Re: Kernel debugging - what's the procedure?
On Sun, Apr 07, 2002 at 01:54:08PM -0700, Terry Lambert wrote: Josef Karthauser wrote: You use another machine, and connect the serial ports together. See the handbook for details. I'd love to ;) but no additional machine is available at the moment. Then you install vmware, and Julian's back-to-back serial driver, and then run the kernel to be debuged in the vmware session. This lets you debug the kernel on a virtual machine (single CPU only). Yes I do have vmware, but it's not going to help me debug the usb stack is it? (I know vmware3 does usb, but not if the underlying usb stack on the host machine is also hosed). If this is against a dead kernel, etiher your dump image is bad, or it doesn't match the kernel that made the dump. That's what I'd guess too, but the kernel that crashed is the same as the kernel that I'm debugging it on - I'm pretty sure: compiled today. Well, since this is the -current list... have you updated recently? There were some recent changes by PHK to the dump format that may have broken/fixed things, if your answer to that question is yes/no, respectively. Yes, that's what I implied by the above paragraph. I was up-to-date with sources yesterday. Joe msg37087/pgp0.pgp Description: PGP signature
Re: Kernel debugging - what's the procedure?
Josef Karthauser wrote: On Sun, Apr 07, 2002 at 01:54:08PM -0700, Terry Lambert wrote: Josef Karthauser wrote: You use another machine, and connect the serial ports together. See the handbook for details. I'd love to ;) but no additional machine is available at the moment. Then you install vmware, and Julian's back-to-back serial driver, and then run the kernel to be debuged in the vmware session. This lets you debug the kernel on a virtual machine (single CPU only). Yes I do have vmware, but it's not going to help me debug the usb stack is it? (I know vmware3 does usb, but not if the underlying usb stack on the host machine is also hosed). It's my understanding that you can assign resources to the vmware machine, which it can then permit to be accessed natively. Thus, if you gave your USB ports to the vmware virtual machine entirely, this would resolve the problem. The vmware3 support for usb appears (to me) to be a tunnel driver, like the standard network driver interface (e.g. it pretends to be hardware, and uses a host driver to contact the real hardware, so that its use can be multiplexed and thus the device accessed by etiher the real or the virtual machine). Well, since this is the -current list... have you updated recently? There were some recent changes by PHK to the dump format that may have broken/fixed things, if your answer to that question is yes/no, respectively. Yes, that's what I implied by the above paragraph. I was up-to-date with sources yesterday. This is probably your problem. If you can back up to when the problem first appeared, or back out the kernel dump fule generation modifications, this would probably fix your problems. If you look at the -current list archives, you will see a number of posts which identify the files which have changed, and are probably what is breaking you [ comments on policy related to commits of kernel changes without corresponding commits to user space utilities elided ]. -- Terry To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Re: Kernel debugging - what's the procedure?
On Mon, Apr 08, 2002 at 04:25:59AM -0700, Terry Lambert wrote: Yes, that's what I implied by the above paragraph. I was up-to-date with sources yesterday. This is probably your problem. If you can back up to when the problem first appeared, or back out the kernel dump fule generation modifications, this would probably fix your problems. If you look at the -current list archives, you will see a number of posts which identify the files which have changed, and are probably what is breaking you [ comments on policy related to commits of kernel changes without corresponding commits to user space utilities elided ]. Looking at todays commits it looks like the gdb/kernel coredump problem has been fixed, at bde's bequest. Joe msg37094/pgp0.pgp Description: PGP signature
Kernel debugging - what's the procedure?
What's the current method for debugging kernels? genius# gdb -k /usr/obj/usr/src/sys/GENIUS/kernel.debug 5d079ab35c111057dbbf8f242940bbad.core GNU gdb 4.18 Copyright 1998 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-unknown-freebsd... IdlePTD at phsyical address 0x004b1000 kernel symbol `dumppcb' not found. Attaching to program: /usr/obj/usr/src/sys/GENIUS/kernel.debug, process 5 ptrace: Invalid argument. (kgdb) quit Is this something I'm doing wrong, or it is a bug somewhere? What's 'dumppcb'? Joe msg37044/pgp0.pgp Description: PGP signature
Re: Kernel debugging - what's the procedure?
Josef Karthauser wrote: What's the current method for debugging kernels? genius# gdb -k /usr/obj/usr/src/sys/GENIUS/kernel.debug Running kernels? You use another machine, and connect the serial ports together. See the handbook for details. If this is against a dead kernel, etiher your dump image is bad, or it doesn't match the kernel that made the dump. -- Terry To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Re: Kernel debugging - what's the procedure?
On Sun, Apr 07, 2002 at 03:29:41AM -0700, Terry Lambert wrote: Josef Karthauser wrote: What's the current method for debugging kernels? genius# gdb -k /usr/obj/usr/src/sys/GENIUS/kernel.debug Running kernels? You use another machine, and connect the serial ports together. See the handbook for details. I'd love to ;) but no additional machine is available at the moment. If this is against a dead kernel, etiher your dump image is bad, or it doesn't match the kernel that made the dump. That's what I'd guess too, but the kernel that crashed is the same as the kernel that I'm debugging it on - I'm pretty sure: compiled today. Joe msg37052/pgp0.pgp Description: PGP signature
Re: Kernel debugging - what's the procedure?
On Sun, Apr 07, 2002 at 10:13:35AM +0100, Josef Karthauser wrote: What's the current method for debugging kernels? genius# gdb -k /usr/obj/usr/src/sys/GENIUS/kernel.debug 5d079ab35c111057dbbf8f242940bbad.core ... kernel symbol `dumppcb' not found. Attaching to program: /usr/obj/usr/src/sys/GENIUS/kernel.debug, process 5 ptrace: Invalid argument. (kgdb) quit You also did not read in the core file. To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Re: Kernel debugging - what's the procedure?
Josef Karthauser wrote: You use another machine, and connect the serial ports together. See the handbook for details. I'd love to ;) but no additional machine is available at the moment. Then you install vmware, and Julian's back-to-back serial driver, and then run the kernel to be debuged in the vmware session. This lets you debug the kernel on a virtual machine (single CPU only). If this is against a dead kernel, etiher your dump image is bad, or it doesn't match the kernel that made the dump. That's what I'd guess too, but the kernel that crashed is the same as the kernel that I'm debugging it on - I'm pretty sure: compiled today. Well, since this is the -current list... have you updated recently? There were some recent changes by PHK to the dump format that may have broken/fixed things, if your answer to that question is yes/no, respectively. -- Terry To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Re: Kernel debugging - what's the procedure?
On Sun, 7 Apr 2002, Josef Karthauser wrote: On Sun, Apr 07, 2002 at 03:29:41AM -0700, Terry Lambert wrote: You use another machine, and connect the serial ports together. See the handbook for details. I'd love to ;) but no additional machine is available at the moment. run it in vmware and connect it to the NMDM(4) drivers If this is against a dead kernel, etiher your dump image is bad, or it doesn't match the kernel that made the dump. That's what I'd guess too, but the kernel that crashed is the same as the kernel that I'm debugging it on - I'm pretty sure: compiled today. I'm not sure it all works right now as people have been doing things with the dump code. Joe To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Re: Kernel debugging - what's the procedure?
On Sun, 7 Apr 2002, Josef Karthauser wrote: What's the current method for debugging kernels? genius# gdb -k /usr/obj/usr/src/sys/GENIUS/kernel.debug 5d079ab35c111057dbbf8f242940bbad.core ... kernel symbol `dumppcb' not found. Attaching to program: /usr/obj/usr/src/sys/GENIUS/kernel.debug, process 5 ptrace: Invalid argument. (kgdb) quit Review all commits and note the unreviewed one that blew away dumppcb? Bruce To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message