| Summary: | Trigger happy assert crashes applications | ||
|---|---|---|---|
| Product: | libva | Reporter: | Francois Gouget <fgouget> |
| Component: | core | Assignee: | haihao <haihao.xiang> |
| Status: | RESOLVED FIXED | QA Contact: | Sean V Kelley <seanvk> |
| Severity: | normal | ||
| Priority: | medium | ||
| Version: | unspecified | ||
| Hardware: | Other | ||
| OS: | All | ||
| Whiteboard: | |||
| i915 platform: | i915 features: | ||
| Attachments: | va/x11: Don't crash the application if it does not have access to the DRI2 device. | ||
Could you provide a git formatted patch ? Created attachment 117233 [details] [review] va/x11: Don't crash the application if it does not have access to the DRI2 device. Sorry for the delay, here's a Git-format patch based on the current tip. Applied, thanks. |
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.
After su-ing to another account, any application attempting to use libva crashes with the following message: $ mplayer big_buck_bunny_480p_h264.mov [...] libva info: VA-API version 0.37.0 mplayer: dri2_util.c:199: isDRI2Connected: Assertion `dri_state->base.fd >= 0' failed. $ gst-launch-1.0 filesrc location=big_buck_bunny_480p_h264.mov ! decodebin ! fakesink [...] libva info: VA-API version 0.37.0 gst-launch-1.0: dri2_util.c:199: isDRI2Connected: Assertion `dri_state->base.fd >= 0' failed. This is because the effective user does not have access to /dev/dri/card0. This may or may not be right, this is not the issue here. The issue is that libva unilaterally takes it upon itself to crash the application which robs it of the chance of trying to use other libraries. For instance the above GStreamer commands does not need libva at all but it never gets a chance to try to use libav instead. Assert are meant to verify for properties of the code they are in, not to verify that the environment is set up as the code wants. That's what regular error handling is for, and indeed such code is present just two lines down. So the fix is really simple: --- va/x11/dri2_util.c.orig 2015-06-06 23:56:15.728190780 +0200 +++ va/x11/dri2_util.c 2015-06-06 23:56:31.928154507 +0200 @@ -196,7 +196,6 @@ goto err_out; dri_state->base.fd = open(device_name, O_RDWR); - assert(dri_state->base.fd >= 0); if (dri_state->base.fd < 0) goto err_out;