Closed Bug 1552379 Opened 5 years ago Closed 5 years ago

camera getUserMedia() from jsfiddle.net broken in Nightly (NotReadableError)

Categories

(Core :: Permission Manager, defect, P1)

defect

Tracking

()

RESOLVED FIXED
mozilla68
Tracking Status
firefox-esr60 --- unaffected
firefox66 --- unaffected
firefox67 --- unaffected
firefox68 - fixed

People

(Reporter: jib, Unassigned)

References

(Regression)

Details

(Keywords: regression)

Attachments

(1 obsolete file)

STRs

  1. Open https://jsfiddle.net/jib1/r60bzmrs/
  2. Allow camera permission

Expected result:
Self view.

Actual result:

NotReadableError: Failed to allocate videosource

Regression range:

7:47.26 INFO: No more inbound revisions, bisection finished.
7:47.26 INFO: Last good revision: 55089d0bf0701221a363514e12592e8bc07405db
7:47.26 INFO: First bad revision: 0229d5353d503441ed11643c8e5e4e90011261b8
7:47.26 INFO: Pushlog:
https://hg.mozilla.org/integration/autoland/pushloghtml?fromchange=55089d0bf0701221a363514e12592e8bc07405db&tochange=0229d5353d503441ed11643c8e5e4e90011261b8

Points to Bug 1330467.

Flags: needinfo?(jhofmann)

Note I have privacy.firstparty.isolate set to false (the default).

As I recall NotReadableError can be caused by the parent process permission double-check.

Why is this not covered by tests?

[Tracking Requested - why for this release]:
Can't request WebRTC permission in iframes

We could probably fix this, I haven't looked into it yet, but I wonder if landing bug 1330467 right before soft freeze might have been a mistake. Gary, do you think we should try to back out 1330467 while we still can instead of spending our time uplifting fixes to Beta? I'm kind of on the fence here.

Component: DOM: Security → Permission Manager
Flags: needinfo?(xeonchen)
Flags: needinfo?(jhofmann)
Flags: in-testsuite?
Priority: P2 → P1
Flags: needinfo?(jhofmann)
Flags: needinfo?(xeonchen)
Flags: needinfo?(xeonchen)

(In reply to Johann Hofmann [:johannh] from comment #3)

We could probably fix this, I haven't looked into it yet, but I wonder if landing bug 1330467 right before soft freeze might have been a mistake. Gary, do you think we should try to back out 1330467 while we still can instead of spending our time uplifting fixes to Beta? I'm kind of on the fence here.

OK, I agree this was more risky than I expected while the breakage occurs even when FPI is disabled.
I'll request to back out bug 1330467 and re-land into v59.

Flags: needinfo?(xeonchen)
Status: NEW → RESOLVED
Closed: 5 years ago
Flags: needinfo?(jhofmann)
Resolution: --- → FIXED
Target Milestone: --- → mozilla68
Attachment #9066049 - Attachment is obsolete: true
Has Regression Range: --- → yes
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: