From patchwork Tue Feb 25 19:52:14 2025 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Akhil P Oommen X-Patchwork-Id: 13990646 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from gabe.freedesktop.org (gabe.freedesktop.org [131.252.210.177]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.lore.kernel.org (Postfix) with ESMTPS id C6F0AC021B2 for ; Tue, 25 Feb 2025 19:52:39 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 0A13310E7C0; Tue, 25 Feb 2025 19:52:39 +0000 (UTC) Authentication-Results: gabe.freedesktop.org; dkim=pass (2048-bit key; unprotected) header.d=quicinc.com header.i=@quicinc.com header.b="kMjeqMJA"; dkim-atps=neutral Received: from mx0b-0031df01.pphosted.com (mx0b-0031df01.pphosted.com [205.220.180.131]) by gabe.freedesktop.org (Postfix) with ESMTPS id 4318810E7C0; Tue, 25 Feb 2025 19:52:37 +0000 (UTC) Received: from pps.filterd (m0279868.ppops.net [127.0.0.1]) by mx0a-0031df01.pphosted.com (8.18.1.2/8.18.1.2) with ESMTP id 51PHinoW013301; Tue, 25 Feb 2025 19:52:32 GMT DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=quicinc.com; h= cc:content-transfer-encoding:content-type:date:from:message-id :mime-version:subject:to; s=qcppdkim1; bh=RrPOkSjYbqRM3KAmIfVN6s d80ddpWeBNzJEAuPYvib4=; b=kMjeqMJAeRyUkJVceGzPvk/pEL2c2+wy81Wujd VHM1g1w/bdUAML21OSwy96QL0BrdgGJ6V5WeLidYD1lUWkv7XLFswC3HfKSSj3uw Q2+sipoaVwOrS3ETV77d+58YvJay87V3dfEP0kLYPGInf9N28wHKRFsT5yoE41kc ksdASu1L/RiDQKfL8CNY7thlAJ8Elz2wwsl5fwfQDJwpz4bY0XAlV+F/WcOJenG+ KyGxlvDMIuBMv92NzxwujM9Qj8SXzAMncTDZZRFwif67ztaJqLM3BM9MjeuqsV9j x7GMnac9mOr6JJFOqLM5JajYuZfda3S+cKwrvsptj1UPXZ+g== Received: from nasanppmta04.qualcomm.com (i-global254.qualcomm.com [199.106.103.254]) by mx0a-0031df01.pphosted.com (PPS) with ESMTPS id 44y5wgt44k-1 (version=TLSv1.2 cipher=ECDHE-RSA-AES256-GCM-SHA384 bits=256 verify=NOT); Tue, 25 Feb 2025 19:52:32 +0000 (GMT) Received: from nasanex01b.na.qualcomm.com (nasanex01b.na.qualcomm.com [10.46.141.250]) by NASANPPMTA04.qualcomm.com (8.18.1.2/8.18.1.2) with ESMTPS id 51PJqU6F016222 (version=TLSv1.2 cipher=ECDHE-RSA-AES256-GCM-SHA384 bits=256 verify=NOT); Tue, 25 Feb 2025 19:52:30 GMT Received: from [10.213.111.143] (10.80.80.8) by nasanex01b.na.qualcomm.com (10.46.141.250) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384) id 15.2.1544.9; Tue, 25 Feb 2025 11:52:26 -0800 From: Akhil P Oommen Date: Wed, 26 Feb 2025 01:22:14 +0530 Subject: [PATCH v2] drm/msm/a6xx: Fix stale rpmh votes from GPU MIME-Version: 1.0 Message-ID: <20250226-adreno-sys-suspend-fix-v2-1-cc9628d583ed@quicinc.com> X-B4-Tracking: v=1; b=H4sIAG0fvmcC/3WOwW7CMAyGX6XyeZ7iLGkZp70H4pA4LuRAyuJSg VDffVk5c/x+y5++J6jULAr77glVlqx5Kg3sRwd8DuUkmFNjsMZ6Q2QwpCplQn0o6k2vUhKO+Y7 sIw9RJBh20J6vVdq8iQ/HF1f5vTX//BohBhXk6XLJ874bbDJhdIE5ffc+8ZcdvXHOmTS4dnGJd nEcdgT/rnPWeaqPrXmhTbblWdu/y1sICY13tqcYA5H7aS2cC3+2Ajiu6/oHLYmcDAoBAAA= X-Change-ID: 20250110-adreno-sys-suspend-fix-c5bc7beea0c4 To: Rob Clark , Sean Paul , "Konrad Dybcio" , Abhinav Kumar , Dmitry Baryshkov , Marijn Suijten , David Airlie , "Simona Vetter" , Jordan Crouse CC: , , , , , Akhil P Oommen X-Mailer: b4 0.14.2 X-Developer-Signature: v=1; a=ed25519-sha256; t=1740513146; l=4190; i=quic_akhilpo@quicinc.com; s=20240726; h=from:subject:message-id; bh=oIhL+Qdbu1UaNBgkh+rV4EwGPx0L/bL4d+e/qgMhUKA=; b=092qsXfcOFw3ETYPgMSy1zSqbQyxWUM+17l4lMjbcOTzy8M9zEEc5eWnScMMw1/fLfEi9QJmy S7HVwvjy79vDuHI5llfILigpY8ug9Zo+XgHic3WM8y62flP9kQKVZfk X-Developer-Key: i=quic_akhilpo@quicinc.com; a=ed25519; pk=lmVtttSHmAUYFnJsQHX80IIRmYmXA4+CzpGcWOOsfKA= X-Originating-IP: [10.80.80.8] X-ClientProxiedBy: nasanex01b.na.qualcomm.com (10.46.141.250) To nasanex01b.na.qualcomm.com (10.46.141.250) X-QCInternal: smtphost X-Proofpoint-Virus-Version: vendor=nai engine=6200 definitions=5800 signatures=585085 X-Proofpoint-GUID: 7KmwKPSKBkTJoDMNo38VqcOc9byaNS9R X-Proofpoint-ORIG-GUID: 7KmwKPSKBkTJoDMNo38VqcOc9byaNS9R X-Proofpoint-Virus-Version: vendor=baseguard engine=ICAP:2.0.293,Aquarius:18.0.1057,Hydra:6.0.680,FMLib:17.12.68.34 definitions=2025-02-25_06,2025-02-25_03,2024-11-22_01 X-Proofpoint-Spam-Details: rule=outbound_notspam policy=outbound score=0 priorityscore=1501 impostorscore=0 suspectscore=0 lowpriorityscore=0 bulkscore=0 clxscore=1011 adultscore=0 phishscore=0 mlxlogscore=999 spamscore=0 mlxscore=0 malwarescore=0 classifier=spam adjust=0 reason=mlx scancount=1 engine=8.19.0-2502100000 definitions=main-2502250120 X-BeenThere: dri-devel@lists.freedesktop.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Direct Rendering Infrastructure - Development List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dri-devel-bounces@lists.freedesktop.org Sender: "dri-devel" It was observed on sc7180 (A618 gpu) that GPU votes for GX rail and CNOC BCM nodes were not removed after GPU suspend. This was because we skipped sending 'prepare-slumber' request to gmu during suspend sequence in some cases. So, make sure we always call prepare-slumber hfi during suspend. Also, calling prepare-slumber without a prior oob-gpu handshake messes up gmu firmware's internal state. So, do that when required. Fixes: 4b565ca5a2cb ("drm/msm: Add A6XX device support") Cc: stable@vger.kernel.org Signed-off-by: Akhil P Oommen --- Changes in v2: - Minor update to commit text and CC'ed Stable - Link to v1: https://lore.kernel.org/r/20250226-adreno-sys-suspend-fix-v1-1-054261bba114@quicinc.com --- drivers/gpu/drm/msm/adreno/a6xx_gmu.c | 72 +++++++++++++++++++---------------- 1 file changed, 39 insertions(+), 33 deletions(-) --- base-commit: 72d0af4accd965dc32f504440d74d0a4d18bf781 change-id: 20250110-adreno-sys-suspend-fix-c5bc7beea0c4 Best regards, diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gmu.c b/drivers/gpu/drm/msm/adreno/a6xx_gmu.c index 699b0dd34b18f0ec811e975779ba95991d485098..38c94915d4c9d6d33354502651a77c1f9e4648df 100644 --- a/drivers/gpu/drm/msm/adreno/a6xx_gmu.c +++ b/drivers/gpu/drm/msm/adreno/a6xx_gmu.c @@ -1169,49 +1169,50 @@ static void a6xx_gmu_shutdown(struct a6xx_gmu *gmu) struct a6xx_gpu *a6xx_gpu = container_of(gmu, struct a6xx_gpu, gmu); struct adreno_gpu *adreno_gpu = &a6xx_gpu->base; u32 val; + int ret; /* - * The GMU may still be in slumber unless the GPU started so check and - * skip putting it back into slumber if so + * GMU firmware's internal power state gets messed up if we send "prepare_slumber" hfi when + * oob_gpu handshake wasn't done after the last wake up. So do a dummy handshake here when + * required */ - val = gmu_read(gmu, REG_A6XX_GPU_GMU_CX_GMU_RPMH_POWER_STATE); + if (adreno_gpu->base.needs_hw_init) { + if (a6xx_gmu_set_oob(&a6xx_gpu->gmu, GMU_OOB_GPU_SET)) + goto force_off; - if (val != 0xf) { - int ret = a6xx_gmu_wait_for_idle(gmu); + a6xx_gmu_clear_oob(&a6xx_gpu->gmu, GMU_OOB_GPU_SET); + } - /* If the GMU isn't responding assume it is hung */ - if (ret) { - a6xx_gmu_force_off(gmu); - return; - } + ret = a6xx_gmu_wait_for_idle(gmu); - a6xx_bus_clear_pending_transactions(adreno_gpu, a6xx_gpu->hung); + /* If the GMU isn't responding assume it is hung */ + if (ret) + goto force_off; - /* tell the GMU we want to slumber */ - ret = a6xx_gmu_notify_slumber(gmu); - if (ret) { - a6xx_gmu_force_off(gmu); - return; - } + a6xx_bus_clear_pending_transactions(adreno_gpu, a6xx_gpu->hung); - ret = gmu_poll_timeout(gmu, - REG_A6XX_GPU_GMU_AO_GPU_CX_BUSY_STATUS, val, - !(val & A6XX_GPU_GMU_AO_GPU_CX_BUSY_STATUS_GPUBUSYIGNAHB), - 100, 10000); + /* tell the GMU we want to slumber */ + ret = a6xx_gmu_notify_slumber(gmu); + if (ret) + goto force_off; - /* - * Let the user know we failed to slumber but don't worry too - * much because we are powering down anyway - */ + ret = gmu_poll_timeout(gmu, + REG_A6XX_GPU_GMU_AO_GPU_CX_BUSY_STATUS, val, + !(val & A6XX_GPU_GMU_AO_GPU_CX_BUSY_STATUS_GPUBUSYIGNAHB), + 100, 10000); - if (ret) - DRM_DEV_ERROR(gmu->dev, - "Unable to slumber GMU: status = 0%x/0%x\n", - gmu_read(gmu, - REG_A6XX_GPU_GMU_AO_GPU_CX_BUSY_STATUS), - gmu_read(gmu, - REG_A6XX_GPU_GMU_AO_GPU_CX_BUSY_STATUS2)); - } + /* + * Let the user know we failed to slumber but don't worry too + * much because we are powering down anyway + */ + + if (ret) + DRM_DEV_ERROR(gmu->dev, + "Unable to slumber GMU: status = 0%x/0%x\n", + gmu_read(gmu, + REG_A6XX_GPU_GMU_AO_GPU_CX_BUSY_STATUS), + gmu_read(gmu, + REG_A6XX_GPU_GMU_AO_GPU_CX_BUSY_STATUS2)); /* Turn off HFI */ a6xx_hfi_stop(gmu); @@ -1221,6 +1222,11 @@ static void a6xx_gmu_shutdown(struct a6xx_gmu *gmu) /* Tell RPMh to power off the GPU */ a6xx_rpmh_stop(gmu); + + return; + +force_off: + a6xx_gmu_force_off(gmu); }