch were designed and proposed by
Benoît Cousson <b-cousson@ti.com> and were refined in a discussion
between Thara Gopinath <thara@ti.com>, Kevin Hilman
<khilman@deeprootsystems.com>, and myself.
This patch uses WARN() lines that are longer than 80 characters, as
Kevin noted a broader lkml consensus to increase greppability by
keeping the messages all on one line.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Signed-off-by: Benoît Cousson <b-cousson@ti.com>
Cc: Thara Gopinath <thara@ti.com>
Cc: Kevin Hilman <khilman@deeprootsystems.com>
ib_ud_header_init() first clears header and then fills up the various
fields. Later on, it tests header->immediate_present, which it has
already cleared, so the condition is always false. Fix this by adding
an immediate_present parameter and setting header->immediate_present
as is done with grh_present. Also remove unused calculation of
header_len.
Signed-off-by: Eli Cohen <eli@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
To fix a bug and address the reviewers' comments regarding the ACPI
GPE refcounting patch, do the following additional changes:
o Remove the second argument of acpi_ev_enable_gpe(),
'write_to_hardware', because it is not necessary any more.
o Add the "bad parameter" test against 'type' in
acpi_enable_gpe() and acpi_disable_gpe().
o Make acpi_enable_gpe() only check 'status' for runtime GPEs if
acpi_ev_enable_gpe() was actually called.
o Make acpi_disable_gpe() return 'status' returned by
acpi_ev_disable_gpe() and fix a bug where ACPI_GPE_TYPE_WAKE
and ACPI_GPE_TYPE_RUNTIME were exchanged by mistake.
o Add comments explaining why acpi_set_gpe() is used by the ACPI EC
driver.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Jeff correctly noted that using unsigned ea length is more intuitive.
CC: Jeff Lyaton <jlayton@redhat.com>
Signed-off-by: Steve French <sfrench@us.ibm.com>
The clk_round_rate() functions in the U300 clocking will always
select the lowest clocking frequency due to inverted rounding
comparisons. Fix this.
Signed-off-by: Linus Walleij <linus.walleij@stericsson.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
The timer defines are only used in core.c. Move them so
they will not be globaly exposed.
While here, add additional defines to document the magic
numbers used in the registers. Also, add some comments
for clarification.
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Acked-by: Ryan Mallon <ryan@bluewatersys.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
The GPIO support in core.c handles the interrupt support for GPIO
ports A, B, and F. The gpiolib implementation in gpio.c needs to
access the function ep93xx_gpio_int_mask when a gpio pin is made
an output and ep93xx_gpio_update_int_params in order to update
the registers.
Moving this support from core.c to gpio.c allows making the two
functions static. It also keeps all the GPIO handling together in one
file.
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Acked-by: Ryan Mallon <ryan@bluewatersys.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
