mbox

[GIT,PULL] Samsung fixes for v3.4

Message ID 4FAED92C.4090103@samsung.com
State New
Headers show

Pull-request

git://git.kernel.org/pub/scm/linux/kernel/git/kgene/linux-samsung.git

Message

Kukjin Kim May 12, 2012, 9:42 p.m. UTC
Hi Arnd and Olof,

Sorry for late request for v3.4.
If there is a chance, please pull this for v3.4:
git://git.kernel.org/pub/scm/linux/kernel/git/kgene/linux-samsung.git 
v3.4-samsung-fixes-4

This is including fixes of hotplug for Cortext-A15(EXYNOS5) and build 
failure of S3C2412(smdk2412 test) and bit typo for clk_pdma1 of exynos5 
and build failure of early version of exynos4210 on universalc210 board.

Thanks.

Best regards,
Kgene.
--
Kukjin Kim <kgene.kim@samsung.com>, Senior Engineer,
SW Solution Development Team, Samsung Electronics Co., Ltd.

The following changes since commit d48b97b403d23f6df0b990cee652bdf9a52337a3:

   Linux 3.4-rc6 (2012-05-06 15:07:32 -0700)

are available in the git repository at:
   git://git.kernel.org/pub/scm/linux/kernel/git/kgene/linux-samsung.git 
v3.4-samsung-fixes-4

Changhwan Youn (1):
       ARM: EXYNOS: fix the hotplug for Cortex-A15

Jose Miguel Goncalves (1):
       ARM: SAMSUNG: Fix for S3C2412 EBI memory mapping

Kukjin Kim (1):
       ARM: EXYNOS: fix ctrlbit for exynos5_clk_pdma1

Marek Szyprowski (1):
       ARM: EXYNOS: use s5p-timer for UniversalC210 board

  arch/arm/mach-exynos/Kconfig                 |    3 ++
  arch/arm/mach-exynos/clock-exynos5.c         |    2 +-
  arch/arm/mach-exynos/hotplug.c               |   39 
++++++++++++++++++++++++-
  arch/arm/mach-exynos/mach-universal_c210.c   |    4 ++-
  arch/arm/plat-samsung/include/plat/map-s3c.h |    2 +-
  5 files changed, 45 insertions(+), 5 deletions(-)

Comments

Olof Johansson May 12, 2012, 10:29 p.m. UTC | #1
Hi,


Since we are _so_ close to release, I am going to be picky about what
kind of patches we're going to accept for 3.4. See below.

On Sat, May 12, 2012 at 2:42 PM, Kukjin Kim <kgene.kim@samsung.com> wrote:


> Changhwan Youn (1):
>      ARM: EXYNOS: fix the hotplug for Cortex-A15

This is a feature, not a bug fix -- cpu hotplug has never worked for
A15 based on the patch description. So per definition it's 3.5
material.

Also, I'll send out a comment on the patch itself, it shouldn't check
for SoC version, it should check for core version. See separate email
shortly.

> Jose Miguel Goncalves (1):
>      ARM: SAMSUNG: Fix for S3C2412 EBI memory mapping

This has been broken since 2.6.32. Please queue it up in a fixes
branch for the 3.5 merge window instead, it doesn't qualify as a
recent regression and thus we shouldn't try to squeeze it in before
the merge window closes.

> Kukjin Kim (1):
>      ARM: EXYNOS: fix ctrlbit for exynos5_clk_pdma1
>
> Marek Szyprowski (1):
>      ARM: EXYNOS: use s5p-timer for UniversalC210 board

Those two look appropriate. Please send a new pull request with those two.


-Olof
Olof Johansson May 12, 2012, 10:48 p.m. UTC | #2
On Sat, May 12, 2012 at 4:45 PM, Kukjin Kim <kgene.kim@samsung.com> wrote:
> On 05/13/12 07:29, Olof Johansson wrote:
>>
>> Hi,
>>
>>
>> Since we are _so_ close to release, I am going to be picky about what
>> kind of patches we're going to accept for 3.4. See below.
>>
>> On Sat, May 12, 2012 at 2:42 PM, Kukjin Kim<kgene.kim@samsung.com>  wrote:
>>
>>
>>> Changhwan Youn (1):
>>>      ARM: EXYNOS: fix the hotplug for Cortex-A15
>>
>>
>> This is a feature, not a bug fix -- cpu hotplug has never worked for
>> A15 based on the patch description. So per definition it's 3.5
>> material.
>>
> Hmm, in this case, IMHO, it's hard to say whether it is not a bug fix
> because in current mach-exynos/ problem happens during cpu hotplug on
> exynos5250 without that...But, OK it will be queued for v3.5.

Agreed, it's a borderline case -- and earlier in the -rc cycle I would
have included this one without complaints.

>> Also, I'll send out a comment on the patch itself, it shouldn't check
>> for SoC version, it should check for core version. See separate email
>> shortly.
>>
> OK, thanks for your comments and will update it even though checking SoC
> version is enough in mach-exynos/ :)

Well, sort of -- you will need to add new checks when you add another
SoC with A15, and that's just messy. If you check for A15 instead,
then you won't have to modify that piece of code for every new SoC
(hopefully).

>>> Jose Miguel Goncalves (1):
>>>      ARM: SAMSUNG: Fix for S3C2412 EBI memory mapping
>>
>>
>> This has been broken since 2.6.32. Please queue it up in a fixes
>> branch for the 3.5 merge window instead, it doesn't qualify as a
>> recent regression and thus we shouldn't try to squeeze it in before
>> the merge window closes.
>>
> OK, I agree.
>
>
>>> Kukjin Kim (1):
>>>      ARM: EXYNOS: fix ctrlbit for exynos5_clk_pdma1
>>>
>>> Marek Szyprowski (1):
>>>      ARM: EXYNOS: use s5p-timer for UniversalC210 board
>>
>>
>> Those two look appropriate. Please send a new pull request with those two.
>>
> I sent just now.

Thanks for the quick turnaround!


-Olof
Kukjin Kim May 12, 2012, 11:45 p.m. UTC | #3
On 05/13/12 07:29, Olof Johansson wrote:
> Hi,
>
>
> Since we are _so_ close to release, I am going to be picky about what
> kind of patches we're going to accept for 3.4. See below.
>
> On Sat, May 12, 2012 at 2:42 PM, Kukjin Kim<kgene.kim@samsung.com>  wrote:
>
>
>> Changhwan Youn (1):
>>       ARM: EXYNOS: fix the hotplug for Cortex-A15
>
> This is a feature, not a bug fix -- cpu hotplug has never worked for
> A15 based on the patch description. So per definition it's 3.5
> material.
>
Hmm, in this case, IMHO, it's hard to say whether it is not a bug fix 
because in current mach-exynos/ problem happens during cpu hotplug on 
exynos5250 without that...But, OK it will be queued for v3.5.

> Also, I'll send out a comment on the patch itself, it shouldn't check
> for SoC version, it should check for core version. See separate email
> shortly.
>
OK, thanks for your comments and will update it even though checking SoC 
version is enough in mach-exynos/ :)

>> Jose Miguel Goncalves (1):
>>       ARM: SAMSUNG: Fix for S3C2412 EBI memory mapping
>
> This has been broken since 2.6.32. Please queue it up in a fixes
> branch for the 3.5 merge window instead, it doesn't qualify as a
> recent regression and thus we shouldn't try to squeeze it in before
> the merge window closes.
>
OK, I agree.

>> Kukjin Kim (1):
>>       ARM: EXYNOS: fix ctrlbit for exynos5_clk_pdma1
>>
>> Marek Szyprowski (1):
>>       ARM: EXYNOS: use s5p-timer for UniversalC210 board
>
> Those two look appropriate. Please send a new pull request with those two.
>
I sent just now.

Thanks.

Best regards,
Kgene.
--
Kukjin Kim <kgene.kim@samsung.com>, Senior Engineer,
SW Solution Development Team, Samsung Electronics Co., Ltd.