1293188 - Crash in EnterBaseline (can be defective hardware [Mem or VGA])

Status: RESOLVED WORKSFORME

(Core :: JavaScript Engine, defect, P3)

Description

[Ahmed Bhram]

I've been having this crash for a couple of months now, and it occurs like more than once an hour, I'm not sure if it started with a FF update or with a VGA driver update, I'm doubting those 2 factors.

This bug was filed from the Socorro interface and is 
report bp-50064d25-b5c8-469c-8683-fa2b62160807.
=============================================================

Comment 1

[Jan de Mooij [:jandem]]

Hi Ahmed, this is almost certainly a hardware problem. I'd recommend running Memtest for at least a few hours.

I looked at the crashy JIT code:

ecf2dd71:	48 bf d8 66 2a 91 41 	movabs $0x241912a66d8,%rdi
ecf2dd78:	02 00 00
...
ecf2dd84:	48 bf e8 66 2a 91 41 	movabs $0x241912a66e8,%rdi
ecf2dd8b:	02 00 00
...
ecf2dd9d:	48 bf f8 66 2a 91 41 	movabs $0xff000241912a66f8,%rdi
ecf2dda4:	02 00 ff 
ecf2dda7:	48 8b 3f             	mov    (%rdi),%rdi

We crash at the last instruction, because the value we're loading into %rdi is bogus: that 0xff byte should have been 0x00, like the ones before it.

Comment 2

[Ahmed Bhram]

I ran memtest for over an hour, until it completed one pass with no errors.
Also I must mention here that no other browser crashes except for FF.
Let me know if I can help with any additional info.

Comment 3

[Ahmed Bhram]

I've also noticed in the crash report that it reports 128 TB of virtual memory on my system.
I can't imaging how this can be right.

This still happens up to 48.01 and very frequently.

Again, it appears that only firefox suffers from this problem, there must be a reason for this.

Comment 4

[Ahmed Bhram]

Still a problem on 48.0.2
and still the only browser suffering from this problem is firefox.

Comment 5

[Wayne Mery (:wsmwk)]

Have you tried running memory test again?
Does crash happen with Firefox started in safe mode?

Comment 6

[Ahmed Bhram]

It's really good to see someone is still replying to this problem.
Here is what's going on so far:
1- Browser doesn't crash now, tabs crash but browser is still open, the browser crash has become very rare, I guess with FF 52.
Note: all tabs crash frequently, this is almost the same problem, I have to reload each and every tab again.
2- Can't blame the HW when everything else works fine (AAA games, browsers, videos, ...) only FF has a problem.
3- I tried Safe mode when browser was still crashing (before 52), the crash frequency is certainly less when in safe mode, but it does crash in safe mode.

I do love FF, but this problem is really destroying the whole experience.

Comment 7

[Wayne Mery (:wsmwk)]

Your original crash report has aged off the system.
Do you have any recent crash IDs?

Comment 8

[Ahmed Bhram]

(In reply to Wayne Mery (:wsmwk) from comment #7)
> Your original crash report has aged off the system.
> Do you have any recent crash IDs?

https://crash-stats.mozilla.com/report/index/bp-9fda7415-2125-431c-ba52-469431180102
That's the latest crash report, that's a whole firefox crash.
This whole crash has become less and less a couple of releases before quantum, but instead the frequency of tabs crashing is still the same if not more than before.
Here is a single tab crash report, happened when I tried to open youtube while typing this response:
https://crash-stats.mozilla.com/report/index/bp-abc1abf6-bed7-4389-b2f2-eb90e1180102

A comment: It certainly happens more with a slower connection, I'm not sure if this info is of any help.

Best Regards

Comment 9

[Asif Youssuff]

User on reddit reports this issue as well:

https://crash-stats.mozilla.com/report/index/e95b338c-e923-4564-b0e7-711a80180110
https://crash-stats.mozilla.com/report/index/aaddeba4-fea6-465d-a83a-d261b0180109
https://crash-stats.mozilla.com/report/index/d3adc15e-fc04-49f4-94f9-2bbff0180110

https://www.reddit.com/r/firefox/comments/7pf9pa/help_5704_keeps_crashing_tabs_and_whole_browser/

Comment 10

[Asif Youssuff]

Ignore comment 9, the user commented back that it was indeed bad hardware (memory).

Comment 11

[Ted Campbell [:tcampbell]]

Closing in favor of meta Bug 858032. Issues here seem to be tracked to bad memory.