Skip to content

A
angle
Switch branch/tag
  1. 20 Oct, 2020 2 commits
    • Add ES 3.1 check to ValidateBindImageTexture · 8f57f60f
      Cody Northrop authored 
      This fixes a crash that occured in the Lego trace when run with
an ES 3.0 context.

* At the end of each frame it unbinds all the image units using
  glBindImageTexture, which is an ES 3.1 call.
* During validation, we verify things using ANGLE's internals,
  but they are only set up when the context is >= ES 3.1.
* This led to a crash.
* The fix is to check the ES context early and bail, which
  aligns with other entry points.

Test: LEGO MEC
Bug: b/161716126
Bug: angleproject:4048
Change-Id: I9aee2ee173436f5de5062758a70e0909f6fc9a7f
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/2462163Reviewed-by: 's avatarJamie Madill <jmadill@chromium.org>
Reviewed-by: 's avatarCharlie Lao <cclao@google.com>
Reviewed-by: 's avatarCourtney Goeltzenleuchter <courtneygo@google.com>
Commit-Queue: Cody Northrop <cnorthrop@google.com>
    • iOS: Disable worker contexts, and use sized formats · babce049
      James Darpinian authored 
      Kimmo Kinnunen made this change downstream in WebKit.
https://bugs.webkit.org/show_bug.cgi?id=215908

Disable ANGLE workers until EAGL implementation is more complete.
Current implementation fails to compile any shader, since the
compilation happens in the worker thread and worker EAGL context
which does not use the same sharegroup as the main context.
The shader objects are created in the main context but the shader
source setting and compilation happens in the worker context.
EAGL needs a flush between state changes, and adding that
correctly is a bigger change to be done later.

Use sized formats when calling [EAGLContext -texImageIOSurface]
from EGL_ANGLE_iosurface_client_buffer code. The texImageIOSurface
accepts parameters with glTexImage2D logic. On ES3, some of the
internal formats must be sized formats. The EAGLContext instantiated
by ANGLE is ES3, even if the ANGLE context would be ES2.

No tests added since this should be caught with the many video
related tests. It's unclear why this is not the case -- at
least on real hw. This is to be investigated later, too.

Bug: angleproject:5104
Change-Id: Iebf1a04488c5137d55a278d973a34511e8dc46bb
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/2481850Reviewed-by: 's avatarKenneth Russell <kbr@chromium.org>
Reviewed-by: 's avatarJonah Ryan-Davis <jonahr@google.com>
Commit-Queue: James Darpinian <jdarpinian@chromium.org>
  3. 19 Oct, 2020 15 commits
  5. 18 Oct, 2020 6 commits
  7. 17 Oct, 2020 5 commits
  9. 16 Oct, 2020 12 commits