1974058 - Update wgpu to upstream df272d3c (2025-06-25)

Status: RESOLVED FIXED

(Core :: Graphics: WebGPU, task, P1)

Description

[Jim Blandy :jimb]

What it says on the tin.

Comment 1

[Jim Blandy :jimb]

Attachment: Bug 1974058: Update wgpu to upstream df272d3c (2025-06-25) r=#webgpu-reviewers!

Comment 2

[Jim Blandy :jimb]

try push

Comment 3

[Jim Blandy :jimb]

We're getting some new crashes in backlog tests on Linux and Mac:

modified   testing/web-platform/mozilla/meta/webgpu/cts/webgpu/web_platform/canvas/getCurrentTexture/cts.https.html.ini
modified   testing/web-platform/mozilla/meta/webgpu/cts/webgpu/web_platform/canvas/getCurrentTexture/dedicated.https.html.ini
modified   testing/web-platform/mozilla/meta/webgpu/cts/webgpu/web_platform/canvas/getCurrentTexture/shared.https.html.ini
modified   testing/web-platform/mozilla/meta/webgpu/cts/webgpu/web_platform/copyToTexture/canvas/cts.https.html.ini

I want to make sure these are not a problem.

Comment 4

[Erich Gubler [:ErichDonGubler] (he/him)]

:jimb: Those failures precede the patch stack you've submitted; they're already failing in mozilla-central, as I also independently discovered with testing for bug 1973770. Just about to land those crashes as accepted outcomes in D255027.

Comment 5

[Jim Blandy :jimb]

These crashes are hitting one of the following assertions:

https://searchfox.org/mozilla-central/rev/445973bddca7645f6d8068e9347b4d98b1aeb59e/dom/webgpu/ipc/WebGPUParent.cpp#1105-1113

ipc::IPCResult WebGPUParent::GetFrontBufferSnapshot(
    IProtocol* aProtocol, const layers::RemoteTextureOwnerId& aOwnerId,
    const RawId& aCommandEncoderId, Maybe<Shmem>& aShmem, gfx::IntSize& aSize,
    uint32_t& aByteStride) {
  const auto& lookup = mPresentationDataMap.find(aOwnerId);
  if (lookup == mPresentationDataMap.end()) {
    MOZ_ASSERT_UNREACHABLE("unexpected to be called");
    return IPC_OK();
  }

https://searchfox.org/mozilla-central/rev/445973bddca7645f6d8068e9347b4d98b1aeb59e/dom/webgpu/ipc/WebGPUParent.cpp#1313-1332

void WebGPUParent::SwapChainPresent(
    RawId aTextureId, RawId aCommandEncoderId,
    const layers::RemoteTextureId& aRemoteTextureId,
    const layers::RemoteTextureOwnerId& aOwnerId) {
  ...
  if (data->mUseExternalTextureInSwapChain) {
    auto it = mExternalTextures.find(aTextureId);
    if (it == mExternalTextures.end()) {
      MOZ_ASSERT_UNREACHABLE("unexpected to be called");
      return;
    }

Comment 6

[Jim Blandy :jimb]

(In reply to Erich Gubler [:ErichDonGubler] (he/him) from comment #4)

:jimb: Those failures precede the patch stack you've submitted; they're already failing in mozilla-central, as I also independently discovered with testing for bug 1973770. Just about to land those crashes as accepted outcomes in D255027.

Okay, sounds good.

Comment 7

[Jim Blandy :jimb]

new try push

Comment 8

[Jim Blandy :jimb]

That try push failed for reasons it won't say. new try push

edit: Okay, that one failed too. Something is going on here that I don't understand.

edit: Lando seems to be unhappy today. Other folks are having the same problem with unrelated work, so it seems like the problem has nothing to do with this PR.

Comment 9

[Pulsebot]

Pushed by [e-mail]:
https://github.com/mozilla-firefox/firefox/commit/3e350b9d96c4
https://hg.mozilla.org/integration/autoland/rev/22af877c8192
Update wgpu to upstream df272d3c (2025-06-25) r=webgpu-reviewers,supply-chain-reviewers,nical

Comment 10

[pstanciu]

https://hg.mozilla.org/mozilla-central/rev/22af877c8192