On Thu, Oct 20, 2022 at 3:49 AM Alexey Dobriyan <adobriyan@xxxxxxxxx> wrote:
>
> struct p4_event_bind::cntr[][] should be signed because of
> the following code:
>
> int i, j;
> for (i = 0; i < P4_CNTR_LIMIT; i++) {
> ===> j = bind->cntr[thread][i];
> if (j != -1 && !test_bit(j, used_mask))
> return j;
> }
>
> Making this member unsigned will make "j" 255 and fail "j != -1"
> comparison.
>
> Signed-off-by: Alexey Dobriyan <adobriyan@xxxxxxxxx>
Nice catch.
Reviewed-by: Jason A. Donenfeld <Jason@xxxxxxxxx>
> arch/x86/events/intel/p4.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> --- a/arch/x86/events/intel/p4.c
> +++ b/arch/x86/events/intel/p4.c
> @@ -24,7 +24,7 @@ struct p4_event_bind {
> unsigned int escr_msr[2]; /* ESCR MSR for this event */
> unsigned int escr_emask; /* valid ESCR EventMask bits */
> unsigned int shared; /* event is shared across threads */
> - char cntr[2][P4_CNTR_LIMIT]; /* counter index (offset), -1 on absence */
> + signed char cntr[2][P4_CNTR_LIMIT]; /* counter index (offset), -1 on absence */
> };
This is fine, but I wonder if this is a good occasion to use `s8` instead?
Jason
