Re: [PATCH v2] binfmt_elf: Align eligible read-only PT_LOAD segments to PMD_SIZE for THP

[hev]

Hi,

I ran a quick benchmark on x86_64 as well.

Machine: AMD Ryzen 9 7950X
Binutils: 2.46
GCC: 15.2.1 (built with -z,noseparate-code + --enable-host-pie)

Workload: building Linux v7.0-rc1 with x86_64_defconfig.

Without patch:

 * instructions:     8,246,133,611,932
 * cpu-cycles:       8,001,028,142,928
 * itlb-misses:      3,672,158,331
 * time elapsed:     64.66 s

With patch:

 * instructions:     8,246,025,137,750
 * cpu-cycles:       7,565,925,107,502
 * itlb-misses:      26,821,242
 * time elapsed:     61.97 s

Instructions are basically unchanged. iTLB misses drop from ~3.67B to
~26M (~99.27% reduction), which results in about a ~5.44% reduction in
cycles and ~4.18% shorter wall time for this workload.

Thanks,
Rui