Rebased on top of Linux 5.18-rc1

This is the mm part of the series that converts powerpc to default
topdown mmap layout, for merge into v5.18

powerpc requires its own arch_get_unmapped_area() only when
slices are needed, which is only for book3s/64. First part of
the series moves slices into book3s/64 specific directories
and cleans up other subarchitectures.

The actual convertion of powerpc to default topdown mmap layout will
then be resent in a follow-up series for application on v5.19

First patch modifies the core mm to allow powerpc to still provide its
own arch_randomize_brk()

Second patch modifies core mm to give len and flags to
arch_get_mmap_end() as powerpc needs it. 

Third patch modifies core mm to allow powerpc to use generic versions
of get_unmapped_area functions for Radix while still providing its own
implementation for Hash, the selection between Radix and Hash being
done at runtime.

Fourth patch is a complement/fix of f6795053dac8 ("mm: mmap: Allow for
"high" userspace addresses") for hugetlb. It adds support for "high"
userspace addresses that are optionally supported on the system and
have to be requested via a hint mechanism ("high" addr parameter to mmap).

Previous version of the series is available at
https://patchwork.ozlabs.org/project/linuxppc-dev/list/?state=*&series=289718

Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>

Changes in v9:
- Sending mm patches and powerpc patches separately with the objective
of mm patches going into kernel v5.18 and powerpc patches folling up
into kernel v5.19

Changes in v8:
- Moved patch "sizes.h: Add SZ_1T macro" up from which is already in linux-next but not in Linus tree yet.
- Rebased on today's powerpc/next

Changes in v7:
- Taken into account comments from Catalin (patches 3 and 4)

Changes in v6:
- New patch (patch 4) to take arch_get_mmap_base() and arch_get_mmap_end() into account in generic hugetlb_get_unmapped_area()
- Get back arch_randomize_brk() simplification as it relies on default topdown mmap layout.
- Fixed precedence between || and && in powerpc's arch_get_mmap_end() (patch 9)

Changes in v5:
- Added patch 3
- Added arch_get_mmap_base() and arch_get_mmap_end() to patch 7 to better match original powerpc behaviour
- Switched patched 10 and 11 and performed full randomisation in patch 10 just before switching to default implementation, as suggested by Nic.

Changes in v4:
- Move arch_randomize_brk() simplification out of this series
- Add a change to core mm to enable using generic implementation
while providing arch specific one at the same time.
- Reworked radix get_unmapped_area to use generic implementation
- Rebase on top of Nic's series v6

Changes in v3:
- Fixed missing <linux/elf-randomize.h> in last patch
- Added a patch to move SZ_1T out of drivers/pci/controller/pci-xgene.c

Changes in v2:
- Moved patch 4 before patch 2
- Make generic arch_randomize_brk() __weak
- Added patch 9

Christophe Leroy (4):
  mm: Allow arch specific arch_randomize_brk() with
    CONFIG_ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT
  mm, hugetlbfs: Allow an arch to always use generic versions of
    get_unmapped_area functions
  mm: Add len and flags parameters to arch_get_mmap_end()
  mm, hugetlbfs: Allow for "high" userspace addresses

 arch/arm64/include/asm/processor.h |  4 +--
 fs/hugetlbfs/inode.c               | 26 ++++++++++++------
 include/linux/hugetlb.h            |  5 ++++
 include/linux/sched/mm.h           | 17 ++++++++++++
 mm/mmap.c                          | 43 ++++++++++++++++++------------
 mm/util.c                          |  2 +-
 6 files changed, 69 insertions(+), 28 deletions(-)

-- 
2.35.1

On Fri,  8 Apr 2022 09:24:58 +0200 Christophe Leroy <christophe.leroy@csgroup.eu> wrote:

> Rebased on top of Linux 5.18-rc1
> 
> This is the mm part of the series that converts powerpc to default
> topdown mmap layout, for merge into v5.18

We're at 5.18-rc1.  The 5.18 merge window has closed and we're in
fixes-only mode.

If there's a case to be made that these patches are needed by 5.18
users then please let's make that case.  Otherwise, this is 5.19-rc1 material.

And if it is indeed all 5.19-rc1 material, then please carry all four
in the powerpc tree with Acked-by: Andrew Morton
<akpm@linux-foundation.org>.

Also, [4/4] has a cc:stable.  This is a bit odd because -stable
candidates should be standalone patches, staged ahead of all
for-next-merge-window material, so we can get them merged up quickly.

More oddly, [4/4]'s changelog provides no explanation for why the patch
should be considered for backporting.

Le 09/04/2022 à 05:25, Andrew Morton a écrit :
> On Fri,  8 Apr 2022 09:24:58 +0200 Christophe Leroy <christophe.leroy@csgroup.eu> wrote:
> 
>> Rebased on top of Linux 5.18-rc1
>>
>> This is the mm part of the series that converts powerpc to default
>> topdown mmap layout, for merge into v5.18
> 
> We're at 5.18-rc1.  The 5.18 merge window has closed and we're in
> fixes-only mode.

Umm ... There must have been a misunderstanding then.

I contacted you before the merge window, and your answer was:

Le 11/03/2022 à 05:49, Andrew Morton a écrit :
>
> 5.18 isn't a problem.  Perhaps you meant 5.17, which would be real tough.
>
> Can we take a look after 5.18-rc1?


> 
> If there's a case to be made that these patches are needed by 5.18
> users then please let's make that case.  Otherwise, this is 5.19-rc1 material.

It's not really needed for 5.18. The idea was to merge that common part 
in 5.18 in order to minimise risks on conflicts. As far as I understand 
it often happens that changes of that kind get merged at the very end of 
the merge window or between rc1 and rc2. I was therefore not surprised 
that you offered to handle it past rc1.

History at:
https://patchwork.ozlabs.org/project/linuxppc-dev/cover/cover.1646847561.git.christophe.leroy@csgroup.eu/#2856080


> 
> And if it is indeed all 5.19-rc1 material, then please carry all four
> in the powerpc tree with Acked-by: Andrew Morton
> <akpm@linux-foundation.org>.

Well, Michael was a bit unconfortable with doing it that way, see below:


Le 11/03/2022 à 05:26, Michael Ellerman a écrit :
 >
 > Yeah I didn't pick it up because the mm changes don't have many acks and
 > I'm always nervous about carrying generic mm changes.
 >
 > It would be my preference if Andrew could take 2-5 through mm for v5.18,
 > but it is quite late, so I'm not sure how he will feel about that.
 >
 > Arguably 2, 3, 4 do very little. It's only patch 5 that has much effect,
 > and it has a reviewed-by from Catalin at least.

Michael, is it now ok for you to merge it via powerpc tree with Andrew's 
Ack ?

> 
> Also, [4/4] has a cc:stable.  This is a bit odd because -stable
> candidates should be standalone patches, staged ahead of all
> for-next-merge-window material, so we can get them merged up quickly.
> 
> More oddly, [4/4]'s changelog provides no explanation for why the patch
> should be considered for backporting.
> 

That was a request from Catalin from ARM64:

Le 04/01/2022 à 17:21, Catalin Marinas a écrit :
 > I wonder whether we should add a fixes tag (or at least the cc stable):
 >
 > Fixes: f6795053dac8 ("mm: mmap: Allow for "high" userspace addresses")
 > Cc: <stable@vger.kernel.org> # 5.0.x
 >
 > I think the original commit should have changed
 > hugetlb_get_unmapped_area() to have the same behaviour as
 > arch_get_unmapped_area(). Steve, any thoughts?
 >
 > FWIW,
 >
 > Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>

 From 
https://patchwork.ozlabs.org/project/linuxppc-dev/patch/db238c1ca2d46e33c57328f8d450f2563e92f8c2.1639736449.git.christophe.leroy@csgroup.eu/

I can try and see whether this can be moved in front of the other patches.

Thanks
Christophe

Christophe Leroy <christophe.leroy@csgroup.eu> writes:
> Le 09/04/2022 à 05:25, Andrew Morton a écrit :
>> On Fri,  8 Apr 2022 09:24:58 +0200 Christophe Leroy <christophe.leroy@csgroup.eu> wrote:
>> 
>>> Rebased on top of Linux 5.18-rc1
>>>
>>> This is the mm part of the series that converts powerpc to default
>>> topdown mmap layout, for merge into v5.18
>> 
>> We're at 5.18-rc1.  The 5.18 merge window has closed and we're in
>> fixes-only mode.
>
> Umm ... There must have been a misunderstanding then.

That's probably my fault for not getting back to Andrew.

> Le 11/03/2022 à 05:26, Michael Ellerman a écrit :
>  >
>  > Yeah I didn't pick it up because the mm changes don't have many acks and
>  > I'm always nervous about carrying generic mm changes.
>  >
>  > It would be my preference if Andrew could take 2-5 through mm for v5.18,
>  > but it is quite late, so I'm not sure how he will feel about that.
>  >
>  > Arguably 2, 3, 4 do very little. It's only patch 5 that has much effect,
>  > and it has a reviewed-by from Catalin at least.
>
> Michael, is it now ok for you to merge it via powerpc tree with Andrew's 
> Ack ?

Yes.

>> Also, [4/4] has a cc:stable.  This is a bit odd because -stable
>> candidates should be standalone patches, staged ahead of all
>> for-next-merge-window material, so we can get them merged up quickly.
>> 
>> More oddly, [4/4]'s changelog provides no explanation for why the patch
>> should be considered for backporting.
 
Yeah it's just a bit too politely worded :)

It says it's "a complement of f6795053dac8", but it's actually a fix for
a bug in that commit, that commit should have updated hugetlb behaviour.

> That was a request from Catalin from ARM64:
>
> Le 04/01/2022 à 17:21, Catalin Marinas a écrit :
>  > I wonder whether we should add a fixes tag (or at least the cc stable):
>  >
>  > Fixes: f6795053dac8 ("mm: mmap: Allow for "high" userspace addresses")
>  > Cc: <stable@vger.kernel.org> # 5.0.x
>  >
>  > I think the original commit should have changed
>  > hugetlb_get_unmapped_area() to have the same behaviour as
>  > arch_get_unmapped_area(). Steve, any thoughts?
>  >
>  > FWIW,
>  >
>  > Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
>
>  From 
> https://patchwork.ozlabs.org/project/linuxppc-dev/patch/db238c1ca2d46e33c57328f8d450f2563e92f8c2.1639736449.git.christophe.leroy@csgroup.eu/
>
> I can try and see whether this can be moved in front of the other patches.

Thanks, that would be preferable.

cheers