== Linus Walleij linusw ==
=== Highlights ===
* pinctrl fixes sent and pulled by Torvalds.
* Reviewed and applied a bunch of pinctrl patches from Stephen,
Dong, and a newcomer: John Crispin from OpenWRT. We have
good traction in this subsystem.
* Sent pull requests for ux500-core, ux500-devicetree and
ux500-gpio-pins to bring the machine up to date with the recent
patches. Not yet pulled into ARM SoC
* Brought ux500 pinctrl forward, trying to find a smooth transition
from the existing codebase to a pinctrl-based approach. This is
taking some time since it requires quite a bit of upfront code.
* Reviewing device tree from Lee Jones.
* Russell accepted PL190 VIC cleanups and applied them.
* Discussing power domain issues. Andrea and our internal
power management team as well as Vincent Guittot from Linaro
PM are all involved. We might need to propose modifications to
runtime PM.
* Ran the ESC platform perimeter presentation twice internally
discussing and presenting new features in kernels 3.0 thru 3.4+
roughly. It appears to have been appreciated.
http://www.df.lth.se/~triad/papers/ESC-400Slides_Walleij.pdf
=== Plans ===
* Review incoming pinctrl patches.
* Convert ux500 (gpio-nomadik) to pinctrl
for the kernel 3.5 timeframe.
* Test the PL08x patches on the Ericsson Research
PB11MPCore and submit platform data for using
pl08x DMA on that platform.
* Look into other Ux500 stuff in need of mainlining...
like
- Ux500 clocks (looks like a new assignee might look into this)
- the HWMON stuff.
=== Issues ===
* Reviewing-and applying pinctrl and ux500 patches cuts
into ux500 pin control development and disturbs focus.
Thanks,
Linus Walleij
== Deepak Saxena <dsaxena> ==
=== Highlights ===
* Researching new roadmap cards
* Talk on "One zImage to Rule them All" accepted to LinuCon Japan!
* Doing lots of back research on single zImage. Focusing on
information research right now,
going through all the emails I can find on the topic back to last
May, chasing down discussions,
digging through the kernel, syncing up with Nico and Arnd as
needed, and getting my head
around it all. My goal is to have a clear idea of what is left to
do and also a plan of what
we should focus on at Connect.
* 1:1's
* Interviewed Rajinakanth from ST-E for KWG
* Synced up with Scott and Vicki on training sessions at Connect.
=== Plans ===
* Usual 1:1's and meetings
* ST-E DT sync up with Raj and Lee
* Connect training planning
* Interviewing Ulf from ST-E for KWG position
* Continue work on new roadmap cards
* Continue single zImage deep dive
* Dig out my timex.h patches if I have time
=== Issues ===
* None at the time
=== Travel/Time Off ===
* Possibly out Monday, April 30th.
* Off May 4th and following week, back to work on Monday the 13th.
Will have no interwebs and limited phone access.
* Connect Q2, with follow up travel to Linux Con Japan the week after
and going to Austin before Connect to sync up with Mounir.
=== Highlights ===
* Sent the mach-mxs common clk conversion series based on v3.4-rc3 plus
various fixup/cleanup patches that Mike confirmed pick-up on his
-next branch and the clkdev registration patch.
* Reviewed imx pinctrl patch from Dong Aisheng.
* Added pinctrl support for mach-mxs in using a different binding from
imx. Discussing the pros and cros between these two the bindings.
* Reviewed initial spi-mxs support patch from Fabio.
* Reviewed the mach-mxs changes of Marek's mxs usb series, and push him
to use device tree.
* Reviewed mxs-dma device tree support series from Dong Aisheng.
* Reviewed gpmi-nand device tree support series from Huang Shijie.
* Reviewed and merged the initial imx6q-sabresd support patch from
Fabio.
* Reviewed imx5/6 cpuidle patches from Rob Lee.
--
Regards,
Shawn
=== Pinctrl ===
* sent two patch to improve pinctrl core debug information
* sent out pinctrl-imx v2 patch
Addressed many comments from Stephen, Sascha and etc.
The main changes are allow pin based config.
Still not received comments.
* Sent out handling pinctrl dummy state in core patch.
Got a lot discussions, the finally decision would be provide a common
api for platforms to call
to decide whether it needs using dummy state.
* sent out a patch series to convert mxs-dma driver to be more generic to
support both mxs and mx6
and add dt support
=== Plan ===
* send pincgtrl handling dummy state v2 patch, then we can start convert
drivers to use pinctrl
without breaking exist code.
* pinctrl handle dummy gpio support which avoid breaking exist code when
using pinctrl gpio support
* pinctrl-imx gpio support which depends on pinctrl handle dummy gpio patch
...otherwise the we're getting the wrong numbers in usage_in_bytes.
On Mon, Mar 05, 2012 at 09:19:34AM +0900, KAMEZAWA Hiroyuki wrote:
[...]
> > diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> > index 228d646..c8abdc5 100644
> > --- a/mm/memcontrol.c
> > +++ b/mm/memcontrol.c
> > @@ -3812,6 +3812,9 @@ static inline u64 mem_cgroup_usage(struct mem_cgroup *memcg, bool swap)
> >
> > val = mem_cgroup_recursive_stat(memcg, MEM_CGROUP_STAT_CACHE);
> > val += mem_cgroup_recursive_stat(memcg, MEM_CGROUP_STAT_RSS);
> > + val += mem_cgroup_recursive_stat(memcg, MEM_CGROUP_STAT_FILE_MAPPED);
> >
> > 1. Is there any particular reason we don't currently account file mapped
> > memory in usage_in_bytes?
> >
> > To me, MEM_CGROUP_STAT_FILE_MAPPED hunk seems logical even if we
> > don't use it for lowmemory notifications.
> >
> > Plus, it seems that FILE_MAPPED _is_ accounted for the non-root
> > cgroups, so I guess it's clearly a bug for the root memcg?
>
> CACHE includes all file caches. Why do you think FILE_MAPPED is not included in CACHE ?
There were tons of changes in the memcg lately, but I believe the issue
is still there.
For example, looking into this code flow:
-> page_add_file_rmap() (mm/rmap.c)
-> mem_cgroup_inc_page_stat(page, MEMCG_NR_FILE_MAPPED) (include/linux/memcontrol.h)
-> void mem_cgroup_update_page_stat(page, MEMCG_NR_FILE_MAPPED, 1) (mm/memcontrol.c)
And then:
void mem_cgroup_update_page_stat(struct page *page,
enum mem_cgroup_page_stat_item idx, int val)
{
...
switch (idx) {
case MEMCG_NR_FILE_MAPPED:
idx = MEM_CGROUP_STAT_FILE_MAPPED;
break;
default:
BUG();
}
this_cpu_add(memcg->stat->count[idx], val);
...
}
So, clearly, this function only bothers updating _FILE_MAPPED only,
leaving _CACHE alone.
If you're saying that _CACHE meant to include _FILE_MAPPED, then
I guess the patch down below would be a proper fix then... Otherwise
we need to be consistent on stats reporting, and either fall-back
to my original fix (in mem_cgroup_usage()), or think about doing it
some other way...
Signed-off-by: Anton Vorontsov <anton.vorontsov(a)linaro.org>
---
The patch is against current -next.
Thanks,
mm/memcontrol.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 884e936..760ecf5 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -1958,6 +1958,8 @@ void mem_cgroup_update_page_stat(struct page *page,
switch (idx) {
case MEMCG_NR_FILE_MAPPED:
+ idx = MEM_CGROUP_STAT_CACHE;
+ this_cpu_add(memcg->stat->count[idx], val);
idx = MEM_CGROUP_STAT_FILE_MAPPED;
break;
default:
--
1.7.9.2
=== Highlights ===
* Reworked the Android ashmem code so it uses my fadvise volatile work
* Handled a few issues for Andrey w/ the linaro-android-3.4 tree.
* Synced up with gregkh and tglx, then sent the Android Alarm timer
mending patches to staging for 3.5, and got them merged.
* Converted Android Alarm driver to using wakeup sources instead of
wakelocks, sent out an RFC to lkml and the Android team
* Spent more time looking at Rafael's userland wakelock api.
* Worked on reported RTC regression. Looks to be a hardware quirk, but
still waiting on feedback from the owner.
* Dug in a bit on a community clocksource watchdog issue. Still no
solution yet.
* Gave Deepak feedback on upstream tracking ideas.
* Did bi-weekly Android upstreaming subteam email
=== Plans ===
* Haven't gotten much feedback recently, so I'm planning to submit
volatile range code for inclusion. This likely will get detractors to
come out of the woodwork. But hey, maybe I'll get lucky and it will
just go in?
* See if Rafael's userland wakelock api can be readied for 3.5
=== Issues ===
* NA.
== Linus Walleij linusw ==
=== Highlights ===
* Brought pending ux500 GPIO and pinctrl fixes forward. I have
a patch set with all internal changes from ST-Ericsson that
will be mailed off today. This is a prerequisite for the ux500
pinctrl work.
* Sent pull request for ux500 fixes for the -rc series to ARM SoC,
Olof @lixom has pulled it in.
* Reviewed a patch-storm from Lee Jones. Keeps me on the
move that guy :-)
* Finalized NO_IRQ removal issues and irq_domain migration for
the FPGA IRQ controller in the ARM Integrator and Versatile.
Patch is in Russell's patch tracker.
* Wrote four patches to the PL190 VIC due to bugs found during
the FPGA IRQ conversion. One makes the VIC respect the mask
of valid interrupt sources, one uses the map iterator, one bumps
the U300 IRQ list and one bumps the Nomadik IRQ list, both to
avoid using IRQ 0 which is now "NO_IRQ".
* Discussing power domain issues. Now on the I2C controller in
ux500. It looks like the maintainers want a centralized approach
to power domains, not distributed.
* Proposed patches to delete support for the obsolete DB5500
ASIC. Need to be rebased to a patch from MFD before it can
be pulled into ARM SoC.
* Booked hotel for Linaro Connect 2012.2
=== Plans ===
* Send pinctrl fixes to Torvalds.
* Review incoming pinctrl patches.
* Convert ux500 (gpio-nomadik) to pinctrl
for the kernel 3.5 timeframe.
* Test the PL08x patches on the Ericsson Research
PB11MPCore and submit platform data for using
pl08x DMA on that platform.
* Look into other Ux500 stuff in need of mainlining...
like
- Ux500 clocks
- the HWMON stuff.
* Presentation for a dept. here about the platform perimeter
from the ESC two weeks ago. Seems useful for getting some
news in v3.4+. Today actually!
=== Issues ===
* Too much internal fuzz and advisory for different internal
projects have been a bit stressful, so I dropped off IRC for
that reason. The people visiting me in person are distracting
enough as it is. I will currently only use IRC when working from
home. This is one of the few environmental actions I can take
to get some focus from time to time.
Thanks,
Linus Walleij
== Thomas Abraham <thomas-ab> ==
=== Highlights ===
* Submitted initial device tree support patches for Exynos5 Rev1.0.
* Prepared initial device tree support for SPI controller driver. This will
be submitted after additional testing.
=== Plans ===
* Complete device tree support for SPI controller driver.
* Rework combiner and wakeup interrupt controller device tree patches.
* Start with device tree support for Samsung pin-ctrl driver.
=== Highlight ===
* Reviewed common clock bindings from Rob Herring, and raised concern
on multiple clock outputs support and the question on how the entire
clock tree will be represented in the combination of clock driver
and device tree.
* Converted mach-mxs clock to common clk, and waiting for a stable
branch with all recent fixup/cleanup integrated to post the patches.
* Discovered a couple of bugs existing in common clk and previous imx6
porting during mach-mxs porting and testing. Sent fixing to Mike and
Sascha respectively for them to integrate.
* Reviewed the v6 of pwm framework series and tested it with mxs pwm
backlight drivers.
* Fixed the gpmi-nand breakage caused by dma cookie series in the way
that Russell suggested, enabling dma channel in issue_pending call.
* Reviewed DA9052 mfd and regulator device tree support from Paul Liu.
* Reviewed a couple of imx6q dts patches from Fabio, including the
initial SabreSD dts file.
--
Regards,
Shawn
=== Highlights ===
* Resolved the volatile persistence issue, and finally took a crack at
resolving the calling vmtruncate_range from a shrinker lockdep problem.
Sent out another revision (v7) of the volatile range code to lkml.
* Notified Andrey about my rebased 3.4-rc1 androidization patch set, and
immediately afterwards the Google developers released their own. Did
some basic testing and mirrored it out for 12.04. Tixy pointed out a
missing patch and I re-added it.
* Pinged Rafael and Arve about the reworked userland wakelocks code.
Ends up the userland wakelocks api is not reworked, Arve's just removed
the dependencies on the wakelock core, and that there are still
outstanding issues blocking upstream w/ the existing implementation.
Rafael's userland wakelock implementation is likely the future, but
needs further work/testing.
* Ran Android upstreaming subteam call
=== Plans ===
* Punted on a few community issues this week, so I'll likely need to
revisit them.
* Read up on sysrq-t and why its holding irqs off for very long periods
of time.
* See what feedback I get from volatile range, and if its minimal, drop
the RFC and submit it for inclusion.
=== Issues ===
* NA.