[Bug 96289] all OpenGL application with feenableexcept(FE_DIVBYZERO|FE_OVERFLOW|FE_INVALID) raise SIGFPE

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=96289

            Bug ID: 96289
           Summary: all OpenGL application with
                    feenableexcept(FE_DIVBYZERO|FE_OVERFLOW|FE_INVALID)
                    raise SIGFPE
           Product: Mesa
           Version: 11.2
          Hardware: x86-64 (AMD64)
                OS: Linux (All)
            Status: NEW
          Severity: normal
          Priority: medium
         Component: Drivers/DRI/i965
          Assignee: idr at freedesktop.org
          Reporter: joanbe at gmail.com
        QA Contact: intel-3d-bugs at lists.freedesktop.org

Created attachment 124210
  --> https://bugs.freedesktop.org/attachment.cgi?id=124210&action=edit
gdb backtrace

After Ubuntu upgraded to Mesa 11.2 all OpenGL applications with floating point
exceptions enabled (FE_INVALID) raise a SIGFPE signal at the first drawing
command.
Oibaf Mesa 11.3-git ppa as of (31-may-2016) also shows this behavior.

To test it's necessary to patch glxgears.
---
--- glxgears.c  2016-05-31 12:18:57.175490491 +0200
+++ glxgears_fpe.c      2016-05-31 12:28:30.854798058 +0200
@@ -37,6 +37,9 @@
 #include <GL/glx.h>
 #include <GL/glxext.h>

+#define __USE_GNU
+#include <fenv.h>
+
 #ifndef GLX_MESA_swap_control
 #define GLX_MESA_swap_control 1
 typedef int (*PFNGLXGETSWAPINTERVALMESAPROC)(void);
@@ -731,6 +734,8 @@
    VisualID visId;
    int i;

+   feenableexcept(FE_DIVBYZERO | FE_OVERFLOW | FE_INVALID);
+
    for (i = 1; i < argc; i++) {
       if (strcmp(argv[i], "-display") == 0) {
          dpyName = argv[i+1];
---

My guess
line 301/302 has a 0 by 0 division
unsigned hs_additional = (unsigned)
round(hs_wants * (((double) remaining_space) / total_wants));

Probably it's really uncommon to enable exceptions as neither glxgears, tux
racer, xonotic or web browsers or plasma desktop have this problem.

But until now (Mesa 11.1), neither mesa-radeon or mesa-intel or
nvidia-proprietary-blob had this issue.

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/intel-3d-bugs/attachments/20160531/ba4a0e76/attachment.html>