Nightly fails to start in macOS Sonoma virtual machine | Crash in [@ mozilla::widget::GfxInfo::GetDeviceInfo]
Categories
(Core :: Widget: Cocoa, defect, P3)
Tracking
()
Tracking | Status | |
---|---|---|
firefox121 | --- | fixed |
People
(Reporter: sam, Assigned: ran1, NeedInfo)
References
(Blocks 1 open bug)
Details
(Whiteboard: [tbird crash])
Crash Data
Attachments
(2 files, 2 obsolete files)
48 bytes,
text/x-phabricator-request
|
Details | Review | |
1.99 KB,
patch
|
Details | Diff | Splinter Review |
Hello,
I installed macOS Sonoma in a virtual machine, using VirtualBuddy, which uses Apple's built-in virtualization frameworks. However, Nightly does not start in it, with this crash. Given the nature of the assert, I should note a few items:
- All transparency effects in the OS are working, indicating graphics acceleration is working.
- System Information does not list any GPUs.
Crash report: https://crash-stats.mozilla.org/report/index/dfc67b44-dffb-4b83-91ff-b776f0230610
MOZ_CRASH Reason: MOZ_DIAGNOSTIC_ASSERT(mNumGPUsDetected > 0) (Failed to detect any GPUs)
Top 10 frames of crashing thread:
0 XUL mozilla::widget::GfxInfo::GetDeviceInfo widget/cocoa/GfxInfo.mm
0 XUL mozilla::widget::GfxInfo::Init widget/cocoa/GfxInfo.mm:176
1 XUL mozilla::xpcom::CreateInstanceImpl xpcom/components/StaticComponents.cpp:12795
2 XUL mozilla::xpcom::StaticModule::CreateInstance const xpcom/components/StaticComponents.cpp:13161
2 XUL xpcom/components/nsComponentManager.cpp:184
2 XUL nsComponentManagerImpl::GetServiceLocked xpcom/components/nsComponentManager.cpp:971
2 XUL nsComponentManagerImpl::GetServiceByContractID xpcom/components/nsComponentManager.cpp:1160
3 XUL CallGetService xpcom/components/nsComponentManagerUtils.cpp:61
3 XUL xpc::CIGSHelper js/xpconnect/src/XPCJSID.cpp:585
4 XUL CallJSNative js/src/vm/Interpreter.cpp:486
Reporter | ||
Updated•1 year ago
|
Comment 1•1 year ago
|
||
The bug has a crash signature, thus the bug will be considered confirmed.
Comment 2•1 year ago
|
||
The severity field is not set for this bug.
:spohl, could you have a look please?
For more information, please visit BugBot documentation.
Updated•11 months ago
|
Comment 3•11 months ago
|
||
Two different users, basically confirming comment 0:
- Attempted to run in a macOS VM (Parallels Desktop 18, with macOS Sonoma beta 1; appears to be an issue where the underlying macOS doesn’t find a GPU — and accordingly, Firefox doesn’t run. It appears that while there isn’t hardware graphics, Parallels is somewhat filling in the gap.
- First launch of Firefox Nightly on a fresh new macOS Sonoma beta 1 virtual machine within macOS Ventura.
Detect GPUs on macOS, including VMs, by searching for IOAccelerators.
This replaces checking for VGA devices, and AGXAccelerator devices on aarch64.
This also removes an unused item from sandbox IOKit property lists.
Updated•11 months ago
|
Hello, I found a fix to detect virtualization framework GPUs and have submitted a patch for it. This should work for Parallels, though I haven't tested it.
Updated•10 months ago
|
Comment 6•10 months ago
|
||
The bug is linked to a topcrash signature, which matches the following criteria:
- Top 20 desktop browser crashes on beta (startup)
- Top 5 desktop browser crashes on Mac on beta (startup)
For more information, please visit BugBot documentation.
The registryID property used to get a Metal GPU's IOService is not available before macOS 10.13.
Updated•10 months ago
|
Updated•10 months ago
|
Adds a search for AppleParavirtGPUs found on macOS virtual machines.
Comment 9•9 months ago
|
||
Based on the topcrash criteria, the crash signature linked to this bug is not a topcrash signature anymore.
For more information, please visit BugBot documentation.
Comment 10•9 months ago
|
||
Sorry for removing the keyword earlier but there is a recent change in the ranking, so the bug is again linked to a topcrash signature, which matches the following criterion:
- Top 5 desktop browser crashes on Mac on beta (startup)
For more information, please visit BugBot documentation.
Comment 11•9 months ago
•
|
||
Interestingly, these crashes don't happen in the Firefox release version, in a UTM virtual machine made using Apple's Virtualization framework. I tested using UTM 4.3.5 (the current version), with Firefox 117, on an Apple Silicon Mac running macOS 14 Beta 7 (build 23A5337a, also the current version). My VM was created using the same version of macOS (macOS 14 Beta 7).
The first page with actual web content takes a long time to load, though -- about five seconds.
Edit: But they do happen with today's mozilla-central nightly:
bp-941c9ca8-b763-4746-8be6-9cd2a0230905
bp-04c36830-6338-40cc-9088-715900230905
Comment 12•9 months ago
|
||
(In reply to Steven Michaud [:smichaud] (Retired) from comment #11)
Interestingly, these crashes don't happen in the Firefox release version, in a UTM virtual machine made using Apple's Virtualization framework. I tested using UTM 4.3.5 (the current version), with Firefox 117, on an Apple Silicon Mac running macOS 14 Beta 7 (build 23A5337a, also the current version). My VM was created using the same version of macOS (macOS 14 Beta 7).
The first page with actual web content takes a long time to load, though -- about five seconds.
Edit: But they do happen with today's mozilla-central nightly:
bp-941c9ca8-b763-4746-8be6-9cd2a0230905
bp-04c36830-6338-40cc-9088-715900230905
The crash is a MOZ_DIAGNOSTIC_ASSERT which is only compiled in on DEBUG, Nightly, or early Beta builds. Does the posted patch work for you?
Comment 13•9 months ago
|
||
Does the posted patch work for you?
Yes, it does. It doesn't crash when I run it in my macOS 14 VM.
Comment 14•9 months ago
•
|
||
But RAN1's patch doesn't apply to current trunk. Here's an updated version.
Comment 15•8 months ago
|
||
Based on the topcrash criteria, the crash signature linked to this bug is not a topcrash signature anymore.
For more information, please visit BugBot documentation.
Comment 16•8 months ago
|
||
Sorry for removing the keyword earlier but there is a recent change in the ranking, so the bug is again linked to a topcrash signature, which matches the following criteria:
- Top 20 desktop browser crashes on beta (startup)
- Top 5 desktop browser crashes on Mac on beta (startup)
For more information, please visit BugBot documentation.
Comment 17•8 months ago
|
||
Based on the topcrash criteria, the crash signature linked to this bug is not a topcrash signature anymore.
For more information, please visit BugBot documentation.
Comment 18•7 months ago
|
||
This keeps falling in and out of the Mac topcrasher list. We have a patch for it, and it's been approved. Can we get it landed? :-)
Updated•7 months ago
|
Comment 19•7 months ago
|
||
Thanks for flagging this, Steven.
@RAN1, the patch needs a rebase to merge cleanly. Are you available to do that today? If not, I should be able to do the rebase and push the patch attributing it to you.
Updated•7 months ago
|
Comment 20•7 months ago
|
||
Pushed by haftandilian@mozilla.com: https://hg.mozilla.org/integration/autoland/rev/75cb4eab7ed3 Nightly fails to start in macOS Sonoma virtual machine r=jrmuizel,haik
Comment 21•7 months ago
|
||
bugherder |
Updated•7 months ago
|
Updated•7 months ago
|
Updated•7 months ago
|
Comment 22•6 months ago
|
||
These crashes also happen on Intel Macs, when running in a VMware VM:
bp-6342a45d-b51e-40f8-a710-a9e150231128
bp-c8b13c24-ca4b-4f71-99f0-88f420231128
My VM is running macOS 12.7.1 (build 21G920).
Comment 23•6 months ago
|
||
Or maybe I should just open a new bug. I'll do that.
Description
•