From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.1 (2015-04-28) on sa.local.altlinux.org X-Spam-Level: X-Spam-Status: No, score=-4.3 required=5.0 tests=ALL_TRUSTED,BAYES_00, RP_MATCHES_RCVD autolearn=unavailable autolearn_force=no version=3.4.1 From: Alexey Sheplyakov To: devel-kernel@lists.altlinux.org Date: Wed, 14 Dec 2022 17:19:11 +0400 Message-Id: <20221214131919.681481-24-asheplyakov@basealt.ru> X-Mailer: git-send-email 2.33.5 In-Reply-To: <20221214131919.681481-1-asheplyakov@basealt.ru> References: <20221214131919.681481-1-asheplyakov@basealt.ru> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Cc: =?UTF-8?q?=D0=A0=D0=BE=D0=BC=D0=B0=D0=BD=20=D0=A1=D1=82=D0=B0=D0=B2=D1=86=D0=B5=D0=B2?= , =?UTF-8?q?=D0=98=D0=B3=D0=BE=D1=80=D1=8C=20=D0=A7=D1=83=D0=B4=D0=BE=D0=B2?= , =?UTF-8?q?=D0=95=D0=B2=D0=B3=D0=B5=D0=BD=D0=B8=D0=B9=20=D0=A1=D0=B8=D0=BD=D0=B5=D0=BB=D1=8C=D0=BD=D0=B8=D0=BA=D0=BE=D0=B2?= , =?UTF-8?q?=D0=94=D0=BC=D0=B8=D1=82=D1=80=D0=B8=D0=B9=20=D0=A2=D0=B5=D1=80=D1=91=D1=85=D0=B8=D0=BD?= Subject: [d-kernel] [PATCH 24/32] drm/panfrost: forcibly set dma-coherent on Baikal-M X-BeenThere: devel-kernel@lists.altlinux.org X-Mailman-Version: 2.1.12 Precedence: list Reply-To: ALT Linux kernel packages development List-Id: ALT Linux kernel packages development List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 14 Dec 2022 13:20:17 -0000 Archived-At: List-Archive: List-Post: With memattr 0x888d88 (set by arm_mali_lpae_alloc_pgtable) GPU (Mali T628 r1p0) experiences a lot of DATA_INVALID faults, unhandled page faults, and other errors. Also the screen goes black almost immediately. On the other hand with memattr 0x484d48 (as set by mali_kbase) the GPU appears to work just fine. Robin Murphy explains: > using the outer-cacheable attribute is deliberate because it is necessary > for I/O-coherent GPUs to work properly (and should be irrelevant for > non-coherent integrations) > I'd note that panfrost has been working OK - to the extent that Mesa > supports its older ISA - on the T624 (single core group) in Arm's > Juno SoC for over a year now since commit 268af50f38b1. > If you have to force outer non-cacheable to avoid getting translation > faults and other errors that look like the GPU is inexplicably seeing > the wrong data, I'd check whether you have the same thing where your > integration is actually I/O-coherent and you're missing the "dma-coherent" > property in your DT. Indeed setting the "dma-coherent" property (and adjusting jobs affinity for dual core group GPU) makes panfrost work just fine on Baikal-M. However on Baikal-M the FDT is passed to the kernel by the firmware, and replacing the FDT in the firmware is tricky. Therefore set `coherent` property when running on Baikal-M (even if the `dma-coherent` property is missing in the FDT). Signed-off-by: Alexey Sheplyakov X-DONTUPSTREAM X-feature-Baikal-M --- drivers/gpu/drm/panfrost/panfrost_drv.c | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/drivers/gpu/drm/panfrost/panfrost_drv.c b/drivers/gpu/drm/panfrost/panfrost_drv.c index 2fa5afe21288..38ba755a0c3d 100644 --- a/drivers/gpu/drm/panfrost/panfrost_drv.c +++ b/drivers/gpu/drm/panfrost/panfrost_drv.c @@ -563,6 +563,11 @@ static int panfrost_probe(struct platform_device *pdev) return -ENODEV; pfdev->coherent = device_get_dma_attr(&pdev->dev) == DEV_DMA_COHERENT; + if (!pfdev->coherent && (of_device_is_compatible(of_root, "baikal,baikal-m") || + of_device_is_compatible(of_root, "baikal,bm1000"))) { + pfdev->coherent = true; + dev_warn(&pdev->dev, "marking as DMA coherent on BE-M1000"); + } /* Allocate and initialize the DRM device. */ ddev = drm_dev_alloc(&panfrost_drm_driver, &pdev->dev); -- 2.33.5