[Bug backends/25832] New: aarch64 fails to compile with musl

[Bug backends/25832] New: aarch64 fails to compile with musl

[luizluca at gmail dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=25832

            Bug ID: 25832
           Summary: aarch64 fails to compile with musl
           Product: elfutils
           Version: unspecified
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: backends
          Assignee: unassigned at sourceware dot org
          Reporter: luizluca at gmail dot com
                CC: elfutils-devel at sourceware dot org
  Target Milestone: ---

Hello

gibc has user_fpsimd_struct.vregs as __uint128_t
https://github.com/bminor/glibc/blob/master/sysdeps/unix/sysv/linux/aarch64/sys/user.h#L32

while musl has it as long double:
https://github.com/AssemblyScript/musl/blob/master/arch/aarch64/bits/user.h#L9

While building 0.179 with musl, I got this error:


aarch64_initreg.c: In function 'aarch64_set_initial_registers_tid':
aarch64_initreg.c:85:37: error: invalid operands to binary & (have 'long
double' and 'unsigned int')
     dwarf_fregs[r] = fregs.vregs[r] & 0xFFFFFFFF;
                      ~~~~~~~~~~~~~~ ^
make[6]: *** [Makefile:682: aarch64_initreg.o] Error 1
make[5]: *** [Makefile:499: all-recursive] Error 1

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug backends/25832] aarch64 fails to compile with musl

[mark at klomp dot org]

https://sourceware.org/bugzilla/show_bug.cgi?id=25832

Mark Wielaard <mark at klomp dot org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mark at klomp dot org

--- Comment #1 from Mark Wielaard <mark at klomp dot org> ---
Looks like glibc is correct.
The linux uapi headers also declare arm64 vregs as an uint128_t array.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug backends/25832] aarch64 fails to compile with musl

[anarchy at gentoo dot org]

https://sourceware.org/bugzilla/show_bug.cgi?id=25832

Jory A. Pratt <anarchy at gentoo dot org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |anarchy at gentoo dot org

--- Comment #2 from Jory A. Pratt <anarchy at gentoo dot org> ---
Created attachment 12534
  --> https://sourceware.org/bugzilla/attachment.cgi?id=12534&action=edit
properly fix aarch64 support for both glibc/musl

Patch is self explanatory

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug backends/25832] aarch64 fails to compile with musl

[mark at klomp dot org]

https://sourceware.org/bugzilla/show_bug.cgi?id=25832

--- Comment #3 from Mark Wielaard <mark at klomp dot org> ---
(In reply to Jory A. Pratt from comment #2)
> Created attachment 12534 [details]
> properly fix aarch64 support for both glibc/musl
> 
> Patch is self explanatory

Could you for those of us who don't know why there is a difference between
glibc and musl here and don't know how __uint128_t and/or long double translate
into to a Dwarf_Word (which is a 64bit value)? I am not sure I understand how
the original code is correct. Shouldn't these values be represented by 2 dwarf
registers?

-- 
You are receiving this mail because:
You are on the CC list for the bug.