Re: crash in entry.S restore_all, 2.6.12-rc2, x86, PAGEALLOC

[Stas Sergeev]

Hi Linus.

Linus Torvalds wrote:
> The NMI code had better be really careful, and yeah, I suspect it needs 
> fixing.
And since the NMI return will need a
ESP fixup too, it will require the
"branched" version of the restore_all
checks I suppose.

> > 2. How can one be sure there are no more
> > of the like places where the stack is left
> > empty?
> That's a good argument, and may be the strongest reason for _not_ doing 
> the speculation.
I haven't said that explicitly, only
implied:) My another idea was to adjust
the tss.esp0 to always point 8 bytes below
the real top of the stack, so that even in
case of an NMI we are still safe. And in
case of another such instance - too.
This may look more like a hack than shifting
the "sti", but it is probably more reliable.
At least something to consider as soon as we
are at it.

-
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/