Floating Point Exception Crashes

[Rickyroko]

Hi All, Ever since installing the new Mac OS X 10.5.2 from Kalyway for AMD, I've been having some trouble with stabability. I'm having random crashes for all apps, including iTunes, WindowServer, Firefox, Console, the Kernel, and more.

It seems all crash reports include a Floating Point Exception (Division by Zero). Also, they all involve a call to the subroutine __gettimeofday, perhaps when starting a new thread. Running on:

Mac OS X Kalyway 10.5.2 AMD

Kernel 9.2.2

nvInject 0.2.0

Using netkas SMBIOS from Here.

 

AMD Athlon 4400+ Dual-Core

2 GB DDR2 PC3200 RAM

Nvidia 7600 GT Video

MSI MS-7093 Mobo

 

More info Here.

 

It seems that after some time after starting OSX, crashes start happening. I'd estimate about 20-30 Hours later. Also, all crashes seem to occur at the same time, maybe seperated my a few minutes or so.

 

Any ideas would be appreciated.

 

Jun 25 21:34:58 Chimera mDNSResponder[20]: mDNSPlatformRawTime went backwards by 1 ticks; setting correction factor to -1473993656
Jun 25 21:34:59 Chimera ReportCrash[1246]: Formulating crash report for process WindowServer[81]
Jun 25 21:34:59 Chimera SecurityAgent[1205]: HIToolbox: received notification of WindowServer event port death.
Jun 25 21:34:59 Chimera synergys[124]: HIToolbox: received notification of WindowServer event port death.
Jun 25 21:34:59 Chimera fseventsd[53]: callback_client: ERROR: d2f_callback_rpc() => (ipc/send) invalid destination port (268435459) for pid 141
Jun 25 21:34:59 Chimera DashboardClient[825]: HIToolbox: received notification of WindowServer event port death.
Jun 25 21:34:59 Chimera DashboardClient[826]: HIToolbox: received notification of WindowServer event port death.
Jun 25 21:34:59 Chimera synergys[124]: port matched the WindowServer port created in BindCGSToRunLoop
Jun 25 21:34:59 Chimera DashboardClient[825]: port matched the WindowServer port created in BindCGSToRunLoop
Jun 25 21:34:59 Chimera com.apple.launchd[1] (com.apple.WindowServer[81]): Exited abnormally: Floating point exception
Jun 25 21:34:59 Chimera kernel[0]: CODE SIGNING: cs_invalid_page: p=1248[loginwindow] clearing CS_VALID
Jun 25 21:34:59 Chimera /System/Library/CoreServices/loginwindow.app/Contents/MacOS/loginwindow[1248]: Login Window Application Started
Jun 25 21:34:59 Chimera com.apple.launchd[123] ([0x0-0x10010].com.blacktree.Quicksilver[149]): Exited: Terminated
Jun 25 21:34:59 Chimera com.apple.launchd[123] ([0x0-0x6b06b].org.mozilla.firefox[1244]): Exited: Terminated
Jun 25 21:34:59 Chimera com.apple.launchd[123] ([0x0-0x18018].org.mozilla.thunderbird[162]): Exited: Terminated
Jun 25 21:34:59 Chimera com.apple.launchd[123] (com.apple.AirPortBaseStationAgent[128]): Exited: Terminated
Jun 25 21:34:59 Chimera com.apple.launchd[123] ([0x0-0xc00c].com.apple.finder[141]): Exited: Terminated
Jun 25 21:34:59 Chimera com.apple.launchd[123] ([0x0-0x16016].com.codeweavers.crossover CD helper[159]): Exited: Terminated
Jun 25 21:34:59 Chimera com.apple.launchd[123] ([0x0-0x1e01e].com.apple.iTunes[181]): Exited: Terminated
Jun 25 21:34:59 Chimera com.apple.launchd[123] (0x108420.rcd[584]): Exited: Terminated
Jun 25 21:34:59 Chimera com.apple.launchd[123] (com.apple.pboard[139]): Exited: Terminated
Jun 25 21:34:59 Chimera com.apple.launchd[123] (com.apple.Spotlight[132]): Exited: Terminated
Jun 25 21:34:59 Chimera com.apple.launchd[123] ([0x0-0xf00f].com.apple.iTunesHelper[148]): Exited: Terminated
Jun 25 21:34:59 Chimera SecurityAgent[1205]: port matched the WindowServer port created in BindCGSToRunLoop
Jun 25 21:34:59 Chimera DashboardClient[826]: port matched the WindowServer port created in BindCGSToRunLoop
Jun 25 21:34:59 Chimera com.apple.launchd[123] ([0x0-0x8008].com.intego.IntegoStatusItemHelper[136]): Exited: Terminated
Jun 25 21:34:59 Chimera com.apple.launchd[123] ([0x0-0x13013].com.intego.WashingMachineHelper[154]): Exited: Terminated
Jun 25 21:34:59 Chimera com.apple.launchd[123] (com.apple.UserEventAgent-Aqua[133]): Exited: Terminated
Jun 25 21:34:59 Chimera com.apple.launchd[123] ([0x0-0x1d01d].com.Growl.GrowlHelperApp[180]): Exited: Terminated
Jun 25 21:34:59 Chimera com.apple.launchd[123] ([0x0-0xb00b].com.apple.systemuiserver[140]): Exited: Terminated
Jun 25 21:34:59 Chimera com.apple.launchd[123] ([0x0-0x6006].com.apple.dock[135]): Exited: Terminated
Jun 25 21:34:59 Chimera ReportCrash[1246]: Saved crashreport to /Library/Logs/CrashReporter/WindowServer_2008-06-25-213457_Chimera.crash using uid: 0 gid: 0, euid: 0 egid: 0
Jun 25 21:35:00 Chimera loginwindow[1248]: Login Window Started Security Agent
Jun 25 21:35:00 Chimera kernel[0]: CODE SIGNING: cs_invalid_page: p=1257[ATSServer] clearing CS_VALID
Jun 25 21:35:00 Chimera SecurityAgent[1256]: NSExceptionHandler has recorded the following exception:\nNSRangeException -- *** -[NSCFArray objectAtIndex:]: index (0) beyond bounds (0)\nStack trace:  0x3719a  0x9699c0fb  0x96d7302b  0x96d7306a  0x959d23df  0x95950218  0x76568  0x76da7  0x60451  0x74825  0x68549  0x6987e  0x6b2ea  0x690d0  0x67bfe  0x6d3d1  0x7d165  0xd648  0x12c40  0x129f3  0xd18a  0x9598f4d3  0x96cd5555  0x96cf9921  0x96cf9d18  0x931096a0  0x931094b9  0x9310932d  0x924e57d9  0x924e508e  0x924de0c5  0x10fc7  0x202a  0x1
Jun 25 21:35:05 Chimera authorizationhost[1255]: MechanismInvoke 0x12d660 retainCount 2
Jun 25 21:35:05 Chimera SecurityAgent[1256]: MechanismInvoke 0x1311b0 retainCount 1
Jun 25 21:35:05 Chimera SecurityAgent[1256]: NSSecureTextFieldCell detected a field editor ((null)) that is not a NSTextView subclass designed to work with the cell. Ignoring...
Jun 25 21:35:05 Chimera SecurityAgent[1256]: NSExceptionHandler has recorded the following exception:\nNSRangeException -- *** -[NSCFArray objectAtIndex:]: index (0) beyond bounds (0)\nStack trace:  0x3719a  0x9699c0fb  0x96d7302b  0x96d7306a  0x959d23df  0x95950218  0x76568  0x76da7  0x60451  0x74825  0x68549  0x747bc  0x6d44e  0x7d165  0xd648  0x12c40  0x129f3  0xd18a  0x9598f4d3  0x96cd5555  0x96cf9921  0x96cf9d18  0x931096a0  0x931094b9  0x9310932d  0x924e57d9  0x924e508e  0x924de0c5  0x10fc7  0x202a  0x1
Jun 25 21:35:05 Chimera SecurityAgent[1256]: MechanismDestroy 0x1311b0 retainCount 1
Jun 25 21:35:05 Chimera loginwindow[1248]: Login Window - Returned from Security Agent
Jun 25 21:35:05 Chimera authorizationhost[1255]: MechanismDestroy 0x12d660 retainCount 2
Jun 25 21:35:05 Chimera /Library/Intego/integod[97]: launching intego menu
Jun 25 21:35:05 Chimera loginwindow[1248]: USER_PROCESS: 1248 console
Jun 25 21:35:05 Chimera com.apple.launchd[1] (com.apple.UserEventAgent-LoginWindow[1252]): Exited: Terminated
Jun 25 21:35:05 Chimera com.apple.loginwindow[1248]: INFO: Synergy server 1.3.1 on Darwin 9.2.2 Darwin Kernel Version 9.2.2: Tue Mar  4 21:17:34 PST 2008; root:xnu-1228.4.31~1/RELEASE_I386 i386
Jun 25 21:35:06 Chimera kernel[0]: CODE SIGNING: cs_invalid_page: p=1271[Dock] clearing CS_VALID
Jun 25 21:35:06 Chimera synergys[1262]: Can't register screen saver connection 'com.apple.ScreenSaverDaemon'
Jun 25 21:35:06 Chimera com.apple.loginwindow[1248]: 2008-06-25 21:35:06.145 synergys[1262:10b] Can't register screen saver connection 'com.apple.ScreenSaverDaemon'
Jun 25 21:35:06 Chimera kernel[0]: CODE SIGNING: cs_invalid_page: p=1272[ATSServer] clearing CS_VALID
Jun 25 21:35:06 Chimera kernel[0]: CODE SIGNING: cs_invalid_page: p=1274[systemUIServer] clearing CS_VALID
Jun 25 21:35:06 Chimera kernel[0]: CODE SIGNING: cs_invalid_page: p=1275[Finder] clearing CS_VALID
Jun 25 21:35:07 Chimera mDNSResponder[20]: mDNSPlatformRawTime: last_mach_absolute_time 8992FE1A00001758
Jun 25 21:35:07 Chimera mDNSResponder[20]: mDNSPlatformRawTime: this_mach_absolute_time 89605E8900001758
Jun 25 21:35:07 Chimera mDNSResponder[20]: mDNSPlatformRawTime: last_mach_absolute_time 8BAA6E8800001758
Jun 25 21:35:07 Chimera mDNSResponder[20]: mDNSPlatformRawTime: this_mach_absolute_time 8B868DDD00001758
Jun 25 21:35:08 Chimera kernel[0]: CODE SIGNING: cs_invalid_page: p=1281[lssave] clearing CS_VALID
Jun 25 21:35:09 Chimera com.apple.loginwindow[1248]: WARNING: could not map hotkey id=ef14 mask=0000
Jun 25 21:35:09 Chimera com.apple.loginwindow[1248]: NOTE: started server
Jun 25 21:35:09 Chimera com.apple.loginwindow[1248]: INFO: screen "Chimera" shape changed
Jun 25 21:35:16 Chimera kernel[0]: CODE SIGNING: cs_invalid_page: p=1298[Console] clearing CS_VALID
Jun 25 21:35:26 Chimera IntegoStatusItemHelper[1270]: NetUpdateISIPlugin: Failed to get settings. Cannot initialize.
Jun 25 21:35:35 Chimera mDNSResponder[20]: mDNSPlatformRawTime: last_mach_absolute_time 026A59540000175F
Jun 25 21:35:35 Chimera mDNSResponder[20]: mDNSPlatformRawTime: this_mach_absolute_time 0244E9310000175F
Jun 25 21:35:35 Chimera mDNSResponder[20]: mDNSPlatformRawTime went backwards by 1 ticks; setting correction factor to -1473993655
Jun 25 21:35:26 Chimera [0x0-0x70070].com.intego.IntegoStatusItemHelper[1270]: NetUpdate - NetU
Jun 25 21:35:26 Chimera [0x0-0x70070].com.intego.IntegoStatusItemHelper[1270]: pdateGetSettings CFMessagePortSen
Jun 25 21:35:26 Chimera [0x0-0x70070].com.intego.IntegoStatusItemHelper[1270]: dRequest error : -2
Jun 25 21:35:35 Chimera /Library/Intego/integod[97]: Intego menu launched
Jun 25 21:36:00 Chimera mDNSResponder[20]: mDNSPlatformRawTime: last_mach_absolute_time D05B75C300001764
Jun 25 21:36:00 Chimera mDNSResponder[20]: mDNSPlatformRawTime: this_mach_absolute_time D03E010800001764

 

 

 

WindowServer_2008_06_25_213457_Chimera.crash.zip

 

 

Thanks,

Ricky

[x2PhaNToM]

Same problem here. Maybe this is caused by the dual core AMD cpu? Never had something like this with my single core Athlon 3500+.

Will try cpus=1 next time.

[tehnick]

Exact same problem here... and I've experienced this on my system since I started using OS X - through all the different installs I've had.

 

The problem with using cpus=1 (besides using only 1 core) is that this really is random it seems, and it can take a day or two for it to occur again, a few hours, or even just a few minutes after booting up. So it's hard to determine if the problem really has disappeared, and who wants to run their computer for an extended period of time with 1 core disabled?

 

AMD 64 X2 5000+ running 10.5.4 & latest kernel (9.4.0) right now. I have the Asus M2N-SLI motherboard, which is similar to yours aswell Phantom.

[Rickyroko]

Well guys,

 

I know this may sound a little like giving up, but I uninstalled Mac OS X and went back to my old ways with Gentoo Linux.

 

I still have Mac OS X installed on my Laptop, It has a Dual Core Intel processor.

 

Post a solution if you find one because I really enjoyed using OS X as the main OS, but with that instability I couldn't really count on my Computer.

 

Good Luck,

 

Rickyroko

[tehnick]

A little update for me, and a temporary solution:

 

After doing some more research, I came across this topic:

 

http://forum.infinitemac.com/discussion/75...-aka-mouse-bug/

 

They've been analyzing this "issue" for quite a while, it's a 15-page thread dating back to March. This indeed deals with AMD dual-core.

 

After lots of contemplating, I finally decided to disable one of my cores. I really wasn't sure if this would fix the problems I was having so I held off doing so for a long time. Results? Improved responsiveness, and this is responsiveness that I never realized was supposed to be there. I realized this issue has a lot more to do with then just application crashes! EVERY thing you do seems to be bogged down by this bug; now that I've disabled the other core everything from typing, moving the mouse, cycling through menus, is MUCH more responsive and "faster" (and I never had a problem with it before, nor did I suspect). Now the obvious disadvantages is that you're using only 1 core; but IMHO the difference without having this bug is well worth the trade off. This should be a temporary solution, but so far no one has come up with a direct answer to fixing the bug...

 

To set to OS X to use only one core, edit your /Library/Preferences/SystemConfiguration/com.apple.Boot.plist; Add the kernel flag "cpus=1"; Reboot

 

I'm awfully tired of these application crashes, system crashes, having to reboot all the time from the "symptoms", etc. so for me this is my only choice. In the future I plan on going with an Intel build, but I hope there is a fix for this by then. I think this affects ALL AMD dual core users, just in more ways for some then others (and most don't even realize it).

[Rickyroko]

That was a very good find tehnick, thanks for finding a temporary solution to the issue.

My temporary "solution" was to download and install a seperate "driver" for managing the mouse issue.

 

I can't really remember the name of this program, but It was written by a Japanese author.

 

This was a fix to the symptoms, not the problem. I decided that I cannot use a computer that I cannot depend on to be stable. At any point, my Documents would close or WindowServer would crash. It seems that the crashes would all occur at around the same time, and they all involve the subroutine that I discussed earlier.

 

Unfortunately, all my data is in HFS+ formatted partitions, so I will have to transfer that data later.

 

Hopefully, an improvement to the modified kernels will allow our AMD Dual Cores to run on OS X.

 

Rickyroko

[x2PhaNToM]

After disabling one core i realized almost no difference in responsiveness, but my mous was a lot smooter while moving.

Now i've found something about that OSX can't manage X.5 multiplicators well. Now trying a 12x multiplicator instead of 12.5 with my Athlon 4800+. Will report the results later.

[x2PhaNToM]

Same problem as before ... only cups=1 helps ... :censored2:

[tehnick]

Changing CPU speeds isn't going to make a difference. It's the way multi-processors are used in OS X that brings issues with AMD's dual core; no matter what speed they're running at.

 

The "mouse" issue is one of MANY symptoms of this bug; you may or may not have that problem (I never did) but the application crashes really did me in. The one with lesser symptoms are likely to be the ones who never realize something is wrong. This has been going on since the beginning of OSX86, and I fear we may not see any fix for it because it's not as noticable for some as it is others - but it's still there. It was mentioned that JaS discovered this back in the earliest releases of Tiger.