Re: Possible bug in softfloat

2014-11-30 Thread Julian Elischer

On 11/29/14, 4:59 AM, Adrian Chadd wrote:

You can easily fire up a mips32 / mips64 emulator build - cross-build
a world+kernel, build an image, then run qemu-devel to boot it.

https://wiki.freebsd.org/FreeBSD/MipsEmulation

You should be able to get a 32 bit soft-float mips environment inside
there which you can use to trigger it.
(And also run whatever other floating point validation suite you may hvae.)


I suspect this is one for bde to look at.





-adrian


On 28 November 2014 at 11:07, Steve Kargl
s...@troutmask.apl.washington.edu wrote:

On Fri, Nov 28, 2014 at 10:54:25AM -0800, Adrian Chadd wrote:

On 28 November 2014 at 10:34, Steve Kargl
s...@troutmask.apl.washington.edu wrote:

In a thread on comp.lang.c, it was pointed out that softfloat
has a bug and in checking src/lib/libc/softfloat I confimed
the issue is present in FreeBSD.  What I hae not confirmed
is whether or not it is possible to hit this bug.  In fact,
it may only hit arm and mips.  Anyway, here's the patch

So we should just commit this?


I suspect the answer is yes, but I have no idea on how
to trigger this code path.  I also have no access to
arm or mips hardware where the problem may manifest only.

It may also be appropriate to have someone else confirm
that the patch is indeed correct.

--
steve

___
freebsd-current@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-current
To unsubscribe, send any mail to freebsd-current-unsubscr...@freebsd.org




___
freebsd-current@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-current
To unsubscribe, send any mail to freebsd-current-unsubscr...@freebsd.org


Re: Possible bug in softfloat

2014-11-28 Thread Adrian Chadd
On 28 November 2014 at 10:34, Steve Kargl
s...@troutmask.apl.washington.edu wrote:
 In a thread on comp.lang.c, it was pointed out that softfloat
 has a bug and in checking src/lib/libc/softfloat I confimed
 the issue is present in FreeBSD.  What I hae not confirmed
 is whether or not it is possible to hit this bug.  In fact,
 it may only hit arm and mips.  Anyway, here's the patch

So we should just commit this?


-a



 Index: softfloat/bits64/softfloat-macros
 ===
 --- softfloat/bits64/softfloat-macros   (revision 275211)
 +++ softfloat/bits64/softfloat-macros   (working copy)
 @@ -157,7 +157,7 @@
  z0 = a0count;
  }
  else {
 -z1 = ( count  64 ) ? ( a0( count  63 ) ) : 0;
 +z1 = ( count  128 ) ? ( a0( count  63 ) ) : 0;
  z0 = 0;
  }
  *z1Ptr = z1;

 --
 Steve
 ___
 freebsd-current@freebsd.org mailing list
 http://lists.freebsd.org/mailman/listinfo/freebsd-current
 To unsubscribe, send any mail to freebsd-current-unsubscr...@freebsd.org
___
freebsd-current@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-current
To unsubscribe, send any mail to freebsd-current-unsubscr...@freebsd.org


Re: Possible bug in softfloat

2014-11-28 Thread Steve Kargl
On Fri, Nov 28, 2014 at 10:54:25AM -0800, Adrian Chadd wrote:
 On 28 November 2014 at 10:34, Steve Kargl
 s...@troutmask.apl.washington.edu wrote:
  In a thread on comp.lang.c, it was pointed out that softfloat
  has a bug and in checking src/lib/libc/softfloat I confimed
  the issue is present in FreeBSD.  What I hae not confirmed
  is whether or not it is possible to hit this bug.  In fact,
  it may only hit arm and mips.  Anyway, here's the patch
 
 So we should just commit this?
 

I suspect the answer is yes, but I have no idea on how
to trigger this code path.  I also have no access to 
arm or mips hardware where the problem may manifest only.  

It may also be appropriate to have someone else confirm
that the patch is indeed correct.

-- 
steve
___
freebsd-current@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-current
To unsubscribe, send any mail to freebsd-current-unsubscr...@freebsd.org


Re: Possible bug in softfloat

2014-11-28 Thread Adrian Chadd
You can easily fire up a mips32 / mips64 emulator build - cross-build
a world+kernel, build an image, then run qemu-devel to boot it.

https://wiki.freebsd.org/FreeBSD/MipsEmulation

You should be able to get a 32 bit soft-float mips environment inside
there which you can use to trigger it.
(And also run whatever other floating point validation suite you may hvae.)



-adrian


On 28 November 2014 at 11:07, Steve Kargl
s...@troutmask.apl.washington.edu wrote:
 On Fri, Nov 28, 2014 at 10:54:25AM -0800, Adrian Chadd wrote:
 On 28 November 2014 at 10:34, Steve Kargl
 s...@troutmask.apl.washington.edu wrote:
  In a thread on comp.lang.c, it was pointed out that softfloat
  has a bug and in checking src/lib/libc/softfloat I confimed
  the issue is present in FreeBSD.  What I hae not confirmed
  is whether or not it is possible to hit this bug.  In fact,
  it may only hit arm and mips.  Anyway, here's the patch

 So we should just commit this?


 I suspect the answer is yes, but I have no idea on how
 to trigger this code path.  I also have no access to
 arm or mips hardware where the problem may manifest only.

 It may also be appropriate to have someone else confirm
 that the patch is indeed correct.

 --
 steve
___
freebsd-current@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-current
To unsubscribe, send any mail to freebsd-current-unsubscr...@freebsd.org