Re: [PATCH 4/11] i386 vDSO: use VM_ALWAYSDUMP

[Roland McGrath]

> Not that the old code did either, so it's not like it's a new bug, but I 
> thought I'd point it out anyway. As if we care (but you can probably 
> trigger this by having an app that does
> 
> 	munmap(NULL, TASK_SIZE);
>
> which will cause a SIGSEGV on return (because the stack doesn't exist) and 
> then the core-dump should be empty.

Ok, damn you.  I admit I thought of exactly this case and then glossed it
over because I didn't want to complicate the patch and discuss the arcane
justification.  And I was doing a good job of repressing the memory of it
completely before you dredged it up.

> Not that I tested anythign that evil anyway, nor do I think we really care 
> if it means that the gate_vma doesn't get shown in the core-dump either.

I didn't test it either.  And I do sort of think the vDSO should be (the
only thing) in that case's core dump on anal principle.  Now that you've
brought it up, I feel all dirty for ever having sent in code that doesn't
cover the case.  (Not that we really care, since CONFIG_COMPAT_VDSO won't
actually be turned on in practice.)

Starting the loops with "vma = current->mm->mmap ?: gate_vma" should do it.
But I guess Andrew would prefer it:

static struct vm_area_struct *first_vma(struct task_struct *tsk,
					struct vm_area_struct *gate_vma) 
{
	struct vm_area_struct *ret = tsk->mm->mmap;
	if (ret)
		return ret;
	return gate_vma;
}

and:

	for (vma = first_vma(current); vma; vma = next_vma(vma, gate_vma)) {


Thanks,
Roland
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/