`__builtin_memcpy` for large blocks - #596

Open

xavierleroy wants to merge 9 commits into

big-memcpy-fixes

Commits on Sep 14, 2026

lib/Camlcoq: export Z.{shl,shr,logand,logor,logxor}
xavierleroy
committed
4dac4a4
x86_64 `__builtin_memcpy`: handle the case `sz` >= 2^34
xavierleroy
committed
51f035a
x86_64 `__builtin_memcpy`: use `rep movsb` for large blocks
xavierleroy
committed
630b4f4
AArch64 `__builtin_memcpy`: handle the case `sz` >= 2^36
xavierleroy
committed
da819bf
RISC-V `__builtin_memcpy`: handle the case `sz` >= 2^32 * chunksize
xavierleroy
committed
6221478
ARM `__builtin_memcpy`: handle the case `sz` >= 2^30
xavierleroy
committed
17347a0
PowerPC `__builtin_memcpy`: handle the case `sz` >= 2^30
xavierleroy
committed
eabf245
Update the test suite with a test for __builtin_memcpy_aligned
xavierleroy
committed
0af5f48
`__builtin_memcpy_aligned`: clamp the alignment to at most 8
xavierleroy
committed
2f28fd5