This patch enables building compiler-rt builtins for ARM targets that only support single-precision floating point instructions (e.g., those with -mfpu=fpv4-sp-d16).
This fixes PR42838.
Differential D90698
[builtins] Add support for single-precision-only-FPU ARM targets. chaosdefinition on Nov 3 2020, 11:12 AM. Authored by
Details This patch enables building compiler-rt builtins for ARM targets that only support single-precision floating point instructions (e.g., those with -mfpu=fpv4-sp-d16). This fixes PR42838.
Diff Detail
Event TimelineComment Actions This looks good to me. The __ARM_FP feature test macro is defined in the ACLE and 0x8 is the right flag to check for Double precision support. I've tested at least one single precision only build. I suggest waiting a day or two before committing to see if anyone else has any objections or alternative implementation suggestions. Comment Actions If there are no more comments, can someone please commit this for me? I don't have commit access to LLVM. Comment Actions That's a little complicated for me as I need to go through various internal company procedures to commit something I didn't write. I can do this but it will take time, if there are any others that can do it faster then please do. Comment Actions Assuming there are no problems with the bots or other downstream builds, please update the PR with a link to this review. Comment Actions Thank you for taking time doing this! I'll consider getting myself commit access for future patches. |