[PATCH v2] ARM: EXYNOS: Fix failed second suspend on Exynos4

Wed Mar 18 03:29:39 PDT 2015

Hi,

On Wednesday, March 18, 2015 09:57:27 AM Krzysztof Kozlowski wrote:
> On śro, 2015-03-18 at 03:05 +0900, Kukjin Kim wrote:
> > On 03/11/15 19:29, Krzysztof Kozlowski wrote:
> > > On śro, 2015-03-11 at 11:20 +0100, Krzysztof Kozlowski wrote:
> > >> On Exynos4412 boards (Trats2, Odroid U3) after enabling L2 cache in
> > >> 56b60b8bce4a ("ARM: 8265/1: dts: exynos4: Add nodes for L2 cache
> > >> controller") the second suspend to RAM failed. First suspend worked fine
> > >> but the next one hang just after powering down of secondary CPUs (system
> > >> consumed energy as it would be running but was not responsive).
> > >>
> > >> The issue was caused by enabling delayed reset assertion for CPU0 just
> > >> after issuing power down of cores. This was introduced for Exynos4 in
> > >> 13cfa6c4f7fa ("ARM: EXYNOS: Fix CPU idle clock down after CPU off").
> > >>
> > >> The whole behavior is not well documented but after checking with vendor
> > >> code this should be done like this (on Exynos4):
> > >> 1. Enable delayed reset assertion when system is running (for all CPUs).
> > >> 2. Disable delayed reset assertion before suspending the system.
> > >>    This can be done after powering off secondary CPUs.
> > >> 3. Re-enable the delayed reset assertion when system is resumed.
> > >>
> > >> Signed-off-by: Krzysztof Kozlowski <k.kozlowski at samsung.com>
> > >> Fixes: 13cfa6c4f7fa ("ARM: EXYNOS: Fix CPU idle clock down after CPU off")
> > >> Cc: <stable at vger.kernel.org>
> > >> Tested-by: Bartlomiej Zolnierkiewicz <b.zolnierkie at samsung.com>
> > >> Tested-by: Chanwoo Choi <cw00.choi at samsung.com>
> > > 
> > > Dear Kukjin,
> > > 
> > > This patch was first sent on 3rd of February. It could enter before
> > > opening 4.0 merge window. I did not receive any response from you in
> > > that time.
> > > 
> > > So let me point next steps:
> > > 1. The Exynos4412 suspend on 4.0 is broken and now this patch applies as
> > > a fix.
> > > 2. I resent it on 18th of February.
> > > 3. I received tested-by from Bartlomiej and Chanwoo.
> > > 4. Bartlomiej pinged you on 3rd March.
> > > 
> > > Still no response. If the patch does not look good then please share
> > > your comments. I'll fix it.
> > > If this patch looks good, why does it take so much time?
> > > 
> > 
> > Please use another way something like check ARM core rather than use
> > 'soc_is_xxx()', as you know it is not acceptable now even it is just
> > moving/modifying exist function though.

Kukjin, could you please explain why 'soc_is_xxx()' usage inside
arch/arm/mach-exynos/ code is not acceptable?  I know that it should
not be used outisde of this directory because of multiplatform support
but what is wrong with using it for arch/arm/mach-exynos/ code?

I'm also not sure if -rc4 is a desirable time to be doing such changes
(especially given that the patch in question is moving an existing code,
not adding new 'soc_is_xxx()' users).

[ Moreover of_machine_is_compatible() can be sometimes harmful as could
  have been seen in commit ca489c58ef0b81 ("ARM: EXYNOS: Don't use LDREX
  and STREX after disabling cache coherency" fix. ]

> Probably of_machine_is_compatible() could be used here but such change
> should be done in separate patch. This is fix for wrong usage of
> use_delayed_assertion so it should not mix with other changes in the
> code. This fixes one thing at a time. Fixing many things in one patch
> often leads to new errors or difficulties in debugging.
> 
> I can prepare a separate patch for changing this to
> of_machine_is_compatible().

IMHO this would be the best solution if there is an agreement on
'soc_is_xxx()' removal.

> > 
> > And please make sure your updates don't hurt other exynos5 stuff. Any
> > tests on exynos5 platforms would be helpful.

The patch is quite obvious and only affects Exynos4 SoCs.  Extra testing
on Exynos5 SoCs won't hurt but they should not be required for merge.

> > And I don't think the fix should be sent to 'stable' because I can't see
> > the 'add node for L2$ controller' in v3.19...looks applied from v4.0-rc...
> 
> You're right. git-describe gave me 3.19-rc1 but this was tag for the
> specific commit, not for merge-commit. The stable can be removed if this
> comes during this RC-cycle.

Yes, the stable tag was a mistake and should be removed.

> > One more if you have any doubts, I'd like to ask you to contact S.LSI
> > guys who have created the vendor codes not assume with the code because
> > maybe the vendor code you mentioned cannot cover all exynos stuff I
> > think. Then we could make more clear pm codes in mainline. To be honest
> > I'm not a Power Management hardware guy so I don't know every regarding
> > PM stuff in exynos SoCs, I can contact them easier though...I mean
> > please don't assume any hardware behavior with just vendor code. Please
> > ask, you have an access in Samsung intranet before posting something
> > like this...Hope let's make a better fix together during -rc.

FWIW I think that -rc4 is too late to be doing 'perfect patch' (we've waited
for 6 weeks on any feedback on this patch from you).  The current solution is
quite simple and has been tested to fix the regression without introducing
other problems.

> As you probably know I work in completely different company within
> Samsung Electronics than System LSI. I don't have access to the LSI
> intranet. I don't have access to guys from LSI. I'll try contacting them
> through my HQ partners.
> 
> Thanks for feedback!
> 
> Best regards,
> Krzysztof

Best regards,
--
Bartlomiej Zolnierkiewicz
Samsung R&D Institute Poland
Samsung Electronics