I've been using your program for months with great results. This weekend I upgraded to Leopard, though, and it looks like I ran across some incompatibilities. DoubleCommand works fine while the machine is running, but the computer ends up crashing when I try to reboot or shut it down. From what I can tell from the stack trace, it might have something to do with unloading of the kernel extension. Here is the stack trace, if that will make troubleshooting easier. Thanks again for an amazingly useful software,
Pedja
I have also gotten two BSODs related to DoubleCommand since upgrading two Leopard and have been forced to remove it for the time being. In my context it might have something to do with the new Apple Wireless Keyboard giving me "Connection Lost" messages from time to time, as both crashes happened right after that icon showed up on my desktop but in the stack trace it made mention of com.???.driver.DoubleCommand. Sorry I didn't think to save you guys the stack traces. I've also had two other BSODs which might or might not have been related to this. Cheers!
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I've a similar problem with Tiger. I use a PC IBM 1989 Keyboard with ps2 to usb converter with dual G4. In my case, it's not a DoubleCommand problem. I got crash on boot or reboot if I haven't Apple keyboard installed. With Apple keyboard attached, the boot is fine and all works ok.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
i.e. at KeyBehaviorManager::removeKeyboard(IOHIKeyboard*)+64 (i.e. in the DoubleCommand kext) it tried to access 0x61657bc4. This is an invalid address. Note this also the value found in register ebx.
Hopefully this will help diagnose the problem further
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Thanks for the problem reports. With the information that has come in so far (and special thanks to Brian for the decoded backtrace!), I've created a build of DoubleCommand identical to 1.6.5, except for the removal of a patch that could be causing this issue. If you decide to try it, please let me know how it goes for you.
It seems that the provided version does not have the problem anymore. At least the computer rebooted once without problem and shut down correctly twice. Tanks Jerome
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I downloaded the latest version from svn ( the version where you took out the two patches ), compiled the kext and replaced it, seems to work fine now.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
The first entry on each line is the stack frame. The second entry is the return address, which is the interesting part for our purposes. We need to grab each of those second entries and decode them, to see the backtrace with symbols.
I've done this enough so I can easily edit the line with a BBEdit script; the resulting gdb commands which display the backtrace look like this:
x/i 0x12b0e1
x/i 0x1a7bed
x/i 0x19e517
x/i 0x4dd59a86
x/i 0x4dd58563
x/i 0x4dd585f0
x/i 0x40aa2d
x/i 0x40b622
x/i 0x40edd8
x/i 0x40ff5c
x/i 0x40958e
x/i 0x6cc3ab
x/i 0x6e2591
x/i 0x6e45b6
x/i 0x6e4c7b
x/i 0x6e4fdc
you could do this manually as well, of course.
I will feed those gdb commands into gdb after I set things up. The set up requires that I get the symbols for DoubleCommand from the kext, and the mach_kernel from Apple.
From the lines:
Kernel loadable modules in backtrace (with dependencies):
com.baltaks.driver.DoubleCommand(1.6.2)@0x4dd57000->0x4dd5bfff
dependency: com.apple.iokit.IOHIDFamily(1.5)@0x6c0000
com.apple.iokit.IOHIDFamily(1.5)@0x6c0000->0x6f7fff
in the backtrace, we know that DoubleCommand is in the backtrace.
Download the DoubleCommand dmg.
Mount the dmg. This creates /Volumes/DoubleCommand
Make a temporary directory to hold the pax.gz file, and copy it to the temporary directory
mkdir ~/d
cp /Volumes/DoubleCommand/DoubleCommand1.6.5_10.4.pkg/Contents/Archive.pax.gz ~/d/
Unarchive the pax.gz file
cd ~/d
pax -rzf Archive.pax.gz
Switch to the directory with the kext in it.
cd ./Library/StartupItems/DoubleCommand/
build the symbol file for the kext, using kextload and the addresses from the panic log "Kernel loadable modules in backtrace (with dependencies)" listing:
kextload -z -n -a com.baltaks.driver.DoubleCommand@0x4dd57000 -a com.apple.iokit.IOHIDFamily@0x6c0000 -s ~/d -- DoubleCommand.kext
Download the 10.5 Kernel Debug Kit from http://developer.apple.com/sdk (this requires signing up for a free ADC membership)
Mount that dmg. This creates /Volumes/KernelDebugKit
Invoke gdb, passing it the mach_kernel from /Volumes/KernelDebugKit. Read in the kgmacros file, and the two symbol files you just created:
$ gdb /Volumes/KernelDebugKit/mach_kernel
GNU gdb 6.3.50-20050815 (Apple version gdb-768) (Tue Oct 2 04:07:49 UTC 2007)
Copyright 2004 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General Public License, and you are
welcome to change it and/or distribute copies of it under certain conditions.
Type "show copying" to see the conditions.
There is absolutely no warranty for GDB. Type "show warranty" for details.
This GDB was configured as "i386-apple-darwin"...
(gdb) source /Volumes/KernelDebugKit/kgmacros
Loading Kernel GDB Macros package. Type "help kgm" for more info.
(gdb) add-symbol-file ~/d/com.baltaks.driver.DoubleCommand.sym
add symbol table from file "/Volumes/Data/Users/bb/d/com.baltaks.driver.DoubleCommand.sym"? (y or n) y
Reading symbols from /Volumes/Data/Users/bb/d/com.baltaks.driver.DoubleCommand.sym...done.
(gdb) add-symbol-file ~/d/com.apple.iokit.IOHIDFamily.sym
add symbol table from file "/Volumes/Data/Users/bb/d/com.apple.iokit.IOHIDFamily.sym"? (y or n) y
Reading symbols from /Volumes/Data/Users/bb/d/com.apple.iokit.IOHIDFamily.sym...done.
(gdb) x/i 0x12b0e1
0x12b0e1 <panic+422>: mov 0x4cf278,%eax
(gdb) x/i 0x1a7bed
0x1a7bed <kernel_trap+1450>: add $0xcc,%esp
(gdb) x/i 0x19e517
0x19e517 <trap_from_kernel+26>: mov %edi,%esp
(gdb) x/i 0x4dd59a86
0x4dd59a86 <remap(unsigned int*, unsigned int*, unsigned int*, unsigned int*, unsigned int*, unsigned int*, unsigned int*, unsigned int*, bool*, UnsignedWide*)+2890>: add %edi,0x0(%eax,%eax,1)
(gdb) x/i 0x4dd58563
0x4dd58563 <hijack_keyboard(IOHIKeyboard*)+153>: test %eax,%eax
(gdb) x/i 0x4dd585f0
0x4dd585f0 <hijack_keyboard(IOHIKeyboard*)+294>: (bad)
(gdb) x/i 0x40aa2d
0x40aa2d <IOService::invokeNotifer(_IOServiceNotifier*)+127>: mov %eax,%esi
(gdb) x/i 0x40b622
0x40b622 <IOService::deliverNotification(OSSymbol const*, unsigned long, unsigned long)+280>: mov -0x1c(%ebp),%edx
(gdb) x/i 0x40edd8
0x40edd8 <IOService::doServiceMatch(unsigned long)+500>: mov 0x24(%ebx),%edx
(gdb) x/i 0x40ff5c
0x40ff5c <IOService::startMatching(unsigned long)+352>: mov (%ebx),%eax
(gdb) x/i 0x40958e
0x40958e <IOService::registerService(unsigned long)+370>: add $0x2c,%esp
(gdb) x/i 0x6cc3ab
0x6cc3ab <IOHIKeyboard::start(IOService*)+51>: add %al,0x4f(%ecx,%eax,1)
(gdb) x/i 0x6e2591
0x6e2591 <IOHIDKeyboard::start(IOService*)+519>: add %al,(%eax)
(gdb) x/i 0x6e45b6
0x6e45b6 <IOHIDEventService::newKeyboardShim(unsigned long, unsigned long)+536>: add %al,(%eax)
(gdb) x/i 0x6e4c7b
0x6e4c7b <IOHIDEventService::parseSupportedElements(OSArray*, unsigned long)+1155>: add %ch,(%esi)
(gdb) x/i 0x6e4fdc
0x6e4fdc <IOHIDEventService::start(IOService*)+820>: test %al,(%ecx)
(gdb)
I removed the extraneous instruction information from the backtrace I posted, because I haven't actually loaded the code for DoubleCommand; I just loaded symbols and told gdb "pretend that DoubleCommand is loaded at 0x4dd57000, and that IOHIDFamily is at 0x6c0000". So the assembly instructions in my backtrace are useless, but the addresses inside the kext are correct.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I'm also having the kernel panic problem since upgrading (I know, I know) to Leopard. The panic shows up about 1 out of 4 times when I shut down the machine (MacBook Pro), seemingly at random but consistently I completely forgot about having DoubleCommand installed and upgraded with it running. Duh. After seeing DoubleCommand mentioned in the crash report, I uninstalled it. It didn't completely uninstall and is still running (I know that because my reassigned delete key is working). The DoubleCommand icon is gone and I can't find a way to stop it or uninstall it manually. I tried reinstalling over the top of it, but not surprisingly, the install invariably hangs, including the new 1.6.6b1 version. I'm hoping not to have to do a clean Leopard install to clean up the mess. Is there a way?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
The core of DoubleCommand lives in /Library/StartupItems/DoubleCommand. If you delete that folder and restart, DoubleCommand is gone. You can optionaly remove the prefpane as well by removing /Library/PreferencePanes/DoubleCommandPreferences.prefPane.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I experienced same problem while reboot and shutdown.
And it seemed to be fixed by 1.6.6b1 which I have installed about a week ago.
But, I have encountered another problem yesterday that the setting of doublecommand could not be saved.
In fact, it can be saved when I first save on each reboot.
But while running for hours, the setting will be gone and cannot save anymore until reboot again.
I really love this cool software with my MacBook.
I hope it will be fixed soon.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
HW: Mac Pro + new aluminum Apple keyboard
OS: 10.5.1
DoubleCommand: 1.6.6.b1
Since this is not a laptop, I've got a full keyboard - I only tried out DoubleCommand to get working Home and End keys. It worked alright, but with "PC style home and end keys" as the only active setting the Ctrl key stops working (both right and left). After reading the messages in this thread, I didn't try to reboot.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi All, I'm going crazy because I deleted DoubleCommand (great program, btw - used it daily with great effect until Leopard) from startupitems as well as the prefpane, and I am STILL getting kernel panics when I plug in or unplug my USB keyboard (cheap Viewsonic Viewmate, model no. VSAC27936-1M).
I keep getting the message pertaining to com.baltaks.doublecommand.driver causing the crash, and I've spent a fair bit of time looking for said driver to no avail. My two theories are: 1. it's in the kextcache or something to that effect, or 2. it's hidden somewhere within a package or framework or something that isn't searchable and I haven't thought to look.
Someone please help! Thanks!
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
DoubleCommand is not hidden anywhere you've missed, if you've deleted the /Library/StartupItems/DoubleCommand folder and the DoubleCommandPreferences.prefpane and then restarted, it should be gone. If not, then your theory of the kextcache is possible - try deleting it, /System/Library/Extentions.kextcache.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I had to uninstall DoubleCommand 1.6.5 before installing 1.6.6a (like one of the posters, if I didn't uninstall first, the installer would hang... actually, it took up all my CPU and I couldn't get my computer to respond).
1.6.6a works the first time I install it, but the instant I reboot, it doesn't start up automatically. Any ideas?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Sorry to post again, but could this be the issue? When I try to run DoubleCommand from the command line (/Library/StartupItems/DoubleCommand/DoubleCommand), I get this:
touch: ./DoubleCommand.crashed: Permission denied
Loading DoubleCommand
you must be running as root to load modules into the kernel
dc.config: 4 -> 4
I have to use sudo to run as root. Could this be the cause?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
As I mentioned, I can sudo and it runs properly. I just doesn't start when OS X starts up.
Here's what I found in system.log when I reboot:
Feb 21 14:53:13 Allegro sudo[259]: astonlau : TTY=ttys000 ; PWD=/Library/StartupItems/DoubleCommand ; USER=root ; COMMAND=/bin/rm DoubleCommand.crashed
Feb 21 14:53:40 Allegro SystemStarter[18]: "/Library/StartupItems/DoubleCommand" failed sanity check: path was created after boot up
Feb 21 14:54:34 localhost com.apple.loginwindow[25]: Configuring DoubleCommand
Feb 21 14:54:34 localhost com.apple.loginwindow[25]: DoubleCommand configured, it is now safe to close this window.
Feb 21 14:54:37 Allegro SystemStarter[18]: "/Library/StartupItems/DoubleCommand" failed sanity check: path was created after boot up
It's strange -- I wonder what's causing the DoubleCommand directory's date to update?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Aha, the temporary file that we use to make sure we didn't crash last load is kept in the same folder as the kext and the startup items, I'll move that to /tmp and that might help.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi,
I've been using your program for months with great results. This weekend I upgraded to Leopard, though, and it looks like I ran across some incompatibilities. DoubleCommand works fine while the machine is running, but the computer ends up crashing when I try to reboot or shut it down. From what I can tell from the stack trace, it might have something to do with unloading of the kernel extension. Here is the stack trace, if that will make troubleshooting easier. Thanks again for an amazingly useful software,
Pedja
Stack trace from most recent crash:
Sun Oct 28 09:12:56 2007
panic(cpu 0 caller 0x001A7BED): Kernel trap at 0x51248214, type 14=page fault, registers:
CR0: 0x8001003b, CR2: 0x372d2b69, CR3: 0x00d4f000, CR4: 0x00000660
EAX: 0x2e5fc918, EBX: 0x372d2619, ECX: 0x00507444, EDX: 0x04a1e754
CR2: 0x372d2b69, EBP: 0x2e31bcb8, ESI: 0x04070000, EDI: 0x5124a3e8
EFL: 0x00010202, EIP: 0x51248214, CS: 0x00000008, DS: 0x00000010
Error code: 0x00000000
Backtrace, Format - Frame : Return Address (4 potential args on stack)
0x2e31bac8 : 0x12b0e1 (0x4555b4 0x2e31bafc 0x133238 0x0)
0x2e31bb18 : 0x1a7bed (0x45e568 0x51248214 0xe 0x45dd24)
0x2e31bbf8 : 0x19e517 (0x2e31bc10 0x43e3600 0x2e31bcb8 0x51248214)
0x2e31bc08 : 0x51248214 (0xe 0x520048 0x10 0x10)
0x2e31bcb8 : 0x5124676b (0x5124a428 0x4070000 0x2e60efc8 0x3e75aa)
0x2e31bce8 : 0x512467ee (0x4070000 0x2e61a720 0x4070000 0x34564f1c)
0x2e31bd08 : 0x40aa2d (0x4909080 0x0 0x4070000 0x3e81d9)
0x2e31bd38 : 0x40b622 (0x4070000 0x421e740 0x0 0x4070000)
0x2e31bd78 : 0x40f74c (0x4070000 0x3aedc40 0x0 0xffffffff)
0x2e31bdc8 : 0x40f8ff (0x3f9ec00 0x7 0x1 0x3e840c)
0x2e31bde8 : 0x34940606 (0x3f9ec00 0x3 0x1 0x3f34f00)
0x2e31be08 : 0x34d2fb02 (0x3f9ec00 0x3 0x0 0x3f4fc00)
0x2e31be28 : 0x34d2fc89 (0x3f34f00 0x3495a064 0x1 0x349982e0)
0x2e31be68 : 0x34d2be04 (0x3f34f00 0xfe941630 0x2e31bed8 0x1a63e0)
0x2e31bea8 : 0x34d2d8f1 (0x3eec200 0x34d2c6e8 0x1 0x0)
0x2e31bed8 : 0x412cae (0x3eec200 0x0 0x3eec200 0x3eec200)
Backtrace continues...
Kernel loadable modules in backtrace (with dependencies):
com.baltaks.driver.DoubleCommand(1.6.5)@0x51245000->0x5124afff
dependency: com.apple.iokit.IOHIDFamily(1.5)@0x2e5e3000
com.apple.driver.AppleUSBHub(3.0.3)@0x34d2a000->0x34d34fff
dependency: com.apple.iokit.IOUSBFamily(3.0.3)@0x34933000
com.apple.iokit.IOUSBFamily(3.0.3)@0x34933000->0x3495afff
BSD process name corresponding to current thread: kernel_task
Mac OS version:
9A581
Kernel version:
Darwin Kernel Version 9.0.0: Tue Oct 9 21:35:55 PDT 2007; root:xnu-1228~1/RELEASE_I386
System model name: MacBookPro3,1 (Mac-F4238BC8)
I have also gotten two BSODs related to DoubleCommand since upgrading two Leopard and have been forced to remove it for the time being. In my context it might have something to do with the new Apple Wireless Keyboard giving me "Connection Lost" messages from time to time, as both crashes happened right after that icon showed up on my desktop but in the stack trace it made mention of com.???.driver.DoubleCommand. Sorry I didn't think to save you guys the stack traces. I've also had two other BSODs which might or might not have been related to this. Cheers!
Here's a stack trace caused from a crash after removing DoubleCommand but before rebooting, I hope it is of use:
Sun Oct 28 15:19:31 2007
panic(cpu 1 caller 0x001A7BED): Kernel trap at 0x4dd59a86, type 14=page fault, registers:
CR0: 0x8001003b, CR2: 0x00000551, CR3: 0x0105f000, CR4: 0x000006e0
EAX: 0x00000001, EBX: 0x03c0c000, ECX: 0x00507444, EDX: 0x006d9794
CR2: 0x00000551, EBP: 0x2e55f958, ESI: 0x0000000c, EDI: 0x4dd5bac8
EFL: 0x00010202, EIP: 0x4dd59a86, CS: 0x00000004, DS: 0x2e55000c
Error code: 0x00000000
Backtrace, Format - Frame : Return Address (4 potential args on stack)
0x2e55f798 : 0x12b0e1 (0x4555b4 0x2e55f7cc 0x133238 0x0)
0x2e55f7e8 : 0x1a7bed (0x45e568 0x4dd59a86 0xe 0x45dd24)
0x2e55f8c8 : 0x19e517 (0x2e55f8e8 0x0 0x1 0x31c4cd00)
0x2e55f8e0 : 0x4dd59a86 (0xe 0x48 0xc 0x3c0000c)
0x2e55f958 : 0x4dd58563 (0x4dd5bac8 0x3c0c000 0x6ebfc8 0x3e75aa)
0x2e55f988 : 0x4dd585f0 (0x3c0c000 0x6f7720 0x2e55f9b8 0x3e810c)
0x2e55f9a8 : 0x40aa2d (0x4959600 0x0 0x3c0c000 0x3e81d9)
0x2e55f9d8 : 0x40b622 (0x3c0c000 0x40b01c0 0x0 0x40de83)
0x2e55fa18 : 0x40edd8 (0x3c0c000 0x3c00bc0 0x4 0xffffffff)
0x2e55fa68 : 0x40ff5c (0x3c0c000 0x2 0x1 0x3c0c000)
0x2e55faa8 : 0x40958e (0x3c0c000 0x2 0x0 0x3be4b80)
0x2e55fae8 : 0x6cc3ab (0x3c0c000 0x2 0x2e55fb18 0x3c0c000)
0x2e55fb08 : 0x6e2591 (0x3c0c000 0xb0e7000 0x0 0x0)
0x2e55fb38 : 0x6e45b6 (0x3c0c000 0xb0e7000 0x48766e0 0x0)
0x2e55fb78 : 0x6e4c7b (0xb0e7000 0x9e207f 0x1 0x3e810c)
0x2e55fbe8 : 0x6e4fdc (0xb0e7000 0x4091b40 0x0 0x0)
Backtrace continues...
Kernel loadable modules in backtrace (with dependencies):
com.baltaks.driver.DoubleCommand(1.6.2)@0x4dd57000->0x4dd5bfff
dependency: com.apple.iokit.IOHIDFamily(1.5)@0x6c0000
com.apple.iokit.IOHIDFamily(1.5)@0x6c0000->0x6f7fff
BSD process name corresponding to current thread: kernel_task
Mac OS version:
9A581
Kernel version:
Darwin Kernel Version 9.0.0: Tue Oct 9 21:35:55 PDT 2007; root:xnu-1228~1/RELEASE_I386
System model name: MacBookPro1,1 (Mac-F425BEC8)
I've a similar problem with Tiger. I use a PC IBM 1989 Keyboard with ps2 to usb converter with dual G4. In my case, it's not a DoubleCommand problem. I got crash on boot or reboot if I haven't Apple keyboard installed. With Apple keyboard attached, the boot is fine and all works ok.
I just realized I didn't give my computer model in my original email. I am running a MacBook Pro and am using built-in keyboard.
Cheers,
pedja
I can only agree with Pedja's analysis. Here is my trace file that I got with the keyboard of my MacBook Pro (latest revision).
Sun Oct 28 19:34:45 2007
panic(cpu 1 caller 0x001A7BED): Kernel trap at 0x50e82214, type 14=page fault, registers:
CR0: 0x8001003b, CR2: 0x61657bc4, CR3: 0x013cd000, CR4: 0x00000660
EAX: 0x0070b918, EBX: 0x61657674, ECX: 0x00507444, EDX: 0x05463d54
CR2: 0x61657bc4, EBP: 0x344d7cb8, ESI: 0x04406000, EDI: 0x50e843e8
EFL: 0x00010206, EIP: 0x50e82214, CS: 0x00000008, DS: 0x00000010
Error code: 0x00000000
Backtrace, Format - Frame : Return Address (4 potential args on stack)
0x344d7ac8 : 0x12b0e1 (0x4555b4 0x344d7afc 0x133238 0x0)
0x344d7b18 : 0x1a7bed (0x45e568 0x50e82214 0xe 0x45dd24)
0x344d7bf8 : 0x19e517 (0x344d7c10 0x47bea00 0x344d7cb8 0x50e82214)
0x344d7c08 : 0x50e82214 (0xe 0x520048 0x4460010 0x10)
0x344d7cb8 : 0x50e8076b (0x50e84428 0x4406000 0x71dfc8 0x3e75aa)
0x344d7ce8 : 0x50e807ee (0x4406000 0x729720 0x4406000 0x8d4f1c)
0x344d7d08 : 0x40aa2d (0x5490c00 0x0 0x4406000 0x3e81d9)
0x344d7d38 : 0x40b622 (0x4406000 0x4636fc0 0x0 0x4406000)
0x344d7d78 : 0x40f74c (0x4406000 0x416bc40 0x0 0xffffffff)
0x344d7dc8 : 0x40f8ff (0x43fc400 0x7 0x1 0x3e840c)
0x344d7de8 : 0x619606 (0x43fc400 0x3 0x1 0x43ee200)
0x344d7e08 : 0xb2bb02 (0x43fc400 0x3 0x0 0x43ee800)
0x344d7e28 : 0xb2bc89 (0x43ee200 0x633064 0x1 0x8b02e0)
0x344d7e68 : 0xb27e04 (0x43ee200 0x2e9a2000 0x344d7eb8 0x1361b0)
0x344d7ea8 : 0xb298f1 (0x4343200 0xb286e8 0x1 0x0)
0x344d7ed8 : 0x412cae (0x4343200 0x1 0x4343200 0x4343200)
Backtrace continues...
Kernel loadable modules in backtrace (with dependencies):
com.baltaks.driver.DoubleCommand(1.6.5)@0x50e7f000->0x50e84fff
dependency: com.apple.iokit.IOHIDFamily(1.5)@0x6f2000
com.apple.driver.AppleUSBHub(3.0.3)@0xb26000->0xb30fff
dependency: com.apple.iokit.IOUSBFamily(3.0.3)@0x60c000
com.apple.iokit.IOUSBFamily(3.0.3)@0x60c000->0x633fff
BSD process name corresponding to current thread: kernel_task
Mac OS version:
9A581
Kernel version:
Darwin Kernel Version 9.0.0: Tue Oct 9 21:35:55 PDT 2007; root:xnu-1228~1/RELEASE_I386
System model name: MacBookPro3,1 (Mac-F4238BC8)
Well, I'd just like to say that I hope we can get this fixed. Without DoubleCommand, Mac laptops have no Delete key. Incredibly irritating.
The backtrace for this panic decodes as:
0x12b0e1 <panic+422>
0x1a7bed <kernel_trap+1450>
0x19e517 <trap_from_kernel+26>
0x51248214 <KeyBehaviorManager::removeKeyboard(IOHIKeyboard*)+64>
0x5124676b <return_keyboard(IOHIKeyboard*)+179>
0x512467ee <dc_terminated(com_baltaks_driver_DoubleCommand*, void*, IOService*)+34>
0x40aa2d <IOService::invokeNotifer(_IOServiceNotifier*)+127>
0x40b622 <IOService::deliverNotification(OSSymbol const*, unsigned long, unsigned long)+280>
0x40f74c <IOService::terminatePhase1(unsigned long)+274>
0x40f8ff <IOService::terminate(unsigned long)+27>
0x34940606 <IOUSBDevice::terminate(unsigned long)+104>
0x34d2fb02 <AppleUSBHubPort::RemoveDevice()+140>
0x34d2fc89 <AppleUSBHubPort::stop()+329>
0x34d2be04 <AppleUSBHub::StopPorts()+118>
0x34d2d8f1 <AppleUSBHub::HubPowerChange(unsigned long)+675>
0x412cae <IOService::driverSetPowerState()+132>
i.e. at KeyBehaviorManager::removeKeyboard(IOHIKeyboard*)+64 (i.e. in the DoubleCommand kext) it tried to access 0x61657bc4. This is an invalid address. Note this also the value found in register ebx.
Hopefully this will help diagnose the problem further
Thanks for the problem reports. With the information that has come in so far (and special thanks to Brian for the decoded backtrace!), I've created a build of DoubleCommand identical to 1.6.5, except for the removal of a patch that could be causing this issue. If you decide to try it, please let me know how it goes for you.
http://doublecommand.sourceforge.net/files/DoubleCommand1.6.6a1_10.4.dmg
It seems that the provided version does not have the problem anymore. At least the computer rebooted once without problem and shut down correctly twice. Tanks Jerome
I downloaded the latest version from svn ( the version where you took out the two patches ), compiled the kext and replaced it, seems to work fine now.
Brian: How did you decode the backtrace?
The backtrace in the panic log consists of lines which look like this:
Backtrace, Format - Frame : Return Address (4 potential args on stack)
0x2e55f798 : 0x12b0e1 (0x4555b4 0x2e55f7cc 0x133238 0x0)
0x2e55f7e8 : 0x1a7bed (0x45e568 0x4dd59a86 0xe 0x45dd24)
0x2e55f8c8 : 0x19e517 (0x2e55f8e8 0x0 0x1 0x31c4cd00)
0x2e55f8e0 : 0x4dd59a86 (0xe 0x48 0xc 0x3c0000c)
0x2e55f958 : 0x4dd58563 (0x4dd5bac8 0x3c0c000 0x6ebfc8 0x3e75aa)
0x2e55f988 : 0x4dd585f0 (0x3c0c000 0x6f7720 0x2e55f9b8 0x3e810c)
0x2e55f9a8 : 0x40aa2d (0x4959600 0x0 0x3c0c000 0x3e81d9)
0x2e55f9d8 : 0x40b622 (0x3c0c000 0x40b01c0 0x0 0x40de83)
0x2e55fa18 : 0x40edd8 (0x3c0c000 0x3c00bc0 0x4 0xffffffff)
0x2e55fa68 : 0x40ff5c (0x3c0c000 0x2 0x1 0x3c0c000)
0x2e55faa8 : 0x40958e (0x3c0c000 0x2 0x0 0x3be4b80)
0x2e55fae8 : 0x6cc3ab (0x3c0c000 0x2 0x2e55fb18 0x3c0c000)
0x2e55fb08 : 0x6e2591 (0x3c0c000 0xb0e7000 0x0 0x0)
0x2e55fb38 : 0x6e45b6 (0x3c0c000 0xb0e7000 0x48766e0 0x0)
0x2e55fb78 : 0x6e4c7b (0xb0e7000 0x9e207f 0x1 0x3e810c)
0x2e55fbe8 : 0x6e4fdc (0xb0e7000 0x4091b40 0x0 0x0)
The first entry on each line is the stack frame. The second entry is the return address, which is the interesting part for our purposes. We need to grab each of those second entries and decode them, to see the backtrace with symbols.
I've done this enough so I can easily edit the line with a BBEdit script; the resulting gdb commands which display the backtrace look like this:
x/i 0x12b0e1
x/i 0x1a7bed
x/i 0x19e517
x/i 0x4dd59a86
x/i 0x4dd58563
x/i 0x4dd585f0
x/i 0x40aa2d
x/i 0x40b622
x/i 0x40edd8
x/i 0x40ff5c
x/i 0x40958e
x/i 0x6cc3ab
x/i 0x6e2591
x/i 0x6e45b6
x/i 0x6e4c7b
x/i 0x6e4fdc
you could do this manually as well, of course.
I will feed those gdb commands into gdb after I set things up. The set up requires that I get the symbols for DoubleCommand from the kext, and the mach_kernel from Apple.
From the lines:
Kernel loadable modules in backtrace (with dependencies):
com.baltaks.driver.DoubleCommand(1.6.2)@0x4dd57000->0x4dd5bfff
dependency: com.apple.iokit.IOHIDFamily(1.5)@0x6c0000
com.apple.iokit.IOHIDFamily(1.5)@0x6c0000->0x6f7fff
in the backtrace, we know that DoubleCommand is in the backtrace.
Download the DoubleCommand dmg.
Mount the dmg. This creates /Volumes/DoubleCommand
Make a temporary directory to hold the pax.gz file, and copy it to the temporary directory
mkdir ~/d
cp /Volumes/DoubleCommand/DoubleCommand1.6.5_10.4.pkg/Contents/Archive.pax.gz ~/d/
Unarchive the pax.gz file
cd ~/d
pax -rzf Archive.pax.gz
Switch to the directory with the kext in it.
cd ./Library/StartupItems/DoubleCommand/
build the symbol file for the kext, using kextload and the addresses from the panic log "Kernel loadable modules in backtrace (with dependencies)" listing:
kextload -z -n -a com.baltaks.driver.DoubleCommand@0x4dd57000 -a com.apple.iokit.IOHIDFamily@0x6c0000 -s ~/d -- DoubleCommand.kext
Download the 10.5 Kernel Debug Kit from http://developer.apple.com/sdk (this requires signing up for a free ADC membership)
Mount that dmg. This creates /Volumes/KernelDebugKit
Invoke gdb, passing it the mach_kernel from /Volumes/KernelDebugKit. Read in the kgmacros file, and the two symbol files you just created:
$ gdb /Volumes/KernelDebugKit/mach_kernel
GNU gdb 6.3.50-20050815 (Apple version gdb-768) (Tue Oct 2 04:07:49 UTC 2007)
Copyright 2004 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General Public License, and you are
welcome to change it and/or distribute copies of it under certain conditions.
Type "show copying" to see the conditions.
There is absolutely no warranty for GDB. Type "show warranty" for details.
This GDB was configured as "i386-apple-darwin"...
(gdb) source /Volumes/KernelDebugKit/kgmacros
Loading Kernel GDB Macros package. Type "help kgm" for more info.
(gdb) add-symbol-file ~/d/com.baltaks.driver.DoubleCommand.sym
add symbol table from file "/Volumes/Data/Users/bb/d/com.baltaks.driver.DoubleCommand.sym"? (y or n) y
Reading symbols from /Volumes/Data/Users/bb/d/com.baltaks.driver.DoubleCommand.sym...done.
(gdb) add-symbol-file ~/d/com.apple.iokit.IOHIDFamily.sym
add symbol table from file "/Volumes/Data/Users/bb/d/com.apple.iokit.IOHIDFamily.sym"? (y or n) y
Reading symbols from /Volumes/Data/Users/bb/d/com.apple.iokit.IOHIDFamily.sym...done.
(gdb) x/i 0x12b0e1
0x12b0e1 <panic+422>: mov 0x4cf278,%eax
(gdb) x/i 0x1a7bed
0x1a7bed <kernel_trap+1450>: add $0xcc,%esp
(gdb) x/i 0x19e517
0x19e517 <trap_from_kernel+26>: mov %edi,%esp
(gdb) x/i 0x4dd59a86
0x4dd59a86 <remap(unsigned int*, unsigned int*, unsigned int*, unsigned int*, unsigned int*, unsigned int*, unsigned int*, unsigned int*, bool*, UnsignedWide*)+2890>: add %edi,0x0(%eax,%eax,1)
(gdb) x/i 0x4dd58563
0x4dd58563 <hijack_keyboard(IOHIKeyboard*)+153>: test %eax,%eax
(gdb) x/i 0x4dd585f0
0x4dd585f0 <hijack_keyboard(IOHIKeyboard*)+294>: (bad)
(gdb) x/i 0x40aa2d
0x40aa2d <IOService::invokeNotifer(_IOServiceNotifier*)+127>: mov %eax,%esi
(gdb) x/i 0x40b622
0x40b622 <IOService::deliverNotification(OSSymbol const*, unsigned long, unsigned long)+280>: mov -0x1c(%ebp),%edx
(gdb) x/i 0x40edd8
0x40edd8 <IOService::doServiceMatch(unsigned long)+500>: mov 0x24(%ebx),%edx
(gdb) x/i 0x40ff5c
0x40ff5c <IOService::startMatching(unsigned long)+352>: mov (%ebx),%eax
(gdb) x/i 0x40958e
0x40958e <IOService::registerService(unsigned long)+370>: add $0x2c,%esp
(gdb) x/i 0x6cc3ab
0x6cc3ab <IOHIKeyboard::start(IOService*)+51>: add %al,0x4f(%ecx,%eax,1)
(gdb) x/i 0x6e2591
0x6e2591 <IOHIDKeyboard::start(IOService*)+519>: add %al,(%eax)
(gdb) x/i 0x6e45b6
0x6e45b6 <IOHIDEventService::newKeyboardShim(unsigned long, unsigned long)+536>: add %al,(%eax)
(gdb) x/i 0x6e4c7b
0x6e4c7b <IOHIDEventService::parseSupportedElements(OSArray*, unsigned long)+1155>: add %ch,(%esi)
(gdb) x/i 0x6e4fdc
0x6e4fdc <IOHIDEventService::start(IOService*)+820>: test %al,(%ecx)
(gdb)
I removed the extraneous instruction information from the backtrace I posted, because I haven't actually loaded the code for DoubleCommand; I just loaded symbols and told gdb "pretend that DoubleCommand is loaded at 0x4dd57000, and that IOHIDFamily is at 0x6c0000". So the assembly instructions in my backtrace are useless, but the addresses inside the kext are correct.
I'm also having the kernel panic problem since upgrading (I know, I know) to Leopard. The panic shows up about 1 out of 4 times when I shut down the machine (MacBook Pro), seemingly at random but consistently I completely forgot about having DoubleCommand installed and upgraded with it running. Duh. After seeing DoubleCommand mentioned in the crash report, I uninstalled it. It didn't completely uninstall and is still running (I know that because my reassigned delete key is working). The DoubleCommand icon is gone and I can't find a way to stop it or uninstall it manually. I tried reinstalling over the top of it, but not surprisingly, the install invariably hangs, including the new 1.6.6b1 version. I'm hoping not to have to do a clean Leopard install to clean up the mess. Is there a way?
The core of DoubleCommand lives in /Library/StartupItems/DoubleCommand. If you delete that folder and restart, DoubleCommand is gone. You can optionaly remove the prefpane as well by removing /Library/PreferencePanes/DoubleCommandPreferences.prefPane.
I use an iMac G5 and a Mac Book Pro and I appear to be having the same Restart problems on my Mac Book Pro but NOT on my iMac G5.
Both machines have 10.5.1 installed.
All the reports of this problem that I have seen (5 or 6) in this thread and in the discussions on Apple's web page appear to be on Mac Book Pro.
Hope that helps!
Jim
OS 10.5.1
MacBook 2.16GHz
I experienced same problem while reboot and shutdown.
And it seemed to be fixed by 1.6.6b1 which I have installed about a week ago.
But, I have encountered another problem yesterday that the setting of doublecommand could not be saved.
In fact, it can be saved when I first save on each reboot.
But while running for hours, the setting will be gone and cannot save anymore until reboot again.
I really love this cool software with my MacBook.
I hope it will be fixed soon.
HW: Mac Pro + new aluminum Apple keyboard
OS: 10.5.1
DoubleCommand: 1.6.6.b1
Since this is not a laptop, I've got a full keyboard - I only tried out DoubleCommand to get working Home and End keys. It worked alright, but with "PC style home and end keys" as the only active setting the Ctrl key stops working (both right and left). After reading the messages in this thread, I didn't try to reboot.
Hi All, I'm going crazy because I deleted DoubleCommand (great program, btw - used it daily with great effect until Leopard) from startupitems as well as the prefpane, and I am STILL getting kernel panics when I plug in or unplug my USB keyboard (cheap Viewsonic Viewmate, model no. VSAC27936-1M).
I keep getting the message pertaining to com.baltaks.doublecommand.driver causing the crash, and I've spent a fair bit of time looking for said driver to no avail. My two theories are: 1. it's in the kextcache or something to that effect, or 2. it's hidden somewhere within a package or framework or something that isn't searchable and I haven't thought to look.
Someone please help! Thanks!
DoubleCommand is not hidden anywhere you've missed, if you've deleted the /Library/StartupItems/DoubleCommand folder and the DoubleCommandPreferences.prefpane and then restarted, it should be gone. If not, then your theory of the kextcache is possible - try deleting it, /System/Library/Extentions.kextcache.
I had to uninstall DoubleCommand 1.6.5 before installing 1.6.6a (like one of the posters, if I didn't uninstall first, the installer would hang... actually, it took up all my CPU and I couldn't get my computer to respond).
1.6.6a works the first time I install it, but the instant I reboot, it doesn't start up automatically. Any ideas?
Sorry to post again, but could this be the issue? When I try to run DoubleCommand from the command line (/Library/StartupItems/DoubleCommand/DoubleCommand), I get this:
touch: ./DoubleCommand.crashed: Permission denied
Loading DoubleCommand
you must be running as root to load modules into the kernel
dc.config: 4 -> 4
I have to use sudo to run as root. Could this be the cause?
What happens when you run:
sudo /Library/StartupItems/DoubleCommand/DoubleCommand
Both the Startup process and the installer should be running this script as root, but perhaps something subtle has changed with 10.5.
As I mentioned, I can sudo and it runs properly. I just doesn't start when OS X starts up.
Here's what I found in system.log when I reboot:
Feb 21 14:53:13 Allegro sudo[259]: astonlau : TTY=ttys000 ; PWD=/Library/StartupItems/DoubleCommand ; USER=root ; COMMAND=/bin/rm DoubleCommand.crashed
Feb 21 14:53:40 Allegro SystemStarter[18]: "/Library/StartupItems/DoubleCommand" failed sanity check: path was created after boot up
Feb 21 14:54:34 localhost com.apple.loginwindow[25]: Configuring DoubleCommand
Feb 21 14:54:34 localhost com.apple.loginwindow[25]: DoubleCommand configured, it is now safe to close this window.
Feb 21 14:54:37 Allegro SystemStarter[18]: "/Library/StartupItems/DoubleCommand" failed sanity check: path was created after boot up
It's strange -- I wonder what's causing the DoubleCommand directory's date to update?
Aha, the temporary file that we use to make sure we didn't crash last load is kept in the same folder as the kext and the startup items, I'll move that to /tmp and that might help.