Re: stable 1.1.9 & current GIT broken on mips

[Sebastian Gottschall <s.gottschall@dd-wrt.com>]

Am 02.06.2015 um 19:19 schrieb Rich Felker:
> On Tue, Jun 02, 2015 at 05:57:23PM +0200, Sebastian Gottschall wrote:
>> Hello
>>
>> i tested today the current 1.1.9 (and later also current git so see
>> if its the same behaviour)
>> build on a mipsr2 big endian target (atheros ar7xxx) based on my
>> dd-wrt firmware.
>> i found out that mips seems to be broken on musl right now. the
>> behaviour is
>> that a call using execvp will not result in calling the desired
>> application.
>> on a second call and a following return call, the userspace will
>> lock up with no way todo anything anymore.
>>
>> for testing i used the same compiler, just recompiled musl with
>> version 1.1.8 and overwrote the libc library and crt stuff. the
>> result was, that it worked
>> again without any issue
>>
>> it needs to be considered that parts of the code is compiled using
>> -mips16 and -minterlink-mips16 if that matters
> I suspect the regression is related to your use of mips16, which I had
> no idea worked at all with musl and is not something that's officially
> supported. Did you build musl itself as mips16? I don't think this is
> going to work because we have tail calls back and forth between C and
> asm code where the asm is written in normal 32-bit mips instructions.
> There may also be issues with the dynamic linker start code if
> crt_arch.h is assembled in mips16 mode.
no musl itself is build without mips16. i'm aware that this will likelly 
not work. only the applications are partially compiled with mips16 (some 
libs like openssl arent compiled with mips16 for performance reasons)
> Could you provide details on your setup, specifically, which
> components are being compiled as mips16?
all libs and applications which are affected here in this testcase are 
compiled with mips16, except for musl

Sebastian
>
> Rich
>