Xi Ruoyao
Xi Ruoyao
> cc @MaskRay @xry111 > > Unfortunately I can't allocate time for this at present. I don't understand linking process (especially relaxation) very well but I think Fangrui is an...
> With hindsight this might actually be a hardware issue... not sure if this will get fixed in the next revision or confirmed unfixable. I'm not really sure too. Now...
> Is this a grub bug? Is it related to LoongArch? With GRUB 2.06 for x86_64 EFI `(hd0)` is also not removed in `ls` output, but it does not crash....
> * 新增 `*.n`、`*.nu` 操作符宽度后缀，用来表示与当前汇编器所 target 的 `march` 的 GRLen 相同的宽度。 我觉得根据 `-mabi=` 的值更合理，因为我们进行这种操作的时候一般是操作指针(地址)，而不是真的关心硬件寄存器多宽。 那么后缀用 `*.p` 似乎更好？
I don't think using "previously" here is correct. This option makes the building system build target libraries with GCC pass 1, instead of the "just built" GCC pass 2 (if...
I don't think it's the "correct" approach. We should define `MULTILIB_DEFAULTS` for this purpose, instead of introduce those hacks. More serious problem is that the hacks may affect all targets,...
> 2. Making multilib directory layout symmetric (and constant across compilers with different default ABI). It will still be symmetric and constant using `MULTILIB_DEFAULT`. `MULTILIB_DEFAULT` has no effect on `-print-multi-os-directory`....
要不直接丢到 [email protected] 去？ 一些注意事项：https://gcc.gnu.org/contribute.html 我写的某些 testcase 里面有 `scan-assembler "\\\$r4"` 这种东西，如果要默认生成 `$a0` 代替 `$r4` 的话也得改掉 :)
On LoongArch 64-bit the same failures happen too.