Patches for gcc 8

Patches for gcc 8

[Nagakamira]

[-- Attachment #1: Type: text/plain, Size: 133 bytes --]

Hi everyone! I remarked that there is no patches for gcc 8.1 and 8.2. Maybe
there is problems with gcc 8.x?

Best wishes,
protonesso

[-- Attachment #2: Type: text/html, Size: 222 bytes --]

Re: Patches for gcc 8

[Rich Felker]

On Wed, Aug 15, 2018 at 09:11:39PM +0300, Nagakamira wrote:
> Hi everyone! I remarked that there is no patches for gcc 8.1 and 8.2. Maybe
> there is problems with gcc 8.x?

At first gcc 8 was rumored to have significant bugs, but if so they
should be fixed by now. I'm not aware of anyone having prepared
patches yet though.

Rich

Re: Patches for gcc 8

[Sebastian Gottschall]

There are no patches required. i'm using musl with GCC 8.2.0 right now 
and can't find any issues on x86, x64, arm and mips

Sebastian

Am 17.08.2018 um 01:52 schrieb Rich Felker:
> On Wed, Aug 15, 2018 at 09:11:39PM +0300, Nagakamira wrote:
>> Hi everyone! I remarked that there is no patches for gcc 8.1 and 8.2. Maybe
>> there is problems with gcc 8.x?
> At first gcc 8 was rumored to have significant bugs, but if so they
> should be fixed by now. I'm not aware of anyone having prepared
> patches yet though.
>
> Rich
>

Re: Patches for gcc 8

[Szabolcs Nagy]

* Sebastian Gottschall <s.gottschall@dd-wrt.com> [2018-08-17 08:30:36 +0200]:
> There are no patches required. i'm using musl with GCC 8.2.0 right now and
> can't find any issues on x86, x64, arm and mips

the s390, m68k, j2 patches are not upstream (yet)

there are some minor issues with upstream gcc, the most
common one is probably -fstack-protector on i386 and ppc
fails to link a __stack_chk_fail definition (glibc uses
a linker script for this to add libssp_nonshared.a),

and -static -pie produced position independent executable
on musl toolchains, but in gcc 8 you need -static-pie
for that, otherwise it's just normal -static executable
(this is a problem if you are building a distro that
tries to make all executables pie).

(the mcm repo also has patches to improve cross libc
testing so make check in the gcc build directory works
if you built a musl toolchain on a glibc system)

SO_PEERSEC is wrong on MIPS / MIPS64 Systems

[Sebastian Gottschall]

musl defined SO_PEERSEC as 31, but on MIPS and MIPS64 systems it must be 30
and of course this isnt the only wrong one. see SO_SNDBUFFORCE 32 vs 31. 
havent checked them all

Sebastian

Re: SO_PEERSEC is wrong on MIPS / MIPS64 Systems

[Sebastian Gottschall]

forget this email. i have seen its already fixed

Am 19.08.2018 um 20:11 schrieb Sebastian Gottschall:
> musl defined SO_PEERSEC as 31, but on MIPS and MIPS64 systems it must 
> be 30
> and of course this isnt the only wrong one. see SO_SNDBUFFORCE 32 vs 
> 31. havent checked them all
>
> Sebastian
>
>