| Message ID | 1506083673-6041-1-git-send-email-mika.kahola@intel.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
Maybe we are missing a vblank wait somewhere on kernel CRC code?! Or maybe o kernel we read and discard the first for GLK?! :/ Also this is pipe crc right?! Shouldn't it be independent of the panel at the end?! Does it only happen with MIPI/DSI ?! Or it just happen on that particular unity on CI that coincidentally has this panel?! On Fri, Sep 22, 2017 at 5:40 AM Mika Kahola <mika.kahola@intel.com> wrote: > It seems that at least with GLK with MIPI/DSI display, the first collected > CRC is bogus. To fix this, try to collect two CRC's instead of one. > > Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=101309 > Signed-off-by: Mika Kahola <mika.kahola@intel.com> > --- > tests/kms_frontbuffer_tracking.c | 12 +++++++++++- > 1 file changed, 11 insertions(+), 1 deletion(-) > > diff --git a/tests/kms_frontbuffer_tracking.c > b/tests/kms_frontbuffer_tracking.c > index a068c8a..df7cc6e 100644 > --- a/tests/kms_frontbuffer_tracking.c > +++ b/tests/kms_frontbuffer_tracking.c > @@ -1239,8 +1239,18 @@ static void print_crc(const char *str, struct > both_crcs *crc) > > static void collect_crcs(struct both_crcs *crcs, bool mandatory_sink_crc) > { > - igt_pipe_crc_collect_crc(pipe_crc, &crcs->pipe); > + int n; > + igt_crc_t *crc = NULL; > + > + igt_pipe_crc_start(pipe_crc); > + n = igt_pipe_crc_get_crcs(pipe_crc, 2, &crc); > + igt_pipe_crc_stop(pipe_crc); > + igt_assert(n > 0); > + igt_assert_crc_equal(&crc[0], &crc[1]); > + crcs->pipe = crc[0]; > + > get_sink_crc(&crcs->sink, mandatory_sink_crc); > + free(crc); > } > > static void init_blue_crc(enum pixel_format format, bool > mandatory_sink_crc) > -- > 2.7.4 > > _______________________________________________ > Intel-gfx mailing list > Intel-gfx@lists.freedesktop.org > https://lists.freedesktop.org/mailman/listinfo/intel-gfx >
On Fri, 2017-09-22 at 15:25 +0000, Rodrigo Vivi wrote: > > Maybe we are missing a vblank wait somewhere on kernel CRC code?! > Or maybe o kernel we read and discard the first for GLK?! :/ > Also this is pipe crc right?! Shouldn't it be independent of the > panel at the end?! Does it only happen with MIPI/DSI ?! Or it just > happen on that particular unity on CI that coincidentally has this > panel?! Yes, this is pipe crc. It should be independent of the display panel. GLK+MIPI/DSI was the first platform that hit this issue. There are others such as BDW, CFL and CNL that seems to be affected by this https://bugs.freedesktop.org/show_bug.cgi?id=102374 > > On Fri, Sep 22, 2017 at 5:40 AM Mika Kahola <mika.kahola@intel.com> > wrote: > > It seems that at least with GLK with MIPI/DSI display, the first > > collected > > CRC is bogus. To fix this, try to collect two CRC's instead of one. > > > > Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=101309 > > Signed-off-by: Mika Kahola <mika.kahola@intel.com> > > --- > > tests/kms_frontbuffer_tracking.c | 12 +++++++++++- > > 1 file changed, 11 insertions(+), 1 deletion(-) > > > > diff --git a/tests/kms_frontbuffer_tracking.c > > b/tests/kms_frontbuffer_tracking.c > > index a068c8a..df7cc6e 100644 > > --- a/tests/kms_frontbuffer_tracking.c > > +++ b/tests/kms_frontbuffer_tracking.c > > @@ -1239,8 +1239,18 @@ static void print_crc(const char *str, > > struct both_crcs *crc) > > > > static void collect_crcs(struct both_crcs *crcs, bool > > mandatory_sink_crc) > > { > > - igt_pipe_crc_collect_crc(pipe_crc, &crcs->pipe); > > + int n; > > + igt_crc_t *crc = NULL; > > + > > + igt_pipe_crc_start(pipe_crc); > > + n = igt_pipe_crc_get_crcs(pipe_crc, 2, &crc); > > + igt_pipe_crc_stop(pipe_crc); > > + igt_assert(n > 0); > > + igt_assert_crc_equal(&crc[0], &crc[1]); > > + crcs->pipe = crc[0]; > > + > > get_sink_crc(&crcs->sink, mandatory_sink_crc); > > + free(crc); > > } > > > > static void init_blue_crc(enum pixel_format format, bool > > mandatory_sink_crc) > > -- > > 2.7.4 > > > > _______________________________________________ > > Intel-gfx mailing list > > Intel-gfx@lists.freedesktop.org > > https://lists.freedesktop.org/mailman/listinfo/intel-gfx > > > -- > Rodrigo Vivi > Blog: http://blog.vivi.eng.br >
diff --git a/tests/kms_frontbuffer_tracking.c b/tests/kms_frontbuffer_tracking.c index a068c8a..df7cc6e 100644 --- a/tests/kms_frontbuffer_tracking.c +++ b/tests/kms_frontbuffer_tracking.c @@ -1239,8 +1239,18 @@ static void print_crc(const char *str, struct both_crcs *crc) static void collect_crcs(struct both_crcs *crcs, bool mandatory_sink_crc) { - igt_pipe_crc_collect_crc(pipe_crc, &crcs->pipe); + int n; + igt_crc_t *crc = NULL; + + igt_pipe_crc_start(pipe_crc); + n = igt_pipe_crc_get_crcs(pipe_crc, 2, &crc); + igt_pipe_crc_stop(pipe_crc); + igt_assert(n > 0); + igt_assert_crc_equal(&crc[0], &crc[1]); + crcs->pipe = crc[0]; + get_sink_crc(&crcs->sink, mandatory_sink_crc); + free(crc); } static void init_blue_crc(enum pixel_format format, bool mandatory_sink_crc)
It seems that at least with GLK with MIPI/DSI display, the first collected CRC is bogus. To fix this, try to collect two CRC's instead of one. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=101309 Signed-off-by: Mika Kahola <mika.kahola@intel.com> --- tests/kms_frontbuffer_tracking.c | 12 +++++++++++- 1 file changed, 11 insertions(+), 1 deletion(-)