Il 14/02/23 04:11, Yong Wu ha scritto:
After commit f1ad5338a4d5 ("of: Fix "dma-ranges" handling for bus
controllers"), the dma-ranges is not allowed for dts leaf node.
but we still would like to separate to different masters
into different iova regions.
Thus we have to separate it by the HW larbid and portid. For example,
larb1/2 are in region2 and larb3 is in region3. The problem is that
some ports inside a larb are in region4 while some ports inside this
larb are in region5. Therefore I define a "iova_region_larb_msk" to help
record the information for each a port. Take a example for a larb:
[1] = ~0: means all ports in this larb are in region1;
[2] = BIT(3) | BIT(4): means port3/4 in this larb are region2;
[3] = ~(BIT(3) | BIT(4)): means all the other ports except port3/4
in this larb are region3.
This method also avoids the users forget/abuse the iova regions.
Signed-off-by: Yong Wu <yong.wu@xxxxxxxxxxxx>
---
drivers/iommu/mtk_iommu.c | 47 ++++++++++++++++++++++-----------------
1 file changed, 27 insertions(+), 20 deletions(-)
diff --git a/drivers/iommu/mtk_iommu.c b/drivers/iommu/mtk_iommu.c
index ab53edcb221f..7e2cb3b8cac8 100644
--- a/drivers/iommu/mtk_iommu.c
+++ b/drivers/iommu/mtk_iommu.c
@@ -8,7 +8,6 @@
#include <linux/clk.h>
#include <linux/component.h>
#include <linux/device.h>
-#include <linux/dma-direct.h>
#include <linux/err.h>
#include <linux/interrupt.h>
#include <linux/io.h>
@@ -212,6 +211,11 @@ struct mtk_iommu_plat_data {
struct {
unsigned int iova_region_nr;
const struct mtk_iommu_iova_region *iova_region;
+ /*
+ * Indicate the correspondence between larbs/ports and regions.
+ * The index is same with iova_region.
* The index is the same as iova_region and larb port numbers are
* described as bit positions.
* For example, storing BIT(0) at index 2 means "port 0 is in region 2".
That's most probably the best short explanation that we can give, so that nobody
goes crazy with understanding this one.
After fixing that comment, you totally deserve my
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@xxxxxxxxxxxxx>
Cheers!
