[Bug 728129] New: exec_opcodes_sys test fails on Intel i7-2620M

["GStreamer" (bugzilla.gnome.org)]

https://bugzilla.gnome.org/show_bug.cgi?id=728129
GStreamer | orc | git

Summary: exec_opcodes_sys test fails on Intel i7-2620M
Classification: Platform
Product: GStreamer
Version: git
OS/Version: Linux
Status: NEW
Severity: blocker
Priority: Normal
Component: orc
AssignedTo: gstreamer-***@lists.freedesktop.org
ReportedBy: ***@coaxion.net
QAContact: gstreamer-***@lists.freedesktop.org
CC: ***@schleef.org
GNOME version: ---


Created an attachment (id=274209)
--> (https://bugzilla.gnome.org/attachment.cgi?id=274209)
orc-bugreport.log

See attached output of the test and orc-bugreport. Happens with git master, .17
and .18 at least. Didn't test earlier versions.

["GStreamer" (bugzilla.gnome.org)]

https://bugzilla.gnome.org/show_bug.cgi?id=728129
GStreamer | orc | git

--- Comment #1 from Sebastian Dröge (slomo) <***@coaxion.net> 2014-04-13 19:12:18 UTC ---
Created an attachment (id=274210)
--> (https://bugzilla.gnome.org/attachment.cgi?id=274210)
exec_opcodes_sys.log

["GStreamer" (bugzilla.gnome.org)]

https://bugzilla.gnome.org/show_bug.cgi?id=728129
GStreamer | orc | git

Sebastian Dröge (slomo) <slomo> changed:

What |Removed |Added
----------------------------------------------------------------------------
Summary|exec_opcodes_sys test fails |exec_opcodes_sys test fails
|on Intel i7-2620M |on Intel i7-2620M when
| |built with clang 3.5

--- Comment #2 from Sebastian Dröge (slomo) <***@coaxion.net> 2014-04-14 19:51:36 UTC ---
Works when building with gcc 4.9, fails when building with clang 3.5

["GStreamer" (bugzilla.gnome.org)]

https://bugzilla.gnome.org/show_bug.cgi?id=728129
GStreamer | orc | git

Sebastian Dröge (slomo) <slomo> changed:

What |Removed |Added
----------------------------------------------------------------------------
Status|NEW |ASSIGNED

--- Comment #3 from Sebastian Dröge (slomo) <***@coaxion.net> 2014-04-14 21:15:07 UTC ---
Fixed half of it already

["GStreamer" (bugzilla.gnome.org)]

https://bugzilla.gnome.org/show_bug.cgi?id=728129
GStreamer | orc | git

--- Comment #4 from Sebastian Dröge (slomo) <***@coaxion.net> 2014-04-15 07:17:47 UTC ---
The remaining problem here is the convdl and convfl code. The result of casting
a float/double to an integer is undefined if the integral part of the
float/double can't be represented by the target integer type.

gcc apparently casts everything to 0x80000000 according to the existing code,
clang to 0x00000000 or 0x80000000 depending on the phase of moon.

Adding this check to the code

else if (tmp == 0 && (var32.f > 1.0 || var32.f < -1.0)) tmp = 0x7fffffff

fixes it for most cases. The only remaining problem is that clang returns
0x80000000 in some cases for huge negative numbers, and the check above does
not catch that

if (tmp == 0x80000000 && !(var32.i&0x80000000)) tmp = 0x7fffffff

because the sign bit is set in the float.

We could add another check that automatically changes floats > MAX or < MIN to
0x7fffffff but even then I wouldn't be surprised if the behaviour of the
various SIMD instructions to do this conversion is different for the undefined
case.


What should we do here?

["GStreamer" (bugzilla.gnome.org)]

https://bugzilla.gnome.org/show_bug.cgi?id=728129
GStreamer | orc | git

--- Comment #5 from Sebastian Dröge (slomo) <***@coaxion.net> 2014-04-15 07:22:26 UTC ---
(In reply to comment #4)

Post by "GStreamer" (bugzilla.gnome.org)
We could add another check that automatically changes floats > MAX or < MIN to
0x7fffffff but even then I wouldn't be surprised if the behaviour of the
various SIMD instructions to do this conversion is different for the undefined
case.

Actually that already does not work. Huge negative numbers are currently
expected to be cast to -0 and huge positive numbers to 0x7fffffff.

["GStreamer" (bugzilla.gnome.org)]

https://bugzilla.gnome.org/show_bug.cgi?id=728129
GStreamer | orc | git

Sebastian Dröge (slomo) <slomo> changed:

What |Removed |Added
----------------------------------------------------------------------------
Status|ASSIGNED |NEW

--- Comment #6 from Sebastian Dröge (slomo) <***@coaxion.net> 2014-04-15 07:24:57 UTC ---
And even that does not work properly with the audioconvert test, only the
internal orc tests.

else if (%s > 9223372036854775807.0) tmp = 0x7fffffff;
else if (%s < -9223372036854775808.0) tmp = 0;

["GStreamer" (bugzilla.gnome.org)]

https://bugzilla.gnome.org/show_bug.cgi?id=728129
GStreamer | orc | git

Vladimír Čunát <vcunat> changed:

What |Removed |Added
----------------------------------------------------------------------------
CC| |***@gmail.com

--- Comment #7 from Vladimír Čunát <***@gmail.com> 2014-08-23 09:49:28 UTC ---
With the released version, I still had this and other test problem with gcc on
i686-linux. https://bugzilla.gnome.org/show_bug.cgi?id=735273 (unfortunately I
didn't notice this report before creating the other one)

["GStreamer" (bugzilla.gnome.org)]

https://bugzilla.gnome.org/show_bug.cgi?id=728129
GStreamer | orc | git

Mateusz Kowalczyk <fuuzetsu> changed:

What |Removed |Added
----------------------------------------------------------------------------
CC| |***@fuuzetsu.co.uk

["GStreamer" (bugzilla.gnome.org)]

https://bugzilla.gnome.org/show_bug.cgi?id=728129
GStreamer | orc | git

Sebastian Dröge (slomo) <slomo> changed:

What |Removed |Added
----------------------------------------------------------------------------
CC| |***@gmail.com

--- Comment #8 from Sebastian Dröge (slomo) <***@coaxion.net> 2014-11-07 12:25:49 UTC ---
*** Bug 739776 has been marked as a duplicate of this bug. ***

["GStreamer" (bugzilla.gnome.org)]

https://bugzilla.gnome.org/show_bug.cgi?id=728129
GStreamer | orc | git

Daniel Macks <dmacks> changed:

What |Removed |Added
----------------------------------------------------------------------------
CC| |***@netspace.org

["GStreamer" (bugzilla.gnome.org)]

https://bugzilla.gnome.org/show_bug.cgi?id=728129
GStreamer | orc | git

--- Comment #9 from Ivan Aponte <***@gmail.com> 2015-01-02 15:29:00 UTC ---
Created an attachment (id=293606)
--> (https://bugzilla.gnome.org/attachment.cgi?id=293606)
bug report from tegra k1

bug report tegra k1 (neon) with gcc-4.9

["GStreamer" (GNOME Bugzilla)]

https://bugzilla.gnome.org/show_bug.cgi?id=728129

Sebastian Dröge (slomo) <***@coaxion.net> changed:

What |Removed |Added
----------------------------------------------------------------------------
Severity|blocker |critical

["GStreamer" (GNOME Bugzilla)]

https://bugzilla.gnome.org/show_bug.cgi?id=728129

Edward Hervey <***@bilboed.com> changed:

What |Removed |Added
----------------------------------------------------------------------------
Status|NEW |NEEDINFO
CC| |***@bilboed.com

--- Comment #10 from Edward Hervey <***@bilboed.com> ---
Ping

Is this still an issue with current orc master ?

["GStreamer" (GNOME Bugzilla)]

https://bugzilla.gnome.org/show_bug.cgi?id=728129

Sebastian Dröge (slomo) <***@coaxion.net> changed:

What |Removed |Added
----------------------------------------------------------------------------
Status|NEEDINFO |NEW

--- Comment #11 from Sebastian Dröge (slomo) <***@coaxion.net> ---
Yes, nobody fixed anything further since Wim fixed a few of these issues.

["GStreamer" (GNOME Bugzilla)]

https://bugzilla.gnome.org/show_bug.cgi?id=728129

Sebastian Dröge (slomo) <***@coaxion.net> changed:

What |Removed |Added
----------------------------------------------------------------------------
Severity|critical |blocker

--- Comment #12 from Sebastian Dröge (slomo) <***@coaxion.net> ---
I'd consider this a blocker now, with clang becoming much more common and
compilers becoming more irrational with regard to undefined behaviour.

["GStreamer" (GNOME Bugzilla)]

https://bugzilla.gnome.org/show_bug.cgi?id=728129

Edward Hervey <***@bilboed.com> changed:

What |Removed |Added
----------------------------------------------------------------------------
Status|NEW |NEEDINFO

--- Comment #13 from Edward Hervey <***@bilboed.com> ---
no longer the case ?

["GStreamer" (GNOME Bugzilla)]

https://bugzilla.gnome.org/show_bug.cgi?id=728129

Sebastian Dröge (slomo) <***@coaxion.net> changed:

What |Removed |Added
----------------------------------------------------------------------------
Status|NEEDINFO |RESOLVED
Resolution|--- |FIXED

--- Comment #14 from Sebastian Dröge (slomo) <***@coaxion.net> ---
Yes, seems to have been fixed at some point by something. We still make use of
undefined behaviour in the ORC C code though, but there's another bug for that.

["GStreamer" (GNOME Bugzilla)]

https://bugzilla.gnome.org/show_bug.cgi?id=728129

--- Comment #15 from Vladimír Čunát <***@gmail.com> ---
Created attachment 364715
--> https://bugzilla.gnome.org/attachment.cgi?id=364715&action=edit
FAIL exec_opcodes_sys: orc-0.4.28 + i686-linux + gcc-6.4.0

This doesn't happen on any of my machines with direct access (tried a few
times), like https://bugzilla.gnome.org/show_bug.cgi?id=735273#c9

This should be on Dell r815, containing 4x12-core Opterons (most likely 61xx,
k10 arch).