Vineet Gupta comments

Results 22 comments of


Vineet Gupta

__atomic_compare_exchange_n generates extra cmp instruction

well cmp + bnz is equivalent to a BRNE, what is the issue. It seems the overall code size of 2 snippets is the same, I don't use the BNZ_S...

__atomic_compare_exchange_n generates extra cmp instruction

Right, so the compiler guys can indeed optimize the cmp_s + bnz_s with a single BRNE. compilers tend to get a bit conservative since the branhc can technically get out...

[arc64] elf gcc not generating call frame info

Stack unwinding works now !

[arc64] elf gcc not generating call frame info

After the original fix, dwarf info is now being inhibited (elf toolchain) so we are back to square one. I understand this requires some newlib updates but ARCv2 elf toolchain...

ARCv2: objdump incorrectly disassembling jump tables as random instructions (including floating point)

[issue-385.tar.gz](https://github.com/foss-for-synopsys-dwc-arc-processors/toolchain/files/6651786/issue-385.tar.gz)

[gas] gas need to relax ST a, [b, off] into ST.as a, [b, off/8]

@shahab-vahedi what is intrusive about using address scaling ? It should be default w/o any flag whatsoever.

[gas] gas need to relax ST a, [b, off] into ST.as a, [b, off/8]

> The assembly instructions should not change on the fly, be that may for optimization reasons. Why is that. Isn't assembler relaxation supposed to do that - transform instructions to...

Document %U GNU In-line Assembly modifier

FWIW, for ARC64 this is not a documentation issue - but feature missing issue ;-)

[LD] Erroneous R_ARC_NONE in Relocation

FWIW, there's STAR : 9001086764- build linker to prune extraneous R_ARC_NONE dynamic relos

ARC64: fp-int-convert-timode-* fails in Linux

The linker warning is benign - for soft-float builds we always get them (even in ARCv2) and still have no extraneous failures in ARCv2 or ARCv3. You need to triage...