On Tue, Jan 05, 2021 at 07:27:49PM +0100, Andrey Konovalov wrote:
> diff --git a/arch/arm64/mm/fault.c b/arch/arm64/mm/fault.c
> index 3c40da479899..57d3f165d907 100644
> --- a/arch/arm64/mm/fault.c
> +++ b/arch/arm64/mm/fault.c
> @@ -302,12 +302,20 @@ static void die_kernel_fault(const char *msg, unsigned long addr,
> static void report_tag_fault(unsigned long addr, unsigned int esr,
> struct pt_regs *regs)
> {
> - bool is_write = ((esr & ESR_ELx_WNR) >> ESR_ELx_WNR_SHIFT) != 0;
> + static bool reported;
> + bool is_write;
> +
> + if (READ_ONCE(reported))
> + return;
> +
> + if (mte_report_once())
> + WRITE_ONCE(reported, true);
I guess the assumption here is that you don't get any report before the
tests start and temporarily set report_once to false. It's probably
fine, if we get a tag check failure we'd notice in the logs anyway.
> /*
> * SAS bits aren't set for all faults reported in EL1, so we can't
> * find out access size.
> */
> + is_write = ((esr & ESR_ELx_WNR) >> ESR_ELx_WNR_SHIFT) != 0;
I now noticed, you could write this in a shorter way:
is_write = !!(esr & ESR_ELx_WNR);
> kasan_report(addr, 0, is_write, regs->pc);
> }
The patch looks fine to me.
Reviewed-by: Catalin Marinas <catalin.marinas@xxxxxxx>
