[v1] mm/page_alloc: gfp flags cleanups for alloc_contig_*()

[PATCH v1 0/6] mm/page_alloc: gfp flags cleanups for alloc_contig_*()

Posted by David Hildenbrand 1 year, 2 months ago

Let's clean up the gfp flags handling, and support __GFP_ZERO, such that we
can finally remove the TODO in memtrace code.

I did some alloc_contig_*() testing with virtio-mem and hugetlb; I did not
test powernv/memtrace -- I cross-compiled it, though.

Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Zi Yan <ziy@nvidia.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Nicholas Piggin <npiggin@gmail.com>
Cc: Christophe Leroy <christophe.leroy@csgroup.eu>
Cc: Naveen N Rao <naveen@kernel.org>
Cc: Madhavan Srinivasan <maddy@linux.ibm.com>

David Hildenbrand (6):
  mm/page_isolation: don't pass gfp flags to isolate_single_pageblock()
  mm/page_isolation: don't pass gfp flags to start_isolate_page_range()
  mm/page_alloc: make __alloc_contig_migrate_range() static
  mm/page_alloc: sort out the alloc_contig_range() gfp flags mess
  mm/page_alloc: forward the gfp flags from alloc_contig_range() to
    post_alloc_hook()
  powernv/memtrace: use __GFP_ZERO with alloc_contig_pages()

 arch/powerpc/platforms/powernv/memtrace.c | 31 +++---------
 include/linux/page-isolation.h            |  2 +-
 mm/internal.h                             |  4 --
 mm/memory_hotplug.c                       |  3 +-
 mm/page_alloc.c                           | 62 +++++++++++++++++++----
 mm/page_isolation.c                       | 12 ++---
 6 files changed, 63 insertions(+), 51 deletions(-)

-- 
2.47.1

Re: [PATCH v1 0/6] mm/page_alloc: gfp flags cleanups for alloc_contig_*()

Posted by Zi Yan 1 year, 2 months ago

On 2 Dec 2024, at 7:58, David Hildenbrand wrote:

> Let's clean up the gfp flags handling, and support __GFP_ZERO, such that we
> can finally remove the TODO in memtrace code.
>
> I did some alloc_contig_*() testing with virtio-mem and hugetlb; I did not
> test powernv/memtrace -- I cross-compiled it, though.
>
> Cc: Andrew Morton <akpm@linux-foundation.org>
> Cc: Oscar Salvador <osalvador@suse.de>
> Cc: Zi Yan <ziy@nvidia.com>
> Cc: Michael Ellerman <mpe@ellerman.id.au>
> Cc: Nicholas Piggin <npiggin@gmail.com>
> Cc: Christophe Leroy <christophe.leroy@csgroup.eu>
> Cc: Naveen N Rao <naveen@kernel.org>
> Cc: Madhavan Srinivasan <maddy@linux.ibm.com>

FYI, linux-mm does not get any of your emails: https://lore.kernel.org/linux-mm/ECFA727B-F542-42E5-BE32-F9FB27F5DCDB@nvidia.com/, but linux-kernel has them.

Best Regards,
Yan, Zi

Re: [PATCH v1 0/6] mm/page_alloc: gfp flags cleanups for alloc_contig_*()

Posted by David Hildenbrand 1 year, 2 months ago

On 02.12.24 23:07, Zi Yan wrote:
> On 2 Dec 2024, at 7:58, David Hildenbrand wrote:
> 
>> Let's clean up the gfp flags handling, and support __GFP_ZERO, such that we
>> can finally remove the TODO in memtrace code.
>>
>> I did some alloc_contig_*() testing with virtio-mem and hugetlb; I did not
>> test powernv/memtrace -- I cross-compiled it, though.
>>
>> Cc: Andrew Morton <akpm@linux-foundation.org>
>> Cc: Oscar Salvador <osalvador@suse.de>
>> Cc: Zi Yan <ziy@nvidia.com>
>> Cc: Michael Ellerman <mpe@ellerman.id.au>
>> Cc: Nicholas Piggin <npiggin@gmail.com>
>> Cc: Christophe Leroy <christophe.leroy@csgroup.eu>
>> Cc: Naveen N Rao <naveen@kernel.org>
>> Cc: Madhavan Srinivasan <maddy@linux.ibm.com>
> 
> FYI, linux-mm does not get any of your emails: https://lore.kernel.org/linux-mm/ECFA727B-F542-42E5-BE32-F9FB27F5DCDB@nvidia.com/, but linux-kernel has them.

Indeed, linuxppc seems to have them as well.

Guess I'll have to blame it on linux-mm, then ;)

Let me resend with #5 patch description adjusted. Thanks for the review!

-- 
Cheers,

David / dhildenb