Hi Bruno,
On 10/27/2016 12:43 PM, Bruno Herrera wrote:
Hi Alex,
On Wed, Oct 26, 2016 at 7:09 AM, Alexandre Torgue
<alexandre.torgue@xxxxxx> wrote:
Hi Bruno,
On 10/25/2016 11:06 PM, Bruno Herrera wrote:
Hi Alexandre,
stm32f469-disco and stm32f429-eval boards use SDRAM start address
remapping
(to @0) to boost performances. A DMA translation through "dma-ranges"
property was needed for other masters than the M4 CPU.
stm32f429-disco doesn't use remapping so doesn't need this DMA
translation.
This patches moves this DMA translation definition from stm32f429 soc
file
to board files.
Signed-off-by: Alexandre TORGUE <alexandre.torgue@xxxxxx>
diff --git a/arch/arm/boot/dts/stm32429i-eval.dts
b/arch/arm/boot/dts/stm32429i-eval.dts
index 13c7cd2..a763c15 100644
--- a/arch/arm/boot/dts/stm32429i-eval.dts
+++ b/arch/arm/boot/dts/stm32429i-eval.dts
@@ -82,6 +82,10 @@
};
};
+ soc {
+ dma-ranges = <0xc0000000 0x0 0x10000000>;
+ };
+
usbotg_hs_phy: usbphy {
#phy-cells = <0>;
compatible = "usb-nop-xceiv";
Shouldn't also the peripheral dma-ranges property move to board specific
too?
I had this patch for while but I didn't had the time to submit:
Well spot I forgot it. Actually, discussing with Arnd ysterday on IIRC,
empty dma-ranges is not needed. Can you test on your side by removing
dma-ranges in usb node please ?
Unfortunately will take a time for me to set up this environment on
the STM32F4-EVAL board.
And on the discovery boards we dont have this scenario. That was the
main reason I did not submit the patch right away.
My conclusion and I might be wrong but is based on the my tests with
SDIO device at STM32F469I-DISCO board.
I started this issue as discussion at ST Forum but Maxime gave me the hint.
https://my.st.com/public/STe2ecommunities/mcu/Lists/cortex_mx_stm32/Flat.aspx?RootFolder=https%3a%2f%2fmy%2est%2ecom%2fpublic%2fSTe2ecommunities%2fmcu%2fLists%2fcortex_mx_stm32%2fDMA2%20and%20SYSCFG_MEMRMP%20relationship&FolderCTID=0x01200200770978C69A1141439FE559EB459D7580009C4E14902C3CDE46A77F0FFD06506F5B¤tviews=44
I will push a v2 by removing empty dma-ranges if tests are ok in your side.
From my understating/conclusion is: when empty property(dma-ranges) is
the device node, the mapping will be taken in consideration when using
DMA otherwise the mapping is ignored.
And in the SDIO case it is needed for DEV->MEM(SDRAM) and
MEM(SDRAM)->DEV. If it is not the case for the devices in question so
I suppose it can work without the property.
For sure translation has to be done but I'm not sure that an empty
"dma-ranges" is needed in device node to activate it. For Ethernet empty
"dma-ranges" is not needed. I will try with usb.
alex
Thanks in advance
Alex
Author: Bruno Herrera <bruherrera@xxxxxxxxx>
Date: Sun Oct 16 14:50:00 2016 -0200
ARM: DT: STM32: Use dma-ranges property per board not at dtsi file
diff --git a/arch/arm/boot/dts/stm32429i-eval.dts
b/arch/arm/boot/dts/stm32429i-eval.dts
index 6bfc595..2a22a82 100644
--- a/arch/arm/boot/dts/stm32429i-eval.dts
+++ b/arch/arm/boot/dts/stm32429i-eval.dts
@@ -52,6 +52,10 @@
model = "STMicroelectronics STM32429i-EVAL board";
compatible = "st,stm32429i-eval", "st,stm32f429";
+ soc {
+ dma-ranges = <0xC0000000 0x0 0x10000000>;
+ };
+
chosen {
bootargs = "root=/dev/ram rdinit=/linuxrc";
stdout-path = "serial0:115200n8";
@@ -96,6 +100,7 @@
ðernet0 {
status = "okay";
+ dma-ranges;
pinctrl-0 = <ðernet0_mii>;
pinctrl-names = "default";
phy-mode = "mii-id";
@@ -116,6 +121,7 @@
};
&usbotg_hs {
+ dma-ranges;
dr_mode = "host";
phys = <&usbotg_hs_phy>;
phy-names = "usb2-phy";
diff --git a/arch/arm/boot/dts/stm32f429.dtsi
b/arch/arm/boot/dts/stm32f429.dtsi
index 7d624a2..697a133 100644
--- a/arch/arm/boot/dts/stm32f429.dtsi
+++ b/arch/arm/boot/dts/stm32f429.dtsi
@@ -59,7 +59,6 @@
};
soc {
- dma-ranges = <0xc0000000 0x0 0x10000000>;
timer2: timer@40000000 {
compatible = "st,stm32-timer";
@@ -472,13 +471,11 @@
st,syscon = <&syscfg 0x4>;
snps,pbl = <8>;
snps,mixed-burst;
- dma-ranges;
status = "disabled";
};
usbotg_hs: usb@40040000 {
compatible = "snps,dwc2";
- dma-ranges;
reg = <0x40040000 0x40000>;
interrupts = <77>;
clocks = <&rcc 0 29>;
diff --git a/arch/arm/boot/dts/stm32f429.dtsi
b/arch/arm/boot/dts/stm32f429.dtsi
index 0596d60..3a1cfdd 100644
--- a/arch/arm/boot/dts/stm32f429.dtsi
+++ b/arch/arm/boot/dts/stm32f429.dtsi
@@ -59,8 +59,6 @@
};
soc {
- dma-ranges = <0xc0000000 0x0 0x10000000>;
-
timer2: timer@40000000 {
compatible = "st,stm32-timer";
reg = <0x40000000 0x400>;
diff --git a/arch/arm/boot/dts/stm32f469-disco.dts
b/arch/arm/boot/dts/stm32f469-disco.dts
index 9e73656..c2213c0 100644
--- a/arch/arm/boot/dts/stm32f469-disco.dts
+++ b/arch/arm/boot/dts/stm32f469-disco.dts
@@ -64,6 +64,10 @@
aliases {
serial0 = &usart3;
};
+
+ soc {
+ dma-ranges = <0xc0000000 0x0 0x10000000>;
+ };
};
&clk_hse {
--
Br.,
Bruno
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
