Re: [RFC PATCH] ARM: enable irq in translation/section permission fault handlers

[Bernd Edlinger <bernd.edlinger@hotmail.de>]

> But if there is a way to get into the code path do_DataAbort->do_page_fault with
> addr > TASK_SIZE, then I don't see why that works without the RT patch,
> since the interrupts are enabled here:
>
> do_page_fault(unsigned long addr, unsigned int fsr, struct pt_regs *regs)
> {
>         struct task_struct *tsk;
>         struct mm_struct *mm;
>         int sig, code;
>         vm_fault_t fault;
>         unsigned int flags = FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_KILLABLE;
> 
>         if (notify_page_fault(regs, fsr))
>                 return 0;
> 
>         tsk = current;
>         mm  = tsk->mm;
> 
>         /* Enable interrupts if they were enabled in the parent context. */
>         if (interrupts_enabled(regs))
>                 local_irq_enable();
> 
> Is preemption disabled at this point, without the RT patch?
> 
> I still don't quite see why is this no issue without the RT patch.
> Can someone explain that?

Never mind, now I found out how to prove that it is not a RT issue:

$ cat t2.c 
int main() {
  *((char*)0xFFFFFFFF) = 1;
}
$ gcc t2.c -o t2
$ ./t2 
BUG: using smp_processor_id() in preemptible [00000000] code: t2/832
caller is debug_smp_processor_id+0x18/0x24
CPU: 1 PID: 832 Comm: t2 Not tainted 4.18.0-00005-g7523d7f #1
Hardware name: Altera SOCFPGA
[<80112e68>] (unwind_backtrace) from [<8010da28>] (show_stack+0x20/0x24)
[<8010da28>] (show_stack) from [<806bcaec>] (dump_stack+0x78/0x94)
[<806bcaec>] (dump_stack) from [<8043f2d0>] (check_preemption_disabled+0xe4/0x120)
[<8043f2d0>] (check_preemption_disabled) from [<8043f324>] (debug_smp_processor_id+0x18/0x24)
[<8043f324>] (debug_smp_processor_id) from [<80116ba0>] (__do_user_fault+0x48/0x13c)
[<80116ba0>] (__do_user_fault) from [<8011700c>] (do_page_fault+0x2f4/0x328)
[<8011700c>] (do_page_fault) from [<80117200>] (do_DataAbort+0x58/0x100)
[<80117200>] (do_DataAbort) from [<80102460>] (__dabt_usr+0x40/0x60)
Exception stack(0xbf197fb0 to 0xbf197ff8)
7fa0:                                     00000001 7e963d94 00000001 ffffffff
7fc0: 000103f8 00000000 000102e0 00000000 00000000 00000000 76fa9000 7e963c3c
7fe0: 00000000 7e963c3c 76e47bbc 000103e0 60060010 ffffffff
Segmentation fault


What exactly is the reason why we should not apply this patch in the non-RT kernel as well:

https://marc.info/?l=linux-rt-users&m=154997565610392&w=2



Thanks
Bernd.

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel