Re: [Tinycc-devel] i386 build of gawk with tcc now broken

[Thomas Preud'homme]

Le dimanche 8 septembre 2013 17:13:51 Ramsay Jones a écrit :
> On 08/09/13 08:00, address@hidden wrote:
> >>> After pulling the latest on the mob branch, builds of gawk master using
> >>> tcc generate lots of test failures.
> >>> 
> >>> On an i386 linux system:
> >>>   git clone git://git.savannah.gnu.org/gawk.git
> >>>   cd gawk
> >>>   ./bootstrap.sh && configure CC=tcc && make
> >>>   make check      # watch lots of things go kablooey
> >>> 
> >>> This does not occur on x86-64.
> >> 
> >> Does this also happen with tcc 0.9.26 and tcc 0.9.25 or just with the mob
> >> branch?
> > 
> > I'm pretty sure just the mob branch. I can double check tomorrow
> > when I'm at the machine. Almost all the errors are related to floating
> > point exceptions; I strongly suspect it has to do with the most recent
> > changes.
> > 
> > The recipe above will let you reproduce it a lot quicker than waiting
> > for me...
> > 
> > 0.9.25 won't be worth checking as it's tool old, FWIW.
> 
> I have a patch to fix the tests, which fail for me with a Floating
> point exception:
> 
>     address@hidden $ make test
> 
>     [snip]
> 
>     ------------ test3 ------------
>     ../tcc -B.. -I.. -I.. -I../include
> -DCONFIG_LDDIR="\"lib/i386-linux-gnu\""
> -DCONFIG_MULTIARCHDIR="\"i386-linux-gnu\"" -DTCC_TARGET_I386 -DONE_SOURCE
> -run ../tcc.c -B.. -I.. -I.. -I../include
> -DCONFIG_LDDIR="\"lib/i386-linux-gnu\""
> -DCONFIG_MULTIARCHDIR="\"i386-linux-gnu\"" -DTCC_TARGET_I386 -DONE_SOURCE
> -run ../tcc.c -B.. -I.. -I.. -I../include
> -DCONFIG_LDDIR="\"lib/i386-linux-gnu\""
> -DCONFIG_MULTIARCHDIR="\"i386-linux-gnu\"" -DTCC_TARGET_I386 -DONE_SOURCE
> -run ../tcc.c -B.. -I.. -I.. -I../include -run ../tests/tcctest.c >
> test.out3 Floating point exception
>     make[1]: *** [test3] Error 136
>     make[1]: Leaving directory `/home/ramsay/tinycc/tests'
>     make: *** [test] Error 2
>     address@hidden $
> 
> If you run 'git bisect' you will find the culprit is commit 73faaea2
> ("i386-gen: preserve fp control word in gen_cvt_ftoi", 28-08-2013).
> 
> Try the following patch and let me know if it fixes it up for you.
> 
> ATB,
> Ramsay Jones
> 
> --- 8< ---
> From: Ramsay Jones <address@hidden>
> Date: Sat, 7 Sep 2013 22:48:02 +0100
> Subject: [PATCH] libtcc1.c: Fix __asm__() in __tcc_fpinit and __tcc_cvt_ftol
> 
> Signed-off-by: Ramsay Jones <address@hidden>
> ---
>  lib/libtcc1.c | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/lib/libtcc1.c b/lib/libtcc1.c
> index a94a82d..a717701 100644
> --- a/lib/libtcc1.c
> +++ b/lib/libtcc1.c
> @@ -482,7 +482,7 @@ long long __ashldi3(long long a, int b)
>  void __tcc_fpinit(void)
>  {
>      unsigned c = 0x137F;
> -    __asm__ __volatile__ ("fldcw %0" : "=m" (c));
> +    __asm__ __volatile__ ("fldcw %0" : : "m" (c));
>  }
>  #endif
>  long long __tcc_cvt_ftol(long double x)
> @@ -491,9 +491,9 @@ long long __tcc_cvt_ftol(long double x)
>      long long ret;
>      __asm__ __volatile__ ("fnstcw %0" : "=m" (c0));
>      c1 = c0 | 0x0C00;
> -    __asm__ __volatile__ ("fldcw %0" : "=m" (c1));
> +    __asm__ __volatile__ ("fldcw %0" : : "m" (c1));
>      __asm__ __volatile__ ("fistpll %0"  : "=m" (ret));
> -    __asm__ __volatile__ ("fldcw %0" : "=m" (c0));
> +    __asm__ __volatile__ ("fldcw %0" : : "m" (c0));
>      return ret;
>  }
>  #endif /* !__x86_64__ */

The patch looks sound. the memory location operand is indeed an input since 
the output is in a x87 FPU control word. I thus applied the patch.

Thanks Ramsay.

Best regards,

Thomas

signature.asc
Description: This is a digitally signed message part.