[CentOS-devel] Q: method for kernel module debugging

Sun Nov 8 13:19:23 UTC 2015
Ran Shalit <ranshalit at gmail.com>

On Sun, Nov 8, 2015 at 1:22 AM, Jonathan Billings <billings at negate.org> wrote:
> On Nov 7, 2015, at 2:56 PM, Ran Shalit <ranshalit at gmail.com> wrote:
>> Thanks for this value information!
>> As to the steps after prepearing vmlinux for debug:
>> I also find the following important link:
>> https://github.com/AltraMayor/XIA-for-Linux/wiki/Debugging-the-Linux-kernel
>> But it seems that the kernel debugging metod described there is quite
>> complex (qeuires virtualbox , nfs shared folder).
>> I just wander if this is the method which is recommended & used for
>> kernel module debugging.
>
> Odd that it would suggest VirtualBox when kvm-based VMs are native to the kernel.  I’d use virsh to do the same thing, with fewer moving parts.

Right. probably virsh is good enough with CentOS.

>
>> I am intereseted in debugging a new kernel module which I will write
>> (v4l pci express module)
>
> It sounds like you’re going to be debugging a kernel module for physical hardware?  That makes a lot of the above URL useless — you want to use kdbg from a serial console attached to a physical system rather than a VM.
>

Yes, I am testing HW: PCI express device (connected to FPGA) for video stream.
 I think that the above suggestion is useful becuase it is probably
for anyone who want to debug kernel with one machine instead of 2
physical machines.
But at the bottom line, both configuration (2 phisical machines or 1
physical machine and 1 VM)  is seems  complex.
I would probably rather use printk in the end....


Thank you!
Ran