[PATCH v11 0/3] riscv, mm: detect svnapot cpu support at runtime

[Qinglin Pan]

Hi Palmer,

On 2/15/23 8:02 AM, Palmer Dabbelt wrote:
> On Thu, 15 Dec 2022 22:21:06 PST (-0800), panqinglin2020 at iscas.ac.cn wrote:
>> From: Qinglin Pan <panqinglin2020 at iscas.ac.cn>
>>
>> Svnapot is a RISC-V extension for marking contiguous 4K pages as a non-4K
>> page. This patch set is for using Svnapot in hugetlb fs and huge vmap.
>>
>> This patchset adds a Kconfig item for using Svnapot in
>> "Platform type"->"SVNAPOT extension support". Its default value is on,
>> and people can set it off if they don't allow kernel to detect Svnapot
>> hardware support and leverage it.
>>
>> Tested on:
>>   - qemu rv64 with "Svnapot support" off and svnapot=true.
>>   - qemu rv64 with "Svnapot support" on and svnapot=true.
>>   - qemu rv64 with "Svnapot support" off and svnapot=false.
>>   - qemu rv64 with "Svnapot support" on and svnapot=false.
>>
>>
>> Changes in v2:
>>   - detect Svnapot hardware support at boot time.
>> Changes in v3:
>>   - do linear mapping again if has_svnapot
>> Changes in v4:
>>   - fix some errors/warns reported by checkpatch.pl, thanks @Conor
>> Changes in v5:
>>   - modify code according to @Conor and @Heiko
>> Changes in v6:
>>   - use static key insead of alternative errata
>> Changes in v7:
>>   - add napot_cont_order for possible more napot order in the future
>>   - remove linear mapping related code from this patchset to another 
>> patch
>>   - refactor hugetlb code for svnapot according to thanks @Andrew @Conor
>>   - use tools/testing/selftests/vm/map_hugetlb as hugetlb testing program
>>   - support svnapot in huge vmap on newer for-next branch
>> Changes in v8:
>>   - fix compilation errors in rv32_defconfig
>>   - insert some lines of whitespace according to @Conor's suggestion
>> Changes in v9:
>>   - use alternative to avoid using static branches inside heavily used
>>     inline functions
>>   - change napot_cont_mask definition
>>   - post test_vmalloc modification about testing vmalloc_huge
>> Changes in v10:
>>   - fix some nits caught by @Andrew
>>   - collect Reviewed-by/Acked-by
>>   - add memory leak warning in KConfig text
>>   - replace test_vmalloc patch link with the standard one
>> Changes in v11:
>>   - add more detailed warning about HUGETLBFS and SVNAPOT in KConfig text
>>   - fix missing reverse-xmas tree
>>
>>
>> Qinglin Pan (3):
>>   riscv: mm: modify pte format for Svnapot
>>   riscv: mm: support Svnapot in hugetlb page
>>   riscv: mm: support Svnapot in huge vmap
>>
>>  arch/riscv/Kconfig                   |  21 +-
>>  arch/riscv/include/asm/errata_list.h |  15 +-
>>  arch/riscv/include/asm/hugetlb.h     |  34 ++-
>>  arch/riscv/include/asm/hwcap.h       |   2 +-
>>  arch/riscv/include/asm/page.h        |   5 -
>>  arch/riscv/include/asm/pgtable-64.h  |  34 +++
>>  arch/riscv/include/asm/pgtable.h     |  43 +++-
>>  arch/riscv/include/asm/vmalloc.h     |  61 +++++-
>>  arch/riscv/kernel/cpu.c              |   1 +
>>  arch/riscv/kernel/cpufeature.c       |  15 ++
>>  arch/riscv/mm/hugetlbpage.c          | 301 +++++++++++++++++++++++++++
>>  11 files changed, 519 insertions(+), 13 deletions(-)
> 
> This is triggering some build failures.  I sent along a fix before 
> realizing they're actually coming from this patch set: 
> https://lore.kernel.org/all/20230214201358.10647-1-palmer@rivosinc.com/ 
> .  If it's OK I'll just squash that in here, so I can avoid a broken build.

There is a newer v13(in [1]) based on new ISA extension API. But it 
still have conchuod/build_rv64_gcc_allmodconfig check failed in it[2].
I have no idea whether it is caused by my patchset since the output 
seems not related to my code.. Please let me know if there is anything I
need to modify.


[1]: 
https://lore.kernel.org/linux-riscv/20230209131647.17245-1-panqinglin00@gmail.com/
[2]: 
https://patchwork.kernel.org/project/linux-riscv/patch/20230209131647.17245-4-panqinglin00@gmail.com/
[3]: https://gist.github.com/conor-pwbot/0a1850594abdf3aebe4b3026a52a682d