[PATCH] iommu/mediatek-v1: Add def_domain_type

[PATCH] iommu/mediatek-v1: Add def_domain_type

[Yong Wu]

The MediaTek V1 IOMMU is arm32 whose default domain type is
IOMMU_DOMAIN_UNMANAGED. Add this to satisfy the bus_iommu_probe to
enter "probe_finalize".

The iommu framework will create a iommu domain for each a device.
But all the devices share a iommu domain here, thus we skip all the
other domains in the "attach_device" except the domain we create
internally with arm_iommu_create_mapping.

Also a minor change: in the attach_device, "data" always is not null.
Remove "if (!data) return".

Signed-off-by: Yong Wu <yong.wu@mediatek.com>
---
a. rebase on linux-next.
b. After this patch and fixed the mutex issue(locally I only move
   mutex_unlock(&group->mutex) before __iommu_group_dma_attach(group)),
   the mtk_iommu_v1.c could work normally.
---
 drivers/iommu/mtk_iommu_v1.c | 16 +++++++++++-----
 1 file changed, 11 insertions(+), 5 deletions(-)

diff --git a/drivers/iommu/mtk_iommu_v1.c b/drivers/iommu/mtk_iommu_v1.c
index 7bdd74c..f353b07 100644
--- a/drivers/iommu/mtk_iommu_v1.c
+++ b/drivers/iommu/mtk_iommu_v1.c
@@ -265,10 +265,13 @@ static int mtk_iommu_attach_device(struct iommu_domain *domain,
 {
 	struct mtk_iommu_data *data = dev_iommu_priv_get(dev);
 	struct mtk_iommu_domain *dom = to_mtk_domain(domain);
+	struct dma_iommu_mapping *mtk_mapping;
 	int ret;
 
-	if (!data)
-		return -ENODEV;
+	/* Only allow the domain created internally. */
+	mtk_mapping = data->dev->archdata.iommu;
+	if (mtk_mapping->domain != domain)
+		return 0;
 
 	if (!data->m4u_dom) {
 		data->m4u_dom = dom;
@@ -288,9 +291,6 @@ static void mtk_iommu_detach_device(struct iommu_domain *domain,
 {
 	struct mtk_iommu_data *data = dev_iommu_priv_get(dev);
 
-	if (!data)
-		return;
-
 	mtk_iommu_config(data, dev, false);
 }
 
@@ -416,6 +416,11 @@ static int mtk_iommu_create_mapping(struct device *dev,
 	return 0;
 }
 
+static int mtk_iommu_def_domain_type(struct device *dev)
+{
+	return IOMMU_DOMAIN_UNMANAGED;
+}
+
 static struct iommu_device *mtk_iommu_probe_device(struct device *dev)
 {
 	struct iommu_fwspec *fwspec = dev_iommu_fwspec_get(dev);
@@ -525,6 +530,7 @@ static int mtk_iommu_hw_init(const struct mtk_iommu_data *data)
 	.probe_device	= mtk_iommu_probe_device,
 	.probe_finalize = mtk_iommu_probe_finalize,
 	.release_device	= mtk_iommu_release_device,
+	.def_domain_type = mtk_iommu_def_domain_type,
 	.device_group	= generic_device_group,
 	.pgsize_bitmap	= ~0UL << MT2701_IOMMU_PAGE_SHIFT,
 };
-- 
1.9.1
_______________________________________________
iommu mailing list
iommu@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/iommu

Re: [PATCH] iommu/mediatek-v1: Add def_domain_type

[Joerg Roedel]

On Fri, May 15, 2020 at 04:08:43PM +0800, Yong Wu wrote:
> The MediaTek V1 IOMMU is arm32 whose default domain type is
> IOMMU_DOMAIN_UNMANAGED. Add this to satisfy the bus_iommu_probe to
> enter "probe_finalize".
> 
> The iommu framework will create a iommu domain for each a device.
> But all the devices share a iommu domain here, thus we skip all the
> other domains in the "attach_device" except the domain we create
> internally with arm_iommu_create_mapping.
> 
> Also a minor change: in the attach_device, "data" always is not null.
> Remove "if (!data) return".
> 
> Signed-off-by: Yong Wu <yong.wu@mediatek.com>
> ---
> a. rebase on linux-next.
> b. After this patch and fixed the mutex issue(locally I only move
>    mutex_unlock(&group->mutex) before __iommu_group_dma_attach(group)),
>    the mtk_iommu_v1.c could work normally.
> ---
>  drivers/iommu/mtk_iommu_v1.c | 16 +++++++++++-----
>  1 file changed, 11 insertions(+), 5 deletions(-)

Applied, thanks.
_______________________________________________
iommu mailing list
iommu@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/iommu