Closed Bug 1651756 Opened 4 years ago Closed 4 years ago

Content processes don't work in a local (unpackaged) developer build on Windows

Categories

(Core :: Security: Process Sandboxing, defect)

Unspecified
Windows
defect

Tracking

()

RESOLVED DUPLICATE of bug 1650419
Tracking Status
firefox-esr68 --- unaffected
firefox-esr78 --- unaffected
firefox78 --- unaffected
firefox79 --- unaffected
firefox80 --- fixed

People

(Reporter: jfkthame, Unassigned)

References

(Regression)

Details

(Keywords: regression)

Since bug 1639030 landed, it appears that content processes don't work in my local dev build on Windows. (They work fine in an official Nightly updated to today's build, which AFAICS should include those changesets.)

When trying to ./mach run a build that was compiled with debugging enabled, I see a lot of warnings "Failed to duplicate file handle for other process", which sounds like the sandbox is blocking communication that needs to happen.

Flags: needinfo?(tkikuchi)

If a symptom is no content process is launched at all and only a blank page is shown, it might be caused by Clang's bug (bug 1650419). I saw it on PGO build, but it's possible that happens with a different compiler option. To fix it, I've integrated clang's patch to our version of Clang 9.0.1. Can you update our compiler (I believe mach bootstrap does it) and rebuild the code?

Flags: needinfo?(tkikuchi)

Hmm, given that's about filesystem_interception, it sounds like it might be relevant. I'll try updating the compiler and see if that helps.... thanks!

I just tried (opt build) and had the same issue.
bootstrap fixed it for me.
toshi: probably worth a post to dev-platform, to let people know.

(In reply to Bob Owen (:bobowen) from comment #3)

I just tried (opt build) and had the same issue.
bootstrap fixed it for me.
toshi: probably worth a post to dev-platform, to let people know.

Good point. I'll post an announcement.

Set release status flags based on info from the regressing bug 1639030

Just to confirm, updating tools with mach bootstrap seems to have fixed this for me, too (though getting through mach bootstrap successfully was no easy matter either, thanks to bug 1636797).

Thank you for the update! Glad to hear you finally made it :).

Status: NEW → RESOLVED
Closed: 4 years ago
Resolution: --- → DUPLICATE
Has Regression Range: --- → yes
You need to log in before you can comment on or make changes to this bug.