From patchwork Wed Mar 25 20:16:03 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Thierry Reding X-Patchwork-Id: 1261638 Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Authentication-Results: ozlabs.org; spf=none (no SPF record) smtp.mailfrom=vger.kernel.org (client-ip=209.132.180.67; helo=vger.kernel.org; envelope-from=linux-tegra-owner@vger.kernel.org; receiver=) Authentication-Results: ozlabs.org; dmarc=pass (p=none dis=none) header.from=gmail.com Authentication-Results: ozlabs.org; dkim=pass (2048-bit key; unprotected) header.d=gmail.com header.i=@gmail.com header.a=rsa-sha256 header.s=20161025 header.b=sj7XVrax; dkim-atps=neutral Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by ozlabs.org (Postfix) with ESMTP id 48nfWs1X7mz9sRf for ; Thu, 26 Mar 2020 07:16:09 +1100 (AEDT) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1727374AbgCYUQI (ORCPT ); Wed, 25 Mar 2020 16:16:08 -0400 Received: from mail-wm1-f68.google.com ([209.85.128.68]:50585 "EHLO mail-wm1-f68.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1727328AbgCYUQI (ORCPT ); Wed, 25 Mar 2020 16:16:08 -0400 Received: by mail-wm1-f68.google.com with SMTP id d198so4073496wmd.0 for ; Wed, 25 Mar 2020 13:16:07 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20161025; h=from:to:cc:subject:date:message-id:mime-version :content-transfer-encoding; bh=yPLnagV0XBnTWT+nGjtRaD+LnSq2BqmeAJnp8U+CWPw=; b=sj7XVraxdwiyRAeepEQ0wy1nLUUH6vcloNotxoFwaAZmvU2GILePtp+OM8VZxzmSg1 qVjos+BzgdtxI0QGYvlsRwZJmw1PdwfTDzM8kMKmP2AfXDgnFG7LZsGZnzTmdPqErqG6 RfQwpZiPunHplEvI/epnPHACQlV9HoX+teAIWP9gyJkMYwBCVOirkfv4yGqGZWyEciZ2 yM5mGeUZ/OprHtVVEEuF5yb50CJm8cBEHBMr2ooS+0jm+avVEG8DKe9QM2nWgJB7+TXH 7+iryK1A4PDr9L6syw0p6sAbkFd2+P/p44d/rqsKPWTQG0lkd0cgRHx9fVPls/P4Snyr JwCA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:from:to:cc:subject:date:message-id:mime-version :content-transfer-encoding; bh=yPLnagV0XBnTWT+nGjtRaD+LnSq2BqmeAJnp8U+CWPw=; b=HWu2t1YnW/GoMLlkfp6ZQha8CvUnfMi/OK1zsN3hDtTtMLwVQL9YBFPvXYfAASIGzA qXmgdbIdQmwOXRxlDmgcXk8KcOJmvnJTSoE+GPeLrKGVq9h2c6XLINshs7RDWqY7//GM /NMVkESX/sVh5qVQYVzsQOBWAsLkwpVAmt3lJ81XrCGdA/L5aN2FWOftTWJWoStgtHuB 9N27ffBkV8/72gDCcGxM/lJlfxMBcfPIEMDGWlErsl2U/EPtF+e5AH1kF9/a+lImxa1h vBlXvgfPKazfOLm1jA809U0QJrCy5bmTOJsaLqnkLPNJRyvlY6JZqk8a1Wc4u6l44uoI 4l3g== X-Gm-Message-State: ANhLgQ0GzmzHn/uC4G4GzXRW/D8i6fcQ7Y04Wxx+yBOvoeixp0lD9PYD 9Q7E3Ezt7uCnfh5D41Ym8jY= X-Google-Smtp-Source: ADFU+vvV+Qjqcd+wksczhsC9MSisSEM36LfhftNulFkmYxqwCfpDcq22YDEoWHYpgjaXwwZC4lgCyg== X-Received: by 2002:a7b:c842:: with SMTP id c2mr5416219wml.154.1585167366416; Wed, 25 Mar 2020 13:16:06 -0700 (PDT) Received: from localhost (p200300E41F4A9B0076D02BFFFE273F51.dip0.t-ipconnect.de. [2003:e4:1f4a:9b00:76d0:2bff:fe27:3f51]) by smtp.gmail.com with ESMTPSA id i4sm132568wrm.32.2020.03.25.13.16.05 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Wed, 25 Mar 2020 13:16:05 -0700 (PDT) From: Thierry Reding To: Thierry Reding Cc: dri-devel@lists.freedesktop.org, linux-tegra@vger.kernel.org Subject: [PATCH 1/2] drm/tegra: Fix SMMU support on Tegra124 and Tegra210 Date: Wed, 25 Mar 2020 21:16:03 +0100 Message-Id: <20200325201604.833898-1-thierry.reding@gmail.com> X-Mailer: git-send-email 2.24.1 MIME-Version: 1.0 Sender: linux-tegra-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-tegra@vger.kernel.org From: Thierry Reding When testing whether or not to enable the use of the SMMU, consult the supported DMA mask rather than the actually configured DMA mask, since the latter might already have been restricted. Fixes: 2d9384ff9177 ("drm/tegra: Relax IOMMU usage criteria on old Tegra") Signed-off-by: Thierry Reding Tested-by: Jon Hunter --- drivers/gpu/drm/tegra/drm.c | 3 ++- drivers/gpu/host1x/dev.c | 13 +++++++++++++ include/linux/host1x.h | 3 +++ 3 files changed, 18 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/tegra/drm.c b/drivers/gpu/drm/tegra/drm.c index bd268028fb3d..583cd6e0ae27 100644 --- a/drivers/gpu/drm/tegra/drm.c +++ b/drivers/gpu/drm/tegra/drm.c @@ -1039,6 +1039,7 @@ void tegra_drm_free(struct tegra_drm *tegra, size_t size, void *virt, static bool host1x_drm_wants_iommu(struct host1x_device *dev) { + struct host1x *host1x = dev_get_drvdata(dev->dev.parent); struct iommu_domain *domain; /* @@ -1076,7 +1077,7 @@ static bool host1x_drm_wants_iommu(struct host1x_device *dev) * sufficient and whether or not the host1x is attached to an IOMMU * doesn't matter. */ - if (!domain && dma_get_mask(dev->dev.parent) <= DMA_BIT_MASK(32)) + if (!domain && host1x_get_dma_mask(host1x) <= DMA_BIT_MASK(32)) return true; return domain != NULL; diff --git a/drivers/gpu/host1x/dev.c b/drivers/gpu/host1x/dev.c index 388bcc2889aa..40a4b9f8b861 100644 --- a/drivers/gpu/host1x/dev.c +++ b/drivers/gpu/host1x/dev.c @@ -502,6 +502,19 @@ static void __exit tegra_host1x_exit(void) } module_exit(tegra_host1x_exit); +/** + * host1x_get_dma_mask() - query the supported DMA mask for host1x + * @host1x: host1x instance + * + * Note that this returns the supported DMA mask for host1x, which can be + * different from the applicable DMA mask under certain circumstances. + */ +u64 host1x_get_dma_mask(struct host1x *host1x) +{ + return host1x->info->dma_mask; +} +EXPORT_SYMBOL(host1x_get_dma_mask); + MODULE_AUTHOR("Thierry Reding "); MODULE_AUTHOR("Terje Bergstrom "); MODULE_DESCRIPTION("Host1x driver for Tegra products"); diff --git a/include/linux/host1x.h b/include/linux/host1x.h index 62d216ff1097..c230b4e70d75 100644 --- a/include/linux/host1x.h +++ b/include/linux/host1x.h @@ -17,9 +17,12 @@ enum host1x_class { HOST1X_CLASS_GR3D = 0x60, }; +struct host1x; struct host1x_client; struct iommu_group; +u64 host1x_get_dma_mask(struct host1x *host1x); + /** * struct host1x_client_ops - host1x client operations * @init: host1x client initialization code From patchwork Wed Mar 25 20:16:04 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Thierry Reding X-Patchwork-Id: 1261639 Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Authentication-Results: ozlabs.org; spf=none (no SPF record) smtp.mailfrom=vger.kernel.org (client-ip=209.132.180.67; helo=vger.kernel.org; envelope-from=linux-tegra-owner@vger.kernel.org; receiver=) Authentication-Results: ozlabs.org; dmarc=pass (p=none dis=none) header.from=gmail.com Authentication-Results: ozlabs.org; dkim=pass (2048-bit key; unprotected) header.d=gmail.com header.i=@gmail.com header.a=rsa-sha256 header.s=20161025 header.b=XXUz449u; dkim-atps=neutral Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by ozlabs.org (Postfix) with ESMTP id 48nfWw6NvSz9sPk for ; Thu, 26 Mar 2020 07:16:12 +1100 (AEDT) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1727316AbgCYUQM (ORCPT ); Wed, 25 Mar 2020 16:16:12 -0400 Received: from mail-wr1-f65.google.com ([209.85.221.65]:33914 "EHLO mail-wr1-f65.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1727328AbgCYUQM (ORCPT ); Wed, 25 Mar 2020 16:16:12 -0400 Received: by mail-wr1-f65.google.com with SMTP id 65so4990084wrl.1 for ; Wed, 25 Mar 2020 13:16:09 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20161025; h=from:to:cc:subject:date:message-id:in-reply-to:references :mime-version:content-transfer-encoding; bh=aW1zxIHiei+l8kDSE2lVXf/aMBDE/GtIkGFrQXvKkrY=; b=XXUz449uJivXz+1lH6pKa9IvT3vUx61/skXaEyQxpkslFR268FwckKE0ryQDUx701N hFN9ocSGCuE6bKpdgya8YmthXDASOYWZzKV0R5jms1rqgazVMF6jARv+kE4Jaj9Ek4tl 4eTpmnHinx0xIrgGWCQbfltjb+zAE5XOGX8UCX1526r3yQQpu+OQlKZ70Tvq3pdw0zfT URkTU8sfdTa9DCxUSsUukPcK9vKOk6XHkFleL6FisODDvXphdzzLa1TCv9UTGLrUsHSd XDrukLto5efrUE03q5jP6ZN4xbnLDbhY6IkB7PAW1qwSPG/Eg0p0ivpJ58+QwwmBH6zF ByDQ== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:from:to:cc:subject:date:message-id:in-reply-to :references:mime-version:content-transfer-encoding; bh=aW1zxIHiei+l8kDSE2lVXf/aMBDE/GtIkGFrQXvKkrY=; b=DIWKPWCoYx1rnX34DSkRPm2K6lR1SurVvq+IIY5Nrc9uq+E3pmXQcActG0DDAHHK8a SgnziEvuWTeROgrlwONYq+FUZRQ6s1TRR1+qDXqAlRtdebU/cEep+LRvdzJe/qJBpPqd SnSTR3Xntgo7EcyLRj9YqSodasylPt3OzrhuDudfTSQtKZghElLfyJV/tzgwG+OC3TD4 RJAykZ0tgWHy7Bc1UB+z6LovuT/sgcPUSLfNqDehQWqwQeqHqXgFAomUN0CCEr2YdjkT sCpBZPqKtb22FdDWlDiNnEkEmMPA+K4MIWbZL9VuvArjFaaBn6fBxvnX4tAKEcOiKeUy EZXw== X-Gm-Message-State: ANhLgQ1Vj1gSFYKgV/7jV1T3UIwTE5jasGmLOhuuGuWvjBs2xXUgieyz VhNVgYIYU/8R/0Vx9Hv44rw= X-Google-Smtp-Source: ADFU+vtTfrVHW69I+ZhOz8qw8xUje/j42rKoNxAP2wTt+E5WQ5s6QhBcgeHzC4Bw5Q5NdWxjLUtZ/g== X-Received: by 2002:adf:800e:: with SMTP id 14mr5104354wrk.369.1585167368929; Wed, 25 Mar 2020 13:16:08 -0700 (PDT) Received: from localhost (p200300E41F4A9B0076D02BFFFE273F51.dip0.t-ipconnect.de. [2003:e4:1f4a:9b00:76d0:2bff:fe27:3f51]) by smtp.gmail.com with ESMTPSA id e9sm151985wrw.30.2020.03.25.13.16.07 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Wed, 25 Mar 2020 13:16:07 -0700 (PDT) From: Thierry Reding To: Thierry Reding Cc: dri-devel@lists.freedesktop.org, linux-tegra@vger.kernel.org Subject: [PATCH 2/2] gpu: host1x: Use SMMU on Tegra124 and Tegra210 Date: Wed, 25 Mar 2020 21:16:04 +0100 Message-Id: <20200325201604.833898-2-thierry.reding@gmail.com> X-Mailer: git-send-email 2.24.1 In-Reply-To: <20200325201604.833898-1-thierry.reding@gmail.com> References: <20200325201604.833898-1-thierry.reding@gmail.com> MIME-Version: 1.0 Sender: linux-tegra-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-tegra@vger.kernel.org From: Thierry Reding Tegra124 and Tegra210 support addressing more than 32 bits of physical memory. However, since their host1x does not support the wide GATHER opcode, they should use the SMMU if at all possible to ensure that all the system memory can be used for command buffers, irrespective of whether or not the host1x firewall is enabled. Signed-off-by: Thierry Reding Tested-by: Jon Hunter --- drivers/gpu/host1x/dev.c | 46 ++++++++++++++++++++++++++++++++++++---- 1 file changed, 42 insertions(+), 4 deletions(-) diff --git a/drivers/gpu/host1x/dev.c b/drivers/gpu/host1x/dev.c index 40a4b9f8b861..d24344e91922 100644 --- a/drivers/gpu/host1x/dev.c +++ b/drivers/gpu/host1x/dev.c @@ -192,17 +192,55 @@ static void host1x_setup_sid_table(struct host1x *host) } } +static bool host1x_wants_iommu(struct host1x *host1x) +{ + /* + * If we support addressing a maximum of 32 bits of physical memory + * and if the host1x firewall is enabled, there's no need to enable + * IOMMU support. This can happen for example on Tegra20, Tegra30 + * and Tegra114. + * + * Tegra124 and later can address up to 34 bits of physical memory and + * many platforms come equipped with more than 2 GiB of system memory, + * which requires crossing the 4 GiB boundary. But there's a catch: on + * SoCs before Tegra186 (i.e. Tegra124 and Tegra210), the host1x can + * only address up to 32 bits of memory in GATHER opcodes, which means + * that command buffers need to either be in the first 2 GiB of system + * memory (which could quickly lead to memory exhaustion), or command + * buffers need to be treated differently from other buffers (which is + * not possible with the current ABI). + * + * A third option is to use the IOMMU in these cases to make sure all + * buffers will be mapped into a 32-bit IOVA space that host1x can + * address. This allows all of the system memory to be used and works + * within the limitations of the host1x on these SoCs. + * + * In summary, default to enable IOMMU on Tegra124 and later. For any + * of the earlier SoCs, only use the IOMMU for additional safety when + * the host1x firewall is disabled. + */ + if (host1x->info->dma_mask <= DMA_BIT_MASK(32)) { + if (IS_ENABLED(CONFIG_TEGRA_HOST1X_FIREWALL)) + return false; + } + + return true; +} + static struct iommu_domain *host1x_iommu_attach(struct host1x *host) { struct iommu_domain *domain = iommu_get_domain_for_dev(host->dev); int err; /* - * If the host1x firewall is enabled, there's no need to enable IOMMU - * support. Similarly, if host1x is already attached to an IOMMU (via - * the DMA API), don't try to attach again. + * We may not always want to enable IOMMU support (for example if the + * host1x firewall is already enabled and we don't support addressing + * more than 32 bits of physical memory), so check for that first. + * + * Similarly, if host1x is already attached to an IOMMU (via the DMA + * API), don't try to attach again. */ - if (IS_ENABLED(CONFIG_TEGRA_HOST1X_FIREWALL) || domain) + if (!host1x_wants_iommu(host) || domain) return domain; host->group = iommu_group_get(host->dev);