I wrote:
> Noah Misch <n...@leadboat.com> writes:
>> Apparently, valgrind-3.15.0 doesn't complain about undefined input
>> to _mm_crc32_u* functions. We should not be surprised if Valgrind gains the
>> features necessary to complain about the other implementations.
> Perhaps it already has ... I wonder if anyone's tried this on ARMv8
> lately.
I installed Fedora 32/aarch64 on a Raspberry Pi 3B+, and can report that
valgrind 3.16.0 is just as blind to this problem in pg_comp_crc32c_armv8
as it is in pg_comp_crc32c_sse42. Seems odd, but there you have it.
(There are some other issues, but they seem fit for separate threads.)
regards, tom lane
