On Mon, Oct 9, 2023 at 10:00 AM Al Viro <viro@xxxxxxxxxxxxxxxxxx> wrote:
>
> On Sat, Oct 07, 2023 at 11:44:32AM +0300, Amir Goldstein wrote:
>
> > @@ -93,7 +93,8 @@ static void show_faulting_vma(unsigned long address)
> > char *nm = "?";
> >
> > if (vma->vm_file) {
> > - nm = file_path(vma->vm_file, buf, ARC_PATH_MAX-1);
> > + nm = d_path(file_user_path(vma->vm_file), buf,
> > + ARC_PATH_MAX-1);
> > if (IS_ERR(nm))
> > nm = "?";
>
> Umm... At one point I considered this:
> if (vma->vm_file)
> pr_info(" @off 0x%lx in [%pD] VMA: 0x%08lx to 0x%08lx\n",
> vma->vm_start < TASK_UNMAPPED_BASE ?
> address : address - vma->vm_start,
> vma->vm_file, vma->vm_start, vma->vm_end);
> else
> pr_info(" @off 0x%lx in [anon] VMA: 0x%08lx to 0x%08lx\n",
> vma->vm_start < TASK_UNMAPPED_BASE ?
> address : address - vma->vm_start,
> vma->vm_start, vma->vm_end);
> and to hell with that 'buf' thing...
It's fine by me.
That would be consistent with print_bad_pte().
I've never had to debug vma/pte faults, so I don't know
how much of the path is really valuable for debugging.
Thanks,
Amir.
