On Thu, 4 Apr 2013 17:26:48 +1100 Stephen Rothwell
wrote:
> Hi Andrew,
>
> Today's linux-next merge of the akpm tree got a conflict in
> fs/proc/generic.c between several commits from the vfs tree and commit
> "procfs: improve scaling in proc" from the akpm tree.
>
> I just dropped the akpm tr
(2013/04/04 9:38), Yinghai Lu wrote:
> Index: linux-2.6/kernel/kexec.c
> ===
> --- linux-2.6.orig/kernel/kexec.c
> +++ linux-2.6/kernel/kexec.c
> @@ -1360,7 +1360,7 @@ static int __init parse_crashkernel_simp
>
> if (*cur ==
Hey John,
First of all, I should confess I just glanced your code and poped
several questions. If I miss something, please slap me.
On Wed, Apr 03, 2013 at 04:52:19PM -0700, John Stultz wrote:
> This patchset is against Minchan's vrange work here:
> https://lkml.org/lkml/2013/3/12/105
>
>
On 4 April 2013 12:17, stratosk wrote:
> Why do you suspect significant increased power? With ondemand the CPU will
> go down to lowest freq as soon as the load will decreased. And the
> measurement shows that the CPU load will decrease faster (because of faster
> calculation).
I suspect it b
Hi Joonsoo,
On 04/04/2013 06:12 AM, Joonsoo Kim wrote:
> Hello, Preeti.
>
> So, how about extending a sched_period with rq->nr_running, instead of
> cfs_rq->nr_running? It is my quick thought and I think that we can ensure
> to run atleast once in this extending sched_period.
Yeah this seems to
On Wed, Apr 03, 2013 at 01:55:05PM -0600, Toshi Kani wrote:
> On Wed, 2013-04-03 at 13:37 +0800, Ram Pai wrote:
> > On Tue, Apr 02, 2013 at 10:17:29AM -0600, Toshi Kani wrote:
> > > + while ((res = *p)) {
...snip...
> > > + if (res->start > start || res->end < end) {
> >
> > This check l
Hi Viresh,
I never use performance governor, but I want improved performance with ondemand.
Why do you suspect significant increased power? With ondemand the CPU will go
down to lowest freq as soon as the load will decreased. And the measurement
shows that the CPU load will decrease faster (be
Hi Kukjin,
On Fri, Mar 15, 2013 at 1:32 PM, Vivek Gautam wrote:
> This patch-set is in continuation with patch-series:
> [PATCH v4 0/4] Enable ehci, ohci and dwc3 devices on exynos5250
> out of which follwowing patches have been picked up:
> ARM: Exynos5250: Enabling ehci-s5p driver
> ARM: Exyno
Hi Kukjin,
On Fri, Mar 15, 2013 at 1:26 PM, Vivek Gautam wrote:
> Based on 'for-next' of linux-samsung tree with following patches
> from Doug on top:
> usb: Document clocks in samsung, exynos4210-ehci/ohci bindings
> ARM: dts: add usb 2.0 clock references to exynos5250 device tree
>
> Also depe
Hi Andrew,
After merging the akpm tree, today's linux-next build (powerpc
ppc64_defconfig) failed like this:
fs/aio.c: In function 'aio_rw_vect_retry':
fs/aio.c:1106:6: error: 'opcode' undeclared (first use in this function)
Caused by commit 36bd5dc0c4e2 ("lift sb_start_write/sb_end_write out of
(2013/04/03 17:28), oskar.and...@sonymobile.com wrote:
>
> +/* it can take some time ( > 100ms ) to initialise the
> + * blacklist so we delay this until we actually need it
> + */
> +static void init_kprobe_blacklist(void)
> +{
> + int i;
> + unsigned long offset = 0, size = 0;
> + c
On 22.03.2013 16:54, Thierry Reding wrote:
> On Fri, Mar 22, 2013 at 04:34:00PM +0200, Terje Bergstrom wrote:
>> This set of patches adds support for Tegra20 and Tegra30 host1x and
>> 2D. It is based on linux-next-20130322 with RTC fixes applied.
(...)
> For the series:
>
> Reviewed-by: Thierry Re
On Thu, Apr 04, 2013 at 11:19:44, Nori, Sekhar wrote:
> On 4/4/2013 10:09 AM, Philip, Avinash wrote:
> > On Tue, Apr 02, 2013 at 14:05:09, Nori, Sekhar wrote:
> >> On 3/25/2013 1:19 PM, Philip Avinash wrote:
> >>> Add platform support for EHRPWM and ECAP by providing clock nodes and
> >>> device tr
Hi,
On Thu, Apr 4, 2013 at 12:19 AM, Russell King - ARM Linux
wrote:
>
> Still not convinced this is a proper fix. Look, the problem is this:
>
> - Initially, set the CPU percpu offset to zero. This means the boot
> CPU reads and writes to the percpu data section in the kernel image.
>
> - Th
On 02/04/13 17:24, Sergey Yanovich wrote:
> On Tue, 2013-04-02 at 16:36 +0300, Adrian Hunter wrote:
>> On my system it is significant:
>>
>> Before the patch:
>>
>> [1.625623] VFS: Mounted root (ext4 filesystem) readonly on device 179:2.
>>
>> After the patch:
>>
>> [1.935851] VFS: Mounted
On Thu, Apr 4, 2013 at 11:55 AM, Viresh Kumar wrote:
> On 3 April 2013 16:00, Benjamin Herrenschmidt
> wrote:
>> On Wed, 2013-04-03 at 15:00 +0530, Viresh Kumar wrote:
>>> On 31 March 2013 09:33, Viresh Kumar wrote:
>
>>> > Benjamin/Paul/Olof,
>>> >
>>> > Any comments on this?
>>>
>>> Ping!!
>>
On Thu, Apr 04, 2013 at 03:18:45PM +1100, Stephen Rothwell wrote:
> Hi Felipe,
>
> Today's linux-next merge of the usb-gadget tree got a conflict in
> drivers/usb/host/ehci-tegra.c between commit 369a9a9d2af7 ("usb: host:
> ehci-tegra: Fix oops in error cleanup") from the usb tree and commit
> 426
Hi Andrew,
Today's linux-next merge of the akpm tree got a conflict in
fs/proc/generic.c between several commits from the vfs tree and commit
"procfs: improve scaling in proc" from the akpm tree.
I just dropped the akpm tree patch (and the following
"procfs-improve-scaling-in-proc-v5") as the con
On 3 April 2013 16:00, Benjamin Herrenschmidt wrote:
> On Wed, 2013-04-03 at 15:00 +0530, Viresh Kumar wrote:
>> On 31 March 2013 09:33, Viresh Kumar wrote:
>> > Benjamin/Paul/Olof,
>> >
>> > Any comments on this?
>>
>> Ping!!
>
> I'm on vacation until end of April. No objection to the patch but
(2013/04/03 17:28), oskar.and...@sonymobile.com wrote:
> Instead of using printk use pr_info/pr_err/pr_warn. This was
> detected by the checkpatch.pl script.
Thank you for cleaning this up ! :)
Acked-by: Masami Hiramatsu
>
> Signed-off-by: Oskar Andero
> ---
> kernel/kprobes.c | 18 +
Hi Andrew,
Today's linux-next merge of the akpm tree got a conflict in fs/coredump.c
between commit d9fe6ea81efe ("lift sb_start_write() out of ->write()")
from the vfs tree and commit "coredump: sanitize the setting of
signal->group_exit_code" from the akpm tree.
I fixed it up (see below) and ca
(2013/04/03 17:28), oskar.and...@sonymobile.com wrote:
> Some blackpoints are only valid for specific architectures. To let each
> architecture specify its own blackpoints the list has been split in two
> lists: common and arch. The common list is kept in kernel/kprobes.c and
> the arch list is kep
From: Wei Yongjun
Fix to return a negative error code from the error handling
case instead of 0, as returned elsewhere in this function.
Signed-off-by: Wei Yongjun
---
drivers/block/drbd/drbd_main.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/drivers/block/drbd/drbd_main.c b/drivers/
Jingoo Han wrote:
>
> On Friday, March 15, 2013 8:01 AM, Doug Anderson wrote:
> >
> > The exynox4210-ehci and exynos4210-ohci nodes need a clock specified
> > using the common clock framework. Document it.
> >
> > Signed-off-by: Doug Anderson
>
> It looks good.
> Acked-by: Jingoo Han
>
>
App
Hi Andrew,
Today's linux-next merge of the akpm tree got conflicts in mm/filemap.c,
fs/xfs/xfs_file.c, fs/splice.c, fs/ocfs2/file.c, fs/ntfs/file.c,
fs/fuse/file.c, fs/cifs/file.c and fs/btrfs/file.c between commit
36bd5dc0c4e2 ("lift sb_start_write/sb_end_write out of ->aio_write()")
from the vfs
Hi Andrew,
Today's linux-next merge of the akpm tree got a conflict in
sound/core/info.c between commit d1ffe1db6641 ("snd_info_register: switch
to proc_create_data/proc_mkdir_mode") from the vfs tree and commit
"sound: convert snd_info_register() to use proc_create_data()" from the
akpm tree.
It
Hi Alex,
On Thu, 4 Apr 2013 10:00:56 +0800, Alex Shi wrote:
> In power balance, we hope some sched groups are fully empty to save
> CPU power of them. So, we want to move any tasks from them.
>
> Signed-off-by: Alex Shi
> ---
> kernel/sched/fair.c | 7 +--
> 1 file changed, 5 insertions(+),
On 4/4/2013 10:09 AM, Philip, Avinash wrote:
> On Tue, Apr 02, 2013 at 14:05:09, Nori, Sekhar wrote:
>> On 3/25/2013 1:19 PM, Philip Avinash wrote:
>>> Add platform support for EHRPWM and ECAP by providing clock nodes and
>>> device tree nodes.
>>> This series depends on [1] and [2] and is availabl
(2013/04/03 17:24), David Woodhouse wrote:
> On Wed, 2013-04-03 at 16:11 +0900, Takao Indoh wrote:
>> (2013/04/02 23:05), Joerg Roedel wrote:
>>> On Mon, Apr 01, 2013 at 02:45:18PM +0900, Takao Indoh wrote:
enable_IR
intel_enable_irq_remapping
iommu_disable_irq_remapp
On Wed, Apr 3, 2013 at 5:19 PM, Mark Brown
wrote:
> On Wed, Apr 03, 2013 at 05:00:04PM +0530, Girish KS wrote:
>> On Mon, Apr 1, 2013 at 6:42 PM, Mark Brown
>
>> >> - if (!sdd->pdev->dev.of_node) {
>> >> + if (!sdd->pdev->dev.of_node && !is_polling(sdd)) {
>> >> res = platfor
Hi Al,
Today's linux-next merge of the signal tree got a conflict in fs/compat.c
between commit 1fd193fd030a ("switch compat readv/writev variants to
COMPAT_SYSCALL_DEFINE") from the vfs tree and commit 76b021d053ed
("convert vmsplice to COMPAT_SYSCALL_DEFINE") from the signal tree.
I fixed it up
Hi Al,
Today's linux-next merge of the signal tree got conflicts in
fs/read_write.c and fs/read_write.h between commit 1fd193fd030a ("switch
compat readv/writev variants to COMPAT_SYSCALL_DEFINE") from the vfs tree
and commit 19f4fc3aee18 ("convert sendfile{,64} to
COMPAT_SYSCALL_DEFINE") from the
On 04/03/2013 09:48 PM, Stephen Warren wrote:
> On 04/03/2013 01:14 AM, Vineet Gupta wrote:
>> forgot to CC linux-arch
>>
>> On 04/03/2013 12:42 PM, Vineet Gupta wrote:
>>> Currently, for every ARC kernel build I see the following:
>>>
>>> --->8-
>>> DTBarch/arc/bo
On Wed, Apr 03, 2013 at 09:39:44PM -0700, Tony Chung wrote:
>
> Observed that the Watchdog Timer Status bit can be set when the driver is
> loaded. Reset it during initialization. The time-out value must be set to 0
> explicitly in this case to prevent an immediate reset.
>
> Depending on the mot
+-- On Thu, 4 Apr 2013, Jaegeuk Kim wrote --+
| Why should we take unnecessary locks and an f2fs_submit_bio call?
Yep, we should not. I wasn't sure if these are unnecessary when
a_ops->writepage = NULL.
Thank you.
--
Prasad J Pandit / Red Hat Security Response Team
DB7A 84C5 D3F9 7CD1 B5EB C9
On 03/04/13 07:20, Tony Prisk wrote:
V3 Changes:
Tidied up patch 1 as requested by Jean-Christophe - Makes the patch much
simpler.
V2 Changes:
Split the SoC and board portions of the binding into their respective files.
Removed the forced non-interlaced code from wm8505fb.c and vt8500lcdfb.c
Ju
Hi H.Peter,
On 04/04/2013 09:32 AM, H. Peter Anvin wrote:
On 04/03/2013 06:17 PM, Simon Jeons wrote:
e820 also contain mmio, correct?
No.
How to check which address is used by mmio? /proc/iomem, correct?
So cpu should not access address beyond
e820 map(RAM+MMIO).
No.
-hpa
--
Hi Tony,
On Thursday 04 April 2013 05:12 AM, Tony Lindgren wrote:
> Hi,
>
[]..
>> @@ -1663,6 +1664,40 @@ static struct omap_clk omap44xx_clks[] = {
>> CLK(NULL, "cpufreq_ck", &dpll_mpu_ck, CK_443X),
>> };
>>
>> +static struct clk *scrm_clks[] = {
>> +&auxclk0_ck,
>> +&a
On 4 April 2013 10:48, Stephen Rothwell wrote:
> Today's linux-next merge of the imx-mxs tree got a conflict in
> arch/arm/mach-imx/cpufreq.c between commits b43a7ffbf33b ("cpufreq:
> Notify all policy->cpus in cpufreq_notify_transition()") and eb2f50ff93f0
> ("cpufreq: drivers: Remove unnecessary
The following changes since commit 8bb9660418e05bb1845ac1a2428444d78e322cc7:
Linux 3.9-rc4 (2013-03-23 16:52:44 -0700)
are available in the git repository at:
git://server.prisktech.co.nz/git/linuxwmt.git tags/vt8500/pinctrl
for you to fetch changes up to ce63d6d4bb9f601de32d4b99f925a65182
Hi Shawn,
Today's linux-next merge of the imx-mxs tree got a conflict in
arch/arm/mach-imx/cpufreq.c between commits b43a7ffbf33b ("cpufreq:
Notify all policy->cpus in cpufreq_notify_transition()") and eb2f50ff93f0
("cpufreq: drivers: Remove unnecessary assignments of policy-> members")
from the p
Hi Sarah,
On Wed, Apr 3, 2013 at 10:57 PM, Sarah Sharp
wrote:
> Question: Do you still need this patch for 3.10?
Felipe's 'next' is closed for 3.10, so this series won't be making it
to 3.10 now, as a whole. :-(
>
> http://marc.info/?l=linux-usb&m=136057666911621&w=2
>
> Does this patchset bui
Tony Prisk wrote:
...
>Can you just confirm you want:
>
>pinctrl: bcm2835: make use of of_property_read_u32_index()
>
>to be included in the pull with this series. I'll get it sent out once
>you let me know.
That sounds fine, thanks.
--
Sent from my Android phone with K-9 Mail. Please excuse
Ping!
On 03/21/2013 07:36 AM, Simon Jeons wrote:
> Hi Naoya,
> On 03/21/2013 05:53 AM, Naoya Horiguchi wrote:
>> On Wed, Mar 20, 2013 at 07:57:32AM +0800, Simon Jeons wrote:
>>> Hi Naoya,
>>> On 02/22/2013 03:41 AM, Naoya Horiguchi wrote:
When we have a page fault for the address which is back
On Mon, Apr 01, 2013 at 12:16:57PM -0700, Kees Cook wrote:
> On Sun, Mar 31, 2013 at 8:22 PM, Anton Vorontsov wrote:
> > Hi all,
> >
> > Here are a few updates from the Android dev tree. Thanks to Arve Hjønnevåg
> > for the code, and John Stultz for actually preparing commits for
> > submission.
>
Hi Stratos,
On 4 April 2013 05:00, Stratos Karafotis wrote:
> I tried to do some measurements simulating a CPU load with a loop that simply
> counts
> an integer. The first test simulates a CPU load that lasts 2 x sampling_rate
> = ~ 2us.
> The second ~4us and the third ~6us.
> Ther
Hi Greg,
Today's linux-next merge of the staging tree got conflicts in
drivers/staging/ccg/f_fs.c and drivers/staging/ccg/rndis.c between
commits 465bced2e3ed ("ccg: don't bother with fops->owner") and
4dfac87dca02 ("procfs: new helper - PDE_DATA(inode)") from the vfs tree
and commit 515e6dd20b3f
On Tue, Apr 02, 2013 at 14:05:09, Nori, Sekhar wrote:
> On 3/25/2013 1:19 PM, Philip Avinash wrote:
> > Add platform support for EHRPWM and ECAP by providing clock nodes and
> > device tree nodes.
> > This series depends on [1] and [2] and is available for testing at [3].
> > Tested for back light
Hi Greg,
Today's linux-next merge of the staging tree got a conflict in
drivers/staging/dgrp/dgrp_specproc.c between commit 464b4fd02caa ("dgrp
procfs fixes, part 2") and various other commits from the vfs tree and
commit 433121c6ef51 ("staging: dgrp: cleanup sparse warnings") from the
staging tre
Observed that the Watchdog Timer Status bit can be set when the driver is
loaded. Reset it during initialization. The time-out value must be set to 0
explicitly in this case to prevent an immediate reset.
Depending on the motherboard and with watchdog disabled in the BIOS, this reset
problem can
Watchdog should support timeout in minutes.
For example, crash dump will usually require 5+ minutes.
This patch increase the timeout from 255 seconds to (255*60) seconds and should
be good for older LTS releases.
This patch need to be rewritten again when this driver migrate to new watchdog
infr
Hi Greg,
Today's linux-next merge of the staging tree got conflicts in
drivers/staging/dgrp/dgrp_dpa_ops.c, drivers/staging/dgrp/dgrp_mon_ops.c,
drivers/staging/dgrp/dgrp_net_ops.c and
drivers/staging/dgrp/dgrp_ports_ops.c between commit 33633903f16f ("dgrp
procfs fixes, part 5: per-node files") f
On 04/04/13 10:23, Stephen Warren wrote:
On 04/03/2013 07:40 AM, Linus Walleij wrote:
On Wed, Apr 3, 2013 at 8:24 AM, Tony Prisk wrote:
v5 changes:
Changed as requested by Rob Herring:
Added of_find_property_value_of_size() to drivers/of/base.c to remove some
code that was being duplicated.
Hi Linus,
On 19 March 2013 13:58, Patrice CHOTARD wrote:
> On 03/19/2013 07:31 AM, Sachin Kamat wrote:
>
>> These symbols are used only in this file. Without this patch
>> we get the following warnings:
>> drivers/pinctrl/pinctrl-abx500.c:520:5: warning:
>> symbol 'abx500_gpio_request' was not de
Hi Felipe,
Today's linux-next merge of the usb-gadget tree got a conflict in
drivers/usb/host/ehci-tegra.c between commit 369a9a9d2af7 ("usb: host:
ehci-tegra: Fix oops in error cleanup") from the usb tree and commit
4261b8f3538c ("usb: host: ehci-tegra: fix PHY error handling") from the
usb-gadge
Hi Andrew,
On 03/25/2013 09:42 PM, Andrew Shewmaker wrote:
Add an admin_reserve_kbytes knob to allow admins to change the
hardcoded memory reserve to something other than 3%, which
may be multiple gigabytes on large memory systems. Only about
8MB is necessary to enable recovery in the default mod
On Wed, Apr 03, 2013 at 07:45:52PM +0200, Oleg Nesterov wrote:
> On 04/03, Anton Arapov wrote:
...
> Looks fine to me. I am going to add this to
> git://git.kernel.org/pub/scm/linux/kernel/git/oleg/misc uprobes/core
>
> Ananth. "4/9 uretprobes/ppc" looks "obviously correct", but could you
> plea
On Wed, Apr 03, 2013 at 06:00:34PM +0200, Anton Arapov wrote:
> Hijack the return address and replace it with a trampoline address.
> PowerPC implementation.
>
> Signed-off-by: Anton Arapov
Acked-by: Ananth N Mavinakayanahalli
> ---
> arch/powerpc/include/asm/uprobes.h | 1 +
> arch/powerpc/
Now that uids and gids are completely encapsulated in kuid_t
and kgid_t we no longer need to pass struct cred which allowed
us to test both the uid and the user namespace for equality.
Passing struct cred potentially allows us to pass the entire group
list as BSD does but I don't believe the cost
David Howells writes:
> Btw, one thing I've noticed is that sysfs appears to be able to execute code
> in the module by way of parameter alterations before the module is completely
> set up.
It's preceeded by the module parameter parsing, which can do the same
thing, so I don't think it's an issu
Sebastian Wankerl writes:
> Add non-zero module sections to sysfs on architectures unequal to PARISC.
> KGDB needs all module sections for proper module debugging. Therefore, commit
> 35dead4235e2b67da7275b4122fed37099c2f462 is revoked except for PARISC
> architecture.
#ifdef CONFIG_PARISC in th
On 04/03/2013 09:36 PM, Stephen Rothwell wrote:
Hi Larry,
On Wed, 03 Apr 2013 20:17:01 -0500 Larry Finger
wrote:
On 04/03/2013 08:04 PM, Stephen Rothwell wrote:
On Thu, 4 Apr 2013 12:00:43 +1100 Stephen Rothwell
wrote:
After merging the wireless-next tree, today's linux-next build (x86
Hi Minchan,
On 04/01/2013 07:45 AM, Minchan Kim wrote:
__do_huge_pmd_anonymous_page depends on page_add_new_anon_rmap's
spinlock for making sure that clear_huge_page write become visible
after set set_pmd_at() write.
1. There are no pte modify, why take page_table_lock here?
2. What's the meani
Hi Larry,
On Wed, 03 Apr 2013 20:17:01 -0500 Larry Finger
wrote:
>
> On 04/03/2013 08:04 PM, Stephen Rothwell wrote:
> >
> > On Thu, 4 Apr 2013 12:00:43 +1100 Stephen Rothwell
> > wrote:
> >>
> >> After merging the wireless-next tree, today's linux-next build (x86_64
> >> allmodconfig) produce
Andy Lutomirski writes:
> On Wed, Apr 3, 2013 at 5:47 PM, Eric W. Biederman
> wrote:
>>
>> No. The patch is still bogus.
>>
>> If the problem is that we are not coallescing messages in stream_recvmsg
>> we need a different fix.
>>
>> Probably something like:
>>
>> if (check_c
On 04/04/2013 10:14 AM, H. Peter Anvin wrote:
Because git didn't exist before then?
Oh, I see, thanks! :-)
Simon Jeons wrote:
On 04/04/2013 09:32 AM, H. Peter Anvin wrote:
On 04/03/2013 06:17 PM, Simon Jeons wrote:
e820 also contain mmio, correct?
No.
So cpu should not access address
Hi Jens,
Today's linux-next merge of the block tree got a conflict in
drivers/block/drbd/drbd_proc.c between commit 4dfac87dca02 ("procfs: new
helper - PDE_DATA(inode)") from the vfs tree and commit 193d01532a73
("drbd: add module_put() on error path in drbd_proc_open()") from the
block tree.
I f
It was reported that the following LSB test case failed
https://lsbbugs.linuxfoundation.org/attachment.cgi?id=2144 because we
were not coallescing unix stream messages when the application was
expecting us to.
The problem was that the first send was before the socket was accepted
and thus sock->s
Because git didn't exist before then?
Simon Jeons wrote:
>On 04/04/2013 09:32 AM, H. Peter Anvin wrote:
>> On 04/03/2013 06:17 PM, Simon Jeons wrote:
>>> e820 also contain mmio, correct?
>> No.
>>
>>> So cpu should not access address beyond
>>> e820 map(RAM+MMIO).
>> No.
>>
>> -hpa
>>
>>
>
This reverts commit 14134f6584212d585b310ce95428014b653dfaf6.
The problem that the above patch was meant to address is that af_unix
messages are not being coallesced because we are sending unnecesarry
credentials. Not sending credentials in maybe_add_creds totally
breaks unconnected unix domain
On Wed, Apr 03, 2013 at 06:44:51PM -0700, Marcel Holtmann wrote:
> Hi Greg,
>
> I've been approached by a developer at Sony who wants to publish an
> open source driver for a Sony GPS receiver module.
> >>>
> >>> What does the device look like? USB device? UART? Something else?
> >>>
We need initialize the se.avg.{decay_count, load_avg_contrib} for a
new forked task.
Otherwise random values of above variables cause mess when do new task
enqueue:
enqueue_task_fair
enqueue_entity
enqueue_entity_load_avg
and make forking balancing imbalance since incorrect
The scale_rt_power() used to represent the left CPU utilization
after reduce rt utilization. so named it as scale_rt_power has a bit
inappropriate.
Since we need to use the rt utilization in some incoming patches, we
just change return value of this function to rt utilization, and
rename it as sca
When we fail to allocate the node pwq, we can use the default pwq
for the node.
Thus we can avoid failure after allocated default pwq, and remove
some code for failure path.
Signed-off-by: Lai Jiangshan
---
kernel/workqueue.c | 28 +++-
1 files changed, 7 insertions(+)
calculate the node of the pool earlier, and allocate the pool
from the node.
Signed-off-by: Lai Jiangshan
---
kernel/workqueue.c | 29 +++--
1 files changed, 15 insertions(+), 14 deletions(-)
diff --git a/kernel/workqueue.c b/kernel/workqueue.c
index 737646d..3f33077 1
freezing is nothing related to pools, but POOL_FREEZING adds a connection,
and causes freeze_workqueues_begin() and thaw_workqueues() complicated.
Since freezing is workqueue instance attribute, so we introduce __WQ_FREEZING
to wq->flags instead and remove POOL_FREEZING.
Signed-off-by: Lai Jiangs
it is very common wq->dfl_pwq->refcnt > 1.
[7.939873] WARNING: at kernel/workqueue.c:4201
destroy_workqueue+0x6a/0x13e()
[7.943601] Hardware name: 4286C12
[7.947250] Modules linked in: sdhci_pci sdhci mmc_core usb_storage i915
drm_kms_helper drm i2c_algo_bit i2c_core video
[7.951
If we have 4096 CPUs, workqueue_cpu_up_callback() will travel too much CPUs,
to avoid it, we use for_each_cpu_worker_pool() for the cpu pools and
use for_each_unbound_pool() for unbound pools.
After it, for_each_pool() becomes unused, but we keep it for future
possible usage.
Signed-off-by: Lai J
simplify pwq_adjust_max_active().
make freeze_workqueues_begin() and thaw_workqueues() fast skip non-freezable wq.
Signed-off-by: Lai Jiangshan
---
kernel/workqueue.c | 13 ++---
1 files changed, 6 insertions(+), 7 deletions(-)
diff --git a/kernel/workqueue.c b/kernel/workqueue.c
inde
merge the code of clearing POOL_DISASSOCIATED to rebind_workers(), and
rename rebind_workers() to associate_cpu_pool().
It merges high related code together and simplify
workqueue_cpu_up_callback().
Signed-off-by: Lai Jiangshan
---
kernel/workqueue.c | 21 ++---
1 files change
Since the rt task priority is higher than fair tasks, cfs_rq utilization
is just the left of rt utilization.
When there are some cfs tasks in queue, the potential utilization may
be yielded, so mulitiplying cfs task number to get max potential
utilization of cfs. Then the rq utilization is sum of
For power aware balancing, we care the sched domain/group's utilization.
So add: sd_lb_stats.sd_util and sg_lb_stats.group_util.
And want to know which group is busiest but still has capability to
handle more tasks, so add: sd_lb_stats.group_leader
Signed-off-by: Alex Shi
---
kernel/sched/fair.
rq->avg_idle is 'to used to accommodate bursty loads in a dirt simple
dirt cheap manner' -- Mike Galbraith.
With this cheap and smart bursty indicator, we can find the wake up
burst, and just use nr_running as instant utilization only.
The 'sysctl_sched_burst_threshold' used for wakeup burst, sin
If a sched domain is idle enough for regular power balance, LBF_POWER_BAL
will be set, LBF_PERF_BAL will be clean. If a sched domain is busy,
their value will be set oppositely.
If the domain is suitable for power balance, but balance should not
be down by this cpu(this cpu is already idle or full
Sleeping task has no utiliation, when they were bursty waked up, the
zero utilization make scheduler out of balance, like aim7 benchmark.
rq->avg_idle is 'to used to accommodate bursty loads in a dirt simple
dirt cheap manner' -- Mike Galbraith.
With this cheap and smart bursty indicator, we can
Packing tasks among such domain can't save power, just performance
losing. So no power balance on them.
Signed-off-by: Alex Shi
---
kernel/sched/fair.c | 7 ---
1 file changed, 4 insertions(+), 3 deletions(-)
diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 047a1b3..3a0284b 1006
From: Preeti U Murthy
Problem:
select_task_rq_fair() returns a target CPU/ waking CPU if no balancing is
required. However with the current power aware scheduling in this path, an
invalid CPU might be returned.
If get_cpu_for_power_policy() fails to find a new_cpu for the forked task,
then ther
This patch enabled the power aware consideration in load balance.
As mentioned in the power aware scheduler proposal, Power aware
scheduling has 2 assumptions:
1, race to idle is helpful for power saving
2, less active sched_groups will reduce power consumption
The first assumption make performan
Added 4 new members in sb_lb_stats, that will be used in incomming
power aware balance.
group_min; //least utliszation group in domain
min_load_per_task; //load_per_task in group_min
leader_util;// sum utilizations of group_leader
min_util; // sum utilizations of group_min
Signed-o
If the waked task is transitory enough, it will has a chance to be
packed into a cpu which is busy but still has time to care it.
For powersaving policy, only the history util < 25% task has chance to
be packed. If there is no cpu eligible to handle it, will use a idlest
cpu in leader group.
Mort
When active task number in sched domain waves around the power friendly
scheduling creteria, scheduling will thresh between the power friendly
balance and performance balance, bring unnecessary task migration.
The typical benchmark is 'make -j x'.
To remove such issue, introduce a u64 perf_lb_reco
In power balance, we hope some sched groups are fully empty to save
CPU power of them. So, we want to move any tasks from them.
Signed-off-by: Alex Shi
---
kernel/sched/fair.c | 7 +--
1 file changed, 5 insertions(+), 2 deletions(-)
diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
ind
In power aware scheduling, we don't want to balance 'prefer_sibling'
groups just because local group has capacity.
If the local group has no tasks at the time, that is the power
balance hope so.
Signed-off-by: Alex Shi
---
kernel/sched/fair.c | 6 +-
1 file changed, 5 insertions(+), 1 deleti
This patch add power aware scheduling in fork/exec/wake. It try to
select cpu from the busiest while still has utilization group. That's
will save power since it leaves more groups idle in system.
The trade off is adding a power aware statistics collection in group
seeking. But since the collectio
Power aware fork/exec/wake balancing needs both of structs in incoming
patches. So move ahead before it.
Signed-off-by: Alex Shi
---
kernel/sched/fair.c | 99 +++--
1 file changed, 50 insertions(+), 49 deletions(-)
diff --git a/kernel/sched/fair.c
This patch add the power aware scheduler knob into sysfs:
$cat /sys/devices/system/cpu/sched_balance_policy/available_sched_balance_policy
performance powersaving
$cat /sys/devices/system/cpu/sched_balance_policy/current_sched_balance_policy
powersaving
This means the using sched balance policy i
The cpu's utilization is to measure how busy is the cpu.
util = cpu_rq(cpu)->avg.runnable_avg_sum * SCHED_POEWR_SCALE
/ cpu_rq(cpu)->avg.runnable_avg_period;
Since the util is no more than 1, we scale its value with 1024, same as
SCHED_POWER_SCALE and set the FULL_UTIL as 1
Remove CONFIG_FAIR_GROUP_SCHED that covers the runnable info, then
we can use runnable load variables.
Signed-off-by: Alex Shi
---
include/linux/sched.h | 7 +--
kernel/sched/core.c | 7 +--
kernel/sched/fair.c | 13 ++---
kernel/sched/sched.h | 9 +
4 files chang
Current scheduler behavior is just consider for larger performance
of system. So it try to spread tasks on more cpu sockets and cpu cores
To adding the consideration of power awareness, the patchset adds
a powersaving scheduler policy. It will use runnable load util in
scheduler balancing. The cur
Many many thanks for Namhyung, PJT, Vicent and Preeti's comments and suggestion!
This version included the following changes:
a, remove the patch 3th to recover the runnable load avg recording on rt
b, check avg_idle for each cpu wakeup burst not only the waking CPU.
c, fix select_task_rq_fair retu
1 - 100 of 750 matches
Mail list logo
