On Tue, Dec 06, 2005 at 10:18:59 +0530, Saumender Dash wrote: > On Fri, Dec 02, 2005 at 16:56:11 +0530, Saumender Dash wrote: > > >Well, you can freely access the userland memory (ok, you have to use > >copy_from_user and such to handle page faults for you) and you have the > >stack pointer available in the task struct, which you obtain from > >current. The rest is making sense out of that stack. > > I'm geting the stack pointer in the kernel, and unwind the stack till it's > empty or I get the program text > and report the PC. The thing is I should be able to report the correct PC, > for a perticular sys call, in a program. But I'm missing out something in > this, which is my concern. Any pointer on that will > simply be great. What is your definition of 'correct' here? It's certainly not obvious. > >That being said, I don't really think you should do this from kernel. > > My requirement is in the kernel space. I will use the derived PC to be used > in the other part of the > KLM. So doing it from user space is not the idea. Do the other parts need to be in kernel? -- Jan 'Bulb' Hudec <bulb@xxxxxx>
Attachment:
signature.asc
Description: Digital signature