Re: [PATCH] Documentation/arm64: HugeTLB page implementation

[Punit Agrawal]

Hi Randy,

Randy Dunlap <rdunlap@xxxxxxxxxxxxx> writes:

> Hi,
> Just some minor stuff (below).
>
> On 10/5/18 7:34 AM, Punit Agrawal wrote:
>> Arm v8 architecture supports multiple page sizes - 4k, 16k and
>> 64k. Based on the active page size, the Linux port supports
>> corresponding hugepage sizes at PMD and PUD(4k only) levels.
>> 
>> In addition, the architecture also supports caching larger sized
>> ranges (composed of multiple entries) at the PTE and PMD level in the
>> TLBs using the contiguous bit. The Linux port makes use of this
>> architectural support to enable additional hugepage sizes.
>> 
>> Describe the two different types of hugepages supported by the arm64
>> kernel and the hugepage sizes enabled by each.
>> 
>> Signed-off-by: Punit Agrawal <punit.agrawal@xxxxxxx>
>> Cc: Catalin Marinas <catalin.marinas@xxxxxxx>
>> Cc: Will Deacon <will.deacon@xxxxxxx>
>> Cc: Jonathan Corbet <corbet@xxxxxxx>
>> ---
>>  Documentation/arm64/hugetlbpage.txt | 39 +++++++++++++++++++++++++++++
>>  1 file changed, 39 insertions(+)
>>  create mode 100644 Documentation/arm64/hugetlbpage.txt
>> 
>> diff --git a/Documentation/arm64/hugetlbpage.txt b/Documentation/arm64/hugetlbpage.txt
>> new file mode 100644
>> index 000000000000..64ee24b88d27
>> --- /dev/null
>> +++ b/Documentation/arm64/hugetlbpage.txt
>> @@ -0,0 +1,39 @@
>> +HugeTLBpage on ARM64
>> +====================
>> +
>> +Hugepage relies on making efficient use of TLBs to improve performance of
>> +address translations. The benefit depends on both -
>> +
>> +  - the size of hugepages
>> +  - size of entries supported by the TLBs
>> +
>> +The ARM64 port supports two flavours of hugepages.
>> +
>> +1) Block mappings at the pud/pmd level
>> +--------------------------------------
>> +
>> +These are regular hugepages where a pmd or a pud page table entry points to a
>> +block of memory. Regardless of the supported size of entries in TLB, block
>> +mappings reduces the depth of page table walk needed to translate hugepage
>
>             reduce
>
>> +addresses.
>> +
>> +2) Using the Contiguous bit
>> +---------------------------
>> +
>> +The architecture provides a contiguous bit in the translation table entries
>> +(D4.5.3, ARM DDI 0487C.a) that hints to the mmu to indicate that it is one of a
>
> preferably                                     MMU
>
>> +contiguous set of entries that can be cached in a single TLB entry.
>> +
>> +The contiguous bit is used in Linux to increase the mapping size at the pmd and
>> +pte (last) level. The number of supported contiguous entries vary by page size
>
>                                                                 varies
>
>> +and level of the page table.
>> +
>> +
>> +
>> +The following hugepage sizes are supported -
>> +
>> +         CONT PTE    PMD    CONT PMD    PUD
>> +         --------    ---    --------    ---
>> +  4K:         64K     2M         32M     1G
>> +  16K:         2M    32M          1G
>> +  64K:         2M   512M         16G
>> 

I've updated the patch with incorporating your feedback and will post an
update shortly.

Thanks a lot for taking a look.

Punit