public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Sasha Levin <Alexander.Levin@microsoft.com>
To: "stable@vger.kernel.org" <stable@vger.kernel.org>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
Cc: Thierry Reding <treding@nvidia.com>,
	Ben Skeggs <bskeggs@redhat.com>,
	Sasha Levin <Alexander.Levin@microsoft.com>
Subject: [PATCH AUTOSEL 4.9 16/34] drm/nouveau: tegra: Detach from ARM DMA/IOMMU mapping
Date: Sat, 15 Sep 2018 01:34:35 +0000	[thread overview]
Message-ID: <20180915013422.180023-16-alexander.levin@microsoft.com> (raw)
In-Reply-To: <20180915013422.180023-1-alexander.levin@microsoft.com>

From: Thierry Reding <treding@nvidia.com>

[ Upstream commit b59fb482b52269977ee5de205308e5b236a03917 ]

Depending on the kernel configuration, early ARM architecture setup code
may have attached the GPU to a DMA/IOMMU mapping that transparently uses
the IOMMU to back the DMA API. Tegra requires special handling for IOMMU
backed buffers (a special bit in the GPU's MMU page tables indicates the
memory path to take: via the SMMU or directly to the memory controller).
Transparently backing DMA memory with an IOMMU prevents Nouveau from
properly handling such memory accesses and causes memory access faults.

As a side-note: buffers other than those allocated in instance memory
don't need to be physically contiguous from the GPU's perspective since
the GPU can map them into contiguous buffers using its own MMU. Mapping
these buffers through the IOMMU is unnecessary and will even lead to
performance degradation because of the additional translation. One
exception to this are compressible buffers which need large pages. In
order to enable these large pages, multiple small pages will have to be
combined into one large (I/O virtually contiguous) mapping via the
IOMMU. However, that is a topic outside the scope of this fix and isn't
currently supported. An implementation will want to explicitly create
these large pages in the Nouveau driver, so detaching from a DMA/IOMMU
mapping would still be required.

Signed-off-by: Thierry Reding <treding@nvidia.com>
Acked-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Robin Murphy <robin.murphy@arm.com>
Tested-by: Nicolas Chauvet <kwizart@gmail.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
---
 drivers/gpu/drm/nouveau/nvkm/engine/device/tegra.c | 13 +++++++++++++
 1 file changed, 13 insertions(+)

diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/device/tegra.c b/drivers/gpu/drm/nouveau/nvkm/engine/device/tegra.c
index 9b638bd905ff..d370bf8bc409 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/device/tegra.c
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/device/tegra.c
@@ -23,6 +23,10 @@
 #ifdef CONFIG_NOUVEAU_PLATFORM_DRIVER
 #include "priv.h"
 
+#if IS_ENABLED(CONFIG_ARM_DMA_USE_IOMMU)
+#include <asm/dma-iommu.h>
+#endif
+
 static int
 nvkm_device_tegra_power_up(struct nvkm_device_tegra *tdev)
 {
@@ -95,6 +99,15 @@ nvkm_device_tegra_probe_iommu(struct nvkm_device_tegra *tdev)
 	unsigned long pgsize_bitmap;
 	int ret;
 
+#if IS_ENABLED(CONFIG_ARM_DMA_USE_IOMMU)
+	if (dev->archdata.mapping) {
+		struct dma_iommu_mapping *mapping = to_dma_iommu_mapping(dev);
+
+		arm_iommu_detach_device(dev);
+		arm_iommu_release_mapping(mapping);
+	}
+#endif
+
 	if (!tdev->func->iommu_bit)
 		return;
 
-- 
2.17.1

  parent reply	other threads:[~2018-09-15  1:34 UTC|newest]

Thread overview: 37+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-09-15  1:34 [PATCH AUTOSEL 4.9 01/34] binfmt_elf: Respect error return from `regset->active' Sasha Levin
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 02/34] audit: fix use-after-free in audit_add_watch Sasha Levin
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 04/34] mtd: rawnand: sunxi: Add an U suffix to NFC_PAGE_OP definition Sasha Levin
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 03/34] mtdchar: fix overflows in adjustment of `count` Sasha Levin
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 05/34] evm: Don't deadlock if a crypto algorithm is unavailable Sasha Levin
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 06/34] PM / devfreq: use put_device() instead of kfree() Sasha Levin
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 07/34] MIPS: loongson64: cs5536: Fix PCI_OHCI_INT_REG reads Sasha Levin
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 08/34] configfs: fix registered group removal Sasha Levin
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 10/34] ARM: hisi: handle of_iomap and fix missing of_node_put Sasha Levin
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 09/34] efi/esrt: Only call efi_mem_reserve() for boot services memory Sasha Levin
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 11/34] ARM: hisi: fix error handling and missing of_node_put Sasha Levin
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 12/34] ARM: hisi: check of_iomap and fix " Sasha Levin
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 13/34] gpu: ipu-v3: csi: pass back mbus_code_to_bus_cfg error codes Sasha Levin
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 14/34] mmc: tegra: prevent HS200 on Tegra 3 Sasha Levin
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 15/34] mmc: sdhci: do not try to use 3.3V signaling if not supported Sasha Levin
2018-09-15  1:34 ` Sasha Levin [this message]
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 18/34] coresight: Handle errors in finding input/output ports Sasha Levin
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 17/34] parport: sunbpp: fix error return code Sasha Levin
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 19/34] coresight: tpiu: Fix disabling timeouts Sasha Levin
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 20/34] gpio: pxa: Fix potential NULL dereference Sasha Levin
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 21/34] gpiolib: Mark gpio_suffixes array with __maybe_unused Sasha Levin
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 22/34] mfd: 88pm860x-i2c: switch to i2c_lock_bus(..., I2C_LOCK_SEGMENT) Sasha Levin
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 24/34] rcu: Fix grace-period hangs due to race with CPU offline Sasha Levin
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 23/34] input: rohm_bu21023: switch to i2c_lock_bus(..., I2C_LOCK_SEGMENT) Sasha Levin
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 25/34] drm/amdkfd: Fix error codes in kfd_get_process Sasha Levin
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 27/34] ALSA: pcm: Fix snd_interval_refine first/last with open min/max Sasha Levin
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 26/34] rtc: bq4802: add error handling for devm_ioremap Sasha Levin
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 28/34] selftest: timers: Tweak raw_skew to SKIP when ADJ_OFFSET/other clock adjustments are in progress Sasha Levin
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 29/34] drm/panel: type promotion bug in s6e8aa0_read_mtp_id() Sasha Levin
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 30/34] dmaengine: hsu: Support dmaengine_terminate_sync() Sasha Levin
2018-09-20  0:38   ` Vinod
2018-09-24  0:55     ` Sasha Levin
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 31/34] dmaengine: idma64: " Sasha Levin
2018-09-20  0:40   ` Vinod
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 32/34] IB/nes: Fix a compiler warning Sasha Levin
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 33/34] gpiolib: Respect error code of ->get_direction() Sasha Levin
2018-09-15  1:34 ` [PATCH AUTOSEL 4.9 34/34] pinctrl: qcom: spmi-gpio: Fix pmic_gpio_config_get() to be compliant Sasha Levin

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20180915013422.180023-16-alexander.levin@microsoft.com \
    --to=alexander.levin@microsoft.com \
    --cc=bskeggs@redhat.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=stable@vger.kernel.org \
    --cc=treding@nvidia.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox