Skip to contentSkip to conversation

arm/dt: tegra: Remove /memreserve/ from device-tree files

PATCH New Stephen Warren 4 participants 17 messages Nov 21, 2011 – Dec 8, 2011

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/5 arm/dt: tegra: Remove /memreserve/ from device-tree files

There are no drivers in the kernel at present which can make use of the memory reserved by /memreserve/, so there is no point reserving it. Remove /memreserve/ to allow the user more memory. It's also unclear whether any future driver would actually require /memreserve/, or allocate memory through some other mechanism.

---
 arch/arm/boot/dts/tegra-harmony.dts   |    1 -
 arch/arm/boot/dts/tegra-paz00.dts     |    1 -
 arch/arm/boot/dts/tegra-seaboard.dts  |    1 -
 arch/arm/boot/dts/tegra-trimslice.dts |    1 -
 arch/arm/boot/dts/tegra-ventana.dts   |    1 -
 5 files changed, 0 insertions(+), 5 deletions(-)
arch/arm/boot/dts/tegra-harmony.dts +0 -1
index 0e225b8..a802324 100644
@@ -1,6 +1,5 @@
 /dts-v1/;
 
-/memreserve/ 0x1c000000 0x04000000;
 /include/ "tegra20.dtsi"
 
 / {
arch/arm/boot/dts/tegra-paz00.dts +0 -1
index 15a949f..1cd0cfe 100644
@@ -1,6 +1,5 @@
 /dts-v1/;
 
-/memreserve/ 0x1c000000 0x04000000;
 /include/ "tegra20.dtsi"
 
 / {
arch/arm/boot/dts/tegra-seaboard.dts +0 -1
index 88c682a..60724e4 100644
@@ -1,6 +1,5 @@
 /dts-v1/;
 
-/memreserve/ 0x1c000000 0x04000000;
 /include/ "tegra20.dtsi"
 
 / {
arch/arm/boot/dts/tegra-trimslice.dts +0 -1
index d3d0d20..95117d3 100644
@@ -1,6 +1,5 @@
 /dts-v1/;
 
-/memreserve/ 0x1c000000 0x04000000;
 /include/ "tegra20.dtsi"
 
 / {
arch/arm/boot/dts/tegra-ventana.dts +0 -1
index 3f9abd6..2af12fa 100644
@@ -1,6 +1,5 @@
 /dts-v1/;
 
-/memreserve/ 0x1c000000 0x04000000;
 /include/ "tegra20.dtsi"
 
 / {
Signature · 1 line
1.7.0.4
#1 lore raw

PATCH 2/5 arm/dt: tegra: Remove /chosen node

The command-lines present in the existing /chosen node are not necessarily correct for all users. Ideally, we should simply use the command-line supplied by the boot-loader.

In fact, using the boot-loader's cmdline is quite easy; either the bootloader fully supports DT, in which case it can modify the DT passed to the kernel to include its command-line, or CONFIG_APPENDED_DTB can be used in conjunction with CONFIG_ARM_ATAG_DTB_COMPAT, and the kernel will substitute the bootloader's command-line into the DT.

---
 arch/arm/boot/dts/tegra-harmony.dts   |    4 ----
 arch/arm/boot/dts/tegra-paz00.dts     |    4 ----
 arch/arm/boot/dts/tegra-seaboard.dts  |    4 ----
 arch/arm/boot/dts/tegra-trimslice.dts |    4 ----
 arch/arm/boot/dts/tegra-ventana.dts   |    4 ----
 5 files changed, 0 insertions(+), 20 deletions(-)
arch/arm/boot/dts/tegra-harmony.dts +0 -4
index a802324..ca6fa10 100644
@@ -6,10 +6,6 @@
 	model = "NVIDIA Tegra2 Harmony evaluation board";
 	compatible = "nvidia,harmony", "nvidia,tegra20";
 
-	chosen {
-		bootargs = "vmalloc=192M video=tegrafb console=ttyS0,115200n8 root=/dev/mmcblk0p2 rw rootwait";
-	};
-
 	memory at 0 {
 		reg = < 0x00000000 0x40000000 >;
 	};
arch/arm/boot/dts/tegra-paz00.dts +0 -4
index 1cd0cfe..4f6a8aa 100644
@@ -6,10 +6,6 @@
 	model = "Toshiba AC100 / Dynabook AZ";
 	compatible = "compal,paz00", "nvidia,tegra20";
 
-	chosen {
-		bootargs = "console=ttyS0,115200n8 root=/dev/mmcblk0p1";
-	};
-
 	memory at 0 {
 		reg = <0x00000000 0x20000000>;
 	};
arch/arm/boot/dts/tegra-seaboard.dts +0 -4
index 60724e4..d1ada26 100644
@@ -6,10 +6,6 @@
 	model = "NVIDIA Seaboard";
 	compatible = "nvidia,seaboard", "nvidia,tegra20";
 
-	chosen {
-		bootargs = "vmalloc=192M video=tegrafb console=ttyS0,115200n8 root=/dev/mmcblk1p3 rw rootwait";
-	};
-
 	memory {
 		device_type = "memory";
 		reg = < 0x00000000 0x40000000 >;
arch/arm/boot/dts/tegra-trimslice.dts +0 -4
index 95117d3..cf8e561 100644
@@ -6,10 +6,6 @@
 	model = "Compulab TrimSlice board";
 	compatible = "compulab,trimslice", "nvidia,tegra20";
 
-	chosen {
-		bootargs = "vmalloc=192M video=tegrafb console=ttyS0,115200n8 root=/dev/mmcblk0p1 rw rootwait";
-	};
-
 	memory at 0 {
 		reg = < 0x00000000 0x40000000 >;
 	};
arch/arm/boot/dts/tegra-ventana.dts +0 -4
index 2af12fa..1c09485 100644
@@ -6,10 +6,6 @@
 	model = "NVIDIA Tegra2 Ventana evaluation board";
 	compatible = "nvidia,ventana", "nvidia,tegra20";
 
-	chosen {
-		bootargs = "vmalloc=192M video=tegrafb console=ttyS0,115200n8 root=/dev/ram rdinit=/sbin/init";
-	};
-
 	memory {
 		reg = < 0x00000000 0x40000000 >;
 	};
Signature · 1 line
1.7.0.4
#2 lore raw

PATCH 2/5 arm/dt: tegra: Remove /chosen node

Quoted 11 lines · Stephen Warren

The command-lines present in the existing /chosen node are not necessarily correct for all users. Ideally, we should simply use the command-line supplied by the boot-loader.

In fact, using the boot-loader's cmdline is quite easy; either the bootloader fully supports DT, in which case it can modify the DT passed to the kernel to include its command-line, or CONFIG_APPENDED_DTB can be used in conjunction with CONFIG_ARM_ATAG_DTB_COMPAT, and the kernel will substitute the bootloader's command-line into the DT.

Thanks, applied to for-3.3/boards

-Olof

#3 lore raw

PATCH 3/5 arm/dt: tegra: Fix I2C nodes to match board files

In reply to Stephen Warren

With board files, all I2C busses run at 400KHz. Fix the device-tree to be consistent with this. It's possible this is incorrect, but at least it keeps the board files and device-tree consistent.

Also, disable any I2C controllers that the board files don't register, also for consistency.

---
 arch/arm/boot/dts/tegra-seaboard.dts  |   16 ++++++++++++++++
 arch/arm/boot/dts/tegra-trimslice.dts |    4 ++++
 arch/arm/boot/dts/tegra-ventana.dts   |   16 ++++++++++++++++
 3 files changed, 36 insertions(+), 0 deletions(-)
arch/arm/boot/dts/tegra-seaboard.dts +16 -0
index d1ada26..92da1f1 100644
@@ -11,6 +11,22 @@
 		reg = < 0x00000000 0x40000000 >;
 	};
 
+	i2c at 7000c000 {
+		clock-frequency = <400000>;
+	};
+
+	i2c at 7000c400 {
+		clock-frequency = <400000>;
+	};
+
+	i2c at 7000c500 {
+		clock-frequency = <400000>;
+	};
+
+	i2c at 7000d000 {
+		clock-frequency = <400000>;
+	};
+
 	serial at 70006300 {
 		clock-frequency = < 216000000 >;
 	};
arch/arm/boot/dts/tegra-trimslice.dts +4 -0
index cf8e561..b1aaa45 100644
@@ -22,6 +22,10 @@
 		clock-frequency = <400000>;
 	};
 
+	i2c at 7000d000 {
+		status = "disable";
+	};
+
 	serial at 70006000 {
 		clock-frequency = < 216000000 >;
 	};
arch/arm/boot/dts/tegra-ventana.dts +16 -0
index 1c09485..0f14ee3 100644
@@ -10,6 +10,22 @@
 		reg = < 0x00000000 0x40000000 >;
 	};
 
+	i2c at 7000c000 {
+		clock-frequency = <400000>;
+	};
+
+	i2c at 7000c400 {
+		clock-frequency = <400000>;
+	};
+
+	i2c at 7000c500 {
+		clock-frequency = <400000>;
+	};
+
+	i2c at 7000d000 {
+		clock-frequency = <400000>;
+	};
+
 	serial at 70006300 {
 		clock-frequency = < 216000000 >;
 	};
Signature · 1 line
1.7.0.4
#4 lore raw

PATCH 3/5 arm/dt: tegra: Fix I2C nodes to match board files

Quoted 8 lines · Stephen Warren

With board files, all I2C busses run at 400KHz. Fix the device-tree to be consistent with this. It's possible this is incorrect, but at least it keeps the board files and device-tree consistent.

Also, disable any I2C controllers that the board files don't register, also for consistency.

Thanks, applied to for-3.3/boards

-Olof

#5 lore raw

PATCH 4/5 arm/dt: tegra: Fix serial nodes to match board files

In reply to Stephen Warren

Mark any serial ports that aren't registered by the board files as disabled in the device-tree files.

In practice, none of the now-disabled ports ended up succeeding device probing because of the missing clock-frequency property. However, explicitly marking the devices disabled has the advantage of squashing the dev_warn() the failed probe causes, and documenting that we intend the port not to be used, rather than accidentally left out the property.

---
 arch/arm/boot/dts/tegra-harmony.dts   |   16 ++++++++++++++++
 arch/arm/boot/dts/tegra-seaboard.dts  |   16 ++++++++++++++++
 arch/arm/boot/dts/tegra-trimslice.dts |   16 ++++++++++++++++
 arch/arm/boot/dts/tegra-ventana.dts   |   16 ++++++++++++++++
 4 files changed, 64 insertions(+), 0 deletions(-)
arch/arm/boot/dts/tegra-harmony.dts +16 -0
index ca6fa10..ac179a0 100644
@@ -47,10 +47,26 @@
 		ext-mic-en-gpios = <&gpio 185 0>;
 	};
 
+	serial at 70006000 {
+		status = "disable";
+	};
+
+	serial at 70006040 {
+		status = "disable";
+	};
+
+	serial at 70006200 {
+		status = "disable";
+	};
+
 	serial at 70006300 {
 		clock-frequency = < 216000000 >;
 	};
 
+	serial at 70006400 {
+		status = "disable";
+	};
+
 	sdhci at c8000200 {
 		cd-gpios = <&gpio 69 0>; /* gpio PI5 */
 		wp-gpios = <&gpio 57 0>; /* gpio PH1 */
arch/arm/boot/dts/tegra-seaboard.dts +16 -0
index 92da1f1..cbb2f08 100644
@@ -27,10 +27,26 @@
 		clock-frequency = <400000>;
 	};
 
+	serial at 70006000 {
+		status = "disable";
+	};
+
+	serial at 70006040 {
+		status = "disable";
+	};
+
+	serial at 70006200 {
+		status = "disable";
+	};
+
 	serial at 70006300 {
 		clock-frequency = < 216000000 >;
 	};
 
+	serial at 70006400 {
+		status = "disable";
+	};
+
 	sdhci at c8000400 {
 		cd-gpios = <&gpio 69 0>; /* gpio PI5 */
 		wp-gpios = <&gpio 57 0>; /* gpio PH1 */
arch/arm/boot/dts/tegra-trimslice.dts +16 -0
index b1aaa45..6060ce0 100644
@@ -30,6 +30,22 @@
 		clock-frequency = < 216000000 >;
 	};
 
+	serial at 70006040 {
+		status = "disable";
+	};
+
+	serial at 70006200 {
+		status = "disable";
+	};
+
+	serial at 70006300 {
+		status = "disable";
+	};
+
+	serial at 70006400 {
+		status = "disable";
+	};
+
 	sdhci at c8000600 {
 		cd-gpios = <&gpio 121 0>;
 		wp-gpios = <&gpio 122 0>;
arch/arm/boot/dts/tegra-ventana.dts +16 -0
index 0f14ee3..5ac0aa3 100644
@@ -26,10 +26,26 @@
 		clock-frequency = <400000>;
 	};
 
+	serial at 70006000 {
+		status = "disable";
+	};
+
+	serial at 70006040 {
+		status = "disable";
+	};
+
+	serial at 70006200 {
+		status = "disable";
+	};
+
 	serial at 70006300 {
 		clock-frequency = < 216000000 >;
 	};
 
+	serial at 70006400 {
+		status = "disable";
+	};
+
 	sdhci at c8000400 {
 		cd-gpios = <&gpio 69 0>; /* gpio PI5 */
 		wp-gpios = <&gpio 57 0>; /* gpio PH1 */
Signature · 1 line
1.7.0.4
#6 lore raw

PATCH 4/5 arm/dt: tegra: Fix serial nodes to match board files

Quoted 10 lines · Stephen Warren

Mark any serial ports that aren't registered by the board files as disabled in the device-tree files.

In practice, none of the now-disabled ports ended up succeeding device probing because of the missing clock-frequency property. However, explicitly marking the devices disabled has the advantage of squashing the dev_warn() the failed probe causes, and documenting that we intend the port not to be used, rather than accidentally left out the property.

Thanks, applied to for-3.3/boards

-Olof

#7 lore raw

PATCH 5/5 arm/dt: tegra: Fix SDHCI nodes to match board files

In reply to Stephen Warren

Mark any SDHCI controllers that aren't registered by the board files as disabled in the device-tree files.

In practice, these controllers:

* Have nothing hooked up to them at all, or
* For ports intended for SDIO usage, the drivers for anything that might be attached are not in the device-tree yet. If/when drivers appear, the SD/MMC port can be re-enabled.

The only possible exception is TrimSlice's mico SD slot, but that wasn't enabled in the board files before anyway, and doesn't work when all the SDHCI controllers are enabled anyway.

---
 arch/arm/boot/dts/tegra-harmony.dts   |    8 ++++++++
 arch/arm/boot/dts/tegra-seaboard.dts  |    8 ++++++++
 arch/arm/boot/dts/tegra-trimslice.dts |   12 ++++++++++++
 arch/arm/boot/dts/tegra-ventana.dts   |    8 ++++++++
 4 files changed, 36 insertions(+), 0 deletions(-)
arch/arm/boot/dts/tegra-harmony.dts +8 -0
index ac179a0..80afa1b 100644
@@ -67,12 +67,20 @@
 		status = "disable";
 	};
 
+	sdhci at c8000000 {
+		status = "disable";
+	};
+
 	sdhci at c8000200 {
 		cd-gpios = <&gpio 69 0>; /* gpio PI5 */
 		wp-gpios = <&gpio 57 0>; /* gpio PH1 */
 		power-gpios = <&gpio 155 0>; /* gpio PT3 */
 	};
 
+	sdhci at c8000400 {
+		status = "disable";
+	};
+
 	sdhci at c8000600 {
 		cd-gpios = <&gpio 58 0>; /* gpio PH2 */
 		wp-gpios = <&gpio 59 0>; /* gpio PH3 */
arch/arm/boot/dts/tegra-seaboard.dts +8 -0
index cbb2f08..f552bcc 100644
@@ -47,6 +47,14 @@
 		status = "disable";
 	};
 
+	sdhci at c8000000 {
+		status = "disable";
+	};
+
+	sdhci at c8000200 {
+		status = "disable";
+	};
+
 	sdhci at c8000400 {
 		cd-gpios = <&gpio 69 0>; /* gpio PI5 */
 		wp-gpios = <&gpio 57 0>; /* gpio PH1 */
arch/arm/boot/dts/tegra-trimslice.dts +12 -0
index 6060ce0..3b3ee7d 100644
@@ -46,6 +46,18 @@
 		status = "disable";
 	};
 
+	sdhci at c8000000 {
+		status = "disable";
+	};
+
+	sdhci at c8000200 {
+		status = "disable";
+	};
+
+	sdhci at c8000400 {
+		status = "disable";
+	};
+
 	sdhci at c8000600 {
 		cd-gpios = <&gpio 121 0>;
 		wp-gpios = <&gpio 122 0>;
arch/arm/boot/dts/tegra-ventana.dts +8 -0
index 5ac0aa3..c7d3b87 100644
@@ -46,6 +46,14 @@
 		status = "disable";
 	};
 
+	sdhci at c8000000 {
+		status = "disable";
+	};
+
+	sdhci at c8000200 {
+		status = "disable";
+	};
+
 	sdhci at c8000400 {
 		cd-gpios = <&gpio 69 0>; /* gpio PI5 */
 		wp-gpios = <&gpio 57 0>; /* gpio PH1 */
Signature · 1 line
1.7.0.4
#8 lore raw

PATCH 5/5 arm/dt: tegra: Fix SDHCI nodes to match board files

Quoted 5 lines · Stephen Warren

Mark any SDHCI controllers that aren't registered by the board files as disabled in the device-tree files.

In practice, these controllers:

Quoted 8 lines

* Have nothing hooked up to them at all, or
* For ports intended for SDIO usage, the drivers for anything that might be attached are not in the device-tree yet. If/when drivers appear, the SD/MMC port can be re-enabled.

The only possible exception is TrimSlice's mico SD slot, but that wasn't enabled in the board files before anyway, and doesn't work when all the SDHCI controllers are enabled anyway.

it will be better to invert the logic

just enable what you are supposed too

Best Regards,
J.

#9 lore raw

PATCH 5/5 arm/dt: tegra: Fix SDHCI nodes to match board files

Jean-Christophe PLAGNIOL-VILLARD wrote at Tuesday, November 22, 2011 1:38 AM:

Quoted 11 lines · Stephen Warren
Quoted 10 lines

Mark any SDHCI controllers that aren't registered by the board files as disabled in the device-tree files.

In practice, these controllers:

* Have nothing hooked up to them at all, or
* For ports intended for SDIO usage, the drivers for anything that might be attached are not in the device-tree yet. If/when drivers appear, the SD/MMC port can be re-enabled.

Quoted 6 lines
Quoted 3 lines

The only possible exception is TrimSlice's mico SD slot, but that wasn't enabled in the board files before anyway, and doesn't work when all the SDHCI controllers are enabled anyway.

it will be better to invert the logic

just enable what you are supposed too

It used to be that way when the files were first added, but they got switched around this way to match standard practice.

Signature · 1 line
nvpublic
#10 lore raw

PATCH 5/5 arm/dt: tegra: Fix SDHCI nodes to match board files

Quoted 15 lines · Stephen Warren

Jean-Christophe PLAGNIOL-VILLARD wrote at Tuesday, November 22, 2011 1:38 AM:

Quoted 14 lines · Stephen Warren
Quoted 13 lines

Mark any SDHCI controllers that aren't registered by the board files as disabled in the device-tree files.

In practice, these controllers:

* Have nothing hooked up to them at all, or
* For ports intended for SDIO usage, the drivers for anything that might be attached are not in the device-tree yet. If/when drivers appear, the SD/MMC port can be re-enabled.

The only possible exception is TrimSlice's mico SD slot, but that wasn't enabled in the board files before anyway, and doesn't work when all the SDHCI controllers are enabled anyway.

Quoted 6 lines
Quoted 3 lines

it will be better to invert the logic

just enable what you are supposed too

It used to be that way when the files were first added, but they got switched around this way to match standard practice.

this is a bad practice because you can enable stuff that a hardware is not supposed to use

and this silenty without anyone notice until you test the hardware

and you add more line and changeset

Best Regrds,
J.

#11 lore raw

PATCH 5/5 arm/dt: tegra: Fix SDHCI nodes to match board files

In reply to Stephen Warren
Quoted 10 lines · Stephen Warren

Mark any SDHCI controllers that aren't registered by the board files as disabled in the device-tree files.

In practice, these controllers:

* Have nothing hooked up to them at all, or
* For ports intended for SDIO usage, the drivers for anything that might be attached are not in the device-tree yet. If/when drivers appear, the SD/MMC port can be re-enabled.

Quoted 5 lines

The only possible exception is TrimSlice's mico SD slot, but that wasn't enabled in the board files before anyway, and doesn't work when all the SDHCI controllers are enabled anyway.

Thanks, applied to for-3.3/boards

-Olof

#12 lore raw

PATCH 1/5 arm/dt: tegra: Remove /memreserve/ from device-tree files

In reply to Stephen Warren

Hi Stephen,

Quoted 5 lines · On Monday 21 November 2011 14:44:07 Stephen Warren wrote:

There are no drivers in the kernel at present which can make use of the memory reserved by /memreserve/, so there is no point reserving it. Remove /memreserve/ to allow the user more memory. It's also unclear whether any future driver would actually require /memreserve/, or allocate memory through some other mechanism.

while this gives 512 MB of usable memory on my AC100, it causes some funny pictures on the screen ;-) I guess our bootloader (fastboot) already setups the frambuffer there (and enables the backlight). I guess it shouldn't be harmful as nothing else writes to this memory area. Otherwise you may add that the bootloader (or the kernel cmdline supplied by the bootloader) is now responsible for the reservation to the description.

Marc

Quoted 64 lines
---
 arch/arm/boot/dts/tegra-harmony.dts   |    1 -
 arch/arm/boot/dts/tegra-paz00.dts     |    1 -
 arch/arm/boot/dts/tegra-seaboard.dts  |    1 -
 arch/arm/boot/dts/tegra-trimslice.dts |    1 -
 arch/arm/boot/dts/tegra-ventana.dts   |    1 -
 5 files changed, 0 insertions(+), 5 deletions(-)
(quoted diff) +0 -0

b/arch/arm/boot/dts/tegra-harmony.dts index 0e225b8..a802324 100644

arch/arm/boot/dts/tegra-harmony.dts +0 -1
@@ -1,6 +1,5 @@
 /dts-v1/;
 
-/memreserve/ 0x1c000000 0x04000000;
 /include/ "tegra20.dtsi"
 
 / {
(quoted diff) +0 -0

b/arch/arm/boot/dts/tegra-paz00.dts index 15a949f..1cd0cfe 100644

arch/arm/boot/dts/tegra-paz00.dts +0 -1
@@ -1,6 +1,5 @@
 /dts-v1/;
 
-/memreserve/ 0x1c000000 0x04000000;
 /include/ "tegra20.dtsi"
 
 / {
(quoted diff) +0 -0

b/arch/arm/boot/dts/tegra-seaboard.dts index 88c682a..60724e4 100644

arch/arm/boot/dts/tegra-seaboard.dts +0 -1
@@ -1,6 +1,5 @@
 /dts-v1/;
 
-/memreserve/ 0x1c000000 0x04000000;
 /include/ "tegra20.dtsi"
 
 / {
(quoted diff) +0 -0

b/arch/arm/boot/dts/tegra-trimslice.dts index d3d0d20..95117d3 100644

arch/arm/boot/dts/tegra-trimslice.dts +0 -1
@@ -1,6 +1,5 @@
 /dts-v1/;
 
-/memreserve/ 0x1c000000 0x04000000;
 /include/ "tegra20.dtsi"
 
 / {
(quoted diff) +0 -0

b/arch/arm/boot/dts/tegra-ventana.dts index 3f9abd6..2af12fa 100644

arch/arm/boot/dts/tegra-ventana.dts +0 -1
@@ -1,6 +1,5 @@
 /dts-v1/;
 
-/memreserve/ 0x1c000000 0x04000000;
 /include/ "tegra20.dtsi"
 
 / {
#13 lore raw

PATCH 1/5 arm/dt: tegra: Remove /memreserve/ from device-tree files

Marc Dietrich wrote at Tuesday, November 22, 2011 3:42 AM:

Quoted 7 lines · On Monday 21 November 2011 14:44:07 Stephen Warren wrote:
Quoted 5 lines

There are no drivers in the kernel at present which can make use of the memory reserved by /memreserve/, so there is no point reserving it. Remove /memreserve/ to allow the user more memory. It's also unclear whether any future driver would actually require /memreserve/, or allocate memory through some other mechanism.

Quoted 6 lines

while this gives 512 MB of usable memory on my AC100, it causes some funny pictures on the screen ;-) I guess our bootloader (fastboot) already setups the frambuffer there (and enables the backlight). I guess it shouldn't be harmful as nothing else writes to this memory area. Otherwise you may add that the bootloader (or the kernel cmdline supplied by the bootloader) is now responsible for the reservation to the description.

Ah yes, I guess that would happen. I'm only testing with mainline U-Boot...

I'd suggest this is acceptable fallout for now; I don't believe it will cause any active problems, and as you say, you can get the bootloader to set a command-line that excludes the frame-buffer if you want to solve this. Do you agree?

Signature · 1 line
nvpublic
#14 lore raw

PATCH 1/5 arm/dt: tegra: Remove /memreserve/ from device-treefiles

Quoted 17 lines · On Tuesday 22 November 2011 08:25:03 Stephen Warren wrote:

Marc Dietrich wrote at Tuesday, November 22, 2011 3:42 AM:

Quoted 8 lines · On Monday 21 November 2011 14:44:07 Stephen Warren wrote:
Quoted 6 lines

There are no drivers in the kernel at present which can make use of the
memory reserved by /memreserve/, so there is no point reserving it. Remove /memreserve/ to allow the user more memory. It's also unclear whether any future driver would actually require /memreserve/, or allocate memory through some other mechanism.

Quoted 7 lines

while this gives 512 MB of usable memory on my AC100, it causes some funny pictures on the screen ;-) I guess our bootloader (fastboot) already setups the frambuffer there (and enables the backlight). I guess it shouldn't be harmful as nothing else writes to this memory area. Otherwise you may add that the bootloader (or the kernel cmdline supplied by the bootloader) is now responsible for the reservation to the description.

Quoted 6 lines

Ah yes, I guess that would happen. I'm only testing with mainline U-Boot...

I'd suggest this is acceptable fallout for now; I don't believe it will cause any active problems, and as you say, you can get the bootloader to set a command-line that excludes the frame-buffer if you want to solve this. Do you agree?

Ok. Hopefully we can also use u-Boot soon.

Marc

#15 lore raw

PATCH 1/5 arm/dt: tegra: Remove /memreserve/ from device-tree files

In reply to Stephen Warren
Quoted 15 lines · Stephen Warren

Marc Dietrich wrote at Tuesday, November 22, 2011 3:42 AM:

Quoted 7 lines · On Monday 21 November 2011 14:44:07 Stephen Warren wrote:
Quoted 5 lines

There are no drivers in the kernel at present which can make use of the memory reserved by /memreserve/, so there is no point reserving it. Remove /memreserve/ to allow the user more memory. It's also unclear whether any future driver would actually require /memreserve/, or allocate memory through some other mechanism.

Quoted 6 lines

while this gives 512 MB of usable memory on my AC100, it causes some funny pictures on the screen ;-) I guess our bootloader (fastboot) already setups the frambuffer there (and enables the backlight). I guess it shouldn't be harmful as nothing else writes to this memory area. Otherwise you may add that the bootloader (or the kernel cmdline supplied by the bootloader) is now responsible for the reservation to the description.

Quoted 6 lines

Ah yes, I guess that would happen. I'm only testing with mainline U-Boot...

I'd suggest this is acceptable fallout for now; I don't believe it will cause any active problems, and as you say, you can get the bootloader to set a command-line that excludes the frame-buffer if you want to solve this. Do you agree?

Agreed. It would make more sense to teach u-boot to reserve the memory.

-Olof

#16 lore raw

PATCH 1/5 arm/dt: tegra: Remove /memreserve/ from device-tree files

In reply to Stephen Warren
Quoted 7 lines · Stephen Warren

There are no drivers in the kernel at present which can make use of the memory reserved by /memreserve/, so there is no point reserving it. Remove /memreserve/ to allow the user more memory. It's also unclear whether any future driver would actually require /memreserve/, or allocate memory through some other mechanism.

Thanks, applied to for-3.3/boards

-Olof