On 22/10/2024 17:29, Tobias Burnus wrote:
Andrew Stubbs wrote:
I'm going to push the base patch shortly.
… which happened in commit r15-4540-ga6b26e5ea09779.
Updated patch attached.
Some more testing showed that there was an issue with the builtin
defines, which has been fixed and those have been extended as well.
In particular, it now also includes the __GFX11__ macro define as
architecture family.
Thanks, this looks OK now.
Tobias
PS: I find it surprising which out-of-bounds writes remain undetected on
one machine and show up prominently on the other.
The sizeof/strlen thing? I never saw any trouble from that. :(
PPS: If more macros are desired, clang also defines
__AMDGCN_WAVEFRONT_SIZE__ (i.e. 32 or 64) and __AMDGCN_CUMODE__ (1 = CU
mode, 0 = WGP).
I don't know what these are for, but if was want to have them then I
suppose it's harmless.
Andrew
