Runaway Renderer Helper Mac OSX Current version

Description of the issue:
Bug: Current Release Brave (just upgraded) foxnews.com to foxbusiness.com (internal links) causes runaway Brave Renderer Helper
How can this issue be reproduced?

  1. foxnews.com select article on web page going to foxbusiness.com article
  2. Brave Browser Renderer (helper) consumes CPU and leads to substantial cpu heating and power consumption until killed, causing originating page to fail.

Expected result:
Renderer should not run wild and consume and heat the CPU.

Brave Version( check About Brave):

Version 1.36.112 Chromium: 99.0.4844.51 (Official Build) (x86_64)

Additional Information:
OSX 11.5.2 current, 16 Gb memory

Process sampling gives the following:
Sampling process 56826 for 3 seconds with 1 millisecond of run time between samples
CoreSymbolication warning: fde.addrRange.location 0x103667e70 isn’t in the the __TEXT segment [0x10366f000 - 0x00036000)
CoreSymbolication warning: fde.addrRange.location 0x103667e80 isn’t in the the __TEXT segment [0x10366f000 - 0x00036000)
CoreSymbolication warning: fde.addrRange.location 0x103667e90 isn’t in the the __TEXT segment [0x10366f000 - 0x00036000)
CoreSymbolication warning: fde.addrRange.location 0x103667eb0 isn’t in the the __TEXT segment [0x10366f000 - 0x00036000)
CoreSymbolication warning: fde.addrRange.location 0x103667ed0 isn’t in the the __TEXT segment [0x10366f000 - 0x00036000)
CoreSymbolication warning: fde.addrRange.location 0x103667f00 isn’t in the the __TEXT segment [0x10366f000 - 0x00036000)
CoreSymbolication warning: fde.addrRange.location 0x103667f40 isn’t in the the __TEXT segment [0x10366f000 - 0x00036000)
CoreSymbolication warning: fde.addrRange.location 0x103667f50 isn’t in the the __TEXT segment [0x10366f000 - 0x00036000) . . .

Analysis of sampling Brave Browser Helper (Renderer) (pid 56826) every 1 millisecond
Process: Brave Browser Helper (Renderer) [56826]
Path: /Applications/Brave Browser.app/Contents/Frameworks/Brave Browser Framework.framework/Versions/99.1.36.112/Helpers/Brave Browser Helper (Renderer).app/Contents/MacOS/Brave Browser Helper (Renderer)
Load Address: 0x10366f000
Identifier: Brave Browser Helper (Renderer)
Version: ???
Code Type: X86-64
Platform: macOS
Parent Process: Brave Browser [56328]

Date/Time: 2022-03-23 13:10:52.896 -0500
Launch Time: 2022-03-23 13:04:15.810 -0500
OS Version: macOS 11.5.2 (20G95)
Report Version: 7
Analysis Tool: /usr/bin/sample

Physical footprint: 1.4G
Physical footprint (peak): 1.4G

Total number in stack (recursive counted multiple, when >=5):
629 ChromeMain (in Brave Browser Framework) + 0 [0x109592e10]
25 mach_msg (in libsystem_kernel.dylib) + 60 [0x7fff2044f62c]
25 mach_msg_trap (in libsystem_kernel.dylib) + 0 [0x7fff2044f2b0]
23 ??? (in ) [0x4907f078f8]
22 ChromeMain (in Brave Browser Framework) + 67410485 [0x10d5dc845]
22 _platform_memmove$VARIANT$Haswell (in libsystem_platform.dylib) + 0 [0x7fff204c7000]
22 _pthread_start (in libsystem_pthread.dylib) + 224 [0x7fff204848fc]
22 thread_start (in libsystem_pthread.dylib) + 15 [0x7fff20480443]
18 ChromeMain (in Brave Browser Framework) + 67474018 [0x10d5ec072]
17 ChromeMain (in Brave Browser Framework) + 67483531 [0x10d5ee59b]
15 ChromeMain (in Brave Browser Framework) + 58034148 [0x10cceb5f4]
14 ChromeMain (in Brave Browser Framework) + 58012239 [0x10cce605f]
13 ??? (in ) [0x4907e8df6e]
12 ChromeMain (in Brave Browser Framework) + 69252780 [0x10d79e4bc]
11 ChromeMain (in Brave Browser Framework) + 120665900 [0x1108a653c]
10 ChromeMain (in Brave Browser Framework) + 47556079 [0x10c2ed3ff]
9 ChromeMain (in Brave Browser Framework) + 31185813 [0x10b3509a5]
9 ChromeMain (in Brave Browser Framework) + 43646386 [0x10bf32bc2]
9 ChromeMain (in Brave Browser Framework) + 58034116 [0x10cceb5d4]
9 ChromeMain (in Brave Browser Framework) + 66904980 [0x10d5611a4]
9 ChromeMain (in Brave Browser Framework) + 67007060 [0x10d57a064]
9 ChromeMain (in Brave Browser Framework) + 67156970 [0x10d59e9fa]
9 ChromeMain (in Brave Browser Framework) + 76311330 [0x10de59932]
9 _platform_memcmp (in libsystem_platform.dylib) + 0 [0x7fff204c7aa0]
8 ChromeMain (in Brave Browser Framework) + 117626203 [0x1105c036b]
8 ChromeMain (in Brave Browser Framework) + 47557623 [0x10c2eda07]
8 ChromeMain (in Brave Browser Framework) + 47558972 [0x10c2edf4c]
8 ChromeMain (in Brave Browser Framework) + 47564224 [0x10c2ef3d0]
7 ChromeMain (in Brave Browser Framework) + 113803255 [0x11021ae07]
7 ChromeMain (in Brave Browser Framework) + 129239231 [0x1110d36cf]
7 ChromeMain (in Brave Browser Framework) + 43566263 [0x10bf1f2c7]
7 ChromeMain (in Brave Browser Framework) + 44750105 [0x10c040329]
7 ChromeMain (in Brave Browser Framework) + 67254637 [0x10d5b677d]
7 ChromeMain (in Brave Browser Framework) + 67255976 [0x10d5b6cb8]
7 ChromeMain (in Brave Browser Framework) + 67256611 [0x10d5b6f33]
7 ChromeMain (in Brave Browser Framework) + 67486520 [0x10d5ef148]
6 ChromeMain (in Brave Browser Framework) + 105569147 [0x10fa4098b]
6 ChromeMain (in Brave Browser Framework) + 128057225 [0x110fb2d99]
6 ChromeMain (in Brave Browser Framework) + 43479837 [0x10bf0a12d]
6 ChromeMain (in Brave Browser Framework) + 44749480 [0x10c0400b8]
6 ChromeMain (in Brave Browser Framework) + 47554368 [0x10c2ecd50]
6 ChromeMain (in Brave Browser Framework) + 58012207 [0x10cce603f]
6 ChromeMain (in Brave Browser Framework) + 58034134 [0x10cceb5e6]
6 ChromeMain (in Brave Browser Framework) + 67192329 [0x10d5a7419]
6 ChromeMain (in Brave Browser Framework) + 67193619 [0x10d5a7923]
6 ChromeMain (in Brave Browser Framework) + 67197501 [0x10d5a884d]
6 ChromeMain (in Brave Browser Framework) + 67407658 [0x10d5dbd3a]
6 ChromeMain (in Brave Browser Framework) + 67473630 [0x10d5ebeee]
6 ChromeMain (in Brave Browser Framework) + 68982681 [0x10d75c5a9]
6 ChromeMain (in Brave Browser Framework) + 69110548 [0x10d77b924]
6 ChromeMain (in Brave Browser Framework) + 77385117 [0x10df5fbad]
6 __psynch_mutexwait (in libsystem_kernel.dylib) + 0 [0x7fff204514c0]
6 _platform_strlen (in libsystem_platform.dylib) + 0 [0x7fff204c7540]
6 _pthread_mutex_firstfit_lock_slow (in libsystem_pthread.dylib) + 204 [0x7fff20480192]
6 _pthread_mutex_firstfit_lock_wait (in libsystem_pthread.dylib) + 76 [0x7fff204822ab]
5 ChromeMain (in Brave Browser Framework) + 114625870 [0x1102e3b5e]
5 ChromeMain (in Brave Browser Framework) + 117630024 [0x1105c1258]
5 ChromeMain (in Brave Browser Framework) + 117630126 [0x1105c12be]
5 ChromeMain (in Brave Browser Framework) + 120668862 [0x1108a70ce]
5 ChromeMain (in Brave Browser Framework) + 120668954 [0x1108a712a]
5 ChromeMain (in Brave Browser Framework) + 138483775 [0x1119a464f]
5 ChromeMain (in Brave Browser Framework) + 31144372 [0x10b3467c4]
5 ChromeMain (in Brave Browser Framework) + 42276203 [0x10bde437b]
5 ChromeMain (in Brave Browser Framework) + 42403277 [0x10be033dd]
5 ChromeMain (in Brave Browser Framework) + 43549782 [0x10bf1b266]
5 ChromeMain (in Brave Browser Framework) + 43610154 [0x10bf29e3a]
5 ChromeMain (in Brave Browser Framework) + 43611433 [0x10bf2a339]
5 ChromeMain (in Brave Browser Framework) + 47325780 [0x10c2b5064]
5 ChromeMain (in Brave Browser Framework) + 47333880 [0x10c2b7008]
5 ChromeMain (in Brave Browser Framework) + 47564188 [0x10c2ef3ac]
5 ChromeMain (in Brave Browser Framework) + 48780816 [0x10c418420]
5 ChromeMain (in Brave Browser Framework) + 66745672 [0x10d53a358]
5 ChromeMain (in Brave Browser Framework) + 67283464 [0x10d5bd818]
5 ChromeMain (in Brave Browser Framework) + 67283865 [0x10d5bd9a9]
5 ChromeMain (in Brave Browser Framework) + 69267654 [0x10d7a1ed6]
5 ChromeMain (in Brave Browser Framework) + 77385768 [0x10df5fe38]
5 __psynch_cvwait (in libsystem_kernel.dylib) + 0 [0x7fff20451cd4]
5 _platform_bzero$VARIANT$Haswell (in libsystem_platform.dylib) + 0 [0x7fff204c74a0]
5 _pthread_cond_wait (in libsystem_pthread.dylib) + 1298 [0x7fff20484e49]
5 tlv_get_addr (in libdyld.dylib) + 0 [0x7fff204964cc]

Sort by top of stack, same collapsed (when >= 5):
mach_msg_trap (in libsystem_kernel.dylib) 18764
__psynch_cvwait (in libsystem_kernel.dylib) 6233
kevent64 (in libsystem_kernel.dylib) 2454
__workq_kernreturn (in libsystem_kernel.dylib) 1249
ChromeMain (in Brave Browser Framework) 808
??? (in ) [0x4900b13254] 71
??? (in ) [0x4900b13249] 44
??? (in ) [0x4900b13240] 40
_platform_memmove$VARIANT$Haswell (in libsystem_platform.dylib) 26
??? (in ) [0x4900b13258] 21
_platform_memcmp (in libsystem_platform.dylib) 12
__psynch_mutexwait (in libsystem_kernel.dylib) 7
_platform_strlen (in libsystem_platform.dylib) 7
_kernelrpc_mach_port_insert_member_trap (in libsystem_kernel.dylib) 5
_platform_bzero$VARIANT$Haswell (in libsystem_platform.dylib) 5
tlv_get_addr (in libdyld.dylib) 5

I have saved the entire call stack which is lengthy and can include it, if desired. (/usr/bin/sample output)

Can you try disabling Hardware Acceleration in settings to see if this resolves the issue?

You’ll find this in Settings --> Additional Settings --> System --> Hardware Acceleration

Mac OSX does not have a hardware acceleration and does not use the same Command GUI structure as MSWin.

It does have a switchable graphics acceleration and some macbooks (not this one) had trouble with the graphics card burning out from around 2011.

It does have a System Preferences Selection to disable Graphics acceleration in the Battery section called Automatic Graphics Switching to preserve battery life.

If this is useful, I have disabled it and will report back if it makes a difference.

Update: Per Apple, the Battery/Graphics option will only enable external graphics processor exclusively (at the expense of the battery charge). It will not disable the external. I’m not sure there’s a way to do this on the Mac.

The HWA setting I’m referring to is in the browser itself, not on your OS:

Sorry about the confusion.

The setting is OFF. So, the problem has occurred with the HWA off. To the best of my knowledge this setting has not been turned on.

For what it’s worth, my MacBook Air has been doing the same thing, I believe for the last couple weeks since the last update.

1 Like

Hearing nothing substantial, after three updates, the problem has remained or possibly worsened. Renderer has launched multiple tasks completely compute-binding a quad-core i7, running cpu temps to 100c and slaughtering battery life. I will be migrating to Safari over the next few weeks. This morning an animated, interactive educational task repeated the behavior (developed by Univ Arizona). Sorry guys, I hate to go back to commercial technology, but I can’t live with the cpu utilization problems and intermittent “unresponsive page” messages which resolve themselves when Brave releases some CPU.
Let me know if someone has an idea on how to solve this problem and I’ll be back. It is a good browser and has great ideas, but I need to be productive without burning out cpus.

Appreciate everyone’s help.

@avei , @ryanjruff ,

Recommend:

–

CPU and Memory

  • Apple Activity Monitor
  • Brave Browser Task Manager

–

Tabs - Windows

potato - potatoe

https://www.chromium.org/developers/memory-usage-backgrounder/

Multi-Process Model Background

To understand Chromium’s memory usage, let’s understand the multi-process model. Unlike other browsers, Chromium is divided into multiple processes. When Chromium starts up, it will initially have two processes. One process is the browser process which controls the main browser functionality, and the other is the initial renderer process , which runs the Blink rendering engine and JavaScript (V8). Each time you open a new tab in Chromium, you’ll likely get a new renderer process. With typical browsing, it is common to see 5-7 chrome.exe processes active. Further, if you utilize plugins, apps, or extensions, they may also execute within independent processes. All of Chromium’s processes, whether it is a browser process, a renderer process, or a plugin process, will show under the Task Manager as “chrome.exe”.

–

The problem is not cooling. The fans run properly. Brave Renderer drives the CPU rapidly to 90C and battery consumption rates rise at the same time. This is not a hardware issue and I cannot live with a program (Brave/Renderer) that overloads an Intel i7 quad 2.66 GHz processor with 16 Gb (max) memory. It is not a swapping issue nor an apparent memory issue. The problem happens when the machine is freshly booted or when it is heavily loaded with 3D Parametric CAD/FEA work.

Brave causes a compute bound condition, maxing out CPU. This is NOT a hardware issue.
Running FreeCAD with complex models and compute intensive modeling does not saturate this computer and neither should a web browser. There is something causing Brave/Renderer to enter a run-away condition which ceases when the offending process is killed, sometimes breaking the web sites involved, and sometimes not. The difference is FreeCAD is not designed the same way and does not use parallel processing which chromium/Brave does. There is something not working right in the Brave control program/dispatcher and the Renderer interface/stacks causing Renderer to runaway.

If there were even a Brave setting to have it look for and kill runaway renderer processes, that would be an improvement.

Since’ I’ve started the switch to Safari, using the same websites, I do not have the problem. I do have to put up with the snooping ads, which is a significant reason to run Brave, if I can. This seems to suggest that the web sites are probably not the sole source of the problem. I have not run Firefox in years so I haven’t tried that one. We have few MS/Win machines here so I haven’t tried to reproduce it on MS OS.

Runaway programs (guessing due to params / stacks getting corrupted in passing between modules) are a show stopper. But something is causing a runaway CPU without associated IO and Paging activity.

@avei ,

Macs Fan Control is a good utility, for better management of cooling fans. I did not say, that your Mac has a hardware problem. My interest is, in helping you keep your Mac cool, during your struggles with the problems you state.

–

The main Brave Browser window’s “hamburger” menu [button] pop-down window, has a “Report a Broken Site” selection.

And, going to: brave://settings/privacy . . . you can Enable “Automatically send diagnostic reports”

On the Mac, there is DiagnosticReports folder where Brave Browser places some reports:

/Users/your_username/Library/Logs/DiagnosticReports

–

OK thanks. I did have that information. With the fans running full speed, the processor temps reach 100 and then the system throttles back to protect itself. Power consumption skyrockets and the situation remains until the renderer processes causing the runaway are killed manually. The crash info is in the OP.

I have gone through this on the only Win PC we have here. I cannot replicate this on the Win machine running an I5 with 8 Gb memory. I do not know the MS internals very well as I am a BSD/Unix guy. Mac is built on BSD Unix (a FreeBSD variant).

The problem does appear to be localized to the renderer processes. I don’t know if they are getting bad params, memory is being corrupted, or a stack is bad. This problem is localized to Brave.

The machine is used for scientific monte carlo calcs which will max out the processors but with little memory use until the calculations are completed. The temps on this process rarely reach above 80C with Apple’s utility reporting 367% processor utilization. The bulk of these computations are double precision FP. So, I think the machine is working properly and others have reported this problem on other machines. I have seen strange hardware problems with intel chips like this decades ago.