On Thu, Jan 21, 2021 at 05:04:16PM +0100, Thomas Bogendoerfer wrote:
> When there's a fatal signal pending, MIPS's do_page_fault()
> implementation returns. The intent is that we'll return to the
> faulting userspace instruction, delivering the signal on the way.
>
> However, if we take a fatal signal during fixing up a uaccess, this
> results in a return to the faulting kernel instruction, which will be
> instantly retried, resulting in the same fault being taken forever. As
> the task never reaches userspace, the signal is not delivered, and the
> task is left unkillable. While the task is stuck in this state, it can
> inhibit the forward progress of the system.
>
> To avoid this, we must ensure that when a fatal signal is pending, we
> apply any necessary fixup for a faulting kernel instruction. Thus we
> will return to an error path, and it is up to that code to make forward
> progress towards delivering the fatal signal.
>
> [ Description taken from commit 746a272e4414 ("ARM: 8692/1: mm: abort
> uaccess retries upon fatal signal") ]
>
> Signed-off-by: Thomas Bogendoerfer <tsbogend@xxxxxxxxxxxxxxxx>
FWIW, this looks right to me, from a scan of the no_context path. I
don't have any MIPS system to test on, but FWIW:
Acked-by: Mark Rutland <mark.rutland@xxxxxxx>
Thanks for spinning this!
Mark.
> ---
> arch/mips/mm/fault.c | 5 ++++-
> 1 file changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/arch/mips/mm/fault.c b/arch/mips/mm/fault.c
> index 7c871b14e74a..e7abda9c013f 100644
> --- a/arch/mips/mm/fault.c
> +++ b/arch/mips/mm/fault.c
> @@ -156,8 +156,11 @@ static void __kprobes __do_page_fault(struct pt_regs *regs, unsigned long write,
> */
> fault = handle_mm_fault(vma, address, flags, regs);
>
> - if (fault_signal_pending(fault, regs))
> + if (fault_signal_pending(fault, regs)) {
> + if (!user_mode(regs))
> + goto no_context;
> return;
> + }
>
> if (unlikely(fault & VM_FAULT_ERROR)) {
> if (fault & VM_FAULT_OOM)
> --
> 2.29.2
>
