Hi Krzysztof,
On 22/07/2021 08:43, Krzysztof Hałasa wrote:
> Also, I have spotted an interesting loop condition in
> drivers/media/v4l2-core/v4l2-dv-timings.c (dated 2013,
> d1c65ad6a44b0ff79d2f0bf726fa6fd9248991f4).
> It obviously works (unless the table grows to 600+ entries), but I guess
> I should make it "standard", shouldn't I?
Thanks for catching this!
I've posted a patch fixing it. Luckily it never broke anything, but it definitely
had to be fixed.
Regards,
Hans
>
> The first case is pretty normal:
> @@ -159,10 +161,10 @@ int v4l2_enum_dv_timings_cap(struct v4l2_enum_dv_timings *t,
> u32 i, idx;
>
> memset(t->reserved, 0, sizeof(t->reserved));
> - for (i = idx = 0; i < ARRAY_SIZE(timings); i++) {
> - if (v4l2_dv_valid_timings(timings + i, cap) &&
> + for (i = idx = 0; v4l2_dv_timings_presets[i].bt.width; i++) {
> + if (v4l2_dv_valid_timings(v4l2_dv_timings_presets + i, cap) &&
> idx++ == t->index) {
> - t->timings = timings[i];
> + t->timings = v4l2_dv_timings_presets[i];
> return 0;
> }
> }
>
> This is the interesting part:
> @@ -179,10 +181,10 @@ bool v4l2_find_dv_timings_cap(struct v4l2_dv_timings *t,
> if (!v4l2_dv_valid_timings(t, cap))
> return false;
>
> - for (i = 0; i < ARRAY_SIZE(timings); i++) {
> - if (v4l2_dv_valid_timings(timings + i, cap) &&
> - v4l2_match_dv_timings(t, timings + i, pclock_delta)) {
> - *t = timings[i];
> + for (i = 0; i < v4l2_dv_timings_presets[i].bt.width; i++) {
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> + if (v4l2_dv_valid_timings(v4l2_dv_timings_presets + i, cap) &&
> + v4l2_match_dv_timings(t, v4l2_dv_timings_presets + i, pclock_delta)) {
> + *t = v4l2_dv_timings_presets[i];
> return true;
> }
> }
>
