SwithDrawn Posted December 7, 2009 Share Posted December 7, 2009 [solved - see post 8] Update 6.1.10: I'm bringing this back from the dead because I still have this problem: Kernel panics 5-10 minutes after each boot unless I run a disk permissions repair. Sometimes it panics while the permissions repair is in progress (always at 70%). The system will lock and the harddrive audibly stops accessing - sometimes it panics, sometimes it recovers. The panics are either pmap_flush_tlbs() timeout: "cpu(s) failing to respond to interrupts" as described above OR "Spinlock acquisition timed out". The panics are not related to anything in the permissions repair log. For awhile the log was empty, it repaired no files but it would still panic. I've done multiple clean installs using multiple harddrives. I've also tried using both 32 and 64 bit modes for Snow Leopard. I read a thread about the spinlock timeout problem. It seems setting a boot-arg that effectively sets the timeout to a really high number might solve this. The boot-arg is slto_us=0xffffffff. I put this in my com.apple.boot.plist under Kernel Flags but that caused a panic while booting - I'm not sure if this is the correct way to set this boot-arg. Any ideas?? Current thought is it might be AHCI or motherboard related... I will test this next. Original Post -------------------------------- I've installed Snow Leopard (10.6.2) on my eight core xeon system. As with Leopard, everything went pretty smoothly. It works 100 percent, except for the fact that unless I repair permissions immediately after booting, a multi-language shutdown screen will appear invariably after 5-10 minutes. The Event Log shows nothing useful. However, the repair permissions always repairs the same file, IOATAFamily.kext/Contents/PlugIns/AppleIntelPIIXATA.kext: Repairing permissions for “SnowLeopard”Warning: SUID file "System/Library/CoreServices/RemoteManagement/ARDAgent.app/Contents/MacOS/ARDAgent" has been modified and will not be repaired. Permissions differ on "System/Library/Extensions/IOATAFamily.kext/Contents/PlugIns/AppleIntelPIIXATA.kext/Contents/CodeDirectory", should be lrwxr-xr-x , they are -rwxr-xr-x . Repaired "System/Library/Extensions/IOATAFamily.kext/Contents/PlugIns/AppleIntelPIIXATA.kext/Contents/CodeDirectory". Permissions differ on "System/Library/Extensions/IOATAFamily.kext/Contents/PlugIns/AppleIntelPIIXATA.kext/Contents/CodeRequirements", should be lrwxr-xr-x , they are -rwxr-xr-x . Repaired "System/Library/Extensions/IOATAFamily.kext/Contents/PlugIns/AppleIntelPIIXATA.kext/Contents/CodeRequirements". Permissions differ on "System/Library/Extensions/IOATAFamily.kext/Contents/PlugIns/AppleIntelPIIXATA.kext/Contents/CodeResources", should be lrwxr-xr-x , they are -rwxr-xr-x . Repaired "System/Library/Extensions/IOATAFamily.kext/Contents/PlugIns/AppleIntelPIIXATA.kext/Contents/CodeResources". Permissions differ on "System/Library/Extensions/IOATAFamily.kext/Contents/PlugIns/AppleIntelPIIXATA.kext/Contents/CodeSignature", should be lrwxr-xr-x , they are -rwxr-xr-x . Repaired "System/Library/Extensions/IOATAFamily.kext/Contents/PlugIns/AppleIntelPIIXATA.kext/Contents/CodeSignature". Permissions repair complete Has anyone ran into this? It's somewhat annoying to have to spend four minutes each boot repairing permissions, and sometimes it crashes before the repair finishes and I have to start over. My system is as follows: Dual Quad Zeon E5405 Harpertown Procs (8 Cores) Intel 5000X Northbridge and ESB2 Southbridge EVGA Geforce 9800GTX 8GB Buffered ECC Ram Creative Audigy 2 ZS 10.6.2 with Chameleon. Link to comment Share on other sites More sharing options...
SwithDrawn Posted December 13, 2009 Author Share Posted December 13, 2009 This issue is getting worse. It doesn't make it through the permissions repairs about half of the time now. What's curious is that if you run the repair again, after it's finished, it repairs the same files with the same messages - "should be lrwxr-xr-x , they are -rwxr-xr-x". I've tried manually chmoding these files at startup instead of running the repair, it makes no difference - no matter what, it always corrects those particular permissions. I think just the *act* of running the repair is, for some reason, averting the crashes. Are there different levels of verbosity for permissions repair? Is it doing something here behind the scenes that's fixing everything? Link to comment Share on other sites More sharing options...
c.craigen Posted December 15, 2009 Share Posted December 15, 2009 What it looks like it's saying is that those files should be symbolic links (that's what the "l" in lrwxr-xr-x means vs. -rwxr-xr-x). That can't be fixed by a simple permission repair although it will show up as a permissions issue. You've got to go in and fix those manually. Find the original file off of your install disc (or flash drive) and try that one instead. If you use a patched version, try a fresh copy of the patched copy. As for why this is causing kernel panics, I have no idea. I've seen a similar problem when the kext is initially loading, but not relating to those specific files. Link to comment Share on other sites More sharing options...
SwithDrawn Posted December 18, 2009 Author Share Posted December 18, 2009 You were right, they weren't linked correctly. I copied the files from the installation dvd and repair permissions doesn't complain about it anymore. HOWEVER, the freezing problem still remains, I guess those files had nothing to do with it. I noticed it always freezes at 70 percent. It's definitely doing something. Link to comment Share on other sites More sharing options...
c.craigen Posted December 19, 2009 Share Posted December 19, 2009 Well, that could be another problem related to the same kext. I'm having my own set of issues regarding the IOATAFamily kext right now... What exactly is it saying when it panics? Boot with -v option, that way when it panics it'll output a log to the screen so we can see what's causing the crash. Link to comment Share on other sites More sharing options...
SwithDrawn Posted December 21, 2009 Author Share Posted December 21, 2009 I got the log. Wow, I didn't know booting verbosely would enable outputting logs after boot. Learn something new everyday... Here's the log: panic(cpu 2 caller 0xffffff80002247a0) : "pmap_flush_tlbs() timeout: "cpu(s) failing to respond to interrupts. pmap=0xffffff80116a32b0 cpus_to_respond=8x40"@/SourceCache/xnu/xnu-1486.2.11/osfmk/x86_64/pmap.c:4267Debugger called: <panic> Backtrace (CPU 2), Frame : Return Address 0xffffff80ae9f3970 : 0xffffff8000204ae6 0xffffff80ae9f3a70 : 0xffffff80002247a0 0xffffff80ae9f3ae0 : 0xffffff8000225645 0xffffff80ae9f3b60 : 0xffffff800021ca9b 0xffffff80ae9f3bc0 : 0xffffff800021d08f 0xffffff80ae9f3c30 : 0xffffff80002a4102 0xffffff80ae9f3d40 : 0xffffff80002a4ed4 0xffffff80ae9f3df0 : 0xffffff800028030c 0xffffff80ae9f3e30 : 0xffffff80004769c9 0xffffff80ae9f3e60 : 0xffffff8000476af3 0xffffff80ae9f3ea0 : 0xffffff8000476ee8 0xffffff80ae9f3f00 : 0xffffff800047706f 0xffffff80ae9f3f40 : 0xffffff80004df530 0xffffff80ae9f3fa0 : 0xffffff80002df7a4 BSD process name corresponding to current thread: coreservicesd Mac OS version: 10C540 Kernel version: Darwin Kernel Version 10.2.0: Tue Nov 3 10:35:19 PST 2009; root:xnu-1486.2.11~1/RELEASE_X86_64 System model name: MacPro2,1 System uptime in nanoseconds: 111516530403 Second crash log, very similar message: panic(cpu 5 caller 0xffffff80002247a0) : "pmap_flush_tlbs() timeout: "cpu(s) failing to respond to interrupts. pmap=0xffffff800067d8a0 cpus_to_respond=0x2"@/SourceCache/xnu/xnu-1486.2.11/osfmk/x86_64/pmap.c:4267Debugger called: <panic> Backtrace (CPU 5), Frame : Return Address 0xffffff80aea03a50 : 0xffffff8000204ae6 0xffffff80aea03b50 : 0xffffff80002247a0 0xffffff80aea03bc0 : 0xffffff8000224bd4 0xffffff80aea03c40 : 0xffffff8000225889 0xffffff80aea03c90 : 0xffffff8000217d96 0xffffff80aea03d90 : 0xffffff8000218184 0xffffff80aea03dc0 : 0xffffff80005483a0 0xffffff80aea03e20 : 0xffffff80002ba8b6 0xffffff80aea03e60 : 0xffffff8000205306 0xffffff80aea03e90 : 0xffffff8000202961 0xffffff80aea03ef0 : 0xffffff800027158e 0xffffff80aea03f60 : 0xffffff80002bf389 0xffffff80aea03fa0: 0xffffff80002df7c4 BSD process name corresponding to current thread: WindowServer Mac OS version: 10C540 Kernel version: Darwin Kernel Version 10.2.0: Tue Nov 3 10:35:19 PST 2009; root:xnu-1486.2.11~1/RELEASE_X86_64 System model name: MacPro2,1 System uptime in nanoseconds: 134673553705 All googling suggests faulty hardware, which I doubt in my case because of the fact that a permissions repair somehow solves everything. Link to comment Share on other sites More sharing options...
SwithDrawn Posted June 1, 2010 Author Share Posted June 1, 2010 I'm bringing this back from the dead because I still have this problem: Kernel panics 5-10 minutes after each boot unless I run a disk permissions repair. Sometimes it panics while the permissions repair is in progress (always at 70%). The system will lock and the harddrive audibly stops accessing - sometimes it panics, sometimes it recovers. The panics are either pmap_flush_tlbs() timeout: "cpu(s) failing to respond to interrupts" as described above OR "Spinlock acquisition timed out". The panics are not related to anything in the permissions repair log. For awhile the log was empty, it repaired no files but it would still panic. I've done multiple clean installs using multiple harddrives. I've also tried using both 32 and 64 bit modes for Snow Leopard. I read a thread about the spinlock timeout problem. It seems setting a boot-arg that effectively sets the timeout to a really high number might solve this. The boot-arg is slto_us=0xffffffff. I put this in my com.apple.boot.plist under Kernel Flags but that caused a panic while booting - I'm not sure if this is the correct way to set this boot-arg. Any ideas?? Current thought is it might be AHCI or motherboard related... I will test this next. I'm on 10.6.3 with Chameleon, using EvOreboot, myHack's IOATAFamily.kext, IONetworkingFamily.kext, and NullCPUPOwerManagement.kext. System: Dual Quad Zeon E5405 Harpertown Procs (8 Cores) Intel 5000X Northbridge and ESB2 Southbridge EVGA Geforce 9800GTX 8GB Buffered ECC Ram Creative Audigy 2 ZS Link to comment Share on other sites More sharing options...
SwithDrawn Posted May 30, 2011 Author Share Posted May 30, 2011 This has been solved for those interested - it turns out to be related to the graphics card. I had suspected hardware because (for a time) similar symptoms were appearing in Windows as well. The 9800GTX died recently (artifacting, freezes in both Win and OSX). It was replaced with a GTX260. Problem solved - no more having to repair permissions on each boot. I have NO IDEA why repairing permissions would have fixed a hardware incongruity with the video card. Link to comment Share on other sites More sharing options...
Recommended Posts