From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from out3-smtp.messagingengine.com (out3-smtp.messagingengine.com [66.111.4.27]) by gabe.freedesktop.org (Postfix) with ESMTPS id A412810F005 for ; Fri, 10 Jun 2022 12:16:49 +0000 (UTC) From: Maxime Ripard To: igt-dev@lists.freedesktop.org, Petri Latvala , Arkadiusz Hiler Date: Fri, 10 Jun 2022 14:16:44 +0200 Message-Id: <20220610121647.975942-1-maxime@cerno.tech> Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Subject: [igt-dev] [PATCH i-g-t v3 0/3] Writeback fixes and improvements List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: Pekka Paalanen , Maxime Ripard Errors-To: igt-dev-bounces@lists.freedesktop.org Sender: "igt-dev" List-ID: Hi, Here's a series addressing the writeback-check-output test not passing on vc4. This is due to the fact that the writeback controller will force the X component to 0, while we expect the output and input buffers to be identical, and the input buffer has 0xff. This series tackles this by adding some support to dump the faulty frames in PNG and raw, which proved to be useful while debugging this series. Then, we rework the hash function to ignore the X component of our buffer and only hash the pixels themselves. Finally, we change the input buffer X value to a garbage, different from 0 and 0xff, value to make sure the hardware properly ignores the component and doesn't treat it as alpha. Let me know what you think, Maxime Changes from v2: - Rebased on master - Dropped the frame dumping patches Changes from v1: - Rebased on master - Added maintainers to the recipients Maxime Ripard (3): tests/kms_writeback: Use endianness accessor to fill pixels lib/igt_fb: Ignore the X component when computing CRC tests/kms_writeback: Use a garbage X value to create fill our test buffer lib/igt_core.h | 9 +++++++++ lib/igt_fb.c | 20 ++++++++++++++++---- tests/kms_writeback.c | 4 ++-- 3 files changed, 27 insertions(+), 6 deletions(-) -- 2.36.1