Gelbpunkt wrote:
Ping, is there anything I can do to help this get reviewed?
https://github.com/llvm/llvm-project/pull/149725
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
alexrp wrote:
> Fair point, ideally I'd like both the "proper" ELFv2 with BE to have its
> target triple and the "loosened" ELFv2 to have one. Currently, LLVM will
> enable the VSX and POWER8-specific parts of the ABI based on the endianess or
> CPU, not the ABI. That should probably be change
Gelbpunkt wrote:
> This one claims to support old processors too, so seems like also a case of
> "loosened" ELFv2?
Correct, it isn't strictly ELFv2.
> I'm just concerned about what downstream projects seem to be doing with the
> term `elfv2`...
Fair point, ideally I'd like both the "proper"
alexrp wrote:
> Arch Linux POWER
This one claims to support old processors too, so seems like also a case of
"loosened" ELFv2?
> Also, this isn't exactly a fix limited to glibc target triples. It fixes the
> data layout on any target triple that defaults to ELFv1 when `-mabi=elfv2` is
> spec
Gelbpunkt wrote:
> Is this no longer the case?
The page is probably correct in so far that glibc doesn't explicitly support
it. However, out of the 64-bit big endian Linux distributions *using* glibc
that I'm aware of (Arch Linux POWER, Void Linux for PowerPC, Gentoo), the first
two use ELFv2
alexrp wrote:
The only Linux distro I can find that actually claims support for `powerpc64` +
`elfv2` + glibc is [Void](https://voidlinux-ppc.org):
> Void-ppc uses a variant of the modern ELFv2 ABI (loosened not to impose the
> POWER8/VSX requirements of the specification) on 64-bit glibc, as
alexrp wrote:
From https://sourceware.org/glibc/wiki/ABIList#powerpc (emphasis mine):
> (The GCC distinction for 64-bit is actually ELFv1/ELFv2, **but the BE ELFv2
> and LE ELFv1 combinations aren't supported**.)
Is this no longer the case?
https://github.com/llvm/llvm-project/pull/149725
___
llvmbot wrote:
@llvm/pr-subscribers-backend-powerpc
Author: Jens Reidel (Gelbpunkt)
Changes
Prior to this change, the data layout calculation would not account for
explicitly set `-mabi=elfv2` on `powerpc64-unknown-linux-gnu`, a target that
defaults to `elfv1`.
This is loosely inspired
llvmbot wrote:
@llvm/pr-subscribers-clang
Author: Jens Reidel (Gelbpunkt)
Changes
Prior to this change, the data layout calculation would not account for
explicitly set `-mabi=elfv2` on `powerpc64-unknown-linux-gnu`, a target that
defaults to `elfv1`.
This is loosely inspired by the equ
https://github.com/Gelbpunkt created
https://github.com/llvm/llvm-project/pull/149725
Prior to this change, the data layout calculation would not account for
explicitly set `-mabi=elfv2` on `powerpc64-unknown-linux-gnu`, a target that
defaults to `elfv1`.
This is loosely inspired by the equiv
10 matches
Mail list logo
