<html><head> <meta http-equiv="Content-Type" content="text/html; charset=utf-8"> </head> <body> <div class="moz-cite-prefix"> <p class="MsoPlainText">Hi,</p> This is a new issue which is seen with this patch.<br> We are seeing a failure in pipe_crc_start when this patch is added.<br> <br> (kms_plane_multiple:5716) igt_debugfs-CRITICAL: Test assertion failure function igt_pipe_crc_start, file ../../../usr/src/igt-gpu-tools/lib/igt_debugfs.c:884:<br> (kms_plane_multiple:5716) igt_debugfs-CRITICAL: Failed assertion: pipe_crc->crc_fd != -1<br> <br> This is because the pipe is off when we're trying to do pipe_crc_start with this patch.<br> This is hitting only on i915 because of the below code, where we start pipe crc before the next commit specifically for i915 devices. <br> For other devices, the pipe_crc_start is after the next commit and this is the reason we don’t see any issue there.<br> </div> <div class="moz-cite-prefix"><br> </div> <div class="moz-cite-prefix"> /* Intel devices need it here, timing sensitive on few devices */<br> if (is_i915_device(data->drm_fd))<br> igt_pipe_crc_start(data->pipe_crc);<br> <br> /* randomize planes and set up the holes */<br> prepare_planes(data, pipe, &blue, modifier, c, output);<br> <br> igt_display_commit2(&data->display, COMMIT_ATOMIC);<br> if (!is_i915_device(data->drm_fd))<br> igt_pipe_crc_start(data->pipe_crc);<br> <p class="MsoPlainText">IMHO, this patch needs to be reworked to work with i915 devices.</p> Regards,<br> Karthik.B.S</div> <div class="moz-cite-prefix"><br> </div> <div class="moz-cite-prefix">On 10/7/2022 11:03 AM, Vudum, Lakshminarayana wrote:<br> </div> <blockquote type="cite" cite="mid:DM4PR11MB6213D37BA33E02D2AE17A13DEC5F9@DM4PR11MB6213.namprd11.prod.outlook.com"> <pre class="moz-quote-pre" wrap="">+Karthik and Vandita. Lakshmi. -----Original Message----- From: Mark Yacoub <a class="moz-txt-link-rfc2396E" href="mailto:markyacoub@chromium.org"><markyacoub@chromium.org></a> Sent: Thursday, October 6, 2022 8:34 AM To: Vudum, Lakshminarayana <a class="moz-txt-link-rfc2396E" href="mailto:lakshminarayana.vudum@intel.com"><lakshminarayana.vudum@intel.com></a> Cc: <a class="moz-txt-link-abbreviated" href="mailto:igt-dev@lists.freedesktop.org">igt-dev@lists.freedesktop.org</a>; <a class="moz-txt-link-abbreviated" href="mailto:michel.daenzer@mailbox.org">michel.daenzer@mailbox.org</a>; Alex Hung <a class="moz-txt-link-rfc2396E" href="mailto:alex.hung@amd.com"><alex.hung@amd.com></a> Subject: Re: [igt-dev] [PATCH][V3] tests/kms_plane_multiple: turn off pipe if all planes are off Hi Lakshminarayana, I have a feeling that this might be an intel bug - the code is pretty straightforward and works on amdpgu. we have a feeling that i915 might be missing something. On Mon, Oct 3, 2022 at 12:25 PM Alex Hung <a class="moz-txt-link-rfc2396E" href="mailto:alex.hung@amd.com"><alex.hung@amd.com></a> wrote: </pre> <blockquote type="cite"> <pre class="moz-quote-pre" wrap=""> Hi Mark, As expected, the rev 3 has kms_plane_multiple failures on CI.IGT unlike rev 2 which has "is_amdgpu_device()" on the first set_pipe, and the failures are as below. The first set_pipe required by amdgpu is not needed on all devices. (kms_plane_multiple:1581) igt_debugfs-CRITICAL: Test assertion failure function igt_pipe_crc_start, file ../../../usr/src/igt-gpu-tools/lib/igt_debugfs.c:884: (kms_plane_multiple:1581) igt_debugfs-CRITICAL: Failed assertion: pipe_crc->crc_fd != -1 (kms_plane_multiple:1581) igt_debugfs-CRITICAL: Last errno: 5, Input/output error The Fi.CI.IGT results are available on <a class="moz-txt-link-freetext" href="https://patchwork.freedesktop.org/series/109086/">https://patchwork.freedesktop.org/series/109086/</a> On 2022-09-29 14:06, Alex Hung wrote: </pre> <blockquote type="cite"> <pre class="moz-quote-pre" wrap="">amdgpu rejects when crtc is on + all planes are off, and it is necessary to turn off crtc when all planes are off in the subtest "tiling-none". This is revised from <a class="moz-txt-link-freetext" href="https://patchwork.freedesktop.org/series/80904/">https://patchwork.freedesktop.org/series/80904/</a> Signed-off-by: Alex Hung <a class="moz-txt-link-rfc2396E" href="mailto:alex.hung@amd.com"><alex.hung@amd.com></a> --- V2 - remove second is_amdgpu_device() before setting PIPE_NONE V3 - remove both is_amdgpu_device() before setting PIPE_NONE tests/kms_plane_multiple.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/tests/kms_plane_multiple.c b/tests/kms_plane_multiple.c index cbe8c189..6c0bc36d 100644 --- a/tests/kms_plane_multiple.c +++ b/tests/kms_plane_multiple.c @@ -311,6 +311,7 @@ test_plane_position_with_output(data_t *data, enum pipe pipe, for_each_plane_on_pipe(&data->display, pipe, plane) igt_plane_set_fb(plane, NULL); + igt_output_set_pipe(output, PIPE_NONE); igt_display_commit2(&data->display, COMMIT_ATOMIC); for (int x = 0; x < c; x++) @@ -344,6 +345,7 @@ test_plane_position_with_output(data_t *data, enum pipe pipe, for_each_plane_on_pipe(&data->display, pipe, plane) igt_plane_set_fb(plane, NULL); + igt_output_set_pipe(output, PIPE_NONE); igt_display_commit2(&data->display, COMMIT_ATOMIC); for (int x = 0; x < c; x++) </pre> </blockquote> </blockquote> </blockquote> <p><br> </p> </body> </html>