Skip to contentSkip to conversation

Add support for loading FPGA bitstream

PATCH Superseded tien.fong.chee at intel.com 4 participants 55 messages Feb 13, 2019 – Feb 15, 2019

Subsystems (4)

v8 of 13 Superseded Latest is v13 →

lore
Share & tools
lore
#0 lore raw

v8 cover Add support for loading FPGA bitstream

From: Tien Fong Chee <[email protected]>

This version mainly resolved comments from Marek and Dalon L Westergreen in [v7].

Additonal note:
---------------
There are a few solutions at this moment for solving the performance issue
1. Using absolute data position to allign the core bistream in fitImage.
2. SPL program periph bitstream, then using fpga loadmk for loading the core bitstream from DDR. This is work because there is no allignment performance issue when reading whole fitImage instead of offset to the fitImage.

This series is working on top of u-boot.git -
http://git.denx.de/u-boot.git .

[v7]: https://www.mail-archive.com/u-boot at lists.denx.de/msg314511.html

Tien Fong Chee (8):
ARM: socfpga: Description on FPGA bitstream type and file name for
Arria 10
ARM: socfpga: Add default FPGA bitstream fitImage for Arria10 SoCDK
fit: Add function declarations to the header file
ARM: socfpga: Add FPGA drivers for Arria 10 FPGA bitstream loading
ARM: socfpga: Add the configuration for FPGA SoCFPGA A10 SoCDK spl : socfpga: Implement fpga bitstream loading with socfpga loadfs
ARM: socfpga: Synchronize the configuration for A10 SoCDK
ARM: socfpga: Increase Malloc pool size to support FAT filesystem in
SPL

 arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts       |  17 +
 .../include/mach/fpga_manager_arria10.h            |  39 +-
 arch/arm/mach-socfpga/spl_a10.c                    |  41 +-
 board/altera/arria10-socdk/fit_spl_fpga.its        |  39 ++
 configs/socfpga_arria10_defconfig                  |  21 +-
 .../fpga/altera-socfpga-a10-fpga-mgr.txt           |  26 +-
 drivers/fpga/socfpga_arria10.c                     | 467 ++++++++++++++++++++-
 include/configs/socfpga_common.h                   |   4 +-
 include/image.h                                    |   4 +
 9 files changed, 626 insertions(+), 32 deletions(-)
 create mode 100644 board/altera/arria10-socdk/fit_spl_fpga.its
Signature · 1 line
2.2.0
#1 lore raw

PATCH v8 1/8 ARM: socfpga: Description on FPGA bitstream type and file name for Arria 10

From: Tien Fong Chee <[email protected]>

This patch adds description on properties about file name used for both peripheral bitstream and core bitstream.

---

changes for v8
- Removed explanation about support for altr,bitstream-core

changes for v7
- Provided example of setting FPGA FIT image for both early IO release and full release FPGA configuration.

---
 .../fpga/altera-socfpga-a10-fpga-mgr.txt           | 26 +++++++++++++++++++++-
 1 file changed, 25 insertions(+), 1 deletion(-)
doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +25 -1
index 2fd8e7a..da210bf 100644
@@ -7,8 +7,31 @@ Required properties:
                - The second index is for writing FPGA configuration data.
 - resets     : Phandle and reset specifier for the device's reset.
 - clocks     : Clocks used by the device.
+- altr,bitstream : Fit image file name for both FPGA peripheral bitstream,
+		   FPGA core bitstream and full bitstream.
 
-Example:
+		   Full bitstream, consist of peripheral bitstream and core
+		   bitstream.
+
+		   FPGA peripheral bitstream is used to initialize FPGA IOs,
+		   PLL, IO48 and DDR. This bitstream is required to get DDR up
+		   running.
+
+		   FPGA core bitstream contains FPGA design which is used to
+		   program FPGA CRAM and ERAM.
+
+Example: Bundles both peripheral bitstream and core bitstream into FIT image
+	 called fit_spl_fpga.itb. This FIT image can be created through running
+	 this command: tools/mkimage
+		       -E -p 400
+		       -f board/altera/arria10-socdk/fit_spl_fpga.its
+		       fit_spl_fpga.itb
+
+	 For details of describing structure and contents of the FIT image,
+	 please refer board/altera/arria10-socdk/fit_spl_fpga.its
+
+- Examples for booting with full release or booting with early IO release, then
+  follow by entering early user mode:
 
 	fpga_mgr: fpga-mgr at ffd03000 {
 		compatible = "altr,socfpga-a10-fpga-mgr";
@@ -16,4 +39,5 @@ Example:
 		       0xffcfe400 0x20>;
 		clocks = <&l4_mp_clk>;
 		resets = <&rst FPGAMGR_RESET>;
+		altr,bitstream = "fit_spl_fpga.itb";
 	};
Signature · 1 line
2.2.0
#2 lore raw

PATCH v8 1/8 ARM: socfpga: Description on FPGA bitstream type and file name for Arria 10

Quoted 24 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

This patch adds description on properties about file name used for both peripheral bitstream and core bitstream.

---

changes for v8
- Removed explanation about support for altr,bitstream-core

changes for v7
- Provided example of setting FPGA FIT image for both early IO release and full release FPGA configuration.

---
 .../fpga/altera-socfpga-a10-fpga-mgr.txt           | 26 +++++++++++++++++++++-
 1 file changed, 25 insertions(+), 1 deletion(-)
doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +0 -0
index 2fd8e7a..da210bf 100644

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt Required properties: · line 11

Quoted 5 lines
doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +2 -0
@@ -7,8 +7,31 @@ Required properties:
                - The second index is for writing FPGA configuration data.
 - resets     : Phandle and reset specifier for the device's reset.
 - clocks     : Clocks used by the device.
+- altr,bitstream : Fit image file name for both FPGA peripheral bitstream,
+		   FPGA core bitstream and full bitstream.

So the file contains three bitstreams ? I thought we can load only the core here.

Quoted 5 lines
doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +4 -1
-Example:
+		   Full bitstream, consist of peripheral bitstream and core
+		   bitstream.
+
+		   FPGA peripheral bitstream is used to initialize FPGA IOs,

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt Required properties: · line 26

Quoted 10 lines
doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +10 -0
+		   PLL, IO48 and DDR. This bitstream is required to get DDR up
+		   running.
+
+		   FPGA core bitstream contains FPGA design which is used to
+		   program FPGA CRAM and ERAM.
+
+Example: Bundles both peripheral bitstream and core bitstream into FIT image
+	 called fit_spl_fpga.itb. This FIT image can be created through running
+	 this command: tools/mkimage
+		       -E -p 400

Is the padding still required ?

Quoted 18 lines
doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +9 -0
+		       -f board/altera/arria10-socdk/fit_spl_fpga.its
+		       fit_spl_fpga.itb
+
+	 For details of describing structure and contents of the FIT image,
+	 please refer board/altera/arria10-socdk/fit_spl_fpga.its
+
+- Examples for booting with full release or booting with early IO release, then
+  follow by entering early user mode:
 
 	fpga_mgr: fpga-mgr at ffd03000 {
 		compatible = "altr,socfpga-a10-fpga-mgr";
@@ -16,4 +39,5 @@ Example:
 		       0xffcfe400 0x20>;
 		clocks = <&l4_mp_clk>;
 		resets = <&rst FPGAMGR_RESET>;
+		altr,bitstream = "fit_spl_fpga.itb";
 	};
Signature · 2 lines
Best regards,
Marek Vasut
#3 lore raw

PATCH v8 1/8 ARM: socfpga: Description on FPGA bitstream type and file name for Arria 10

Quoted 33 lines · Marek Vasut
Quoted 32 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

This patch adds description on properties about file name used for both
peripheral bitstream and core bitstream.

---

changes for v8
- Removed explanation about support for altr,bitstream-core

changes for v7
- Provided example of setting FPGA FIT image for both early IO release
  and full release FPGA configuration.
---
 .../fpga/altera-socfpga-a10-fpga-mgr.txt           | 26
+++++++++++++++++++++-

 1 file changed, 25 insertions(+), 1 deletion(-)
(quoted diff) +0 -0

mgr.txt b/doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga- mgr.txt
index 2fd8e7a..da210bf 100644

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +0 -0
@@ -7,8 +7,31 @@ Required properties:

                - The second index is for writing FPGA configuration data.

Quoted 8 lines

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt Required properties: · line 11

Quoted 5 lines

 - resets     : Phandle and reset specifier for the device's reset.
 - clocks     : Clocks used by the device.
+- altr,bitstream : Fit image file name for both FPGA peripheral bitstream,

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +1 -0
+		   FPGA core bitstream and full bitstream.

So the file contains three bitstreams ? I thought we can load only the
core here.

Here is for telling FPGA driver which fitImage file gonna be processed. You can put whatever bitstreams in fitImage. Then, in the default configuration, you can tell SPL FPGA driver program which bitstream, it could be both periph.rbf and core.rbf or just programming periph.rbf only.

Quoted 16 lines
Quoted 16 lines
doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +1 -1
-Example:
+		   Full bitstream, consist of peripheral bitstream

and core

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +3 -0
+		   bitstream.
+
+		   FPGA peripheral bitstream is used to initialize

FPGA IOs,

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +1 -0
+		   PLL, IO48 and DDR. This bitstream is required

to get DDR up

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +3 -0
+		   running.
+
+		   FPGA core bitstream contains FPGA design which

is used to

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +1 -0
+		   program FPGA CRAM and ERAM.
Quoted 8 lines

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt Required properties: · line 26

Quoted 7 lines
doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +2 -0
+
+Example: Bundles both peripheral bitstream and core bitstream into

FIT image

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +1 -0
+	 called fit_spl_fpga.itb. This FIT image can be created

through running

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +2 -0
+	 this command: tools/mkimage
+		       -E -p 400

Is the padding still required ?

Yes, i think that padding method should be sufficient for all use cases, i guess both NAND and QSPI may need this also.

You want me to support data offset(without padding) also?

Quoted 24 lines
Quoted 23 lines
doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +1 -0
+		       -f board/altera/arria10-

socdk/fit_spl_fpga.its

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +3 -0
+		       fit_spl_fpga.itb
+
+	 For details of describing structure and contents of the

FIT image,

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +3 -0
+	 please refer board/altera/arria10-socdk/fit_spl_fpga.its
+
+- Examples for booting with full release or booting with early IO

release, then
+  follow by entering early user mode:

  fpga_mgr: fpga-mgr at ffd03000 { compatible = "altr,socfpga-a10-fpga-mgr";

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +1 -0
@@ -16,4 +39,5 @@ Example:
  		       0xffcfe400 0x20>;
  		clocks = <&l4_mp_clk>;
  		resets = <&rst FPGAMGR_RESET>;
+		altr,bitstream = "fit_spl_fpga.itb";
  	};
#4 lore raw

PATCH v8 1/8 ARM: socfpga: Description on FPGA bitstream type and file name for Arria 10

Quoted 42 lines · Tien Fong
Quoted 33 lines · Marek Vasut
Quoted 32 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

This patch adds description on properties about file name used for both
peripheral bitstream and core bitstream.

---

changes for v8
- Removed explanation about support for altr,bitstream-core

changes for v7
- Provided example of setting FPGA FIT image for both early IO release
  and full release FPGA configuration.
---
 .../fpga/altera-socfpga-a10-fpga-mgr.txt           | 26
+++++++++++++++++++++-

 1 file changed, 25 insertions(+), 1 deletion(-)
(quoted diff) +0 -0

mgr.txt b/doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga- mgr.txt
index 2fd8e7a..da210bf 100644

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +0 -0
@@ -7,8 +7,31 @@ Required properties:

                - The second index is for writing FPGA configuration data.

Quoted 8 lines

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt Required properties: · line 11

Quoted 5 lines

 - resets     : Phandle and reset specifier for the device's reset.
 - clocks     : Clocks used by the device.
+- altr,bitstream : Fit image file name for both FPGA peripheral bitstream,

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +1 -0
+		   FPGA core bitstream and full bitstream.

So the file contains three bitstreams ? I thought we can load only the
core here.

Quoted 5 lines

Here is for telling FPGA driver which fitImage file gonna be processed. You can put whatever bitstreams in fitImage. Then, in the default configuration, you can tell SPL FPGA driver program which bitstream, it could be both periph.rbf and core.rbf or just programming periph.rbf only.

Ah, OK

Quoted 22 lines
Quoted 14 lines
Quoted 14 lines
doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +1 -1
-Example:
+		   Full bitstream, consist of peripheral bitstream

and core

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +3 -0
+		   bitstream.
+
+		   FPGA peripheral bitstream is used to initialize

FPGA IOs,

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +1 -0
+		   PLL, IO48 and DDR. This bitstream is required

to get DDR up

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +3 -0
+		   running.
+
+		   FPGA core bitstream contains FPGA design which

is used to

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +1 -0
+		   program FPGA CRAM and ERAM.
Quoted 8 lines

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt Required properties: · line 26

Quoted 7 lines
doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +2 -0
+
+Example: Bundles both peripheral bitstream and core bitstream into

FIT image

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +1 -0
+	 called fit_spl_fpga.itb. This FIT image can be created

through running

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +2 -0
+	 this command: tools/mkimage
+		       -E -p 400

Is the padding still required ?

Quoted 4 lines

Yes, i think that padding method should be sufficient for all use cases, i guess both NAND and QSPI may need this also.

You want me to support data offset(without padding) also?

I think you should drop the padding, it seems to be workaround ?

Signature · 2 lines
Best regards,
Marek Vasut
#5 lore raw

PATCH v8 1/8 ARM: socfpga: Description on FPGA bitstream type and file name for Arria 10

Quoted 95 lines · Marek Vasut
Quoted 53 lines · Tien Fong
Quoted 43 lines · Marek Vasut
Quoted 41 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

This patch adds description on properties about file name used for
both
peripheral bitstream and core bitstream.

---

changes for v8
- Removed explanation about support for altr,bitstream-core

changes for v7
- Provided example of setting FPGA FIT image for both early IO release
  and full release FPGA configuration.
---
 .../fpga/altera-socfpga-a10-fpga-mgr.txt           | 26
+++++++++++++++++++++-

 1 file changed, 25 insertions(+), 1 deletion(-)
(quoted diff) +0 -0

fpga-
mgr.txt b/doc/device-tree-bindings/fpga/altera-socfpga-a10- fpga-
mgr.txt
index 2fd8e7a..da210bf 100644
--- a/doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga- mgr.txt
+++ b/doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga- mgr.txt

(quoted diff) +0 -0
@@ -7,8 +7,31 @@ Required properties:

                - The second index is for writing FPGA configuration data.
 - resets     : Phandle and reset specifier for the device's reset.
 - clocks     : Clocks used by the device.

Quoted 8 lines

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt Required properties: · line 11

Quoted 4 lines

+- altr,bitstream : Fit image file name for both FPGA peripheral
bitstream,

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +1 -0
+		   FPGA core bitstream and full bitstream.

So the file contains three bitstreams ? I thought we can load only
the
core here.

Quoted 8 lines

Here is for telling FPGA driver which fitImage file gonna be processed.
You can put whatever bitstreams in fitImage. Then, in the default configuration, you can tell SPL FPGA driver program which bitstream, it
could be both periph.rbf and core.rbf or just programming periph.rbf
only.

Ah, OK

Quoted 31 lines
Quoted 31 lines
Quoted 28 lines
(quoted diff) +1 -1
-Example:
+		   Full bitstream, consist of peripheral

bitstream
and core

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +3 -0
+		   bitstream.
+
+		   FPGA peripheral bitstream is used to

initialize
FPGA IOs,

(quoted diff) +1 -0
+		   PLL, IO48 and DDR. This bitstream is

required
to get DDR up

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +3 -0
+		   running.
+
+		   FPGA core bitstream contains FPGA design

which
is used to

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +3 -0
+		   program FPGA CRAM and ERAM.
+
+Example: Bundles both peripheral bitstream and core bitstream

into
FIT image

(quoted diff) +1 -0
+	 called fit_spl_fpga.itb. This FIT image can be

created
through running

doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt Required properties: · line 26

Quoted 2 lines
doc/device-tree-bindings/fpga/altera-socfpga-a10-fpga-mgr.txt +2 -0
+	 this command: tools/mkimage
+		       -E -p 400

Is the padding still required ?

Quoted 5 lines
Quoted 4 lines

Yes, i think that padding method should be sufficient for all use cases, i guess both NAND and QSPI may need this also.

You want me to support data offset(without padding) also?

I think you should drop the padding, it seems to be workaround ?

I can add the data offset support, user is free to use either one of them.

Quoted 1 line
#6 lore raw

PATCH v8 2/8 ARM: socfpga: Add default FPGA bitstream fitImage for Arria10 SoCDK

In reply to tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add default fitImage file bundling FPGA bitstreams for Arria10.

---

changes for v8
- Changed the FPGA node name to fpga-core and fpga-periph for both core and periph bitstreams respectively.

---
 board/altera/arria10-socdk/fit_spl_fpga.its | 39 +++++++++++++++++++++++++++++
 1 file changed, 39 insertions(+)
 create mode 100644 board/altera/arria10-socdk/fit_spl_fpga.its
board/altera/arria10-socdk/fit_spl_fpga.its new file +39 -0
new file mode 100644
index 0000000..8ce175b
@@ -0,0 +1,39 @@
+// SPDX-License-Identifier: GPL-2.0
+ /*
+ * Copyright (C) 2019 Intel Corporation <www.intel.com>
+ *
+ */
+
+/dts-v1/;
+
+/ {
+	description = "FIT image with FPGA bistream";
+	#address-cells = <1>;
+
+	images {
+		fpga-core at 1 {
+			description = "FPGA core bitstream";
+			data = /incbin/("../../../ghrd_10as066n2.core.rbf");
+			type = "fpga";
+			arch = "arm";
+			compression = "none";
+			load = <0x400>;
+		};
+
+		fpga-periph at 2 {
+			description = "FPGA peripheral bitstream";
+			data = /incbin/("../../../ghrd_10as066n2.periph.rbf");
+			type = "fpga";
+			arch = "arm";
+			compression = "none";
+		};
+	};
+
+	configurations {
+		default = "config-1";
+		config-1 {
+			description = "Boot with FPGA early IO release config";
+			fpga = "fpga-periph at 2", "fpga-core at 1";
+		};
+	};
+};
Signature · 1 line
2.2.0
#7 lore raw

PATCH v8 2/8 ARM: socfpga: Add default FPGA bitstream fitImage for Arria10 SoCDK

Quoted 32 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add default fitImage file bundling FPGA bitstreams for Arria10.

---

changes for v8
- Changed the FPGA node name to fpga-core and fpga-periph for both core and periph bitstreams respectively.

---
 board/altera/arria10-socdk/fit_spl_fpga.its | 39 +++++++++++++++++++++++++++++
 1 file changed, 39 insertions(+)
 create mode 100644 board/altera/arria10-socdk/fit_spl_fpga.its
board/altera/arria10-socdk/fit_spl_fpga.its new file +10 -0
new file mode 100644
index 0000000..8ce175b
@@ -0,0 +1,39 @@
+// SPDX-License-Identifier: GPL-2.0
+ /*
+ * Copyright (C) 2019 Intel Corporation <www.intel.com>
+ *
+ */
+
+/dts-v1/;
+
+/ {
+	description = "FIT image with FPGA bistream";

board/altera/arria10-socdk/fit_spl_fpga.its line 20

Quoted 10 lines
board/altera/arria10-socdk/fit_spl_fpga.its new file +10 -0
+	#address-cells = <1>;
+
+	images {
+		fpga-core at 1 {
+			description = "FPGA core bitstream";
+			data = /incbin/("../../../ghrd_10as066n2.core.rbf");
+			type = "fpga";
+			arch = "arm";
+			compression = "none";
+			load = <0x400>;

Is the load address required ?

Quoted 6 lines
board/altera/arria10-socdk/fit_spl_fpga.its new file +6 -0
+		};
+
+		fpga-periph at 2 {
+			description = "FPGA peripheral bitstream";
+			data = /incbin/("../../../ghrd_10as066n2.periph.rbf");
+			type = "fpga";

board/altera/arria10-socdk/fit_spl_fpga.its line 36

Quoted 10 lines
board/altera/arria10-socdk/fit_spl_fpga.its new file +10 -0
+			arch = "arm";
+			compression = "none";
+		};
+	};
+
+	configurations {
+		default = "config-1";
+		config-1 {
+			description = "Boot with FPGA early IO release config";
+			fpga = "fpga-periph at 2", "fpga-core at 1";

Don't you need to load the core first ?

Quoted 4 lines
board/altera/arria10-socdk/fit_spl_fpga.its new file +3 -0
+		};
+	};
+};
Signature · 2 lines
Best regards,
Marek Vasut
#8 lore raw

PATCH v8 2/8 ARM: socfpga: Add default FPGA bitstream fitImage for Arria10 SoCDK

Quoted 58 lines · Marek Vasut
Quoted 34 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add default fitImage file bundling FPGA bitstreams for Arria10.

---

changes for v8
- Changed the FPGA node name to fpga-core and fpga-periph for both core and periph bitstreams respectively.

---
 board/altera/arria10-socdk/fit_spl_fpga.its | 39
+++++++++++++++++++++++++++++
 1 file changed, 39 insertions(+)
 create mode 100644 board/altera/arria10-socdk/fit_spl_fpga.its
board/altera/arria10-socdk/fit_spl_fpga.its new file +10 -0
new file mode 100644
index 0000000..8ce175b
@@ -0,0 +1,39 @@
+// SPDX-License-Identifier: GPL-2.0
+ /*
+ * Copyright (C) 2019 Intel Corporation <www.intel.com>
+ *
+ */
+
+/dts-v1/;
+
+/ {
+	description = "FIT image with FPGA bistream";

board/altera/arria10-socdk/fit_spl_fpga.its line 20

Quoted 10 lines
board/altera/arria10-socdk/fit_spl_fpga.its new file +10 -0
+	#address-cells = <1>;
+
+	images {
+		fpga-core at 1 {
+			description = "FPGA core bitstream";
+			data = /incbin/("../../../ghrd_10as066n2.core.rbf");
+			type = "fpga";
+			arch = "arm";
+			compression = "none";
+			load = <0x400>;

Is the load address required ?

Quoted 10 lines
board/altera/arria10-socdk/fit_spl_fpga.its new file +10 -0
+		};
+
+		fpga-periph at 2 {
+			description = "FPGA peripheral bitstream";
+			data = /incbin/("../../../ghrd_10as066n2.periph.rbf");
+			type = "fpga";
+			arch = "arm";
+			compression = "none";
+		};
+	};
Quoted 8 lines

board/altera/arria10-socdk/fit_spl_fpga.its line 36

Quoted 6 lines
board/altera/arria10-socdk/fit_spl_fpga.its new file +6 -0
+
+	configurations {
+		default = "config-1";
+		config-1 {
+			description = "Boot with FPGA early IO release config";
+			fpga = "fpga-periph at 2", "fpga-core at 1";

Don't you need to load the core first ?

No, the periphery is first. This brings up the dram and i/o.

--dalon

Quoted 7 lines
Quoted 5 lines
board/altera/arria10-socdk/fit_spl_fpga.its new file +3 -0
+		};
+	};
+};
#9 lore raw

PATCH v8 2/8 ARM: socfpga: Add default FPGA bitstream fitImage for Arria10 SoCDK

Quoted 68 lines · Dalon L Westergreen
Quoted 58 lines · Marek Vasut
Quoted 34 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add default fitImage file bundling FPGA bitstreams for Arria10.

---

changes for v8
- Changed the FPGA node name to fpga-core and fpga-periph for both core and periph bitstreams respectively.

---
 board/altera/arria10-socdk/fit_spl_fpga.its | 39
+++++++++++++++++++++++++++++
 1 file changed, 39 insertions(+)
 create mode 100644 board/altera/arria10-socdk/fit_spl_fpga.its
board/altera/arria10-socdk/fit_spl_fpga.its new file +10 -0
new file mode 100644
index 0000000..8ce175b
@@ -0,0 +1,39 @@
+// SPDX-License-Identifier: GPL-2.0
+ /*
+ * Copyright (C) 2019 Intel Corporation <www.intel.com>
+ *
+ */
+
+/dts-v1/;
+
+/ {
+	description = "FIT image with FPGA bistream";

board/altera/arria10-socdk/fit_spl_fpga.its line 20

Quoted 10 lines
board/altera/arria10-socdk/fit_spl_fpga.its new file +10 -0
+	#address-cells = <1>;
+
+	images {
+		fpga-core at 1 {
+			description = "FPGA core bitstream";
+			data = /incbin/("../../../ghrd_10as066n2.core.rbf");
+			type = "fpga";
+			arch = "arm";
+			compression = "none";
+			load = <0x400>;

Is the load address required ?

Quoted 10 lines
board/altera/arria10-socdk/fit_spl_fpga.its new file +10 -0
+		};
+
+		fpga-periph at 2 {
+			description = "FPGA peripheral bitstream";
+			data = /incbin/("../../../ghrd_10as066n2.periph.rbf");
+			type = "fpga";
+			arch = "arm";
+			compression = "none";
+		};
+	};
Quoted 8 lines

board/altera/arria10-socdk/fit_spl_fpga.its line 36

Quoted 6 lines
board/altera/arria10-socdk/fit_spl_fpga.its new file +6 -0
+
+	configurations {
+		default = "config-1";
+		config-1 {
+			description = "Boot with FPGA early IO release config";
+			fpga = "fpga-periph at 2", "fpga-core at 1";

Don't you need to load the core first ?

Quoted 1 line

No, the periphery is first. This brings up the dram and i/o.

Then why do we have periph at 2 above ? Shouldn't those two images be swapped to make this look less confusing ?

Signature · 2 lines
Best regards,
Marek Vasut
#10 lore raw

PATCH v8 2/8 ARM: socfpga: Add default FPGA bitstream fitImage for Arria10 SoCDK

Quoted 45 lines · Marek Vasut
Quoted 44 lines · Dalon L Westergreen
Quoted 42 lines · Marek Vasut
Quoted 40 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add default fitImage file bundling FPGA bitstreams for Arria10.

---

changes for v8
- Changed the FPGA node name to fpga-core and fpga-periph for both core and
  periph bitstreams respectively.
---
 board/altera/arria10-socdk/fit_spl_fpga.its | 39
+++++++++++++++++++++++++++++

 1 file changed, 39 insertions(+)

 create mode 100644 board/altera/arria10-socdk/fit_spl_fpga.its

board/altera/arria10-socdk/fit_spl_fpga.its new file +14 -0
new file mode 100644
index 0000000..8ce175b
@@ -0,0 +1,39 @@
+// SPDX-License-Identifier: GPL-2.0
+ /*
+ * Copyright (C) 2019 Intel Corporation <www.intel.com>
+ *
+ */
+
+/dts-v1/;
+
+/ {
+	description = "FIT image with FPGA bistream";
+	#address-cells = <1>;
+
+	images {
+		fpga-core at 1 {
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

board/altera/arria10-socdk/fit_spl_fpga.its line 20

Quoted 7 lines
board/altera/arria10-socdk/fit_spl_fpga.its new file +2 -0
+			description = "FPGA core bitstream";
+			data =

/incbin/("../../../ghrd_10as066n2.core.rbf");

board/altera/arria10-socdk/fit_spl_fpga.its new file +4 -0
+			type = "fpga";
+			arch = "arm";
+			compression = "none";
+			load = <0x400>;

Is the load address required ?

It is optional for telling destination address of DDR where this core.rbf going to be loaded. If load property, the default OCRAM buffer would be used, bad for performance when loading chunk by chunk.

Quoted 22 lines
Quoted 22 lines
Quoted 22 lines
Quoted 20 lines
board/altera/arria10-socdk/fit_spl_fpga.its new file +4 -0
+		};
+
+		fpga-periph at 2 {
+			description = "FPGA peripheral

bitstream";

board/altera/arria10-socdk/fit_spl_fpga.its new file +1 -0
+			data =

/incbin/("../../../ghrd_10as066n2.periph.rbf");

board/altera/arria10-socdk/fit_spl_fpga.its new file +10 -0
+			type = "fpga";
+			arch = "arm";
+			compression = "none";
+		};
+	};
+
+	configurations {
+		default = "config-1";
+		config-1 {
+			description = "Boot with FPGA early IO

release config";

board/altera/arria10-socdk/fit_spl_fpga.its line 36

Quoted 1 line
board/altera/arria10-socdk/fit_spl_fpga.its new file +1 -0
+			fpga = "fpga-periph at 2", "fpga-core at 1";

Don't you need to load the core first ?

Quoted 3 lines
Quoted 1 line

No, the periphery is first.  This brings up the dram and i/o.

Then why do we have periph at 2 above ? Shouldn't those two images be swapped to make this look less confusing ?

The ordering in configuration fpga property doesn't matter, the driver smart enough to determine what bitstream need to be programmed at what FPGA mode.

The image fpga-core at 1 is alligned at 1st just for avoiding the performance penalty.

Quoted 1 line
#11 lore raw

PATCH v8 2/8 ARM: socfpga: Add default FPGA bitstream fitImage for Arria10 SoCDK

Quoted 54 lines · Tien Fong
Quoted 45 lines · Marek Vasut
Quoted 44 lines · Dalon L Westergreen
Quoted 42 lines · Marek Vasut
Quoted 40 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add default fitImage file bundling FPGA bitstreams for Arria10.

---

changes for v8
- Changed the FPGA node name to fpga-core and fpga-periph for both core and
  periph bitstreams respectively.
---
 board/altera/arria10-socdk/fit_spl_fpga.its | 39
+++++++++++++++++++++++++++++

 1 file changed, 39 insertions(+)

 create mode 100644 board/altera/arria10-socdk/fit_spl_fpga.its

board/altera/arria10-socdk/fit_spl_fpga.its new file +14 -0
new file mode 100644
index 0000000..8ce175b
@@ -0,0 +1,39 @@
+// SPDX-License-Identifier: GPL-2.0
+ /*
+ * Copyright (C) 2019 Intel Corporation <www.intel.com>
+ *
+ */
+
+/dts-v1/;
+
+/ {
+	description = "FIT image with FPGA bistream";
+	#address-cells = <1>;
+
+	images {
+		fpga-core at 1 {
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

board/altera/arria10-socdk/fit_spl_fpga.its line 20

Quoted 7 lines
board/altera/arria10-socdk/fit_spl_fpga.its new file +2 -0
+			description = "FPGA core bitstream";
+			data =

/incbin/("../../../ghrd_10as066n2.core.rbf");

board/altera/arria10-socdk/fit_spl_fpga.its new file +4 -0
+			type = "fpga";
+			arch = "arm";
+			compression = "none";
+			load = <0x400>;

Is the load address required ?

Quoted 3 lines

It is optional for telling destination address of DDR where this core.rbf going to be loaded. If load property, the default OCRAM buffer would be used, bad for performance when loading chunk by chunk.

So if I have something at 0x400 in DRAM and use this example in U-Boot, it will be overwritten ?

How is OCRAM bad for performance ?

Quoted 22 lines
Quoted 22 lines
Quoted 22 lines
Quoted 22 lines
Quoted 20 lines
board/altera/arria10-socdk/fit_spl_fpga.its new file +4 -0
+		};
+
+		fpga-periph at 2 {
+			description = "FPGA peripheral

bitstream";

board/altera/arria10-socdk/fit_spl_fpga.its new file +1 -0
+			data =

/incbin/("../../../ghrd_10as066n2.periph.rbf");

board/altera/arria10-socdk/fit_spl_fpga.its new file +10 -0
+			type = "fpga";
+			arch = "arm";
+			compression = "none";
+		};
+	};
+
+	configurations {
+		default = "config-1";
+		config-1 {
+			description = "Boot with FPGA early IO

release config";

board/altera/arria10-socdk/fit_spl_fpga.its line 36

Quoted 1 line
board/altera/arria10-socdk/fit_spl_fpga.its new file +1 -0
+			fpga = "fpga-periph at 2", "fpga-core at 1";

Don't you need to load the core first ?

Quoted 6 lines
Quoted 3 lines
Quoted 1 line

No, the periphery is first.  This brings up the dram and i/o.

Then why do we have periph at 2 above ? Shouldn't those two images be swapped to make this look less confusing ?

The ordering in configuration fpga property doesn't matter, the driver smart enough to determine what bitstream need to be programmed at what FPGA mode.

Good, then please order it naturally, @1 then @2 etc .

Quoted 2 lines

The image fpga-core at 1 is alligned at 1st just for avoiding the performance penalty.

I thought we concluded this should be fixed elsewhere ?

Signature · 2 lines
Best regards,
Marek Vasut
#12 lore raw

PATCH v8 2/8 ARM: socfpga: Add default FPGA bitstream fitImage for Arria10 SoCDK

Quoted 64 lines · Marek Vasut
Quoted 63 lines · Tien Fong
Quoted 61 lines · Marek Vasut
Quoted 59 lines · Dalon L Westergreen
Quoted 56 lines · Marek Vasut
Quoted 48 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add default fitImage file bundling FPGA bitstreams for Arria10.

---

changes for v8
- Changed the FPGA node name to fpga-core and fpga-periph for
both core and
  periph bitstreams respectively.
---
 board/altera/arria10-socdk/fit_spl_fpga.its | 39
+++++++++++++++++++++++++++++

 1 file changed, 39 insertions(+)

 create mode 100644 board/altera/arria10-
socdk/fit_spl_fpga.its

board/altera/arria10-socdk/fit_spl_fpga.its new file +15 -0
new file mode 100644
index 0000000..8ce175b
@@ -0,0 +1,39 @@
+// SPDX-License-Identifier: GPL-2.0
+ /*
+ * Copyright (C) 2019 Intel Corporation <www.intel.com>
+ *
+ */
+
+/dts-v1/;
+
+/ {
+	description = "FIT image with FPGA bistream";
+	#address-cells = <1>;
+
+	images {
+		fpga-core at 1 {
+			description = "FPGA core

bitstream";

board/altera/arria10-socdk/fit_spl_fpga.its new file +1 -0
+			data =

/incbin/("../../../ghrd_10as066n2.core.rbf");

board/altera/arria10-socdk/fit_spl_fpga.its line 20

Quoted 4 lines
board/altera/arria10-socdk/fit_spl_fpga.its new file +4 -0
+			type = "fpga";
+			arch = "arm";
+			compression = "none";
+			load = <0x400>;

Is the load address required ?

Quoted 7 lines
Quoted 4 lines

It is optional for telling destination address of DDR where this core.rbf going to be loaded. If load property, the default OCRAM buffer
would be used, bad for performance when loading chunk by chunk.

So if I have something at 0x400 in DRAM and use this example in U- Boot,
it will be overwritten ?

This is used for SPL only, at least for now, so that is before loading the U-Boot, DDR is blank.
But, you can define the blank location.
If load property is not defined, the driver would use small buffer from OCRAM.

Quoted 2 lines

How is OCRAM bad for performance ?

With DDR, the performance can up to 85-90%.
It is because very limited space in OCRAM, so very small buffer is used for loading bitstream, so the bitstream has to be loaded chunk by chunk.
For DDR, where whole bitstream can be loaded.

Quoted 34 lines
Quoted 34 lines
Quoted 34 lines
Quoted 30 lines
Quoted 22 lines
Quoted 22 lines
board/altera/arria10-socdk/fit_spl_fpga.its new file +4 -0
+		};
+
+		fpga-periph at 2 {
+			description = "FPGA peripheral

bitstream";

board/altera/arria10-socdk/fit_spl_fpga.its new file +1 -0
+			data =

/incbin/("../../../ghrd_10as066n2.periph.rbf");

board/altera/arria10-socdk/fit_spl_fpga.its new file +7 -0
+			type = "fpga";
+			arch = "arm";
+			compression = "none";
+		};
+	};
+
+	configurations {
Quoted 8 lines
Quoted 7 lines
board/altera/arria10-socdk/fit_spl_fpga.its new file +3 -0
+		default = "config-1";
+		config-1 {
+			description = "Boot with FPGA

early IO
release config";

board/altera/arria10-socdk/fit_spl_fpga.its new file +1 -0
+			fpga = "fpga-periph at 2", "fpga-core

@1";

Don't you need to load the core first ?

Quoted 1 line

No, the periphery is first.  This brings up the dram and i/o.

Then why do we have periph at 2 above ? Shouldn't those two images be
swapped to make this look less confusing ?

Quoted 6 lines
Quoted 5 lines

The ordering in configuration fpga property doesn't matter, the driver
smart enough to determine what bitstream need to be programmed at what
FPGA mode.

Good, then please order it naturally, @1 then @2 etc .

Okay.

Quoted 5 lines
Quoted 4 lines

The image fpga-core at 1 is alligned at 1st just for avoiding the performance penalty.

I thought we concluded this should be fixed elsewhere ?

May be we can wait until there is a solution? So user can benefit the good performance with this default fitIMage?

Quoted 1 line
#13 lore raw

PATCH v8 2/8 ARM: socfpga: Add default FPGA bitstream fitImage for Arria10 SoCDK

Quoted 78 lines · Tien Fong
Quoted 64 lines · Marek Vasut
Quoted 63 lines · Tien Fong
Quoted 61 lines · Marek Vasut
Quoted 59 lines · Dalon L Westergreen
Quoted 56 lines · Marek Vasut
Quoted 48 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add default fitImage file bundling FPGA bitstreams for Arria10.

---

changes for v8
- Changed the FPGA node name to fpga-core and fpga-periph for
both core and
  periph bitstreams respectively.
---
 board/altera/arria10-socdk/fit_spl_fpga.its | 39
+++++++++++++++++++++++++++++

 1 file changed, 39 insertions(+)

 create mode 100644 board/altera/arria10-
socdk/fit_spl_fpga.its

board/altera/arria10-socdk/fit_spl_fpga.its new file +15 -0
new file mode 100644
index 0000000..8ce175b
@@ -0,0 +1,39 @@
+// SPDX-License-Identifier: GPL-2.0
+ /*
+ * Copyright (C) 2019 Intel Corporation <www.intel.com>
+ *
+ */
+
+/dts-v1/;
+
+/ {
+	description = "FIT image with FPGA bistream";
+	#address-cells = <1>;
+
+	images {
+		fpga-core at 1 {
+			description = "FPGA core

bitstream";

board/altera/arria10-socdk/fit_spl_fpga.its new file +1 -0
+			data =

/incbin/("../../../ghrd_10as066n2.core.rbf");

board/altera/arria10-socdk/fit_spl_fpga.its

Quoted 4 lines
board/altera/arria10-socdk/fit_spl_fpga.its new file +4 -0
+			type = "fpga";
+			arch = "arm";
+			compression = "none";
+			load = <0x400>;

Is the load address required ?

Quoted 7 lines
Quoted 4 lines

It is optional for telling destination address of DDR where this core.rbf going to be loaded. If load property, the default OCRAM buffer
would be used, bad for performance when loading chunk by chunk.

So if I have something at 0x400 in DRAM and use this example in U- Boot,
it will be overwritten ?

This is used for SPL only, at least for now, so that is before loading the U-Boot, DDR is blank.
But, you can define the blank location.
If load property is not defined, the driver would use small buffer from OCRAM.

Quoted 1 line
Quoted 6 lines
Quoted 1 line

How is OCRAM bad for performance ?

With DDR, the performance can up to 85-90%.
It is because very limited space in OCRAM, so very small buffer is used for loading bitstream, so the bitstream has to be loaded chunk by chunk.
For DDR, where whole bitstream can be loaded.

Shouldn't the logic to determine where the buffer is be in the code ? I mean, once you have DRAM available, you have all the malloc space, so you can use larger buffer. Why encode that knowledge into the fitImage ?

Quoted 33 lines
Quoted 27 lines
Quoted 27 lines
Quoted 27 lines
Quoted 23 lines
Quoted 15 lines
Quoted 15 lines
board/altera/arria10-socdk/fit_spl_fpga.its new file +4 -0
+		};
+
+		fpga-periph at 2 {
+			description = "FPGA peripheral

bitstream";

board/altera/arria10-socdk/fit_spl_fpga.its new file +1 -0
+			data =

/incbin/("../../../ghrd_10as066n2.periph.rbf");

board/altera/arria10-socdk/fit_spl_fpga.its new file +7 -0
+			type = "fpga";
+			arch = "arm";
+			compression = "none";
+		};
+	};
+
+	configurations {
Quoted 8 lines
Quoted 7 lines
board/altera/arria10-socdk/fit_spl_fpga.its new file +3 -0
+		default = "config-1";
+		config-1 {
+			description = "Boot with FPGA

early IO
release config";

board/altera/arria10-socdk/fit_spl_fpga.its new file +1 -0
+			fpga = "fpga-periph at 2", "fpga-core

@1";

Don't you need to load the core first ?

Quoted 1 line

No, the periphery is first.  This brings up the dram and i/o.

Then why do we have periph at 2 above ? Shouldn't those two images be
swapped to make this look less confusing ?

Quoted 6 lines
Quoted 5 lines

The ordering in configuration fpga property doesn't matter, the driver
smart enough to determine what bitstream need to be programmed at what
FPGA mode.

Good, then please order it naturally, @1 then @2 etc .

Quoted 8 lines

Okay.

Quoted 5 lines
Quoted 4 lines

The image fpga-core at 1 is alligned at 1st just for avoiding the performance penalty.

I thought we concluded this should be fixed elsewhere ?

May be we can wait until there is a solution? So user can benefit the good performance with this default fitIMage?

This will only work for specific cases of specific storage devices and filesystems .

Signature · 2 lines
Best regards,
Marek Vasut
#14 lore raw

PATCH v8 2/8 ARM: socfpga: Add default FPGA bitstream fitImage for Arria10 SoCDK

Quoted 103 lines · Marek Vasut
Quoted 95 lines · Tien Fong
Quoted 80 lines · Marek Vasut
Quoted 72 lines · Tien Fong
Quoted 61 lines · Marek Vasut
Quoted 58 lines · Dalon L Westergreen
Quoted 54 lines · Marek Vasut
Quoted 50 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add default fitImage file bundling FPGA bitstreams for Arria10.

Quoted 1 line

---

changes for v8
- Changed the FPGA node name to fpga-core and fpga- periph
for
both core and
  periph bitstreams respectively.
---
 board/altera/arria10-socdk/fit_spl_fpga.its | 39
+++++++++++++++++++++++++++++

 1 file changed, 39 insertions(+)

 create mode 100644 board/altera/arria10-
socdk/fit_spl_fpga.its

(quoted diff) +0 -0

socdk/fit_spl_fpga.its
b/board/altera/arria10-socdk/fit_spl_fpga.its
new file mode 100644
index 0000000..8ce175b

board/altera/arria10-socdk/fit_spl_fpga.its new file +3 -0
@@ -0,0 +1,39 @@
+// SPDX-License-Identifier: GPL-2.0
+ /*
+ * Copyright (C) 2019 Intel Corporation <www.intel.com
Quoted 1 line

+ *
+ */

board/altera/arria10-socdk/fit_spl_fpga.its new file +10 -0
+
+/dts-v1/;
+
+/ {
+	description = "FIT image with FPGA bistream";
+	#address-cells = <1>;
+
+	images {
+		fpga-core at 1 {
+			description = "FPGA core
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

board/altera/arria10-socdk/fit_spl_fpga.its

Quoted 7 lines

bitstream";

board/altera/arria10-socdk/fit_spl_fpga.its new file +1 -0
+			data =

/incbin/("../../../ghrd_10as066n2.core.rbf");

board/altera/arria10-socdk/fit_spl_fpga.its new file +4 -0
+			type = "fpga";
+			arch = "arm";
+			compression = "none";
+			load = <0x400>;

Is the load address required ?

Quoted 6 lines

It is optional for telling destination address of DDR where this
core.rbf going to be loaded. If load property, the default OCRAM
buffer
would be used, bad for performance when loading chunk by chunk.

Quoted 4 lines

So if I have something at 0x400 in DRAM and use this example in U-
Boot,
it will be overwritten ?

This is used for SPL only, at least for now, so that is before loading
the U-Boot, DDR is blank.
But, you can define the blank location.
If load property is not defined, the driver would use small buffer from
OCRAM.

Quoted 2 lines
Quoted 7 lines
Quoted 1 line

How is OCRAM bad for performance ?

With DDR, the performance can up to 85-90%.
It is because very limited space in OCRAM, so very small buffer is used
for loading bitstream, so the bitstream has to be loaded chunk by chunk.
For DDR, where whole bitstream can be loaded.

Quoted 5 lines

Shouldn't the logic to determine where the buffer is be in the code ? I mean, once you have DRAM available, you have all the malloc space, so
you can use larger buffer.Why encode that knowledge into the fitImage ?

In previosly, we hard code the dest location for loading the core bitstream in SPL, because by that time DDR up running, but malloc is still pointed to OCRAM.
Now, we let user to define the dest location through the load property.

Quoted 60 lines
Quoted 50 lines
Quoted 38 lines
Quoted 38 lines
Quoted 38 lines
Quoted 32 lines
Quoted 24 lines
Quoted 24 lines
board/altera/arria10-socdk/fit_spl_fpga.its new file +4 -0
+		};
+
+		fpga-periph at 2 {
+			description = "FPGA peripheral

bitstream";

board/altera/arria10-socdk/fit_spl_fpga.its new file +1 -0
+			data =

/incbin/("../../../ghrd_10as066n2.periph.rbf");

board/altera/arria10-socdk/fit_spl_fpga.its new file +8 -0
+			type = "fpga";
+			arch = "arm";
+			compression = "none";
+		};
+	};
+
+	configurations {
+		default = "config-1";
Quoted 8 lines
Quoted 7 lines
board/altera/arria10-socdk/fit_spl_fpga.its new file +2 -0
+		config-1 {
+			description = "Boot with FPGA

early IO
release config";

board/altera/arria10-socdk/fit_spl_fpga.its new file +1 -0
+			fpga = "fpga-periph at 2", "fpga-

core
@1";

Don't you need to load the core first ?

Quoted 2 lines

No, the periphery is first.  This brings up the dram and i/o.

Then why do we have periph at 2 above ? Shouldn't those two images
be
swapped to make this look less confusing ?

Quoted 7 lines
Quoted 6 lines

The ordering in configuration fpga property doesn't matter, the driver
smart enough to determine what bitstream need to be programmed at
what
FPGA mode.

Good, then please order it naturally, @1 then @2 etc .

Okay.

Quoted 4 lines
Quoted 4 lines
Quoted 6 lines
Quoted 3 lines
Quoted 2 lines

The image fpga-core at 1 is alligned at 1st just for avoiding the performance penalty.

I thought we concluded this should be fixed elsewhere ?

May be we can wait until there is a solution? So user can benefit the
good performance with this default fitIMage?

This will only work for specific cases of specific storage devices and
filesystems .

#15 lore raw

PATCH v8 2/8 ARM: socfpga: Add default FPGA bitstream fitImage for Arria10 SoCDK

Quoted 109 lines · Tien Fong
Quoted 103 lines · Marek Vasut
Quoted 95 lines · Tien Fong
Quoted 80 lines · Marek Vasut
Quoted 72 lines · Tien Fong
Quoted 61 lines · Marek Vasut
Quoted 58 lines · Dalon L Westergreen
Quoted 54 lines · Marek Vasut
Quoted 50 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add default fitImage file bundling FPGA bitstreams for Arria10.

Quoted 1 line

---

changes for v8
- Changed the FPGA node name to fpga-core and fpga- periph
for
both core and
  periph bitstreams respectively.
---
 board/altera/arria10-socdk/fit_spl_fpga.its | 39
+++++++++++++++++++++++++++++

 1 file changed, 39 insertions(+)

 create mode 100644 board/altera/arria10-
socdk/fit_spl_fpga.its

(quoted diff) +0 -0

socdk/fit_spl_fpga.its
b/board/altera/arria10-socdk/fit_spl_fpga.its
new file mode 100644
index 0000000..8ce175b

board/altera/arria10-socdk/fit_spl_fpga.its new file +3 -0
@@ -0,0 +1,39 @@
+// SPDX-License-Identifier: GPL-2.0
+ /*
+ * Copyright (C) 2019 Intel Corporation <www.intel.com
Quoted 1 line

+ *
+ */

board/altera/arria10-socdk/fit_spl_fpga.its new file +10 -0
+
+/dts-v1/;
+
+/ {
+	description = "FIT image with FPGA bistream";
+	#address-cells = <1>;
+
+	images {
+		fpga-core at 1 {
+			description = "FPGA core
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

board/altera/arria10-socdk/fit_spl_fpga.its

Quoted 7 lines

bitstream";

board/altera/arria10-socdk/fit_spl_fpga.its new file +1 -0
+			data =

/incbin/("../../../ghrd_10as066n2.core.rbf");

board/altera/arria10-socdk/fit_spl_fpga.its new file +4 -0
+			type = "fpga";
+			arch = "arm";
+			compression = "none";
+			load = <0x400>;

Is the load address required ?

Quoted 6 lines

It is optional for telling destination address of DDR where this
core.rbf going to be loaded. If load property, the default OCRAM
buffer
would be used, bad for performance when loading chunk by chunk.

Quoted 4 lines

So if I have something at 0x400 in DRAM and use this example in U-
Boot,
it will be overwritten ?

This is used for SPL only, at least for now, so that is before loading
the U-Boot, DDR is blank.
But, you can define the blank location.
If load property is not defined, the driver would use small buffer from
OCRAM.

Quoted 2 lines
Quoted 7 lines
Quoted 1 line

How is OCRAM bad for performance ?

With DDR, the performance can up to 85-90%.
It is because very limited space in OCRAM, so very small buffer is used
for loading bitstream, so the bitstream has to be loaded chunk by chunk.
For DDR, where whole bitstream can be loaded.

Quoted 5 lines

Shouldn't the logic to determine where the buffer is be in the code ? I mean, once you have DRAM available, you have all the malloc space, so
you can use larger buffer.Why encode that knowledge into the fitImage ?

Quoted 4 lines

In previosly, we hard code the dest location for loading the core bitstream in SPL, because by that time DDR up running, but malloc is still pointed to OCRAM.
Now, we let user to define the dest location through the load property.

But the user shouldn't care where the buffer is, the buffer should be in the best possible location and the SPL can determine that (if DRAM is running, in DRAM, otherwise in OCRAM), no ?

Signature · 2 lines
Best regards,
Marek Vasut
#16 lore raw

PATCH v8 2/8 ARM: socfpga: Add default FPGA bitstream fitImage for Arria10 SoCDK

Quoted 139 lines · Marek Vasut
Quoted 132 lines · Tien Fong
Quoted 122 lines · Marek Vasut
Quoted 112 lines · Tien Fong
Quoted 92 lines · Marek Vasut
Quoted 82 lines · Tien Fong
Quoted 70 lines · Marek Vasut
Quoted 66 lines · Dalon L Westergreen
Quoted 61 lines · Marek Vasut
Quoted 56 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add default fitImage file bundling FPGA bitstreams for
Arria10.

.com

Quoted 2 lines

---

changes for v8
- Changed the FPGA node name to fpga-core and fpga- periph
for
both core and
  periph bitstreams respectively.
---
 board/altera/arria10-socdk/fit_spl_fpga.its | 39
+++++++++++++++++++++++++++++

 1 file changed, 39 insertions(+)

 create mode 100644 board/altera/arria10-
socdk/fit_spl_fpga.its

(quoted diff) +0 -0

socdk/fit_spl_fpga.its
b/board/altera/arria10-socdk/fit_spl_fpga.its
new file mode 100644
index 0000000..8ce175b

board/altera/arria10-socdk/fit_spl_fpga.its new file +3 -0
@@ -0,0 +1,39 @@
+// SPDX-License-Identifier: GPL-2.0
+ /*
+ * Copyright (C) 2019 Intel Corporation <www.intel

.com

Quoted 2 lines

+ *
+ */

board/altera/arria10-socdk/fit_spl_fpga.its new file +5 -0
+
+/dts-v1/;
+
+/ {
+	description = "FIT image with FPGA

bistream";

board/altera/arria10-socdk/fit_spl_fpga.its new file +5 -0
+	#address-cells = <1>;
+
+	images {
+		fpga-core at 1 {
+			description = "FPGA core
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

board/altera/arria10-socdk/fit_spl_fpga.its

Quoted 7 lines

bitstream";

board/altera/arria10-socdk/fit_spl_fpga.its new file +1 -0
+			data =

/incbin/("../../../ghrd_10as066n2.core.rbf");

board/altera/arria10-socdk/fit_spl_fpga.its new file +4 -0
+			type = "fpga";
+			arch = "arm";
+			compression = "none";
+			load = <0x400>;

Is the load address required ?

Quoted 7 lines

It is optional for telling destination address of DDR where this
core.rbf going to be loaded. If load property, the default OCRAM
buffer
would be used, bad for performance when loading chunk by chunk.

Quoted 5 lines

So if I have something at 0x400 in DRAM and use this example in
U-
Boot,
it will be overwritten ?

This is used for SPL only, at least for now, so that is before loading
the U-Boot, DDR is blank.
But, you can define the blank location.
If load property is not defined, the driver would use small buffer
from
OCRAM.

Quoted 4 lines

How is OCRAM bad for performance ?

Quoted 8 lines

With DDR, the performance can up to 85-90%.
It is because very limited space in OCRAM, so very small buffer is
used
for loading bitstream, so the bitstream has to be loaded chunk by
chunk.
For DDR, where whole bitstream can be loaded.

Quoted 8 lines

Shouldn't the logic to determine where the buffer is be in the code ?
I mean, once you have DRAM available, you have all the malloc space,
so
you can use larger buffer.Why encode that knowledge into the fitImage
?

Quoted 6 lines

In previosly, we hard code the dest location for loading the core bitstream in SPL, because by that time DDR up running, but malloc is
still pointed to OCRAM.
Now, we let user to define the dest location through the load property.

Quoted 4 lines

But the user shouldn't care where the buffer is, the buffer should be in
the best possible location and the SPL can determine that (if DRAM is running, in DRAM, otherwise in OCRAM), no ?

Yes, most of the time user don't need to change the dest value at load property, unless they have very good reason to do so.
SPL would smart enough to read the load property after the DDR is up running, and loading the bitstream to dest location, and programming FPGA from there.

Quoted 1 line
#17 lore raw

PATCH v8 2/8 ARM: socfpga: Add default FPGA bitstream fitImage for Arria10 SoCDK

Quoted 144 lines · Tien Fong
Quoted 139 lines · Marek Vasut
Quoted 132 lines · Tien Fong
Quoted 122 lines · Marek Vasut
Quoted 112 lines · Tien Fong
Quoted 92 lines · Marek Vasut
Quoted 82 lines · Tien Fong
Quoted 70 lines · Marek Vasut
Quoted 66 lines · Dalon L Westergreen
Quoted 61 lines · Marek Vasut
Quoted 56 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add default fitImage file bundling FPGA bitstreams for
Arria10.

.com

Quoted 2 lines

---

changes for v8
- Changed the FPGA node name to fpga-core and fpga- periph
for
both core and
  periph bitstreams respectively.
---
 board/altera/arria10-socdk/fit_spl_fpga.its | 39
+++++++++++++++++++++++++++++

 1 file changed, 39 insertions(+)

 create mode 100644 board/altera/arria10-
socdk/fit_spl_fpga.its

(quoted diff) +0 -0

socdk/fit_spl_fpga.its
b/board/altera/arria10-socdk/fit_spl_fpga.its
new file mode 100644
index 0000000..8ce175b

board/altera/arria10-socdk/fit_spl_fpga.its new file +3 -0
@@ -0,0 +1,39 @@
+// SPDX-License-Identifier: GPL-2.0
+ /*
+ * Copyright (C) 2019 Intel Corporation <www.intel

.com

Quoted 2 lines

+ *
+ */

board/altera/arria10-socdk/fit_spl_fpga.its new file +5 -0
+
+/dts-v1/;
+
+/ {
+	description = "FIT image with FPGA

bistream";

board/altera/arria10-socdk/fit_spl_fpga.its new file +5 -0
+	#address-cells = <1>;
+
+	images {
+		fpga-core at 1 {
+			description = "FPGA core
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

board/altera/arria10-socdk/fit_spl_fpga.its

Quoted 7 lines

bitstream";

board/altera/arria10-socdk/fit_spl_fpga.its new file +1 -0
+			data =

/incbin/("../../../ghrd_10as066n2.core.rbf");

board/altera/arria10-socdk/fit_spl_fpga.its new file +4 -0
+			type = "fpga";
+			arch = "arm";
+			compression = "none";
+			load = <0x400>;

Is the load address required ?

Quoted 7 lines

It is optional for telling destination address of DDR where this
core.rbf going to be loaded. If load property, the default OCRAM
buffer
would be used, bad for performance when loading chunk by chunk.

Quoted 5 lines

So if I have something at 0x400 in DRAM and use this example in
U-
Boot,
it will be overwritten ?

This is used for SPL only, at least for now, so that is before loading
the U-Boot, DDR is blank.
But, you can define the blank location.
If load property is not defined, the driver would use small buffer
from
OCRAM.

Quoted 4 lines

How is OCRAM bad for performance ?

Quoted 8 lines

With DDR, the performance can up to 85-90%.
It is because very limited space in OCRAM, so very small buffer is
used
for loading bitstream, so the bitstream has to be loaded chunk by
chunk.
For DDR, where whole bitstream can be loaded.

Quoted 8 lines

Shouldn't the logic to determine where the buffer is be in the code ?
I mean, once you have DRAM available, you have all the malloc space,
so
you can use larger buffer.Why encode that knowledge into the fitImage
?

Quoted 6 lines

In previosly, we hard code the dest location for loading the core bitstream in SPL, because by that time DDR up running, but malloc is
still pointed to OCRAM.
Now, we let user to define the dest location through the load property.

Quoted 4 lines

But the user shouldn't care where the buffer is, the buffer should be in
the best possible location and the SPL can determine that (if DRAM is running, in DRAM, otherwise in OCRAM), no ?

Quoted 5 lines

Yes, most of the time user don't need to change the dest value at load property, unless they have very good reason to do so.
SPL would smart enough to read the load property after the DDR is up running, and loading the bitstream to dest location, and programming FPGA from there.

Then we don't need the 'load' property by default ?

Signature · 2 lines
Best regards,
Marek Vasut
#18 lore raw

PATCH v8 2/8 ARM: socfpga: Add default FPGA bitstream fitImage for Arria10 SoCDK

Quoted 188 lines · Marek Vasut
Quoted 179 lines · Tien Fong
Quoted 171 lines · Marek Vasut
Quoted 161 lines · Tien Fong
Quoted 149 lines · Marek Vasut
Quoted 136 lines · Tien Fong
Quoted 112 lines · Marek Vasut
Quoted 98 lines · Tien Fong
Quoted 85 lines · Marek Vasut
Quoted 80 lines · Dalon L Westergreen
Quoted 73 lines · Marek Vasut
Quoted 66 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add default fitImage file bundling FPGA
bitstreams
for
Arria10.

ntel
.com

Quoted 3 lines

---

changes for v8
- Changed the FPGA node name to fpga-core and
fpga-
periph
for
both core and
  periph bitstreams respectively.
---
 board/altera/arria10-socdk/fit_spl_fpga.its |
39
+++++++++++++++++++++++++++++

 1 file changed, 39 insertions(+)

 create mode 100644 board/altera/arria10-
socdk/fit_spl_fpga.its

(quoted diff) +0 -0

socdk/fit_spl_fpga.its
b/board/altera/arria10-socdk/fit_spl_fpga.its
new file mode 100644
index 0000000..8ce175b

board/altera/arria10- new file +0 -0

socdk/fit_spl_fpga.its

board/altera/arria10- new file +3 -0
@@ -0,0 +1,39 @@
+// SPDX-License-Identifier: GPL-2.0
+ /*
+ * Copyright (C) 2019 Intel Corporation <www.i

ntel
.com

Quoted 3 lines

+ *
+ */

board/altera/arria10- new file +5 -0
+
+/dts-v1/;
+
+/ {
+	description = "FIT image with FPGA

bistream";

board/altera/arria10- new file +5 -0
+	#address-cells = <1>;
+
+	images {
+		fpga-core at 1 {
+			description = "FPGA

core

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

board/altera/arria10-

Quoted 7 lines

bitstream";

board/altera/arria10- new file +1 -0
+			data =

/incbin/("../../../ghrd_10as066n2.core.rbf");

board/altera/arria10- new file +4 -0
+			type = "fpga";
+			arch = "arm";
+			compression = "none";
+			load = <0x400>;

Is the load address required ?

Quoted 10 lines

It is optional for telling destination address of DDR where
this
core.rbf going to be loaded. If load property, the default
OCRAM
buffer
would be used, bad for performance when loading chunk by
chunk.

Quoted 6 lines

So if I have something at 0x400 in DRAM and use this example
in
U-
Boot,
it will be overwritten ?

This is used for SPL only, at least for now, so that is before
loading
the U-Boot, DDR is blank.
But, you can define the blank location.
If load property is not defined, the driver would use small buffer
from
OCRAM.

Quoted 5 lines

How is OCRAM bad for performance ?

Quoted 10 lines

With DDR, the performance can up to 85-90%.
It is because very limited space in OCRAM, so very small buffer
is
used
for loading bitstream, so the bitstream has to be loaded chunk
by
chunk.
For DDR, where whole bitstream can be loaded.

Quoted 9 lines

Shouldn't the logic to determine where the buffer is be in the
code ?
I mean, once you have DRAM available, you have all the malloc space,
so
you can use larger buffer.Why encode that knowledge into the fitImage
?

Quoted 8 lines

In previosly, we hard code the dest location for loading the core
bitstream in SPL, because by that time DDR up running, but malloc
is
still pointed to OCRAM.
Now, we let user to define the dest location through the load property.

Quoted 6 lines

But the user shouldn't care where the buffer is, the buffer should be
in
the best possible location and the SPL can determine that (if DRAM is
running, in DRAM, otherwise in OCRAM), no ?

Quoted 8 lines

Yes, most of the time user don't need to change the dest value at load
property, unless they have very good reason to do so. SPL would smart enough to read the load property after the DDR is up
running, and loading the bitstream to dest location, and programming
FPGA from there.

Quoted 1 line

Then we don't need the 'load' property by default ?

In previosly, we hide this from user, the dest location is defined in the code. But now, we expose this to user, so user can change that through the load property.

If this load property is not defined, then OCRAM buffer would be used instead.

Quoted 1 line
#19 lore raw

PATCH v8 2/8 ARM: socfpga: Add default FPGA bitstream fitImage for Arria10 SoCDK

In reply to Marek Vasut
Quoted 188 lines · Marek Vasut
Quoted 179 lines · Tien Fong
Quoted 171 lines · Marek Vasut
Quoted 161 lines · Tien Fong
Quoted 149 lines · Marek Vasut
Quoted 136 lines · Tien Fong
Quoted 112 lines · Marek Vasut
Quoted 98 lines · Tien Fong
Quoted 85 lines · Marek Vasut
Quoted 80 lines · Dalon L Westergreen
Quoted 73 lines · Marek Vasut
Quoted 66 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add default fitImage file bundling FPGA
bitstreams
for
Arria10.

ntel
.com

Quoted 3 lines

---

changes for v8
- Changed the FPGA node name to fpga-core and
fpga-
periph
for
both core and
  periph bitstreams respectively.
---
 board/altera/arria10-socdk/fit_spl_fpga.its |
39
+++++++++++++++++++++++++++++

 1 file changed, 39 insertions(+)

 create mode 100644 board/altera/arria10-
socdk/fit_spl_fpga.its

(quoted diff) +0 -0

socdk/fit_spl_fpga.its
b/board/altera/arria10-socdk/fit_spl_fpga.its
new file mode 100644
index 0000000..8ce175b

board/altera/arria10- new file +0 -0

socdk/fit_spl_fpga.its

board/altera/arria10- new file +3 -0
@@ -0,0 +1,39 @@
+// SPDX-License-Identifier: GPL-2.0
+ /*
+ * Copyright (C) 2019 Intel Corporation <www.i

ntel
.com

Quoted 3 lines

+ *
+ */

board/altera/arria10- new file +5 -0
+
+/dts-v1/;
+
+/ {
+	description = "FIT image with FPGA

bistream";

board/altera/arria10- new file +5 -0
+	#address-cells = <1>;
+
+	images {
+		fpga-core at 1 {
+			description = "FPGA

core

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

board/altera/arria10-

Quoted 7 lines

bitstream";

board/altera/arria10- new file +1 -0
+			data =

/incbin/("../../../ghrd_10as066n2.core.rbf");

board/altera/arria10- new file +4 -0
+			type = "fpga";
+			arch = "arm";
+			compression = "none";
+			load = <0x400>;

Is the load address required ?

Quoted 10 lines

It is optional for telling destination address of DDR where
this
core.rbf going to be loaded. If load property, the default
OCRAM
buffer
would be used, bad for performance when loading chunk by
chunk.

Quoted 6 lines

So if I have something at 0x400 in DRAM and use this example
in
U-
Boot,
it will be overwritten ?

This is used for SPL only, at least for now, so that is before
loading
the U-Boot, DDR is blank.
But, you can define the blank location.
If load property is not defined, the driver would use small buffer
from
OCRAM.

Quoted 5 lines

How is OCRAM bad for performance ?

Quoted 10 lines

With DDR, the performance can up to 85-90%.
It is because very limited space in OCRAM, so very small buffer
is
used
for loading bitstream, so the bitstream has to be loaded chunk
by
chunk.
For DDR, where whole bitstream can be loaded.

Quoted 9 lines

Shouldn't the logic to determine where the buffer is be in the
code ?
I mean, once you have DRAM available, you have all the malloc space,
so
you can use larger buffer.Why encode that knowledge into the fitImage
?

Quoted 8 lines

In previosly, we hard code the dest location for loading the core
bitstream in SPL, because by that time DDR up running, but malloc
is
still pointed to OCRAM.
Now, we let user to define the dest location through the load property.

Quoted 6 lines

But the user shouldn't care where the buffer is, the buffer should be
in
the best possible location and the SPL can determine that (if DRAM is
running, in DRAM, otherwise in OCRAM), no ?

Quoted 8 lines

Yes, most of the time user don't need to change the dest value at load
property, unless they have very good reason to do so. SPL would smart enough to read the load property after the DDR is up
running, and loading the bitstream to dest location, and programming
FPGA from there.

Quoted 1 line

Then we don't need the 'load' property by default ?

I will add a default ddr load address in the code, and enable the load property to overwrite it if it is defined.

So, that means the load property is optional.

Quoted 1 line
#20 lore raw

PATCH v8 3/8 fit: Add function declarations to the header file

In reply to tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Adding some function declarations to the header file, so these functions can be referred by other C files.

---
 include/image.h | 4 ++++
 1 file changed, 4 insertions(+)
include/image.h +4 -0
index 83a2d41..f839443 100644
@@ -1041,6 +1041,10 @@ int fit_check_format(const void *fit);
 
 int fit_conf_find_compat(const void *fit, const void *fdt);
 int fit_conf_get_node(const void *fit, const char *conf_uname);
+int fit_conf_get_prop_node_count(const void *fit, int noffset,
+		const char *prop_name);
+int fit_conf_get_prop_node_index(const void *fit, int noffset,
+		const char *prop_name, int index);
 
 /**
  * fit_conf_get_prop_node() - Get node refered to by a configuration
Signature · 1 line
2.2.0
#21 lore raw

PATCH v8 3/8 fit: Add function declarations to the header file

Quoted 15 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Adding some function declarations to the header file, so these functions can be referred by other C files.

---
 include/image.h | 4 ++++
 1 file changed, 4 insertions(+)
include/image.h +0 -0
index 83a2d41..f839443 100644

include/image.h int fit_check_format(const void *fit); · line 1047

Quoted 7 lines
include/image.h +4 -0
@@ -1041,6 +1041,10 @@ int fit_check_format(const void *fit);
 
 int fit_conf_find_compat(const void *fit, const void *fdt);
 int fit_conf_get_node(const void *fit, const char *conf_uname);
+int fit_conf_get_prop_node_count(const void *fit, int noffset,
+		const char *prop_name);
+int fit_conf_get_prop_node_index(const void *fit, int noffset,
+		const char *prop_name, int index);

Squash this patch with a patch adding those functions.

Quoted 3 lines
include/image.h +0 -0
 /**
  * fit_conf_get_prop_node() - Get node refered to by a configuration
 
Signature · 2 lines
Best regards,
Marek Vasut
#22 lore raw

PATCH v8 3/8 fit: Add function declarations to the header file

Quoted 12 lines · Marek Vasut
Quoted 11 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Adding some function declarations to the header file, so these functions can be referred by other C files.

---
 include/image.h | 4 ++++

 1 file changed, 4 insertions(+)
Quoted 13 lines

include/image.h int fit_check_format(const void *fit); · line 1047

Quoted 12 lines
include/image.h +0 -0
index 83a2d41..f839443 100644
@@ -1041,6 +1041,10 @@ int fit_check_format(const void *fit);


 int fit_conf_find_compat(const void *fit, const void *fdt); int fit_conf_get_node(const void *fit, const char *conf_uname);

include/image.h +4 -0
+int fit_conf_get_prop_node_count(const void *fit, int noffset,
+		const char *prop_name);
+int fit_conf_get_prop_node_index(const void *fit, int noffset,
+		const char *prop_name, int index);

Squash this patch with a patch adding those functions.

Okay.

Quoted 7 lines
Quoted 6 lines

 /**
  * fit_conf_get_prop_node() - Get node refered to by a configuration

#23 lore raw

PATCH v8 3/8 fit: Add function declarations to the header file

In reply to Marek Vasut
Quoted 12 lines · Marek Vasut
Quoted 11 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Adding some function declarations to the header file, so these functions can be referred by other C files.

---
 include/image.h | 4 ++++

 1 file changed, 4 insertions(+)
Quoted 13 lines

include/image.h int fit_check_format(const void *fit); · line 1047

Quoted 12 lines
include/image.h +0 -0
index 83a2d41..f839443 100644
@@ -1041,6 +1041,10 @@ int fit_check_format(const void *fit);


 int fit_conf_find_compat(const void *fit, const void *fdt); int fit_conf_get_node(const void *fit, const char *conf_uname);

include/image.h +4 -0
+int fit_conf_get_prop_node_count(const void *fit, int noffset,
+		const char *prop_name);
+int fit_conf_get_prop_node_index(const void *fit, int noffset,
+		const char *prop_name, int index);

Squash this patch with a patch adding those functions.

Just to confirm again squashing this patch to the patch calling these function?

Quoted 7 lines
Quoted 6 lines

 /**
  * fit_conf_get_prop_node() - Get node refered to by a configuration

#24 lore raw

PATCH v8 3/8 fit: Add function declarations to the header file

Quoted 26 lines · Tien Fong
Quoted 12 lines · Marek Vasut
Quoted 11 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Adding some function declarations to the header file, so these functions can be referred by other C files.

---
 include/image.h | 4 ++++

 1 file changed, 4 insertions(+)
Quoted 13 lines

include/image.h int fit_check_format(const void *fit); · line 1047

Quoted 12 lines
include/image.h +0 -0
index 83a2d41..f839443 100644
@@ -1041,6 +1041,10 @@ int fit_check_format(const void *fit);


 int fit_conf_find_compat(const void *fit, const void *fdt); int fit_conf_get_node(const void *fit, const char *conf_uname);

include/image.h +4 -0
+int fit_conf_get_prop_node_count(const void *fit, int noffset,
+		const char *prop_name);
+int fit_conf_get_prop_node_index(const void *fit, int noffset,
+		const char *prop_name, int index);

Squash this patch with a patch adding those functions.

Quoted 2 lines

Just to confirm again squashing this patch to the patch calling these function?

Yes

Signature · 2 lines
Best regards,
Marek Vasut
#25 lore raw

PATCH v8 3/8 fit: Add function declarations to the header file

Quoted 31 lines · Marek Vasut
Quoted 30 lines · Tien Fong
Quoted 28 lines · Marek Vasut
Quoted 21 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Adding some function declarations to the header file, so these functions can be referred by other C files.

---
 include/image.h | 4 ++++

 1 file changed, 4 insertions(+)
include/image.h +0 -0
index 83a2d41..f839443 100644
@@ -1041,6 +1041,10 @@ int fit_check_format(const void *fit);


 int fit_conf_find_compat(const void *fit, const void *fdt); int fit_conf_get_node(const void *fit, const char *conf_uname);

include/image.h int fit_check_format(const void *fit); · line 1047

Quoted 4 lines
include/image.h +4 -0
+int fit_conf_get_prop_node_count(const void *fit, int noffset,
+		const char *prop_name);
+int fit_conf_get_prop_node_index(const void *fit, int noffset,
+		const char *prop_name, int index);

Squash this patch with a patch adding those functions.

Quoted 4 lines
Quoted 3 lines

Just to confirm again squashing this patch to the patch calling these
function?

Yes

Okay.

Quoted 1 line
#26 lore raw

PATCH v8 4/8 ARM: socfpga: Add FPGA drivers for Arria 10 FPGA bitstream loading

In reply to tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add FPGA driver to support program FPGA with FPGA bitstream loading from filesystem. The driver are designed based on generic firmware loader framework. The driver can handle FPGA program operation from loading FPGA bitstream in flash to memory and then to program FPGA.

---

changes for v8
- Added codes to discern bitstream type based on fpga node name.

changes for v7
- Restructure the FPGA driver to support both peripheral bitstream and core bitstream bundled into FIT image.
- Support loadable property for core bitstream. User can set loadable in DDR for better performance. This loading would be done in one large chunk instead of chunk by chunk loading with small memory buffer.

---
 arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts       |  17 +
 .../include/mach/fpga_manager_arria10.h            |  39 +-
 drivers/fpga/socfpga_arria10.c                     | 467 ++++++++++++++++++++-
 3 files changed, 500 insertions(+), 23 deletions(-)
arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +17 -0
index 998d811..14f1967 100644
@@ -18,6 +18,23 @@
 /dts-v1/;
 #include "socfpga_arria10_socdk.dtsi"
 
+/ {
+	chosen {
+		firmware-loader = &fs_loader0;
+	};
+
+	fs_loader0: fs-loader at 0 {
+		u-boot,dm-pre-reloc;
+		compatible = "u-boot,fs-loader";
+		phandlepart = <&mmc 1>;
+	};
+};
+
+&fpga_mgr {
+	u-boot,dm-pre-reloc;
+	altr,bitstream = "fit_spl_fpga.itb";
+};
+
 &mmc {
 	u-boot,dm-pre-reloc;
 	status = "okay";
arch/arm/mach-socfpga/include/mach/fpga_manager_arria10.h +37 -2
index 09d13f6..5ef15bb 100644
@@ -1,9 +1,13 @@
 /* SPDX-License-Identifier: GPL-2.0 */
 /*
- * Copyright (C) 2017 Intel Corporation <www.intel.com>
+ * Copyright (C) 2017-2019 Intel Corporation <www.intel.com>
  * All rights reserved.
  */
 
+#include <asm/cache.h>
+#include <altera.h>
+#include <image.h>
+
 #ifndef _FPGA_MANAGER_ARRIA10_H_
 #define _FPGA_MANAGER_ARRIA10_H_
 
@@ -51,6 +55,10 @@
 #define ALT_FPGAMGR_IMGCFG_CTL_02_CFGWIDTH_SET_MSK		BIT(24)
 #define ALT_FPGAMGR_IMGCFG_CTL_02_CDRATIO_LSB			16
 
+#define FPGA_SOCFPGA_A10_RBF_UNENCRYPTED	0xa65c
+#define FPGA_SOCFPGA_A10_RBF_ENCRYPTED		0xa65d
+#define FPGA_SOCFPGA_A10_RBF_PERIPH		0x0001
+#define FPGA_SOCFPGA_A10_RBF_CORE		0x8001
 #ifndef __ASSEMBLY__
 
 struct socfpga_fpga_manager {
@@ -88,12 +96,39 @@ struct socfpga_fpga_manager {
 	u32  imgcfg_fifo_status;
 };
 
+enum rbf_type {
+	unknown,
+	periph_section,
+	core_section
+};
+
+enum rbf_security {
+	invalid,
+	unencrypted,
+	encrypted
+};
+
+struct rbf_info {
+	enum rbf_type section;
+	enum rbf_security security;
+};
+
+struct fpga_loadfs_info {
+	fpga_fs_info *fpga_fsinfo;
+	u32 remaining;
+	u32 offset;
+	struct rbf_info rbfinfo;
+};
+
 /* Functions */
 int fpgamgr_program_init(u32 * rbf_data, size_t rbf_size);
 int fpgamgr_program_finish(void);
 int is_fpgamgr_user_mode(void);
 int fpgamgr_wait_early_user_mode(void);
-
+int is_fpgamgr_early_user_mode(void);
+char *get_fpga_filename(const void *fdt, int *len);
+int socfpga_loadfs(fpga_fs_info *fpga_fsinfo, const void *buf, size_t bsize,
+		  u32 offset);
 #endif /* __ASSEMBLY__ */
 
 #endif /* _FPGA_MANAGER_ARRIA10_H_ */
drivers/fpga/socfpga_arria10.c +446 -21

527 lines not on this page · raw message

Signature · 1 line
2.2.0
#27 lore raw

PATCH v8 4/8 ARM: socfpga: Add FPGA drivers for Arria 10 FPGA bitstream loading

Quoted 31 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add FPGA driver to support program FPGA with FPGA bitstream loading from filesystem. The driver are designed based on generic firmware loader framework. The driver can handle FPGA program operation from loading FPGA bitstream in flash to memory and then to program FPGA.

---

changes for v8
- Added codes to discern bitstream type based on fpga node name.

changes for v7
- Restructure the FPGA driver to support both peripheral bitstream and core bitstream bundled into FIT image.
- Support loadable property for core bitstream. User can set loadable in DDR for better performance. This loading would be done in one large chunk instead of chunk by chunk loading with small memory buffer.

---
 arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts       |  17 +
 .../include/mach/fpga_manager_arria10.h            |  39 +-
 drivers/fpga/socfpga_arria10.c                     | 467 ++++++++++++++++++++-
 3 files changed, 500 insertions(+), 23 deletions(-)
arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +0 -0
index 998d811..14f1967 100644

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts line 23

Quoted 6 lines
arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
@@ -18,6 +18,23 @@
 /dts-v1/;
 #include "socfpga_arria10_socdk.dtsi"
 
+/ {
+	chosen {
+		firmware-loader = &fs_loader0;

Should be a phandle.

Quoted 124 lines
arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +14 -0
+	};
+
+	fs_loader0: fs-loader at 0 {
+		u-boot,dm-pre-reloc;
+		compatible = "u-boot,fs-loader";
+		phandlepart = <&mmc 1>;
+	};
+};
+
+&fpga_mgr {
+	u-boot,dm-pre-reloc;
+	altr,bitstream = "fit_spl_fpga.itb";
+};
+
 &mmc {
 	u-boot,dm-pre-reloc;
 	status = "okay";
arch/arm/mach-socfpga/include/mach/fpga_manager_arria10.h +37 -2
index 09d13f6..5ef15bb 100644
@@ -1,9 +1,13 @@
 /* SPDX-License-Identifier: GPL-2.0 */
 /*
- * Copyright (C) 2017 Intel Corporation <www.intel.com>
+ * Copyright (C) 2017-2019 Intel Corporation <www.intel.com>
  * All rights reserved.
  */
 
+#include <asm/cache.h>
+#include <altera.h>
+#include <image.h>
+
 #ifndef _FPGA_MANAGER_ARRIA10_H_
 #define _FPGA_MANAGER_ARRIA10_H_
 
@@ -51,6 +55,10 @@
 #define ALT_FPGAMGR_IMGCFG_CTL_02_CFGWIDTH_SET_MSK		BIT(24)
 #define ALT_FPGAMGR_IMGCFG_CTL_02_CDRATIO_LSB			16
 
+#define FPGA_SOCFPGA_A10_RBF_UNENCRYPTED	0xa65c
+#define FPGA_SOCFPGA_A10_RBF_ENCRYPTED		0xa65d
+#define FPGA_SOCFPGA_A10_RBF_PERIPH		0x0001
+#define FPGA_SOCFPGA_A10_RBF_CORE		0x8001
 #ifndef __ASSEMBLY__
 
 struct socfpga_fpga_manager {
@@ -88,12 +96,39 @@ struct socfpga_fpga_manager {
 	u32  imgcfg_fifo_status;
 };
 
+enum rbf_type {
+	unknown,
+	periph_section,
+	core_section
+};
+
+enum rbf_security {
+	invalid,
+	unencrypted,
+	encrypted
+};
+
+struct rbf_info {
+	enum rbf_type section;
+	enum rbf_security security;
+};
+
+struct fpga_loadfs_info {
+	fpga_fs_info *fpga_fsinfo;
+	u32 remaining;
+	u32 offset;
+	struct rbf_info rbfinfo;
+};
+
 /* Functions */
 int fpgamgr_program_init(u32 * rbf_data, size_t rbf_size);
 int fpgamgr_program_finish(void);
 int is_fpgamgr_user_mode(void);
 int fpgamgr_wait_early_user_mode(void);
-
+int is_fpgamgr_early_user_mode(void);
+char *get_fpga_filename(const void *fdt, int *len);
+int socfpga_loadfs(fpga_fs_info *fpga_fsinfo, const void *buf, size_t bsize,
+		  u32 offset);
 #endif /* __ASSEMBLY__ */
 
 #endif /* _FPGA_MANAGER_ARRIA10_H_ */
drivers/fpga/socfpga_arria10.c +5 -3
index 114dd91..630d5a3 100644
@@ -1,8 +1,7 @@
 // SPDX-License-Identifier: GPL-2.0
 /*
- * Copyright (C) 2017 Intel Corporation <www.intel.com>
+ * Copyright (C) 2017-2019 Intel Corporation <www.intel.com>
  */
-
 #include <asm/io.h>
 #include <asm/arch/fpga_manager.h>
 #include <asm/arch/reset_manager.h>
@@ -10,8 +9,11 @@
 #include <asm/arch/sdram.h>
 #include <asm/arch/misc.h>
 #include <altera.h>
+#include <asm/arch/pinmux.h>
 #include <common.h>
+#include <dm/ofnode.h>
 #include <errno.h>
+#include <fs_loader.h>
 #include <wait_bit.h>
 #include <watchdog.h>
 
@@ -64,7 +66,7 @@ static int wait_for_user_mode(void)
 		1, FPGA_TIMEOUT_MSEC, false);
 }
 
-static int is_fpgamgr_early_user_mode(void)
+int is_fpgamgr_early_user_mode(void)
 {
 	return (readl(&fpga_manager_base->imgcfg_stat) &
 		ALT_FPGAMGR_IMGCFG_STAT_F2S_EARLY_USERMODE_SET_MSK) != 0;

drivers/fpga/socfpga_arria10.c int fpgamgr_wait_early_user_mode(void) · line 97

Quoted 4 lines
drivers/fpga/socfpga_arria10.c +1 -0
@@ -92,9 +94,10 @@ int fpgamgr_wait_early_user_mode(void)
 				sizeof(sync_data));
 		udelay(FPGA_TIMEOUT_MSEC);
 		i++;
+		WATCHDOG_RESET();

udelay() already triggers watchdog, why is this needed here ?

Quoted 22 lines
drivers/fpga/socfpga_arria10.c +5 -4
 	}
 
-	debug("Additional %i sync word needed\n", i);
+	debug("FPGA: Additional %i sync word needed\n", i);
 
 	/* restoring original CDRATIO */
 	fpgamgr_set_cd_ratio(cd_ratio);
@@ -172,9 +175,10 @@ static int fpgamgr_set_cdratio_cdwidth(unsigned int cfg_width, u32 *rbf_data,
 	compress = (rbf_data[COMPRESSION_OFFSET] >> 1) & 1;
 	compress = !compress;
 
-	debug("header word %d = %08x\n", 69, rbf_data[69]);
-	debug("header word %d = %08x\n", 229, rbf_data[229]);
-	debug("read from rbf header: encrypt=%d compress=%d\n", encrypt, compress);
+	debug("FPGA: Header word %d = %08x.\n", 69, rbf_data[69]);
+	debug("FPGA: Header word %d = %08x.\n", 229, rbf_data[229]);
+	debug("FPGA: Read from rbf header: encrypt=%d compress=%d.\n", encrypt,
+	     compress);
 
 	/*
 	 * from the register map description of cdratio in imgcfg_ctrl_02:

drivers/fpga/socfpga_arria10.c static int fpgamgr_program_poll_cd(void) · line 366

Quoted 4 lines
drivers/fpga/socfpga_arria10.c +1 -0
@@ -359,6 +363,7 @@ static int fpgamgr_program_poll_cd(void)
 			printf("nstatus == 0 while waiting for condone\n");
 			return -EPERM;
 		}
+		WATCHDOG_RESET();

Why is this needed ?

Quoted 25 lines
drivers/fpga/socfpga_arria10.c +6 -5
 	}
 
 	if (i == FPGA_TIMEOUT_CNT)
@@ -432,7 +437,6 @@ int fpgamgr_program_finish(void)
 		printf("FPGA: Poll CD failed with error code %d\n", status);
 		return -EPERM;
 	}
-	WATCHDOG_RESET();
 
 	/* Ensure the FPGA entering user mode */
 	status = fpgamgr_program_poll_usermode();
@@ -447,27 +451,448 @@ int fpgamgr_program_finish(void)
 	return 0;
 }
 
-/*
- * FPGA Manager to program the FPGA. This is the interface used by FPGA driver.
- * Return 0 for sucess, non-zero for error.
- */
+char *get_fpga_filename(const void *fdt, int *len)
+{
+	char *fpga_filename = NULL;
+	int node_offset;
+
+	fdtdec_find_aliases_for_id(gd->fdt_blob, "fpga_mgr",

drivers/fpga/socfpga_arria10.c int fpgamgr_program_finish(void) · line 469

Quoted 10 lines
drivers/fpga/socfpga_arria10.c +10 -0
+				COMPAT_ALTERA_SOCFPGA_FPGA0,
+				&node_offset, 1);
+
+	ofnode fpgamgr_node = offset_to_ofnode(node_offset);
+
+	if (ofnode_valid(fpgamgr_node))
+		fpga_filename = (char *)ofnode_read_string(fpgamgr_node,
+						"altr,bitstream");
+
+

Why is the cast needed ? Drop the two newlines.

Quoted 201 lines
drivers/fpga/socfpga_arria10.c +201 -0
+	return fpga_filename;
+}
+
+static void get_rbf_image_info(struct rbf_info *rbf, u16 *buffer)
+{
+	/*
+	 * Magic ID starting at:
+	 * -> 1st dword[15:0] in periph.rbf
+	 * -> 2nd dword[15:0] in core.rbf
+	 * Note: dword == 32 bits
+	 */
+	u32 word_reading_max = 2;
+	u32 i;
+
+	for (i = 0; i < word_reading_max; i++) {
+		if (*(buffer + i) == FPGA_SOCFPGA_A10_RBF_UNENCRYPTED) {
+			rbf->security = unencrypted;
+		} else if (*(buffer + i) == FPGA_SOCFPGA_A10_RBF_ENCRYPTED) {
+			rbf->security = encrypted;
+		} else if (*(buffer + i + 1) ==
+				FPGA_SOCFPGA_A10_RBF_UNENCRYPTED) {
+			rbf->security = unencrypted;
+		} else if (*(buffer + i + 1) ==
+				FPGA_SOCFPGA_A10_RBF_ENCRYPTED) {
+			rbf->security = encrypted;
+		} else {
+			rbf->security = invalid;
+			continue;
+		}
+
+		/* PERIPH RBF(buffer + i + 1), CORE RBF(buffer + i + 2) */
+		if (*(buffer + i + 1) == FPGA_SOCFPGA_A10_RBF_PERIPH) {
+			rbf->section = periph_section;
+			break;
+		} else if (*(buffer + i + 1) == FPGA_SOCFPGA_A10_RBF_CORE) {
+			rbf->section = core_section;
+			break;
+		} else if (*(buffer + i + 2) == FPGA_SOCFPGA_A10_RBF_PERIPH) {
+			rbf->section = periph_section;
+			break;
+		} else if (*(buffer + i + 2) == FPGA_SOCFPGA_A10_RBF_CORE) {
+			rbf->section = core_section;
+			break;
+		}
+
+		rbf->section = unknown;
+		break;
+
+		WATCHDOG_RESET();
+	}
+}
+
+#ifdef CONFIG_FS_LOADER
+static int first_loading_rbf_to_buffer(struct udevice *dev,
+				struct fpga_loadfs_info *fpga_loadfs,
+				u32 *buffer, size_t *buffer_bsize)
+{
+	u32 *buffer_p = (u32 *)*buffer;
+	u32 *loadable = buffer_p;
+	size_t buffer_size = *buffer_bsize;
+	size_t fit_size;
+	int ret, i, count;
+	int confs_noffset, images_noffset;
+	int rbf_offset;
+	int rbf_size;
+	const char *fpga_node_name = NULL;
+	const char *uname = NULL;
+
+	/* Load image header into buffer */
+	ret = request_firmware_into_buf(dev,
+					fpga_loadfs->fpga_fsinfo->filename,
+					buffer_p,
+					sizeof(struct image_header),
+					0);
+	if (ret < 0) {
+		debug("FPGA: Failed to read image header from flash.\n");
+		return -ENOENT;
+	}
+
+	if (image_get_magic((struct image_header *)buffer_p) != FDT_MAGIC) {
+		debug("FPGA: No FDT magic was found.\n");
+		return -EBADF;
+	}
+
+	fit_size = fdt_totalsize(buffer_p);
+
+	if (fit_size > buffer_size) {
+		debug("FPGA: FIT image is larger than available buffer.\n");
+		debug("Please use FIT external data or increasing buffer.\n");
+		return -ENOMEM;
+	}
+
+	/* Load entire FIT into buffer */
+	ret = request_firmware_into_buf(dev,
+					fpga_loadfs->fpga_fsinfo->filename,
+					buffer_p,
+					fit_size,
+					0);
+
+	if (ret < 0)
+		return ret;
+
+	ret = fit_check_format(buffer_p);
+	if (!ret) {
+		debug("FPGA: No valid FIT image was found.\n");
+		return -EBADF;
+	}
+
+	confs_noffset = fdt_path_offset(buffer_p, FIT_CONFS_PATH);
+	images_noffset = fdt_path_offset(buffer_p, FIT_IMAGES_PATH);
+	if (confs_noffset < 0 || images_noffset < 0) {
+		debug("FPGA: No Configurations or images nodes were found.\n");
+		return -ENOENT;
+	}
+
+	/* Get default configuration unit name from default property */
+	confs_noffset = fit_conf_get_node(buffer_p, NULL);
+	if (confs_noffset < 0) {
+		debug("FPGA: No default configuration was found in config.\n");
+		return -ENOENT;
+	}
+
+	count = fit_conf_get_prop_node_count(buffer_p, confs_noffset,
+					    FIT_FPGA_PROP);
+
+	if (count < 0) {
+		debug("FPGA: Invalid configuration format for FPGA node.\n");
+		return count;
+	} else {
+		debug("FPGA: FPGA node count: %d\n", count);
+	}
+
+	for (i = 0; i < count; i++) {
+		images_noffset = fit_conf_get_prop_node_index(buffer_p,
+							     confs_noffset,
+							     FIT_FPGA_PROP, i);
+		uname = fit_get_name(buffer_p, images_noffset, NULL);
+		if (uname) {
+			debug("FPGA: %s\n", uname);
+
+			if (strstr(uname, "fpga-periph") &&
+				(!is_fpgamgr_early_user_mode() ||
+				is_fpgamgr_user_mode())) {
+				fpga_node_name = uname;
+				printf("FPGA: Start to program ");
+				printf("peripheral/full bitstream ...\n");
+				break;
+			} else if (strstr(uname, "fpga-core") &&
+					(is_fpgamgr_early_user_mode() &&
+					!is_fpgamgr_user_mode())) {
+				fpga_node_name = uname;
+				printf("FPGA: Start to program core ");
+				printf("bitstream ...\n");
+				break;
+			}
+		}
+		WATCHDOG_RESET();
+	}
+
+	if (!fpga_node_name) {
+		debug("FPGA: No suitable bitstream was found, count: %d.\n", i);
+		return 1;
+	}
+
+	images_noffset = fit_image_get_node(buffer_p, fpga_node_name);
+	if (images_noffset < 0) {
+		debug("FPGA: No node '%s' was found in FIT.\n",
+		     fpga_node_name);
+		return -ENOENT;
+	}
+
+	ret = fit_image_get_data_position(buffer_p, images_noffset,
+					 &rbf_offset);
+	if (ret < 0) {
+		debug("FPGA: No data position was found (err=%d).\n", ret);
+		return -ENOENT;
+	}
+
+	ret = fit_image_get_data_size(buffer_p, images_noffset, &rbf_size);
+	if (ret < 0) {
+		debug("FPGA: No data size was found (err=%d).\n", ret);
+		return -ENOENT;
+	}
+
+	ret = fit_image_get_load(buffer_p, images_noffset, (ulong *)loadable);
+	if (ret < 0) {
+		debug("FPGA: No loadable was found (err=%d).\n", ret);
+		debug("FPGA: Using default buffer and size.\n");
+	} else {
+		buffer_p = (u32 *)*loadable;
+		buffer_size = rbf_size;
+		debug("FPGA: Found loadable address = 0x%x.\n", *loadable);
+	}
+
+	debug("FPGA: External data: offset = 0x%x, size = 0x%x.\n",
+	      rbf_offset, rbf_size);
+
+	fpga_loadfs->remaining = rbf_size;
+
+	/*
+	 * Determine buffer size vs bitstream size, and calculating number of

drivers/fpga/socfpga_arria10.c int fpgamgr_program_finish(void) · line 680

Quoted 10 lines
drivers/fpga/socfpga_arria10.c +10 -0
+	 * chunk by chunk transfer is required due to smaller buffer size
+	 * compare to bitstream
+	 */
+	if (rbf_size <= buffer_size) {
+		/* Loading whole bitstream into buffer */
+		buffer_size = rbf_size;
+		fpga_loadfs->remaining = 0;
+	} else {
+		fpga_loadfs->remaining -= buffer_size;
+	}

Shouldn't all this parsing and calculation be done by the firmware loader code ?

Quoted 55 lines
drivers/fpga/socfpga_arria10.c +55 -0
+	fpga_loadfs->offset = rbf_offset;
+	/* Loading bitstream into buffer */
+	ret = request_firmware_into_buf(dev,
+					fpga_loadfs->fpga_fsinfo->filename,
+					buffer_p,
+					buffer_size,
+					fpga_loadfs->offset);
+	if (ret < 0) {
+		debug("FPGA: Failed to read bitstream from flash.\n");
+		return -ENOENT;
+	}
+
+	/* Getting info about bitstream types */
+	get_rbf_image_info(&fpga_loadfs->rbfinfo, (u16 *)buffer_p);
+
+	/* Update next reading bitstream offset */
+	fpga_loadfs->offset += buffer_size;
+
+	/* Update the final addr for bitstream */
+	*buffer = (u32)buffer_p;
+
+	/* Update the size of bitstream to be programmed into FPGA */
+	*buffer_bsize = buffer_size;
+
+	return 0;
+}
+
+static int subsequent_loading_rbf_to_buffer(struct udevice *dev,
+					struct fpga_loadfs_info *fpga_loadfs,
+					u32 *buffer, size_t *buffer_bsize)
+{
+	int ret = 0;
+	u32 *buffer_p = (u32 *)*buffer;
+
+	/* Read the bitstream chunk by chunk. */
+	if (fpga_loadfs->remaining > *buffer_bsize) {
+		fpga_loadfs->remaining -= *buffer_bsize;
+	} else {
+		*buffer_bsize = fpga_loadfs->remaining;
+		fpga_loadfs->remaining = 0;
+	}
+
+	ret = request_firmware_into_buf(dev,
+					fpga_loadfs->fpga_fsinfo->filename,
+					buffer_p,
+					*buffer_bsize,
+					fpga_loadfs->offset);
+	if (ret < 0) {
+		debug("FPGA: Failed to read bitstream from flash.\n");
+		return -ENOENT;
+	}
+
+	/* Update next reading bitstream offset */
+	fpga_loadfs->offset += *buffer_bsize;
+

drivers/fpga/socfpga_arria10.c int fpgamgr_program_finish(void) · line 746

Quoted 10 lines
drivers/fpga/socfpga_arria10.c +10 -0
+	return 0;
+}
+
+int socfpga_loadfs(fpga_fs_info *fpga_fsinfo, const void *buf, size_t bsize,
+			u32 offset)
+{
+	struct fpga_loadfs_info fpga_loadfs;
+	int status = 0;
+	int ret = 0;
+	u32 buffer = (u32)buf;

This will fail on arm64 , look at uintptr_t .

drivers/fpga/socfpga_arria10.c int fpgamgr_program_finish(void) · line 752

Quoted 6 lines
drivers/fpga/socfpga_arria10.c +6 -0
+	size_t buffer_sizebytes = bsize;
+	size_t buffer_sizebytes_ori = bsize;
+	size_t total_sizeof_image = 0;
+	struct udevice *dev;
+
+	ret = uclass_get_device(UCLASS_FS_FIRMWARE_LOADER, 0, &dev);

Shouldn't the firmware loaded instance be obtained via the DT phandle ?

Quoted 156 lines
drivers/fpga/socfpga_arria10.c +135 -9
+	if (ret)
+		return ret;
+
+	memset(&fpga_loadfs, 0, sizeof(fpga_loadfs));
+
+	fpga_loadfs.fpga_fsinfo = fpga_fsinfo;
+	fpga_loadfs.offset = offset;
+
+	printf("FPGA: Checking FPGA configuration setting ...\n");
+
+	/*
+	 * Note: Both buffer and buffer_sizebytes values can be altered by
+	 * function below.
+	 */
+	ret = first_loading_rbf_to_buffer(dev, &fpga_loadfs, &buffer,
+					   &buffer_sizebytes);
+	if (ret == 1) {
+		printf("FPGA: Skipping configuration ...\n");
+		return 0;
+	} else if (ret) {
+		return ret;
+	}
+
+	if (fpga_loadfs.rbfinfo.section == core_section &&
+		!(is_fpgamgr_early_user_mode() && !is_fpgamgr_user_mode())) {
+		debug("FPGA : Must be in Early Release mode to program ");
+		debug("core bitstream.\n");
+		return 0;
+	}
+
+	/* Disable all signals from HPS peripheral controller to FPGA */
+	writel(0, &system_manager_base->fpgaintf_en_global);
+
+	/* Disable all axi bridges (hps2fpga, lwhps2fpga & fpga2hps) */
+	socfpga_bridges_reset();
+
+	if (fpga_loadfs.rbfinfo.section == periph_section) {
+		/* Initialize the FPGA Manager */
+		status = fpgamgr_program_init((u32 *)buffer, buffer_sizebytes);
+		if (status) {
+			debug("FPGA: Init with peripheral bitstream failed.\n");
+			return -EPERM;
+		}
+	}
+
+	/* Transfer bitstream to FPGA Manager */
+	fpgamgr_program_write((void *)buffer, buffer_sizebytes);
+
+	total_sizeof_image += buffer_sizebytes;
+
+	while (fpga_loadfs.remaining) {
+		ret = subsequent_loading_rbf_to_buffer(dev,
+							&fpga_loadfs,
+							&buffer,
+							&buffer_sizebytes_ori);
+
+		if (ret)
+			return ret;
+
+		/* Transfer data to FPGA Manager */
+		fpgamgr_program_write((void *)buffer,
+					buffer_sizebytes_ori);
+
+		total_sizeof_image += buffer_sizebytes_ori;
+
+		WATCHDOG_RESET();
+	}
+
+	if (fpga_loadfs.rbfinfo.section == periph_section) {
+		if (fpgamgr_wait_early_user_mode() != -ETIMEDOUT) {
+			config_pins(gd->fdt_blob, "shared");
+			puts("FPGA: Early Release Succeeded.\n");
+		} else {
+			debug("FPGA: Failed to see Early Release.\n");
+			return -EIO;
+		}
+
+		/* For monolithic bitstream */
+		if (is_fpgamgr_user_mode()) {
+			/* Ensure the FPGA entering config done */
+			status = fpgamgr_program_finish();
+			if (status)
+				return status;
+
+			config_pins(gd->fdt_blob, "fpga");
+			puts("FPGA: Enter user mode.\n");
+		}
+	} else if (fpga_loadfs.rbfinfo.section == core_section) {
+		/* Ensure the FPGA entering config done */
+		status = fpgamgr_program_finish();
+		if (status)
+			return status;
+
+		config_pins(gd->fdt_blob, "fpga");
+		puts("FPGA: Enter user mode.\n");
+	} else {
+		debug("FPGA: Config Error: Unsupported bitstream type.\n");
+		return -ENOEXEC;
+	}
+
+	return (int)total_sizeof_image;
+}
+#endif
+
+/* This function is used to load the core bitstream from the OCRAM. */
 int socfpga_load(Altera_desc *desc, const void *rbf_data, size_t rbf_size)
 {
-	int status;
+	unsigned long status;
+	struct rbf_info rbfinfo;
 
-	/* disable all signals from hps peripheral controller to fpga */
+	memset(&rbfinfo, 0, sizeof(rbfinfo));
+
+	/* Disable all signals from hps peripheral controller to fpga */
 	writel(0, &system_manager_base->fpgaintf_en_global);
 
-	/* disable all axi bridge (hps2fpga, lwhps2fpga & fpga2hps) */
+	/* Disable all axi bridge (hps2fpga, lwhps2fpga & fpga2hps) */
 	socfpga_bridges_reset();
 
-	/* Initialize the FPGA Manager */
-	status = fpgamgr_program_init((u32 *)rbf_data, rbf_size);
-	if (status)
-		return status;
+	/* Getting info about bitstream types */
+	get_rbf_image_info(&rbfinfo, (u16 *)rbf_data);
+
+	if (rbfinfo.section == periph_section) {
+		/* Initialize the FPGA Manager */
+		status = fpgamgr_program_init((u32 *)rbf_data, rbf_size);
+		if (status)
+			return status;
+	}
+
+	if (rbfinfo.section == core_section &&
+		!(is_fpgamgr_early_user_mode() && !is_fpgamgr_user_mode())) {
+		debug("FPGA : Must be in early release mode to program ");
+		debug("core bitstream.\n");
+		return 0;
+	}
 
-	/* Write the RBF data to FPGA Manager */
+	/* Write the bitstream to FPGA Manager */
 	fpgamgr_program_write(rbf_data, rbf_size);
 
-	return fpgamgr_program_finish();
+	status = fpgamgr_program_finish();
+	if (status) {
+		config_pins(gd->fdt_blob, "fpga");
+		puts("FPGA: Enter user mode.\n");
+	}
+
+	return status;
 }
Signature · 2 lines
Best regards,
Marek Vasut
#28 lore raw

PATCH v8 4/8 ARM: socfpga: Add FPGA drivers for Arria 10 FPGA bitstream loading

Quoted 35 lines · Marek Vasut
Quoted 34 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add FPGA driver to support program FPGA with FPGA bitstream loading from
filesystem. The driver are designed based on generic firmware loader
framework. The driver can handle FPGA program operation from loading FPGA
bitstream in flash to memory and then to program FPGA.

---

changes for v8
- Added codes to discern bitstream type based on fpga node name.

changes for v7
- Restructure the FPGA driver to support both peripheral bitstream and core
  bitstream bundled into FIT image.
- Support loadable property for core bitstream. User can set loadable
  in DDR for better performance. This loading would be done in one large
  chunk instead of chunk by chunk loading with small memory buffer.
---
 arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts       |  17 + .../include/mach/fpga_manager_arria10.h            |  39 +- drivers/fpga/socfpga_arria10.c                     | 467
++++++++++++++++++++-

 3 files changed, 500 insertions(+), 23 deletions(-)
Quoted 13 lines

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts line 23

Quoted 12 lines
arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +0 -0
index 998d811..14f1967 100644
@@ -18,6 +18,23 @@
  /dts-v1/;
  #include "socfpga_arria10_socdk.dtsi"

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+/ {
+	chosen {
+		firmware-loader = &fs_loader0;

Should be a phandle.

Can we change this label to phandle stage by stage, may be after fpga driver? This requires time working on firmware loader.

Quoted 130 lines
Quoted 130 lines
arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +14 -0
+	};
+
+	fs_loader0: fs-loader at 0 {
+		u-boot,dm-pre-reloc;
+		compatible = "u-boot,fs-loader";
+		phandlepart = <&mmc 1>;
+	};
+};
+
+&fpga_mgr {
+	u-boot,dm-pre-reloc;
+	altr,bitstream = "fit_spl_fpga.itb";
+};
+

 &mmc {
  u-boot,dm-pre-reloc;
  status = "okay";

(quoted diff) +0 -0

socfpga/include/mach/fpga_manager_arria10.h b/arch/arm/mach- socfpga/include/mach/fpga_manager_arria10.h
index 09d13f6..5ef15bb 100644

arch/arm/mach-socfpga/include/mach/fpga_manager_arria10.h +1 -1
@@ -1,9 +1,13 @@
  /* SPDX-License-Identifier: GPL-2.0 */
  /*
- * Copyright (C) 2017 Intel Corporation <www.intel.com>
+ * Copyright (C) 2017-2019 Intel Corporation <www.intel.com>
   * All rights reserved.
   */

arch/arm/mach-socfpga/include/mach/fpga_manager_arria10.h +4 -0
+#include <asm/cache.h>
+#include <altera.h>
+#include <image.h>
+

 #ifndef _FPGA_MANAGER_ARRIA10_H_
 #define _FPGA_MANAGER_ARRIA10_H_

arch/arm/mach-socfpga/include/mach/fpga_manager_arria10.h +0 -0
@@ -51,6 +55,10 @@
  #define ALT_FPGAMGR_IMGCFG_CTL_02_CFGWIDTH_SET_MSK		
 BIT(24)
  #define ALT_FPGAMGR_IMGCFG_CTL_02_CDRATIO_LSB			

16

arch/arm/mach-socfpga/include/mach/fpga_manager_arria10.h +4 -0
+#define FPGA_SOCFPGA_A10_RBF_UNENCRYPTED	0xa65c
+#define FPGA_SOCFPGA_A10_RBF_ENCRYPTED		0xa65d
+#define FPGA_SOCFPGA_A10_RBF_PERIPH		0x0001
+#define FPGA_SOCFPGA_A10_RBF_CORE		0x8001

 #ifndef __ASSEMBLY__

 struct socfpga_fpga_manager {

arch/arm/mach-socfpga/include/mach/fpga_manager_arria10.h +0 -0
@@ -88,12 +96,39 @@ struct socfpga_fpga_manager {
  	u32  imgcfg_fifo_status;
  };

arch/arm/mach-socfpga/include/mach/fpga_manager_arria10.h +24 -0
+enum rbf_type {
+	unknown,
+	periph_section,
+	core_section
+};
+
+enum rbf_security {
+	invalid,
+	unencrypted,
+	encrypted
+};
+
+struct rbf_info {
+	enum rbf_type section;
+	enum rbf_security security;
+};
+
+struct fpga_loadfs_info {
+	fpga_fs_info *fpga_fsinfo;
+	u32 remaining;
+	u32 offset;
+	struct rbf_info rbfinfo;
+};
+

 /* Functions */
 int fpgamgr_program_init(u32 * rbf_data, size_t rbf_size); int fpgamgr_program_finish(void);
 int is_fpgamgr_user_mode(void);
 int fpgamgr_wait_early_user_mode(void);

arch/arm/mach-socfpga/include/mach/fpga_manager_arria10.h +3 -1
-
+int is_fpgamgr_early_user_mode(void);
+char *get_fpga_filename(const void *fdt, int *len);
+int socfpga_loadfs(fpga_fs_info *fpga_fsinfo, const void *buf,

size_t bsize,

arch/arm/mach-socfpga/include/mach/fpga_manager_arria10.h +1 -0
+		  u32 offset);

 #endif /* __ASSEMBLY__ */

 #endif /* _FPGA_MANAGER_ARRIA10_H_ */

drivers/fpga/socfpga_arria10.c +4 -2
index 114dd91..630d5a3 100644
@@ -1,8 +1,7 @@
  // SPDX-License-Identifier: GPL-2.0
  /*
- * Copyright (C) 2017 Intel Corporation <www.intel.com>
+ * Copyright (C) 2017-2019 Intel Corporation <www.intel.com>
   */
-
  #include <asm/io.h>
  #include <asm/arch/fpga_manager.h>
  #include <asm/arch/reset_manager.h>
@@ -10,8 +9,11 @@
  #include <asm/arch/sdram.h>
  #include <asm/arch/misc.h>
  #include <altera.h>
+#include <asm/arch/pinmux.h>
  #include <common.h>
+#include <dm/ofnode.h>
  #include <errno.h>
+#include <fs_loader.h>
  #include <wait_bit.h>
  #include <watchdog.h>

drivers/fpga/socfpga_arria10.c +0 -0
@@ -64,7 +66,7 @@ static int wait_for_user_mode(void)
  		1, FPGA_TIMEOUT_MSEC, false);
  }

drivers/fpga/socfpga_arria10.c +1 -1
-static int is_fpgamgr_early_user_mode(void)
+int is_fpgamgr_early_user_mode(void)

 {
  return (readl(&fpga_manager_base->imgcfg_stat) &

Quoted 8 lines

drivers/fpga/socfpga_arria10.c int fpgamgr_wait_early_user_mode(void) · line 97

Quoted 7 lines

  ALT_FPGAMGR_IMGCFG_STAT_F2S_EARLY_USERMODE_SET_MSK ) != 0;

drivers/fpga/socfpga_arria10.c +1 -0
@@ -92,9 +94,10 @@ int fpgamgr_wait_early_user_mode(void)
  				sizeof(sync_data));
  		udelay(FPGA_TIMEOUT_MSEC);
  		i++;
+		WATCHDOG_RESET();

udelay() already triggers watchdog, why is this needed here ?

Then i can remove it.

Quoted 27 lines
Quoted 27 lines
drivers/fpga/socfpga_arria10.c +0 -0
  	}

drivers/fpga/socfpga_arria10.c +1 -1
-	debug("Additional %i sync word needed\n", i);
+	debug("FPGA: Additional %i sync word needed\n", i);


  /* restoring original CDRATIO */
  fpgamgr_set_cd_ratio(cd_ratio);

drivers/fpga/socfpga_arria10.c +0 -0
@@ -172,9 +175,10 @@ static int
 fpgamgr_set_cdratio_cdwidth(unsigned int cfg_width, u32 *rbf_data,
  	compress = (rbf_data[COMPRESSION_OFFSET] >> 1) & 1;
  	compress = !compress;

drivers/fpga/socfpga_arria10.c +0 -3
-	debug("header word %d = %08x\n", 69, rbf_data[69]);
-	debug("header word %d = %08x\n", 229, rbf_data[229]);
-	debug("read from rbf header: encrypt=%d compress=%d\n",

encrypt, compress);

drivers/fpga/socfpga_arria10.c +2 -0
+	debug("FPGA: Header word %d = %08x.\n", 69, rbf_data[69]);
+	debug("FPGA: Header word %d = %08x.\n", 229,

rbf_data[229]);

drivers/fpga/socfpga_arria10.c +1 -0
+	debug("FPGA: Read from rbf header: encrypt=%d

compress=%d.\n", encrypt,

drivers/fpga/socfpga_arria10.c +1 -0
+	     compress);


  /*
   * from the register map description of cdratio in

Quoted 8 lines

drivers/fpga/socfpga_arria10.c

Quoted 7 lines

imgcfg_ctrl_02:

drivers/fpga/socfpga_arria10.c +0 -0
@@ -359,6 +363,7 @@ static int fpgamgr_program_poll_cd(void)

  printf("nstatus == 0nwhile waiting for condone\");
  return -EPERM;
  }

drivers/fpga/socfpga_arria10.c +1 -0
+		WATCHDOG_RESET();

Why is this needed ?

This is polling of FPGA configuring done status. The polling could be long enough to reset the watchdog timely, hence we need this in the polling.

Quoted 33 lines
Quoted 33 lines

  }

  if (i == FPGA_TIMEOUT_CNT)

drivers/fpga/socfpga_arria10.c +0 -0
@@ -432,7 +437,6 @@ int fpgamgr_program_finish(void)

  printf("FPGA: Poll CD failed with error code %d\n", status);
  return -EPERM;
  }

drivers/fpga/socfpga_arria10.c +0 -1
-	WATCHDOG_RESET();


  /* Ensure the FPGA entering user mode */
  status = fpgamgr_program_poll_usermode();

drivers/fpga/socfpga_arria10.c +0 -0
@@ -447,27 +451,448 @@ int fpgamgr_program_finish(void)
  	return 0;
  }


-/*
- * FPGA Manager to program the FPGA. This is the interface used by FPGA driver.
- * Return 0 for sucess, non-zero for error.
- */

drivers/fpga/socfpga_arria10.c +10 -0
+char *get_fpga_filename(const void *fdt, int *len)
+{
+	char *fpga_filename = NULL;
+	int node_offset;
+
+	fdtdec_find_aliases_for_id(gd->fdt_blob, "fpga_mgr",
+				COMPAT_ALTERA_SOCFPGA_FPGA0,
+				&node_offset, 1);
+
+	ofnode fpgamgr_node = offset_to_ofnode(node_offset);
Quoted 8 lines

drivers/fpga/socfpga_arria10.c int fpgamgr_program_finish(void) · line 469

Quoted 7 lines
drivers/fpga/socfpga_arria10.c +3 -0
+
+	if (ofnode_valid(fpgamgr_node))
+		fpga_filename = (char

*)ofnode_read_string(fpgamgr_node,

drivers/fpga/socfpga_arria10.c +3 -0
+						"altr,bitstream");
+
+

Why is the cast needed ?

The return string would be eventually set to the char *filename in common struct fpga_fsinfo. So, the cast here is to avoid the warning from compiler.

Quoted 1 line

Drop the two newlines.

Okay.

Quoted 249 lines
Quoted 249 lines
drivers/fpga/socfpga_arria10.c +16 -0
+	return fpga_filename;
+}
+
+static void get_rbf_image_info(struct rbf_info *rbf, u16 *buffer)
+{
+	/*
+	 * Magic ID starting at:
+	 * -> 1st dword[15:0] in periph.rbf
+	 * -> 2nd dword[15:0] in core.rbf
+	 * Note: dword == 32 bits
+	 */
+	u32 word_reading_max = 2;
+	u32 i;
+
+	for (i = 0; i < word_reading_max; i++) {
+		if (*(buffer + i) ==

FPGA_SOCFPGA_A10_RBF_UNENCRYPTED) {

drivers/fpga/socfpga_arria10.c +2 -0
+			rbf->security = unencrypted;
+		} else if (*(buffer + i) ==

FPGA_SOCFPGA_A10_RBF_ENCRYPTED) {

drivers/fpga/socfpga_arria10.c +3 -0
+			rbf->security = encrypted;
+		} else if (*(buffer + i + 1) ==
+				FPGA_SOCFPGA_A10_RBF_UNENCRYPTED)

{

drivers/fpga/socfpga_arria10.c +12 -0
+			rbf->security = unencrypted;
+		} else if (*(buffer + i + 1) ==
+				FPGA_SOCFPGA_A10_RBF_ENCRYPTED) {
+			rbf->security = encrypted;
+		} else {
+			rbf->security = invalid;
+			continue;
+		}
+
+		/* PERIPH RBF(buffer + i + 1), CORE RBF(buffer + i
+ 2) */
+		if (*(buffer + i + 1) ==

FPGA_SOCFPGA_A10_RBF_PERIPH) {

drivers/fpga/socfpga_arria10.c +3 -0
+			rbf->section = periph_section;
+			break;
+		} else if (*(buffer + i + 1) ==

FPGA_SOCFPGA_A10_RBF_CORE) {

drivers/fpga/socfpga_arria10.c +3 -0
+			rbf->section = core_section;
+			break;
+		} else if (*(buffer + i + 2) ==

FPGA_SOCFPGA_A10_RBF_PERIPH) {

drivers/fpga/socfpga_arria10.c +3 -0
+			rbf->section = periph_section;
+			break;
+		} else if (*(buffer + i + 2) ==

FPGA_SOCFPGA_A10_RBF_CORE) {

drivers/fpga/socfpga_arria10.c +14 -0
+			rbf->section = core_section;
+			break;
+		}
+
+		rbf->section = unknown;
+		break;
+
+		WATCHDOG_RESET();
+	}
+}
+
+#ifdef CONFIG_FS_LOADER
+static int first_loading_rbf_to_buffer(struct udevice *dev,
+				struct fpga_loadfs_info

*fpga_loadfs,

drivers/fpga/socfpga_arria10.c +16 -0
+				u32 *buffer, size_t *buffer_bsize)
+{
+	u32 *buffer_p = (u32 *)*buffer;
+	u32 *loadable = buffer_p;
+	size_t buffer_size = *buffer_bsize;
+	size_t fit_size;
+	int ret, i, count;
+	int confs_noffset, images_noffset;
+	int rbf_offset;
+	int rbf_size;
+	const char *fpga_node_name = NULL;
+	const char *uname = NULL;
+
+	/* Load image header into buffer */
+	ret = request_firmware_into_buf(dev,
+					fpga_loadfs->fpga_fsinfo-
Quoted 1 line

filename,

drivers/fpga/socfpga_arria10.c +2 -0
+					buffer_p,
+					sizeof(struct

image_header),

drivers/fpga/socfpga_arria10.c +3 -0
+					0);
+	if (ret < 0) {
+		debug("FPGA: Failed to read image header from

flash.\n");

drivers/fpga/socfpga_arria10.c +4 -0
+		return -ENOENT;
+	}
+
+	if (image_get_magic((struct image_header *)buffer_p) !=

FDT_MAGIC) {

drivers/fpga/socfpga_arria10.c +8 -0
+		debug("FPGA: No FDT magic was found.\n");
+		return -EBADF;
+	}
+
+	fit_size = fdt_totalsize(buffer_p);
+
+	if (fit_size > buffer_size) {
+		debug("FPGA: FIT image is larger than available

buffer.\n");

drivers/fpga/socfpga_arria10.c +1 -0
+		debug("Please use FIT external data or increasing

buffer.\n");

drivers/fpga/socfpga_arria10.c +6 -0
+		return -ENOMEM;
+	}
+
+	/* Load entire FIT into buffer */
+	ret = request_firmware_into_buf(dev,
+					fpga_loadfs->fpga_fsinfo-
Quoted 1 line

filename,

drivers/fpga/socfpga_arria10.c +15 -0
+					buffer_p,
+					fit_size,
+					0);
+
+	if (ret < 0)
+		return ret;
+
+	ret = fit_check_format(buffer_p);
+	if (!ret) {
+		debug("FPGA: No valid FIT image was found.\n");
+		return -EBADF;
+	}
+
+	confs_noffset = fdt_path_offset(buffer_p, FIT_CONFS_PATH);
+	images_noffset = fdt_path_offset(buffer_p,

FIT_IMAGES_PATH);

drivers/fpga/socfpga_arria10.c +2 -0
+	if (confs_noffset < 0 || images_noffset < 0) {
+		debug("FPGA: No Configurations or images nodes

were found.\n");

drivers/fpga/socfpga_arria10.c +4 -0
+		return -ENOENT;
+	}
+
+	/* Get default configuration unit name from default

property */

drivers/fpga/socfpga_arria10.c +3 -0
+	confs_noffset = fit_conf_get_node(buffer_p, NULL);
+	if (confs_noffset < 0) {
+		debug("FPGA: No default configuration was found in

config.\n");

drivers/fpga/socfpga_arria10.c +4 -0
+		return -ENOENT;
+	}
+
+	count = fit_conf_get_prop_node_count(buffer_p,

confs_noffset,

drivers/fpga/socfpga_arria10.c +4 -0
+					    FIT_FPGA_PROP);
+
+	if (count < 0) {
+		debug("FPGA: Invalid configuration format for FPGA

node.\n");

drivers/fpga/socfpga_arria10.c +7 -0
+		return count;
+	} else {
+		debug("FPGA: FPGA node count: %d\n", count);
+	}
+
+	for (i = 0; i < count; i++) {
+		images_noffset =

fit_conf_get_prop_node_index(buffer_p,

drivers/fpga/socfpga_arria10.c +1 -0
+							     confs

_noffset,

drivers/fpga/socfpga_arria10.c +1 -0
+							     FIT_F

PGA_PROP, i);

drivers/fpga/socfpga_arria10.c +1 -0
+		uname = fit_get_name(buffer_p, images_noffset,

NULL);

drivers/fpga/socfpga_arria10.c +9 -0
+		if (uname) {
+			debug("FPGA: %s\n", uname);
+
+			if (strstr(uname, "fpga-periph") &&
+				(!is_fpgamgr_early_user_mode() ||
+				is_fpgamgr_user_mode())) {
+				fpga_node_name = uname;
+				printf("FPGA: Start to program ");
+				printf("peripheral/full bitstream

...\n");

drivers/fpga/socfpga_arria10.c +3 -0
+				break;
+			} else if (strstr(uname, "fpga-core") &&
+					(is_fpgamgr_early_user_mod

e() &&

drivers/fpga/socfpga_arria10.c +1 -0
+					!is_fpgamgr_user_mode()))

{

drivers/fpga/socfpga_arria10.c +2 -0
+				fpga_node_name = uname;
+				printf("FPGA: Start to program

core ");

drivers/fpga/socfpga_arria10.c +9 -0
+				printf("bitstream ...\n");
+				break;
+			}
+		}
+		WATCHDOG_RESET();
+	}
+
+	if (!fpga_node_name) {
+		debug("FPGA: No suitable bitstream was found,

count: %d.\n", i);

drivers/fpga/socfpga_arria10.c +4 -0
+		return 1;
+	}
+
+	images_noffset = fit_image_get_node(buffer_p,

fpga_node_name);

drivers/fpga/socfpga_arria10.c +7 -0
+	if (images_noffset < 0) {
+		debug("FPGA: No node '%s' was found in FIT.\n",
+		     fpga_node_name);
+		return -ENOENT;
+	}
+
+	ret = fit_image_get_data_position(buffer_p,

images_noffset,

drivers/fpga/socfpga_arria10.c +3 -0
+					 &rbf_offset);
+	if (ret < 0) {
+		debug("FPGA: No data position was found

(err=%d).\n", ret);

drivers/fpga/socfpga_arria10.c +4 -0
+		return -ENOENT;
+	}
+
+	ret = fit_image_get_data_size(buffer_p, images_noffset,

&rbf_size);

drivers/fpga/socfpga_arria10.c +2 -0
+	if (ret < 0) {
+		debug("FPGA: No data size was found (err=%d).\n",

ret);

drivers/fpga/socfpga_arria10.c +4 -0
+		return -ENOENT;
+	}
+
+	ret = fit_image_get_load(buffer_p, images_noffset, (ulong

*)loadable);

drivers/fpga/socfpga_arria10.c +2 -0
+	if (ret < 0) {
+		debug("FPGA: No loadable was found (err=%d).\n",

ret);

drivers/fpga/socfpga_arria10.c +5 -0
+		debug("FPGA: Using default buffer and size.\n");
+	} else {
+		buffer_p = (u32 *)*loadable;
+		buffer_size = rbf_size;
+		debug("FPGA: Found loadable address = 0x%x.\n",

*loadable);

drivers/fpga/socfpga_arria10.c +3 -0
+	}
+
+	debug("FPGA: External data: offset = 0x%x, size =

0x%x.\n",

drivers/fpga/socfpga_arria10.c +6 -0
+	      rbf_offset, rbf_size);
+
+	fpga_loadfs->remaining = rbf_size;
+
+	/*
+	 * Determine buffer size vs bitstream size, and

calculating number of

drivers/fpga/socfpga_arria10.c +1 -0
+	 * chunk by chunk transfer is required due to smaller

buffer size

drivers/fpga/socfpga_arria10.c +3 -0
+	 * compare to bitstream
+	 */
+	if (rbf_size <= buffer_size) {
Quoted 8 lines

drivers/fpga/socfpga_arria10.c int fpgamgr_program_finish(void) · line 680

Quoted 6 lines
drivers/fpga/socfpga_arria10.c +6 -0
+		/* Loading whole bitstream into buffer */
+		buffer_size = rbf_size;
+		fpga_loadfs->remaining = 0;
+	} else {
+		fpga_loadfs->remaining -= buffer_size;
+	}

Shouldn't all this parsing and calculation be done by the firmware loader code ?

The calculation here is to determine the available memory size can be used, it could be size from OCRAM buffer, or DDR.

Quoted 69 lines
Quoted 69 lines
drivers/fpga/socfpga_arria10.c +4 -0
+	fpga_loadfs->offset = rbf_offset;
+	/* Loading bitstream into buffer */
+	ret = request_firmware_into_buf(dev,
+					fpga_loadfs->fpga_fsinfo-
Quoted 1 line

filename,

drivers/fpga/socfpga_arria10.c +5 -0
+					buffer_p,
+					buffer_size,
+					fpga_loadfs->offset);
+	if (ret < 0) {
+		debug("FPGA: Failed to read bitstream from

flash.\n");

drivers/fpga/socfpga_arria10.c +5 -0
+		return -ENOENT;
+	}
+
+	/* Getting info about bitstream types */
+	get_rbf_image_info(&fpga_loadfs->rbfinfo, (u16

*)buffer_p);

drivers/fpga/socfpga_arria10.c +8 -0
+
+	/* Update next reading bitstream offset */
+	fpga_loadfs->offset += buffer_size;
+
+	/* Update the final addr for bitstream */
+	*buffer = (u32)buffer_p;
+
+	/* Update the size of bitstream to be programmed into FPGA

*/

drivers/fpga/socfpga_arria10.c +7 -0
+	*buffer_bsize = buffer_size;
+
+	return 0;
+}
+
+static int subsequent_loading_rbf_to_buffer(struct udevice *dev,
+					struct fpga_loadfs_info

*fpga_loadfs,

drivers/fpga/socfpga_arria10.c +1 -0
+					u32 *buffer, size_t

*buffer_bsize)

drivers/fpga/socfpga_arria10.c +14 -0
+{
+	int ret = 0;
+	u32 *buffer_p = (u32 *)*buffer;
+
+	/* Read the bitstream chunk by chunk. */
+	if (fpga_loadfs->remaining > *buffer_bsize) {
+		fpga_loadfs->remaining -= *buffer_bsize;
+	} else {
+		*buffer_bsize = fpga_loadfs->remaining;
+		fpga_loadfs->remaining = 0;
+	}
+
+	ret = request_firmware_into_buf(dev,
+					fpga_loadfs->fpga_fsinfo-
Quoted 1 line

filename,

drivers/fpga/socfpga_arria10.c +5 -0
+					buffer_p,
+					*buffer_bsize,
+					fpga_loadfs->offset);
+	if (ret < 0) {
+		debug("FPGA: Failed to read bitstream from

flash.\n");

drivers/fpga/socfpga_arria10.c +10 -0
+		return -ENOENT;
+	}
+
+	/* Update next reading bitstream offset */
+	fpga_loadfs->offset += *buffer_bsize;
+
+	return 0;
+}
+
+int socfpga_loadfs(fpga_fs_info *fpga_fsinfo, const void *buf,
Quoted 8 lines

drivers/fpga/socfpga_arria10.c int fpgamgr_program_finish(void) · line 746

Quoted 7 lines

size_t bsize,

drivers/fpga/socfpga_arria10.c +6 -0
+			u32 offset)
+{
+	struct fpga_loadfs_info fpga_loadfs;
+	int status = 0;
+	int ret = 0;
+	u32 buffer = (u32)buf;

This will fail on arm64 , look at uintptr_t .

This driver is only used by A10 which is arm32. You want me to use (u32)(uintptr_t)buf?

Quoted 11 lines
Quoted 9 lines
drivers/fpga/socfpga_arria10.c +6 -0
+	size_t buffer_sizebytes = bsize;
+	size_t buffer_sizebytes_ori = bsize;
+	size_t total_sizeof_image = 0;
+	struct udevice *dev;
+
+	ret = uclass_get_device(UCLASS_FS_FIRMWARE_LOADER, 0,

&dev);

Shouldn't the firmware loaded instance be obtained via the DT phandle ?

It just to get the device activated. The firmware loaded itself would go to choosen node(default) for getting the label. I can change to the phandle, may be after this?

Quoted 181 lines
Quoted 180 lines
drivers/fpga/socfpga_arria10.c +12 -0
+	if (ret)
+		return ret;
+
+	memset(&fpga_loadfs, 0, sizeof(fpga_loadfs));
+
+	fpga_loadfs.fpga_fsinfo = fpga_fsinfo;
+	fpga_loadfs.offset = offset;
+
+	printf("FPGA: Checking FPGA configuration setting ...\n");
+
+	/*
+	 * Note: Both buffer and buffer_sizebytes values can be

altered by

drivers/fpga/socfpga_arria10.c +3 -0
+	 * function below.
+	 */
+	ret = first_loading_rbf_to_buffer(dev, &fpga_loadfs,

&buffer,

drivers/fpga/socfpga_arria10.c +10 -0
+					   &buffer_sizebytes);
+	if (ret == 1) {
+		printf("FPGA: Skipping configuration ...\n");
+		return 0;
+	} else if (ret) {
+		return ret;
+	}
+
+	if (fpga_loadfs.rbfinfo.section == core_section &&
+		!(is_fpgamgr_early_user_mode() &&

!is_fpgamgr_user_mode())) {

drivers/fpga/socfpga_arria10.c +1 -0
+		debug("FPGA : Must be in Early Release mode to

program ");

drivers/fpga/socfpga_arria10.c +5 -0
+		debug("core bitstream.\n");
+		return 0;
+	}
+
+	/* Disable all signals from HPS peripheral controller to

FPGA */

drivers/fpga/socfpga_arria10.c +3 -0
+	writel(0, &system_manager_base->fpgaintf_en_global);
+
+	/* Disable all axi bridges (hps2fpga, lwhps2fpga &

fpga2hps) */

drivers/fpga/socfpga_arria10.c +5 -0
+	socfpga_bridges_reset();
+
+	if (fpga_loadfs.rbfinfo.section == periph_section) {
+		/* Initialize the FPGA Manager */
+		status = fpgamgr_program_init((u32 *)buffer,

buffer_sizebytes);

drivers/fpga/socfpga_arria10.c +2 -0
+		if (status) {
+			debug("FPGA: Init with peripheral

bitstream failed.\n");

drivers/fpga/socfpga_arria10.c +12 -0
+			return -EPERM;
+		}
+	}
+
+	/* Transfer bitstream to FPGA Manager */
+	fpgamgr_program_write((void *)buffer, buffer_sizebytes);
+
+	total_sizeof_image += buffer_sizebytes;
+
+	while (fpga_loadfs.remaining) {
+		ret = subsequent_loading_rbf_to_buffer(dev,
+							&fpga_load

fs,

drivers/fpga/socfpga_arria10.c +2 -0
+							&buffer,
+							&buffer_si

zebytes_ori);

drivers/fpga/socfpga_arria10.c +15 -0
+
+		if (ret)
+			return ret;
+
+		/* Transfer data to FPGA Manager */
+		fpgamgr_program_write((void *)buffer,
+					buffer_sizebytes_ori);
+
+		total_sizeof_image += buffer_sizebytes_ori;
+
+		WATCHDOG_RESET();
+	}
+
+	if (fpga_loadfs.rbfinfo.section == periph_section) {
+		if (fpgamgr_wait_early_user_mode() != -ETIMEDOUT)

{

drivers/fpga/socfpga_arria10.c +4 -0
+			config_pins(gd->fdt_blob, "shared");
+			puts("FPGA: Early Release Succeeded.\n");
+		} else {
+			debug("FPGA: Failed to see Early

Release.\n");

drivers/fpga/socfpga_arria10.c +23 -0
+			return -EIO;
+		}
+
+		/* For monolithic bitstream */
+		if (is_fpgamgr_user_mode()) {
+			/* Ensure the FPGA entering config done */
+			status = fpgamgr_program_finish();
+			if (status)
+				return status;
+
+			config_pins(gd->fdt_blob, "fpga");
+			puts("FPGA: Enter user mode.\n");
+		}
+	} else if (fpga_loadfs.rbfinfo.section == core_section) {
+		/* Ensure the FPGA entering config done */
+		status = fpgamgr_program_finish();
+		if (status)
+			return status;
+
+		config_pins(gd->fdt_blob, "fpga");
+		puts("FPGA: Enter user mode.\n");
+	} else {
+		debug("FPGA: Config Error: Unsupported bitstream

type.\n");

drivers/fpga/socfpga_arria10.c +8 -0
+		return -ENOEXEC;
+	}
+
+	return (int)total_sizeof_image;
+}
+#endif
+
+/* This function is used to load the core bitstream from the

OCRAM. */
 int socfpga_load(Altera_desc *desc, const void *rbf_data, size_t rbf_size)
 {

drivers/fpga/socfpga_arria10.c +2 -1
-	int status;
+	unsigned long status;
+	struct rbf_info rbfinfo;

drivers/fpga/socfpga_arria10.c +0 -1
-	/* disable all signals from hps peripheral controller to

fpga */

drivers/fpga/socfpga_arria10.c +3 -0
+	memset(&rbfinfo, 0, sizeof(rbfinfo));
+
+	/* Disable all signals from hps peripheral controller to

fpga */
  writel(0, &system_manager_base->fpgaintf_en_global);

drivers/fpga/socfpga_arria10.c +0 -1
-	/* disable all axi bridge (hps2fpga, lwhps2fpga &

fpga2hps) */

drivers/fpga/socfpga_arria10.c +1 -0
+	/* Disable all axi bridge (hps2fpga, lwhps2fpga &

fpga2hps) */
  socfpga_bridges_reset();

drivers/fpga/socfpga_arria10.c +6 -4
-	/* Initialize the FPGA Manager */
-	status = fpgamgr_program_init((u32 *)rbf_data, rbf_size);
-	if (status)
-		return status;
+	/* Getting info about bitstream types */
+	get_rbf_image_info(&rbfinfo, (u16 *)rbf_data);
+
+	if (rbfinfo.section == periph_section) {
+		/* Initialize the FPGA Manager */
+		status = fpgamgr_program_init((u32 *)rbf_data,

rbf_size);

drivers/fpga/socfpga_arria10.c +6 -0
+		if (status)
+			return status;
+	}
+
+	if (rbfinfo.section == core_section &&
+		!(is_fpgamgr_early_user_mode() &&

!is_fpgamgr_user_mode())) {

drivers/fpga/socfpga_arria10.c +1 -0
+		debug("FPGA : Must be in early release mode to

program ");

drivers/fpga/socfpga_arria10.c +3 -0
+		debug("core bitstream.\n");
+		return 0;
+	}

drivers/fpga/socfpga_arria10.c +1 -1
-	/* Write the RBF data to FPGA Manager */
+	/* Write the bitstream to FPGA Manager */

  fpgamgr_program_write(rbf_data, rbf_size);

drivers/fpga/socfpga_arria10.c +7 -1
-	return fpgamgr_program_finish();
+	status = fpgamgr_program_finish();
+	if (status) {
+		config_pins(gd->fdt_blob, "fpga");
+		puts("FPGA: Enter user mode.\n");
+	}
+
+	return status;

 }

#29 lore raw

PATCH v8 4/8 ARM: socfpga: Add FPGA drivers for Arria 10 FPGA bitstream loading

Quoted 49 lines · Tien Fong
Quoted 35 lines · Marek Vasut
Quoted 34 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add FPGA driver to support program FPGA with FPGA bitstream loading from
filesystem. The driver are designed based on generic firmware loader
framework. The driver can handle FPGA program operation from loading FPGA
bitstream in flash to memory and then to program FPGA.

---

changes for v8
- Added codes to discern bitstream type based on fpga node name.

changes for v7
- Restructure the FPGA driver to support both peripheral bitstream and core
  bitstream bundled into FIT image.
- Support loadable property for core bitstream. User can set loadable
  in DDR for better performance. This loading would be done in one large
  chunk instead of chunk by chunk loading with small memory buffer.
---
 arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts       |  17 + .../include/mach/fpga_manager_arria10.h            |  39 +- drivers/fpga/socfpga_arria10.c                     | 467
++++++++++++++++++++-

 3 files changed, 500 insertions(+), 23 deletions(-)
Quoted 13 lines

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts line 23

Quoted 12 lines
arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +0 -0
index 998d811..14f1967 100644
@@ -18,6 +18,23 @@
  /dts-v1/;
  #include "socfpga_arria10_socdk.dtsi"

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+/ {
+	chosen {
+		firmware-loader = &fs_loader0;

Should be a phandle.

Quoted 2 lines

Can we change this label to phandle stage by stage, may be after fpga driver? This requires time working on firmware loader.

We should fix this as soon as possible, otherwise people might find this bad example and wonder why it doesn't work once this is changed.

Quoted 19 lines
Quoted 19 lines
Quoted 19 lines
arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +14 -0
+	};
+
+	fs_loader0: fs-loader at 0 {
+		u-boot,dm-pre-reloc;
+		compatible = "u-boot,fs-loader";
+		phandlepart = <&mmc 1>;
+	};
+};
+
+&fpga_mgr {
+	u-boot,dm-pre-reloc;
+	altr,bitstream = "fit_spl_fpga.itb";
+};
+

 &mmc {
  u-boot,dm-pre-reloc;
  status = "okay";

[...]

Quoted 22 lines
Quoted 14 lines
Quoted 14 lines

- * FPGA Manager to program the FPGA. This is the interface used by FPGA driver.
- * Return 0 for sucess, non-zero for error.
- */

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +10 -0
+char *get_fpga_filename(const void *fdt, int *len)
+{
+	char *fpga_filename = NULL;
+	int node_offset;
+
+	fdtdec_find_aliases_for_id(gd->fdt_blob, "fpga_mgr",
+				COMPAT_ALTERA_SOCFPGA_FPGA0,
+				&node_offset, 1);
+
+	ofnode fpgamgr_node = offset_to_ofnode(node_offset);
Quoted 8 lines

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts

Quoted 7 lines
arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+
+	if (ofnode_valid(fpgamgr_node))
+		fpga_filename = (char

*)ofnode_read_string(fpgamgr_node,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+						"altr,bitstream");
+
+

Why is the cast needed ?

Quoted 3 lines

The return string would be eventually set to the char *filename in common struct fpga_fsinfo. So, the cast here is to avoid the warning from compiler.

I presume if the compiler generates a warning, it's for a reason. What warning is that ?

Quoted 259 lines
Quoted 1 line

Drop the two newlines.

Okay.

Quoted 249 lines
Quoted 249 lines
arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +16 -0
+	return fpga_filename;
+}
+
+static void get_rbf_image_info(struct rbf_info *rbf, u16 *buffer)
+{
+	/*
+	 * Magic ID starting at:
+	 * -> 1st dword[15:0] in periph.rbf
+	 * -> 2nd dword[15:0] in core.rbf
+	 * Note: dword == 32 bits
+	 */
+	u32 word_reading_max = 2;
+	u32 i;
+
+	for (i = 0; i < word_reading_max; i++) {
+		if (*(buffer + i) ==

FPGA_SOCFPGA_A10_RBF_UNENCRYPTED) {

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +2 -0
+			rbf->security = unencrypted;
+		} else if (*(buffer + i) ==

FPGA_SOCFPGA_A10_RBF_ENCRYPTED) {

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+			rbf->security = encrypted;
+		} else if (*(buffer + i + 1) ==
+				FPGA_SOCFPGA_A10_RBF_UNENCRYPTED)

{

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +12 -0
+			rbf->security = unencrypted;
+		} else if (*(buffer + i + 1) ==
+				FPGA_SOCFPGA_A10_RBF_ENCRYPTED) {
+			rbf->security = encrypted;
+		} else {
+			rbf->security = invalid;
+			continue;
+		}
+
+		/* PERIPH RBF(buffer + i + 1), CORE RBF(buffer + i
+ 2) */
+		if (*(buffer + i + 1) ==

FPGA_SOCFPGA_A10_RBF_PERIPH) {

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+			rbf->section = periph_section;
+			break;
+		} else if (*(buffer + i + 1) ==

FPGA_SOCFPGA_A10_RBF_CORE) {

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+			rbf->section = core_section;
+			break;
+		} else if (*(buffer + i + 2) ==

FPGA_SOCFPGA_A10_RBF_PERIPH) {

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+			rbf->section = periph_section;
+			break;
+		} else if (*(buffer + i + 2) ==

FPGA_SOCFPGA_A10_RBF_CORE) {

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +14 -0
+			rbf->section = core_section;
+			break;
+		}
+
+		rbf->section = unknown;
+		break;
+
+		WATCHDOG_RESET();
+	}
+}
+
+#ifdef CONFIG_FS_LOADER
+static int first_loading_rbf_to_buffer(struct udevice *dev,
+				struct fpga_loadfs_info

*fpga_loadfs,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +16 -0
+				u32 *buffer, size_t *buffer_bsize)
+{
+	u32 *buffer_p = (u32 *)*buffer;
+	u32 *loadable = buffer_p;
+	size_t buffer_size = *buffer_bsize;
+	size_t fit_size;
+	int ret, i, count;
+	int confs_noffset, images_noffset;
+	int rbf_offset;
+	int rbf_size;
+	const char *fpga_node_name = NULL;
+	const char *uname = NULL;
+
+	/* Load image header into buffer */
+	ret = request_firmware_into_buf(dev,
+					fpga_loadfs->fpga_fsinfo-
Quoted 1 line

filename,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +2 -0
+					buffer_p,
+					sizeof(struct

image_header),

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+					0);
+	if (ret < 0) {
+		debug("FPGA: Failed to read image header from

flash.\n");

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +4 -0
+		return -ENOENT;
+	}
+
+	if (image_get_magic((struct image_header *)buffer_p) !=

FDT_MAGIC) {

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +8 -0
+		debug("FPGA: No FDT magic was found.\n");
+		return -EBADF;
+	}
+
+	fit_size = fdt_totalsize(buffer_p);
+
+	if (fit_size > buffer_size) {
+		debug("FPGA: FIT image is larger than available

buffer.\n");

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+		debug("Please use FIT external data or increasing

buffer.\n");

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +6 -0
+		return -ENOMEM;
+	}
+
+	/* Load entire FIT into buffer */
+	ret = request_firmware_into_buf(dev,
+					fpga_loadfs->fpga_fsinfo-
Quoted 1 line

filename,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +15 -0
+					buffer_p,
+					fit_size,
+					0);
+
+	if (ret < 0)
+		return ret;
+
+	ret = fit_check_format(buffer_p);
+	if (!ret) {
+		debug("FPGA: No valid FIT image was found.\n");
+		return -EBADF;
+	}
+
+	confs_noffset = fdt_path_offset(buffer_p, FIT_CONFS_PATH);
+	images_noffset = fdt_path_offset(buffer_p,

FIT_IMAGES_PATH);

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +2 -0
+	if (confs_noffset < 0 || images_noffset < 0) {
+		debug("FPGA: No Configurations or images nodes

were found.\n");

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +4 -0
+		return -ENOENT;
+	}
+
+	/* Get default configuration unit name from default

property */

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+	confs_noffset = fit_conf_get_node(buffer_p, NULL);
+	if (confs_noffset < 0) {
+		debug("FPGA: No default configuration was found in

config.\n");

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +4 -0
+		return -ENOENT;
+	}
+
+	count = fit_conf_get_prop_node_count(buffer_p,

confs_noffset,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +4 -0
+					    FIT_FPGA_PROP);
+
+	if (count < 0) {
+		debug("FPGA: Invalid configuration format for FPGA

node.\n");

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +7 -0
+		return count;
+	} else {
+		debug("FPGA: FPGA node count: %d\n", count);
+	}
+
+	for (i = 0; i < count; i++) {
+		images_noffset =

fit_conf_get_prop_node_index(buffer_p,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+							     confs

_noffset,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+							     FIT_F

PGA_PROP, i);

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+		uname = fit_get_name(buffer_p, images_noffset,

NULL);

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +9 -0
+		if (uname) {
+			debug("FPGA: %s\n", uname);
+
+			if (strstr(uname, "fpga-periph") &&
+				(!is_fpgamgr_early_user_mode() ||
+				is_fpgamgr_user_mode())) {
+				fpga_node_name = uname;
+				printf("FPGA: Start to program ");
+				printf("peripheral/full bitstream

...\n");

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+				break;
+			} else if (strstr(uname, "fpga-core") &&
+					(is_fpgamgr_early_user_mod

e() &&

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+					!is_fpgamgr_user_mode()))

{

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +2 -0
+				fpga_node_name = uname;
+				printf("FPGA: Start to program

core ");

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +9 -0
+				printf("bitstream ...\n");
+				break;
+			}
+		}
+		WATCHDOG_RESET();
+	}
+
+	if (!fpga_node_name) {
+		debug("FPGA: No suitable bitstream was found,

count: %d.\n", i);

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +4 -0
+		return 1;
+	}
+
+	images_noffset = fit_image_get_node(buffer_p,

fpga_node_name);

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +7 -0
+	if (images_noffset < 0) {
+		debug("FPGA: No node '%s' was found in FIT.\n",
+		     fpga_node_name);
+		return -ENOENT;
+	}
+
+	ret = fit_image_get_data_position(buffer_p,

images_noffset,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+					 &rbf_offset);
+	if (ret < 0) {
+		debug("FPGA: No data position was found

(err=%d).\n", ret);

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +4 -0
+		return -ENOENT;
+	}
+
+	ret = fit_image_get_data_size(buffer_p, images_noffset,

&rbf_size);

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +2 -0
+	if (ret < 0) {
+		debug("FPGA: No data size was found (err=%d).\n",

ret);

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +4 -0
+		return -ENOENT;
+	}
+
+	ret = fit_image_get_load(buffer_p, images_noffset, (ulong

*)loadable);

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +2 -0
+	if (ret < 0) {
+		debug("FPGA: No loadable was found (err=%d).\n",

ret);

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +5 -0
+		debug("FPGA: Using default buffer and size.\n");
+	} else {
+		buffer_p = (u32 *)*loadable;
+		buffer_size = rbf_size;
+		debug("FPGA: Found loadable address = 0x%x.\n",

*loadable);

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+	}
+
+	debug("FPGA: External data: offset = 0x%x, size =

0x%x.\n",

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +6 -0
+	      rbf_offset, rbf_size);
+
+	fpga_loadfs->remaining = rbf_size;
+
+	/*
+	 * Determine buffer size vs bitstream size, and

calculating number of

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+	 * chunk by chunk transfer is required due to smaller

buffer size

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+	 * compare to bitstream
+	 */
+	if (rbf_size <= buffer_size) {
Quoted 8 lines

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts

Quoted 6 lines
arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +6 -0
+		/* Loading whole bitstream into buffer */
+		buffer_size = rbf_size;
+		fpga_loadfs->remaining = 0;
+	} else {
+		fpga_loadfs->remaining -= buffer_size;
+	}

Shouldn't all this parsing and calculation be done by the firmware loader code ?

Quoted 2 lines

The calculation here is to determine the available memory size can be used, it could be size from OCRAM buffer, or DDR.

It seems rather that the code is parsing the fitImage structures ? How is that related to such calculations ?

Quoted 76 lines
Quoted 68 lines
Quoted 68 lines
arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +4 -0
+	fpga_loadfs->offset = rbf_offset;
+	/* Loading bitstream into buffer */
+	ret = request_firmware_into_buf(dev,
+					fpga_loadfs->fpga_fsinfo-
Quoted 1 line

filename,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +5 -0
+					buffer_p,
+					buffer_size,
+					fpga_loadfs->offset);
+	if (ret < 0) {
+		debug("FPGA: Failed to read bitstream from

flash.\n");

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +5 -0
+		return -ENOENT;
+	}
+
+	/* Getting info about bitstream types */
+	get_rbf_image_info(&fpga_loadfs->rbfinfo, (u16

*)buffer_p);

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +8 -0
+
+	/* Update next reading bitstream offset */
+	fpga_loadfs->offset += buffer_size;
+
+	/* Update the final addr for bitstream */
+	*buffer = (u32)buffer_p;
+
+	/* Update the size of bitstream to be programmed into FPGA

*/

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +7 -0
+	*buffer_bsize = buffer_size;
+
+	return 0;
+}
+
+static int subsequent_loading_rbf_to_buffer(struct udevice *dev,
+					struct fpga_loadfs_info

*fpga_loadfs,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+					u32 *buffer, size_t

*buffer_bsize)

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +14 -0
+{
+	int ret = 0;
+	u32 *buffer_p = (u32 *)*buffer;
+
+	/* Read the bitstream chunk by chunk. */
+	if (fpga_loadfs->remaining > *buffer_bsize) {
+		fpga_loadfs->remaining -= *buffer_bsize;
+	} else {
+		*buffer_bsize = fpga_loadfs->remaining;
+		fpga_loadfs->remaining = 0;
+	}
+
+	ret = request_firmware_into_buf(dev,
+					fpga_loadfs->fpga_fsinfo-
Quoted 1 line

filename,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +5 -0
+					buffer_p,
+					*buffer_bsize,
+					fpga_loadfs->offset);
+	if (ret < 0) {
+		debug("FPGA: Failed to read bitstream from

flash.\n");

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +10 -0
+		return -ENOENT;
+	}
+
+	/* Update next reading bitstream offset */
+	fpga_loadfs->offset += *buffer_bsize;
+
+	return 0;
+}
+
+int socfpga_loadfs(fpga_fs_info *fpga_fsinfo, const void *buf,
Quoted 8 lines

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts

Quoted 7 lines

size_t bsize,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +6 -0
+			u32 offset)
+{
+	struct fpga_loadfs_info fpga_loadfs;
+	int status = 0;
+	int ret = 0;
+	u32 buffer = (u32)buf;

This will fail on arm64 , look at uintptr_t .

Quoted 2 lines

This driver is only used by A10 which is arm32. You want me to use (u32)(uintptr_t)buf?

If you want to cast pointer to integer type, yes, that's uintptr_t .

Quoted 11 lines
Quoted 11 lines
Quoted 9 lines
arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +6 -0
+	size_t buffer_sizebytes = bsize;
+	size_t buffer_sizebytes_ori = bsize;
+	size_t total_sizeof_image = 0;
+	struct udevice *dev;
+
+	ret = uclass_get_device(UCLASS_FS_FIRMWARE_LOADER, 0,

&dev);

Shouldn't the firmware loaded instance be obtained via the DT phandle ?

Quoted 3 lines

It just to get the device activated. The firmware loaded itself would go to choosen node(default) for getting the label. I can change to the phandle, may be after this?

Except this always activates the first device in the list. [...]

Signature · 2 lines
Best regards,
Marek Vasut
#30 lore raw

PATCH v8 4/8 ARM: socfpga: Add FPGA drivers for Arria 10 FPGA bitstream loading

Quoted 58 lines · Marek Vasut
Quoted 57 lines · Tien Fong
Quoted 55 lines · Marek Vasut
Quoted 49 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add FPGA driver to support program FPGA with FPGA bitstream loading
from
filesystem. The driver are designed based on generic firmware loader
framework. The driver can handle FPGA program operation from loading FPGA
bitstream in flash to memory and then to program FPGA.

---

changes for v8
- Added codes to discern bitstream type based on fpga node name.

changes for v7
- Restructure the FPGA driver to support both peripheral bitstream
and core
  bitstream bundled into FIT image.
- Support loadable property for core bitstream. User can set loadable
  in DDR for better performance. This loading would be done in one
large
  chunk instead of chunk by chunk loading with small memory buffer.
---
 arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts       |  17 + .../include/mach/fpga_manager_arria10.h            |  39 +- drivers/fpga/socfpga_arria10.c                     | 467
++++++++++++++++++++-

 3 files changed, 500 insertions(+), 23 deletions(-)
arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +0 -0
index 998d811..14f1967 100644
@@ -18,6 +18,23 @@
  /dts-v1/;
  #include "socfpga_arria10_socdk.dtsi"

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts line 23

Quoted 3 lines
arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+/ {
+	chosen {
+		firmware-loader = &fs_loader0;

Should be a phandle.

Quoted 6 lines
Quoted 3 lines

Can we change this label to phandle stage by stage, may be after fpga
driver? This requires time working on firmware loader.

We should fix this as soon as possible, otherwise people might find this
bad example and wonder why it doesn't work once this is changed.

Okay.

Quoted 355 lines
Quoted 23 lines
Quoted 23 lines
Quoted 23 lines
arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +14 -0
+	};
+
+	fs_loader0: fs-loader at 0 {
+		u-boot,dm-pre-reloc;
+		compatible = "u-boot,fs-loader";
+		phandlepart = <&mmc 1>;
+	};
+};
+
+&fpga_mgr {
+	u-boot,dm-pre-reloc;
+	altr,bitstream = "fit_spl_fpga.itb";
+};
+

 &mmc {
  u-boot,dm-pre-reloc;
  status = "okay";

[...]

Quoted 27 lines
Quoted 19 lines
Quoted 19 lines

- * FPGA Manager to program the FPGA. This is the interface used by
FPGA driver.
- * Return 0 for sucess, non-zero for error.
- */

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +11 -0
+char *get_fpga_filename(const void *fdt, int *len)
+{
+	char *fpga_filename = NULL;
+	int node_offset;
+
+	fdtdec_find_aliases_for_id(gd->fdt_blob, "fpga_mgr",
+				COMPAT_ALTERA_SOCFPGA_FPGA0,
+				&node_offset, 1);
+
+	ofnode fpgamgr_node = offset_to_ofnode(node_offset);
+
Quoted 8 lines

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts

Quoted 7 lines
arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +2 -0
+	if (ofnode_valid(fpgamgr_node))
+		fpga_filename = (char

*)ofnode_read_string(fpgamgr_node,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+						"altr,bitstrea

m");

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +2 -0
+
+

Why is the cast needed ?

Quoted 4 lines

The return string would be eventually set to the char *filename in common struct fpga_fsinfo. So, the cast here is to avoid the warning
from compiler.

I presume if the compiler generates a warning, it's for a reason. What
warning is that ?

Quoted 295 lines
Quoted 3 lines

 Drop the two newlines.

Okay.

Quoted 291 lines
Quoted 279 lines
arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +4 -0
+	return fpga_filename;
+}
+
+static void get_rbf_image_info(struct rbf_info *rbf, u16

*buffer)

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +12 -0
+{
+	/*
+	 * Magic ID starting at:
+	 * -> 1st dword[15:0] in periph.rbf
+	 * -> 2nd dword[15:0] in core.rbf
+	 * Note: dword == 32 bits
+	 */
+	u32 word_reading_max = 2;
+	u32 i;
+
+	for (i = 0; i < word_reading_max; i++) {
+		if (*(buffer + i) ==

FPGA_SOCFPGA_A10_RBF_UNENCRYPTED) {

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +2 -0
+			rbf->security = unencrypted;
+		} else if (*(buffer + i) ==

FPGA_SOCFPGA_A10_RBF_ENCRYPTED) {

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+			rbf->security = encrypted;
+		} else if (*(buffer + i + 1) ==
+				FPGA_SOCFPGA_A10_RBF_UNENCRYPT

ED)
{

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+			rbf->security = unencrypted;
+		} else if (*(buffer + i + 1) ==
+				FPGA_SOCFPGA_A10_RBF_ENCRYPTED

) {

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +10 -0
+			rbf->security = encrypted;
+		} else {
+			rbf->security = invalid;
+			continue;
+		}
+
+		/* PERIPH RBF(buffer + i + 1), CORE RBF(buffer
+ i
+ 2) */
+		if (*(buffer + i + 1) ==

FPGA_SOCFPGA_A10_RBF_PERIPH) {

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+			rbf->section = periph_section;
+			break;
+		} else if (*(buffer + i + 1) ==

FPGA_SOCFPGA_A10_RBF_CORE) {

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+			rbf->section = core_section;
+			break;
+		} else if (*(buffer + i + 2) ==

FPGA_SOCFPGA_A10_RBF_PERIPH) {

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+			rbf->section = periph_section;
+			break;
+		} else if (*(buffer + i + 2) ==

FPGA_SOCFPGA_A10_RBF_CORE) {

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +14 -0
+			rbf->section = core_section;
+			break;
+		}
+
+		rbf->section = unknown;
+		break;
+
+		WATCHDOG_RESET();
+	}
+}
+
+#ifdef CONFIG_FS_LOADER
+static int first_loading_rbf_to_buffer(struct udevice *dev,
+				struct fpga_loadfs_info

*fpga_loadfs,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+				u32 *buffer, size_t

*buffer_bsize)

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +15 -0
+{
+	u32 *buffer_p = (u32 *)*buffer;
+	u32 *loadable = buffer_p;
+	size_t buffer_size = *buffer_bsize;
+	size_t fit_size;
+	int ret, i, count;
+	int confs_noffset, images_noffset;
+	int rbf_offset;
+	int rbf_size;
+	const char *fpga_node_name = NULL;
+	const char *uname = NULL;
+
+	/* Load image header into buffer */
+	ret = request_firmware_into_buf(dev,
+					fpga_loadfs-
Quoted 3 lines

fpga_fsinfo-

filename,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +2 -0
+					buffer_p,
+					sizeof(struct

image_header),

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+					0);
+	if (ret < 0) {
+		debug("FPGA: Failed to read image header from

flash.\n");

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +4 -0
+		return -ENOENT;
+	}
+
+	if (image_get_magic((struct image_header *)buffer_p)

!=
FDT_MAGIC) {

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +8 -0
+		debug("FPGA: No FDT magic was found.\n");
+		return -EBADF;
+	}
+
+	fit_size = fdt_totalsize(buffer_p);
+
+	if (fit_size > buffer_size) {
+		debug("FPGA: FIT image is larger than

available
buffer.\n");

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+		debug("Please use FIT external data or

increasing
buffer.\n");

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +6 -0
+		return -ENOMEM;
+	}
+
+	/* Load entire FIT into buffer */
+	ret = request_firmware_into_buf(dev,
+					fpga_loadfs-
Quoted 3 lines

fpga_fsinfo-

filename,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +10 -0
+					buffer_p,
+					fit_size,
+					0);
+
+	if (ret < 0)
+		return ret;
+
+	ret = fit_check_format(buffer_p);
+	if (!ret) {
+		debug("FPGA: No valid FIT image was

found.\n");

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +4 -0
+		return -EBADF;
+	}
+
+	confs_noffset = fdt_path_offset(buffer_p,

FIT_CONFS_PATH);

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+	images_noffset = fdt_path_offset(buffer_p,

FIT_IMAGES_PATH);

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +2 -0
+	if (confs_noffset < 0 || images_noffset < 0) {
+		debug("FPGA: No Configurations or images nodes

were found.\n");

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +4 -0
+		return -ENOENT;
+	}
+
+	/* Get default configuration unit name from default

property */

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+	confs_noffset = fit_conf_get_node(buffer_p, NULL);
+	if (confs_noffset < 0) {
+		debug("FPGA: No default configuration was

found in
config.\n");

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +4 -0
+		return -ENOENT;
+	}
+
+	count = fit_conf_get_prop_node_count(buffer_p,

confs_noffset,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +4 -0
+					    FIT_FPGA_PROP);
+
+	if (count < 0) {
+		debug("FPGA: Invalid configuration format for

FPGA
node.\n");

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +7 -0
+		return count;
+	} else {
+		debug("FPGA: FPGA node count: %d\n", count);
+	}
+
+	for (i = 0; i < count; i++) {
+		images_noffset =

fit_conf_get_prop_node_index(buffer_p,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+							     c

onfs
_noffset,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+							     F

IT_F
PGA_PROP, i);

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+		uname = fit_get_name(buffer_p, images_noffset,

NULL);

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +5 -0
+		if (uname) {
+			debug("FPGA: %s\n", uname);
+
+			if (strstr(uname, "fpga-periph") &&
+				(!is_fpgamgr_early_user_mode()

||

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+				is_fpgamgr_user_mode())) {
+				fpga_node_name = uname;
+				printf("FPGA: Start to program

");

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+				printf("peripheral/full

bitstream
...\n");

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +2 -0
+				break;
+			} else if (strstr(uname, "fpga-core")

&&

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+					(is_fpgamgr_early_user

_mod
e() &&

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+					!is_fpgamgr_user_mode(

)))
{

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +2 -0
+				fpga_node_name = uname;
+				printf("FPGA: Start to program

core ");

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +9 -0
+				printf("bitstream ...\n");
+				break;
+			}
+		}
+		WATCHDOG_RESET();
+	}
+
+	if (!fpga_node_name) {
+		debug("FPGA: No suitable bitstream was found,

count: %d.\n", i);

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +4 -0
+		return 1;
+	}
+
+	images_noffset = fit_image_get_node(buffer_p,

fpga_node_name);

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +2 -0
+	if (images_noffset < 0) {
+		debug("FPGA: No node '%s' was found in

FIT.\n",

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +5 -0
+		     fpga_node_name);
+		return -ENOENT;
+	}
+
+	ret = fit_image_get_data_position(buffer_p,

images_noffset,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+					 &rbf_offset);
+	if (ret < 0) {
+		debug("FPGA: No data position was found

(err=%d).\n", ret);

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +4 -0
+		return -ENOENT;
+	}
+
+	ret = fit_image_get_data_size(buffer_p,

images_noffset,
&rbf_size);

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +2 -0
+	if (ret < 0) {
+		debug("FPGA: No data size was found

(err=%d).\n",
ret);

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +4 -0
+		return -ENOENT;
+	}
+
+	ret = fit_image_get_load(buffer_p, images_noffset,

(ulong
*)loadable);

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +2 -0
+	if (ret < 0) {
+		debug("FPGA: No loadable was found

(err=%d).\n",
ret);

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+		debug("FPGA: Using default buffer and

size.\n");

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +4 -0
+	} else {
+		buffer_p = (u32 *)*loadable;
+		buffer_size = rbf_size;
+		debug("FPGA: Found loadable address =

0x%x.\n",
*loadable);

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+	}
+
+	debug("FPGA: External data: offset = 0x%x, size =

0x%x.\n",

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +6 -0
+	      rbf_offset, rbf_size);
+
+	fpga_loadfs->remaining = rbf_size;
+
+	/*
+	 * Determine buffer size vs bitstream size, and

calculating number of

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+	 * chunk by chunk transfer is required due to smaller

buffer size

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts

Quoted 9 lines
arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +9 -0
+	 * compare to bitstream
+	 */
+	if (rbf_size <= buffer_size) {
+		/* Loading whole bitstream into buffer */
+		buffer_size = rbf_size;
+		fpga_loadfs->remaining = 0;
+	} else {
+		fpga_loadfs->remaining -= buffer_size;
+	}

Shouldn't all this parsing and calculation be done by the firmware
loader code ?

Quoted 5 lines
Quoted 3 lines

The calculation here is to determine the available memory size can be
used, it could be size from OCRAM buffer, or DDR.

It seems rather that the code is parsing the fitImage structures ? How is that related to such calculations ?

The driver checking the load property of core image, if there is a destination address is defined, driver would use DDR memory instead of OCRAM buffer, hence driver need to update both buffer size and remaining when DDR is used, so that whole bitstream can be loaded to DDR instead of chunk by chunk loading.

Quoted 87 lines
Quoted 87 lines
Quoted 87 lines
Quoted 80 lines
arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +4 -0
+	fpga_loadfs->offset = rbf_offset;
+	/* Loading bitstream into buffer */
+	ret = request_firmware_into_buf(dev,
+					fpga_loadfs-
Quoted 3 lines

fpga_fsinfo-

filename,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +5 -0
+					buffer_p,
+					buffer_size,
+					fpga_loadfs->offset);
+	if (ret < 0) {
+		debug("FPGA: Failed to read bitstream from

flash.\n");

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +5 -0
+		return -ENOENT;
+	}
+
+	/* Getting info about bitstream types */
+	get_rbf_image_info(&fpga_loadfs->rbfinfo, (u16

*)buffer_p);

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +8 -0
+
+	/* Update next reading bitstream offset */
+	fpga_loadfs->offset += buffer_size;
+
+	/* Update the final addr for bitstream */
+	*buffer = (u32)buffer_p;
+
+	/* Update the size of bitstream to be programmed into

FPGA
*/

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +6 -0
+	*buffer_bsize = buffer_size;
+
+	return 0;
+}
+
+static int subsequent_loading_rbf_to_buffer(struct udevice

*dev,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+					struct

fpga_loadfs_info
*fpga_loadfs,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+					u32 *buffer, size_t

*buffer_bsize)

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +14 -0
+{
+	int ret = 0;
+	u32 *buffer_p = (u32 *)*buffer;
+
+	/* Read the bitstream chunk by chunk. */
+	if (fpga_loadfs->remaining > *buffer_bsize) {
+		fpga_loadfs->remaining -= *buffer_bsize;
+	} else {
+		*buffer_bsize = fpga_loadfs->remaining;
+		fpga_loadfs->remaining = 0;
+	}
+
+	ret = request_firmware_into_buf(dev,
+					fpga_loadfs-
Quoted 3 lines

fpga_fsinfo-

filename,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +5 -0
+					buffer_p,
+					*buffer_bsize,
+					fpga_loadfs->offset);
+	if (ret < 0) {
+		debug("FPGA: Failed to read bitstream from

flash.\n");

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +10 -0
+		return -ENOENT;
+	}
+
+	/* Update next reading bitstream offset */
+	fpga_loadfs->offset += *buffer_bsize;
+
+	return 0;
+}
+
+int socfpga_loadfs(fpga_fs_info *fpga_fsinfo, const void *buf,

size_t bsize,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts

Quoted 6 lines
arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +6 -0
+			u32 offset)
+{
+	struct fpga_loadfs_info fpga_loadfs;
+	int status = 0;
+	int ret = 0;
+	u32 buffer = (u32)buf;

This will fail on arm64 , look at uintptr_t .

Quoted 3 lines
Quoted 2 lines

This driver is only used by A10 which is arm32. You want me to use (u32)(uintptr_t)buf?

If you want to cast pointer to integer type, yes, that's uintptr_t .

Okay.

Quoted 16 lines
Quoted 16 lines
Quoted 16 lines
Quoted 12 lines
arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +6 -0
+	size_t buffer_sizebytes = bsize;
+	size_t buffer_sizebytes_ori = bsize;
+	size_t total_sizeof_image = 0;
+	struct udevice *dev;
+
+	ret = uclass_get_device(UCLASS_FS_FIRMWARE_LOADER, 0,
Quoted 1 line

&dev);

Shouldn't the firmware loaded instance be obtained via the DT phandle
?

Quoted 6 lines
Quoted 5 lines

It just to get the device activated. The firmware loaded itself would
go to choosen node(default) for getting the label. I can change to the
phandle, may be after this?

Except this always activates the first device in the list.

How about this after i change the label to phandle?

chosen_node = ofnode_path("/chosen");
phandle_p = ofnode_get_property(chosen_node, "firmware-loader", &size); phandle = fdt32_to_cpu(*phandle_p);
ret = uclass_get_device_by_phandle_id(UCLASS_FS_FIRMWARE_LOADER,

					     phandle, &dev);
Quoted 1 line

[...]

#31 lore raw

PATCH v8 4/8 ARM: socfpga: Add FPGA drivers for Arria 10 FPGA bitstream loading

In reply to Marek Vasut
Quoted 50 lines · Marek Vasut
Quoted 49 lines · Tien Fong
Quoted 47 lines · Marek Vasut
Quoted 45 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add FPGA driver to support program FPGA with FPGA bitstream loading
from
filesystem. The driver are designed based on generic firmware loader
framework. The driver can handle FPGA program operation from loading FPGA
bitstream in flash to memory and then to program FPGA.

---

changes for v8
- Added codes to discern bitstream type based on fpga node name.

changes for v7
- Restructure the FPGA driver to support both peripheral bitstream
and core
  bitstream bundled into FIT image.
- Support loadable property for core bitstream. User can set loadable
  in DDR for better performance. This loading would be done in one
large
  chunk instead of chunk by chunk loading with small memory buffer.
---
 arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts       |  17 + .../include/mach/fpga_manager_arria10.h            |  39 +- drivers/fpga/socfpga_arria10.c                     | 467
++++++++++++++++++++-

 3 files changed, 500 insertions(+), 23 deletions(-)
arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +0 -0
index 998d811..14f1967 100644

[...]

Quoted 24 lines
Quoted 24 lines
Quoted 24 lines
Quoted 21 lines

- * FPGA Manager to program the FPGA. This is the interface used by
FPGA driver.
- * Return 0 for sucess, non-zero for error.
- */

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +13 -0
+char *get_fpga_filename(const void *fdt, int *len)
+{
+	char *fpga_filename = NULL;
+	int node_offset;
+
+	fdtdec_find_aliases_for_id(gd->fdt_blob, "fpga_mgr",
+				COMPAT_ALTERA_SOCFPGA_FPGA0,
+				&node_offset, 1);
+
+	ofnode fpgamgr_node = offset_to_ofnode(node_offset);
+
+	if (ofnode_valid(fpgamgr_node))
+		fpga_filename = (char

*)ofnode_read_string(fpgamgr_node,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+						"altr,bitstrea

m");

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts

Quoted 2 lines
arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +2 -0
+
+

Why is the cast needed ?

Quoted 7 lines
Quoted 4 lines

The return string would be eventually set to the char *filename in common struct fpga_fsinfo. So, the cast here is to avoid the warning
from compiler.

I presume if the compiler generates a warning, it's for a reason. What
warning is that ?

drivers/fpga/socfpga_arria10.c: In function 'get_fpga_filename':
drivers/fpga/socfpga_arria10.c:466:17: warning: assignment discards 'const' qualifier from pointer target type [-Wdiscarded-qualifiers] fpga_filename = ofnode_read_string(fpgamgr_node,

Quoted 1 line

[...]

#32 lore raw

PATCH v8 4/8 ARM: socfpga: Add FPGA drivers for Arria 10 FPGA bitstream loading

Quoted 81 lines · Tien Fong
Quoted 50 lines · Marek Vasut
Quoted 49 lines · Tien Fong
Quoted 47 lines · Marek Vasut
Quoted 45 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add FPGA driver to support program FPGA with FPGA bitstream loading
from
filesystem. The driver are designed based on generic firmware loader
framework. The driver can handle FPGA program operation from loading FPGA
bitstream in flash to memory and then to program FPGA.

---

changes for v8
- Added codes to discern bitstream type based on fpga node name.

changes for v7
- Restructure the FPGA driver to support both peripheral bitstream
and core
  bitstream bundled into FIT image.
- Support loadable property for core bitstream. User can set loadable
  in DDR for better performance. This loading would be done in one
large
  chunk instead of chunk by chunk loading with small memory buffer.
---
 arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts       |  17 + .../include/mach/fpga_manager_arria10.h            |  39 +- drivers/fpga/socfpga_arria10.c                     | 467
++++++++++++++++++++-

 3 files changed, 500 insertions(+), 23 deletions(-)
arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +0 -0
index 998d811..14f1967 100644

[...]

Quoted 28 lines
Quoted 28 lines
Quoted 16 lines
Quoted 16 lines

- * FPGA Manager to program the FPGA. This is the interface used by
FPGA driver.
- * Return 0 for sucess, non-zero for error.
- */

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +11 -0
+char *get_fpga_filename(const void *fdt, int *len)
+{
+	char *fpga_filename = NULL;
+	int node_offset;
+
+	fdtdec_find_aliases_for_id(gd->fdt_blob, "fpga_mgr",
+				COMPAT_ALTERA_SOCFPGA_FPGA0,
+				&node_offset, 1);
+
+	ofnode fpgamgr_node = offset_to_ofnode(node_offset);
+
Quoted 8 lines

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts

Quoted 7 lines
arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +2 -0
+	if (ofnode_valid(fpgamgr_node))
+		fpga_filename = (char

*)ofnode_read_string(fpgamgr_node,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+						"altr,bitstrea

m");

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +2 -0
+
+

Why is the cast needed ?

The return string would be eventually set to the char *filename in common struct fpga_fsinfo. So, the cast here is to avoid the warning
from compiler.

Quoted 7 lines
Quoted 3 lines

I presume if the compiler generates a warning, it's for a reason. What
warning is that ?

drivers/fpga/socfpga_arria10.c: In function 'get_fpga_filename':
drivers/fpga/socfpga_arria10.c:466:17: warning: assignment discards 'const' qualifier from pointer target type [-Wdiscarded-qualifiers] fpga_filename = ofnode_read_string(fpgamgr_node,

Add missing const then ?

Signature · 2 lines
Best regards,
Marek Vasut
#33 lore raw

PATCH v8 4/8 ARM: socfpga: Add FPGA drivers for Arria 10 FPGA bitstream loading

Quoted 101 lines · Marek Vasut
Quoted 100 lines · Tien Fong
Quoted 59 lines · Marek Vasut
Quoted 57 lines · Tien Fong
Quoted 54 lines · Marek Vasut
Quoted 51 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add FPGA driver to support program FPGA with FPGA bitstream loading
from
filesystem. The driver are designed based on generic firmware
loader
framework. The driver can handle FPGA program operation from
loading FPGA
bitstream in flash to memory and then to program FPGA.

---

changes for v8
- Added codes to discern bitstream type based on fpga node name.

changes for v7
- Restructure the FPGA driver to support both peripheral bitstream
and core
  bitstream bundled into FIT image.
- Support loadable property for core bitstream. User can set
loadable
  in DDR for better performance. This loading would be done in
one
large
  chunk instead of chunk by chunk loading with small memory buffer.
---
 arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts       |  17 + .../include/mach/fpga_manager_arria10.h            |  39

+-
 drivers/fpga/socfpga_arria10.c                     | 467
++++++++++++++++++++-
 3 files changed, 500 insertions(+), 23 deletions(-)
arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +0 -0
index 998d811..14f1967 100644

[...]

Quoted 37 lines
Quoted 37 lines
Quoted 24 lines
Quoted 24 lines

- * FPGA Manager to program the FPGA. This is the interface used by
FPGA driver.
- * Return 0 for sucess, non-zero for error.
- */

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +6 -0
+char *get_fpga_filename(const void *fdt, int *len)
+{
+	char *fpga_filename = NULL;
+	int node_offset;
+
+	fdtdec_find_aliases_for_id(gd->fdt_blob,

"fpga_mgr",

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+				COMPAT_ALTERA_SOCFPGA_FPGA

0,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+				&node_offset, 1);
+
+	ofnode fpgamgr_node =

offset_to_ofnode(node_offset);

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +2 -0
+
+	if (ofnode_valid(fpgamgr_node))
Quoted 8 lines

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts

Quoted 7 lines
arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+		fpga_filename = (char

*)ofnode_read_string(fpgamgr_node,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+						"altr,bits

trea
m");

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +2 -0
+
+

Why is the cast needed ?

The return string would be eventually set to the char *filename in
common struct fpga_fsinfo. So, the cast here is to avoid the warning
from compiler.

Quoted 8 lines
Quoted 7 lines
Quoted 3 lines

I presume if the compiler generates a warning, it's for a reason. What
warning is that ?

drivers/fpga/socfpga_arria10.c: In function 'get_fpga_filename':
drivers/fpga/socfpga_arria10.c:466:17: warning: assignment discards 'const' qualifier from pointer target type [-Wdiscarded-qualifiers] fpga_filename = ofnode_read_string(fpgamgr_node,

Add missing const then ?

Then this requires change on common struct fpga_fsinfo, this would impact to other user using this. Why the cast is not allow as we only reading the filename?

Quoted 1 line
#34 lore raw

PATCH v8 4/8 ARM: socfpga: Add FPGA drivers for Arria 10 FPGA bitstream loading

Quoted 105 lines · Tien Fong
Quoted 104 lines · Marek Vasut
Quoted 103 lines · Tien Fong
Quoted 59 lines · Marek Vasut
Quoted 57 lines · Tien Fong
Quoted 54 lines · Marek Vasut
Quoted 51 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add FPGA driver to support program FPGA with FPGA bitstream loading
from
filesystem. The driver are designed based on generic firmware
loader
framework. The driver can handle FPGA program operation from
loading FPGA
bitstream in flash to memory and then to program FPGA.

---

changes for v8
- Added codes to discern bitstream type based on fpga node name.

changes for v7
- Restructure the FPGA driver to support both peripheral bitstream
and core
  bitstream bundled into FIT image.
- Support loadable property for core bitstream. User can set
loadable
  in DDR for better performance. This loading would be done in
one
large
  chunk instead of chunk by chunk loading with small memory buffer.
---
 arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts       |  17 + .../include/mach/fpga_manager_arria10.h            |  39

+-
 drivers/fpga/socfpga_arria10.c                     | 467
++++++++++++++++++++-
 3 files changed, 500 insertions(+), 23 deletions(-)
arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +0 -0
index 998d811..14f1967 100644

[...]

Quoted 40 lines
Quoted 32 lines
Quoted 24 lines
Quoted 24 lines

- * FPGA Manager to program the FPGA. This is the interface used by
FPGA driver.
- * Return 0 for sucess, non-zero for error.
- */

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +6 -0
+char *get_fpga_filename(const void *fdt, int *len)
+{
+	char *fpga_filename = NULL;
+	int node_offset;
+
+	fdtdec_find_aliases_for_id(gd->fdt_blob,

"fpga_mgr",

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+				COMPAT_ALTERA_SOCFPGA_FPGA

0,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+				&node_offset, 1);
+
+	ofnode fpgamgr_node =

offset_to_ofnode(node_offset);

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +2 -0
+
+	if (ofnode_valid(fpgamgr_node))
Quoted 8 lines

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts

Quoted 7 lines
arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+		fpga_filename = (char

*)ofnode_read_string(fpgamgr_node,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+						"altr,bits

trea
m");

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +2 -0
+
+

Why is the cast needed ?

Quoted 5 lines

The return string would be eventually set to the char *filename in
common struct fpga_fsinfo. So, the cast here is to avoid the warning
from compiler.

I presume if the compiler generates a warning, it's for a reason. What
warning is that ?

Quoted 8 lines
Quoted 5 lines
Quoted 4 lines

drivers/fpga/socfpga_arria10.c: In function 'get_fpga_filename':
drivers/fpga/socfpga_arria10.c:466:17: warning: assignment discards 'const' qualifier from pointer target type [-Wdiscarded-qualifiers] fpga_filename = ofnode_read_string(fpgamgr_node,

Add missing const then ?

Then this requires change on common struct fpga_fsinfo, this would impact to other user using this. Why the cast is not allow as we only reading the filename?

If the string isn't const, someone can write it. If someone writes this string, won't it corrupt the DT ?

Signature · 2 lines
Best regards,
Marek Vasut
#35 lore raw

PATCH v8 4/8 ARM: socfpga: Add FPGA drivers for Arria 10 FPGA bitstream loading

Quoted 141 lines · Marek Vasut
Quoted 140 lines · Tien Fong
Quoted 138 lines · Marek Vasut
Quoted 136 lines · Tien Fong
Quoted 74 lines · Marek Vasut
Quoted 71 lines · Tien Fong
Quoted 67 lines · Marek Vasut
Quoted 63 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add FPGA driver to support program FPGA with FPGA bitstream
loading
from
filesystem. The driver are designed based on generic firmware
loader
framework. The driver can handle FPGA program operation from
loading FPGA
bitstream in flash to memory and then to program FPGA.

Quoted 1 line

---

changes for v8
- Added codes to discern bitstream type based on fpga node
name.

changes for v7
- Restructure the FPGA driver to support both peripheral
bitstream
and core
  bitstream bundled into FIT image.
- Support loadable property for core bitstream. User can
set
loadable
  in DDR for better performance. This loading would be done
in
one
large
  chunk instead of chunk by chunk loading with small memory
buffer.
---
 arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts       | 17 +
 .../include/mach/fpga_manager_arria10.h            | 39

+-
 drivers/fpga/socfpga_arria10.c                     |
467
++++++++++++++++++++-
 3 files changed, 500 insertions(+), 23 deletions(-)

diff --git
a/arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts
b/arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts
index 998d811..14f1967 100644

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +0 -0

[...]

Quoted 46 lines
Quoted 39 lines
Quoted 31 lines
Quoted 31 lines

- * FPGA Manager to program the FPGA. This is the
interface
used by
FPGA driver.
- * Return 0 for sucess, non-zero for error.
- */

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +6 -0
+char *get_fpga_filename(const void *fdt, int *len)
+{
+	char *fpga_filename = NULL;
+	int node_offset;
+
+	fdtdec_find_aliases_for_id(gd->fdt_blob,

"fpga_mgr",

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+				COMPAT_ALTERA_SOCFPGA_

FPGA
0,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+				&node_offset, 1);
+
+	ofnode fpgamgr_node =

offset_to_ofnode(node_offset);

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+
+	if (ofnode_valid(fpgamgr_node))
+		fpga_filename = (char
Quoted 8 lines

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts

Quoted 7 lines

*)ofnode_read_string(fpgamgr_node,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+						"altr,

bits
trea
m");

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +2 -0
+
+

Why is the cast needed ?

Quoted 7 lines

The return string would be eventually set to the char *filename
in
common struct fpga_fsinfo. So, the cast here is to avoid the
warning
from compiler.

Quoted 4 lines

I presume if the compiler generates a warning, it's for a reason.
What
warning is that ?

drivers/fpga/socfpga_arria10.c: In function 'get_fpga_filename':
drivers/fpga/socfpga_arria10.c:466:17: warning: assignment discards
'const' qualifier from pointer target type [-Wdiscarded- qualifiers]
   fpga_filename = ofnode_read_string(fpgamgr_node,

Quoted 8 lines
Quoted 5 lines
Quoted 1 line

Add missing const then ?

Then this requires change on common struct fpga_fsinfo, this would impact to other user using this. Why the cast is not allow as we only
reading the filename?

If the string isn't const, someone can write it. If someone writes this
string, won't it corrupt the DT ?

Yes, but this would not happen in this driver, right?
I don't know also why this common struct declare without the const, may be it supports the write?

Quoted 1 line
#36 lore raw

PATCH v8 4/8 ARM: socfpga: Add FPGA drivers for Arria 10 FPGA bitstream loading

Quoted 147 lines · Tien Fong
Quoted 146 lines · Marek Vasut
Quoted 145 lines · Tien Fong
Quoted 131 lines · Marek Vasut
Quoted 129 lines · Tien Fong
Quoted 74 lines · Marek Vasut
Quoted 71 lines · Tien Fong
Quoted 67 lines · Marek Vasut
Quoted 63 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add FPGA driver to support program FPGA with FPGA bitstream
loading
from
filesystem. The driver are designed based on generic firmware
loader
framework. The driver can handle FPGA program operation from
loading FPGA
bitstream in flash to memory and then to program FPGA.

Quoted 1 line

---

changes for v8
- Added codes to discern bitstream type based on fpga node
name.

changes for v7
- Restructure the FPGA driver to support both peripheral
bitstream
and core
  bitstream bundled into FIT image.
- Support loadable property for core bitstream. User can
set
loadable
  in DDR for better performance. This loading would be done
in
one
large
  chunk instead of chunk by chunk loading with small memory
buffer.
---
 arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts       | 17 +
 .../include/mach/fpga_manager_arria10.h            | 39

+-
 drivers/fpga/socfpga_arria10.c                     |
467
++++++++++++++++++++-
 3 files changed, 500 insertions(+), 23 deletions(-)

diff --git
a/arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts
b/arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts
index 998d811..14f1967 100644

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +0 -0

[...]

Quoted 50 lines
Quoted 39 lines
Quoted 31 lines
Quoted 31 lines

- * FPGA Manager to program the FPGA. This is the
interface
used by
FPGA driver.
- * Return 0 for sucess, non-zero for error.
- */

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +6 -0
+char *get_fpga_filename(const void *fdt, int *len)
+{
+	char *fpga_filename = NULL;
+	int node_offset;
+
+	fdtdec_find_aliases_for_id(gd->fdt_blob,

"fpga_mgr",

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+				COMPAT_ALTERA_SOCFPGA_

FPGA
0,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+				&node_offset, 1);
+
+	ofnode fpgamgr_node =

offset_to_ofnode(node_offset);

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +3 -0
+
+	if (ofnode_valid(fpgamgr_node))
+		fpga_filename = (char
Quoted 8 lines

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts

Quoted 7 lines

*)ofnode_read_string(fpgamgr_node,

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +1 -0
+						"altr,

bits
trea
m");

arch/arm/dts/socfpga_arria10_socdk_sdmmc.dts +2 -0
+
+

Why is the cast needed ?

Quoted 7 lines

The return string would be eventually set to the char *filename
in
common struct fpga_fsinfo. So, the cast here is to avoid the
warning
from compiler.

I presume if the compiler generates a warning, it's for a reason.
What
warning is that ?

Quoted 8 lines
Quoted 7 lines

drivers/fpga/socfpga_arria10.c: In function 'get_fpga_filename':
drivers/fpga/socfpga_arria10.c:466:17: warning: assignment discards
'const' qualifier from pointer target type [-Wdiscarded- qualifiers]
   fpga_filename = ofnode_read_string(fpgamgr_node,

Add missing const then ?

Then this requires change on common struct fpga_fsinfo, this would impact to other user using this. Why the cast is not allow as we only
reading the filename?

Quoted 6 lines
Quoted 3 lines

If the string isn't const, someone can write it. If someone writes this
string, won't it corrupt the DT ?

Yes, but this would not happen in this driver, right?
I don't know also why this common struct declare without the const, may be it supports the write?

I think so, maybe you should check that.

Signature · 2 lines
Best regards,
Marek Vasut
#37 lore raw

PATCH v8 5/8 ARM: socfpga: Add the configuration for FPGA SoCFPGA A10 SoCDK

In reply to tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Update the default configuration file to enable the necessary functionality to get the SoCFPGA loadfs driver support. This would enable the implementation of programming bitstream into FPGA from MMC.

---

changes for v8
- Added FIT related configs

changes for v7
- Removed limit set for CONFIG_FS_FAT_MAX_CLUSTSIZE

---
 configs/socfpga_arria10_defconfig | 8 ++++++++
 1 file changed, 8 insertions(+)
configs/socfpga_arria10_defconfig +8 -0
index 0554f1b..c870543 100644
@@ -27,10 +27,18 @@ CONFIG_MTDIDS_DEFAULT="nor0=ff705000.spi.0"
 # CONFIG_EFI_PARTITION is not set
 CONFIG_DEFAULT_DEVICE_TREE="socfpga_arria10_socdk_sdmmc"
 CONFIG_ENV_IS_IN_MMC=y
+CONFIG_SPL_ENV_SUPPORT=y
 CONFIG_SPL_DM=y
 CONFIG_SPL_DM_SEQ_ALIAS=y
+CONFIG_SPL_DM_MMC=y
+CONFIG_SPL_MMC_SUPPORT=y
+CONFIG_SPL_EXT_SUPPORT=y
 CONFIG_SPL_FS_FAT=y
+CONFIG_SPL_DRIVERS_MISC_SUPPORT=y
+CONFIG_FS_LOADER=y
 CONFIG_FPGA_SOCFPGA=y
+CONFIG_SPL_FIT=y
+CONFIG_FIT=y
 CONFIG_DM_GPIO=y
 CONFIG_DWAPB_GPIO=y
 CONFIG_DM_MMC=y
Signature · 1 line
2.2.0
#38 lore raw

PATCH v8 6/8 spl : socfpga: Implement fpga bitstream loading with socfpga loadfs

In reply to tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add support for loading FPGA bitstream to get DDR up running before U-Boot is loaded into DDR. Boot device initialization, generic firmware loader and SPL FAT support are required for this whole mechanism to work.

---

changes for v7
- Removed casting for get_fpga_filename
- Removed hard coding DDR address for loading core bistream, using loadable property from FIT.
- Added checking for config_pins, return if error.

---
 arch/arm/mach-socfpga/spl_a10.c | 41 ++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 40 insertions(+), 1 deletion(-)
arch/arm/mach-socfpga/spl_a10.c +40 -1
index c97eacb..36003b1 100644
@@ -1,6 +1,6 @@
 // SPDX-License-Identifier: GPL-2.0+
 /*
- *  Copyright (C) 2012 Altera Corporation <www.altera.com>
+ *  Copyright (C) 2012-2019 Altera Corporation <www.altera.com>
  */
 
 #include <common.h>
@@ -23,6 +23,8 @@
 #include <fdtdec.h>
 #include <watchdog.h>
 #include <asm/arch/pinmux.h>
+#include <asm/arch/fpga_manager.h>
+#include <mmc.h>
 
 DECLARE_GLOBAL_DATA_PTR;
 
@@ -68,11 +70,48 @@ u32 spl_boot_mode(const u32 boot_device)
 
 void spl_board_init(void)
 {
+	char buf[16 * 1024] __aligned(ARCH_DMA_MINALIGN);
+
 	/* enable console uart printing */
 	preloader_console_init();
 	WATCHDOG_RESET();
 
 	arch_early_init_r();
+
+	/* If the full FPGA is already loaded, ie.from EPCQ, config fpga pins */
+	if (is_fpgamgr_user_mode()) {
+		int ret = config_pins(gd->fdt_blob, "shared");
+
+		if (ret)
+			return;
+
+		ret = config_pins(gd->fdt_blob, "fpga");
+		if (ret)
+			return;
+	} else if (!is_fpgamgr_early_user_mode()) {
+		/* Program IOSSM(early IO release) or full FPGA */
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename = get_fpga_filename(gd->fdt_blob, &len);
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf, sizeof(buf), 0);
+	}
+
+	/* If the IOSSM/full FPGA is already loaded, start DDR */
+	if (is_fpgamgr_early_user_mode() || is_fpgamgr_user_mode())
+		ddr_calibration_sequence();
+
+	if (!is_fpgamgr_user_mode()) {
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename = get_fpga_filename(gd->fdt_blob, &len);
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf, sizeof(buf), 0);
+	}
 }
 
 void board_init_f(ulong dummy)
Signature · 1 line
2.2.0
#39 lore raw

PATCH v8 6/8 spl : socfpga: Implement fpga bitstream loading with socfpga loadfs

Quoted 41 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add support for loading FPGA bitstream to get DDR up running before U-Boot is loaded into DDR. Boot device initialization, generic firmware loader and SPL FAT support are required for this whole mechanism to work.

---

changes for v7
- Removed casting for get_fpga_filename
- Removed hard coding DDR address for loading core bistream, using loadable property from FIT.
- Added checking for config_pins, return if error.

---
 arch/arm/mach-socfpga/spl_a10.c | 41 ++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 40 insertions(+), 1 deletion(-)
arch/arm/mach-socfpga/spl_a10.c +3 -1
index c97eacb..36003b1 100644
@@ -1,6 +1,6 @@
 // SPDX-License-Identifier: GPL-2.0+
 /*
- *  Copyright (C) 2012 Altera Corporation <www.altera.com>
+ *  Copyright (C) 2012-2019 Altera Corporation <www.altera.com>
  */
 
 #include <common.h>
@@ -23,6 +23,8 @@
 #include <fdtdec.h>
 #include <watchdog.h>
 #include <asm/arch/pinmux.h>
+#include <asm/arch/fpga_manager.h>
+#include <mmc.h>
 
 DECLARE_GLOBAL_DATA_PTR;
 

arch/arm/mach-socfpga/spl_a10.c u32 spl_boot_mode(const u32 boot_device) · line 73

Quoted 4 lines
arch/arm/mach-socfpga/spl_a10.c +1 -0
@@ -68,11 +70,48 @@ u32 spl_boot_mode(const u32 boot_device)
 
 void spl_board_init(void)
 {
+	char buf[16 * 1024] __aligned(ARCH_DMA_MINALIGN);

ALLOC_CACHE_ALIGN_BUFFER()

Quoted 16 lines
arch/arm/mach-socfpga/spl_a10.c +11 -0
+
 	/* enable console uart printing */
 	preloader_console_init();
 	WATCHDOG_RESET();
 
 	arch_early_init_r();
+
+	/* If the full FPGA is already loaded, ie.from EPCQ, config fpga pins */
+	if (is_fpgamgr_user_mode()) {
+		int ret = config_pins(gd->fdt_blob, "shared");
+
+		if (ret)
+			return;
+
+		ret = config_pins(gd->fdt_blob, "fpga");
+		if (ret)

arch/arm/mach-socfpga/spl_a10.c u32 spl_boot_mode(const u32 boot_device) · line 99

Quoted 10 lines
arch/arm/mach-socfpga/spl_a10.c +10 -0
+			return;
+	} else if (!is_fpgamgr_early_user_mode()) {
+		/* Program IOSSM(early IO release) or full FPGA */
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename = get_fpga_filename(gd->fdt_blob, &len);
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf, sizeof(buf), 0);

Why is this code here twice ? The same code seems to be below ...

Quoted 19 lines
arch/arm/mach-socfpga/spl_a10.c +15 -0
+	}
+
+	/* If the IOSSM/full FPGA is already loaded, start DDR */
+	if (is_fpgamgr_early_user_mode() || is_fpgamgr_user_mode())
+		ddr_calibration_sequence();
+
+	if (!is_fpgamgr_user_mode()) {
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename = get_fpga_filename(gd->fdt_blob, &len);
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf, sizeof(buf), 0);
+	}
 }
 
 void board_init_f(ulong dummy)
Signature · 2 lines
Best regards,
Marek Vasut
#40 lore raw

PATCH v8 6/8 spl : socfpga: Implement fpga bitstream loading with socfpga loadfs

Quoted 45 lines · Marek Vasut
Quoted 44 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add support for loading FPGA bitstream to get DDR up running before U-Boot is loaded into DDR. Boot device initialization, generic firmware
loader and SPL FAT support are required for this whole mechanism to work.

---

changes for v7
- Removed casting for get_fpga_filename
- Removed hard coding DDR address for loading core bistream, using loadable
  property from FIT.
- Added checking for config_pins, return if error.
---
 arch/arm/mach-socfpga/spl_a10.c | 41
++++++++++++++++++++++++++++++++++++++++-

 1 file changed, 40 insertions(+), 1 deletion(-)
arch/arm/mach- renamed from arch/arm/mach-socfpga/spl_a10.c +0 -0

socfpga/spl_a10.c
index c97eacb..36003b1 100644

arch/arm/mach-socfpga/spl_a10.c +1 -1
@@ -1,6 +1,6 @@
  // SPDX-License-Identifier: GPL-2.0+
  /*
- *  Copyright (C) 2012 Altera Corporation <www.altera.com>
+ *  Copyright (C) 2012-2019 Altera Corporation <www.altera.com>
   */


 #include <common.h>

arch/arm/mach-socfpga/spl_a10.c +2 -0
@@ -23,6 +23,8 @@
  #include <fdtdec.h>
  #include <watchdog.h>
  #include <asm/arch/pinmux.h>
+#include <asm/arch/fpga_manager.h>
+#include <mmc.h>

Quoted 8 lines

arch/arm/mach-socfpga/spl_a10.c u32 spl_boot_mode(const u32 boot_device) · line 73

Quoted 7 lines

 DECLARE_GLOBAL_DATA_PTR;

arch/arm/mach-socfpga/spl_a10.c +0 -0
@@ -68,11 +70,48 @@ u32 spl_boot_mode(const u32 boot_device)


 void spl_board_init(void)
 {

arch/arm/mach-socfpga/spl_a10.c +1 -0
+	char buf[16 * 1024] __aligned(ARCH_DMA_MINALIGN);

ALLOC_CACHE_ALIGN_BUFFER()

#define ARCH_DMA_MINALIGN CONFIG_SYS_CACHELINE_SIZE
They are not same thing?

Quoted 26 lines
Quoted 26 lines

+
  /* enable console uart printing */
  preloader_console_init();
  WATCHDOG_RESET();

  arch_early_init_r();

arch/arm/mach-socfpga/spl_a10.c +2 -0
+
+	/* If the full FPGA is already loaded, ie.from EPCQ,

config fpga pins */

arch/arm/mach-socfpga/spl_a10.c +15 -0
+	if (is_fpgamgr_user_mode()) {
+		int ret = config_pins(gd->fdt_blob, "shared");
+
+		if (ret)
+			return;
+
+		ret = config_pins(gd->fdt_blob, "fpga");
+		if (ret)
+			return;
+	} else if (!is_fpgamgr_early_user_mode()) {
+		/* Program IOSSM(early IO release) or full FPGA */
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename = get_fpga_filename(gd-
Quoted 6 lines
Quoted 5 lines
Quoted 1 line

fdt_blob, &len);

arch/arm/mach-socfpga/spl_a10.c +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf,

sizeof(buf), 0);

Why is this code here twice ? The same code seems to be below ...

The 1st calling for periph program, then running ddr calibration, then 2nd calling for core program.

Quoted 25 lines
Quoted 24 lines
arch/arm/mach-socfpga/spl_a10.c +4 -0
+	}
+
+	/* If the IOSSM/full FPGA is already loaded, start DDR */
+	if (is_fpgamgr_early_user_mode() ||

is_fpgamgr_user_mode())

arch/arm/mach-socfpga/spl_a10.c +7 -0
+		ddr_calibration_sequence();
+
+	if (!is_fpgamgr_user_mode()) {
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename = get_fpga_filename(gd-
Quoted 1 line

fdt_blob, &len);

arch/arm/mach-socfpga/spl_a10.c +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf,

sizeof(buf), 0);

arch/arm/mach-socfpga/spl_a10.c +1 -0
+	}

 }

 void board_init_f(ulong dummy)

#41 lore raw

PATCH v8 6/8 spl : socfpga: Implement fpga bitstream loading with socfpga loadfs

Quoted 54 lines · Tien Fong
Quoted 45 lines · Marek Vasut
Quoted 44 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add support for loading FPGA bitstream to get DDR up running before U-Boot is loaded into DDR. Boot device initialization, generic firmware
loader and SPL FAT support are required for this whole mechanism to work.

---

changes for v7
- Removed casting for get_fpga_filename
- Removed hard coding DDR address for loading core bistream, using loadable
  property from FIT.
- Added checking for config_pins, return if error.
---
 arch/arm/mach-socfpga/spl_a10.c | 41
++++++++++++++++++++++++++++++++++++++++-

 1 file changed, 40 insertions(+), 1 deletion(-)
arch/arm/mach- renamed from arch/arm/mach-socfpga/spl_a10.c +0 -0

socfpga/spl_a10.c
index c97eacb..36003b1 100644

arch/arm/mach-socfpga/spl_a10.c +1 -1
@@ -1,6 +1,6 @@
  // SPDX-License-Identifier: GPL-2.0+
  /*
- *  Copyright (C) 2012 Altera Corporation <www.altera.com>
+ *  Copyright (C) 2012-2019 Altera Corporation <www.altera.com>
   */


 #include <common.h>

arch/arm/mach-socfpga/spl_a10.c +2 -0
@@ -23,6 +23,8 @@
  #include <fdtdec.h>
  #include <watchdog.h>
  #include <asm/arch/pinmux.h>
+#include <asm/arch/fpga_manager.h>
+#include <mmc.h>

Quoted 8 lines

arch/arm/mach-socfpga/spl_a10.c u32 spl_boot_mode(const u32 boot_device) · line 73

Quoted 7 lines

 DECLARE_GLOBAL_DATA_PTR;

arch/arm/mach-socfpga/spl_a10.c +0 -0
@@ -68,11 +70,48 @@ u32 spl_boot_mode(const u32 boot_device)


 void spl_board_init(void)
 {

arch/arm/mach-socfpga/spl_a10.c +1 -0
+	char buf[16 * 1024] __aligned(ARCH_DMA_MINALIGN);

ALLOC_CACHE_ALIGN_BUFFER()

Quoted 2 lines

#define ARCH_DMA_MINALIGN CONFIG_SYS_CACHELINE_SIZE
They are not same thing?

See include/memalign.h and other drivers, the macro is preferred as it hides the details.

Quoted 25 lines
Quoted 25 lines
Quoted 25 lines

+
  /* enable console uart printing */
  preloader_console_init();
  WATCHDOG_RESET();

  arch_early_init_r();

arch/arm/mach-socfpga/spl_a10.c +2 -0
+
+	/* If the full FPGA is already loaded, ie.from EPCQ,

config fpga pins */

arch/arm/mach-socfpga/spl_a10.c +15 -0
+	if (is_fpgamgr_user_mode()) {
+		int ret = config_pins(gd->fdt_blob, "shared");
+
+		if (ret)
+			return;
+
+		ret = config_pins(gd->fdt_blob, "fpga");
+		if (ret)
+			return;
+	} else if (!is_fpgamgr_early_user_mode()) {
+		/* Program IOSSM(early IO release) or full FPGA */
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename = get_fpga_filename(gd-
Quoted 8 lines
Quoted 6 lines
Quoted 5 lines
Quoted 1 line

fdt_blob, &len);

arch/arm/mach-socfpga/spl_a10.c +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf,

sizeof(buf), 0);

Why is this code here twice ? The same code seems to be below ...

The 1st calling for periph program, then running ddr calibration, then 2nd calling for core program.

Then maybe the code can be deduplicated ?

Quoted 22 lines
Quoted 22 lines
Quoted 22 lines
arch/arm/mach-socfpga/spl_a10.c +4 -0
+	}
+
+	/* If the IOSSM/full FPGA is already loaded, start DDR */
+	if (is_fpgamgr_early_user_mode() ||

is_fpgamgr_user_mode())

arch/arm/mach-socfpga/spl_a10.c +7 -0
+		ddr_calibration_sequence();
+
+	if (!is_fpgamgr_user_mode()) {
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename = get_fpga_filename(gd-
Quoted 1 line

fdt_blob, &len);

arch/arm/mach-socfpga/spl_a10.c +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf,

sizeof(buf), 0);

arch/arm/mach-socfpga/spl_a10.c +1 -0
+	}

 }

 void board_init_f(ulong dummy)

Signature · 2 lines
Best regards,
Marek Vasut
#42 lore raw

PATCH v8 6/8 spl : socfpga: Implement fpga bitstream loading with socfpga loadfs

Quoted 62 lines · Marek Vasut
Quoted 61 lines · Tien Fong
Quoted 59 lines · Marek Vasut
Quoted 55 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add support for loading FPGA bitstream to get DDR up running before
U-Boot is loaded into DDR. Boot device initialization, generic firmware
loader and SPL FAT support are required for this whole mechanism to
work.

---

changes for v7
- Removed casting for get_fpga_filename
- Removed hard coding DDR address for loading core bistream, using
loadable
  property from FIT.
- Added checking for config_pins, return if error.
---
 arch/arm/mach-socfpga/spl_a10.c | 41
++++++++++++++++++++++++++++++++++++++++-

 1 file changed, 40 insertions(+), 1 deletion(-)
arch/arm/mach- renamed from arch/arm/mach-socfpga/spl_a10.c +0 -0

socfpga/spl_a10.c
index c97eacb..36003b1 100644

arch/arm/mach-socfpga/spl_a10.c +1 -1
@@ -1,6 +1,6 @@
  // SPDX-License-Identifier: GPL-2.0+
  /*
- *  Copyright (C) 2012 Altera Corporation <www.altera.com>
+ *  Copyright (C) 2012-2019 Altera Corporation <www.altera.com
Quoted 1 line

  */

 #include <common.h>

arch/arm/mach-socfpga/spl_a10.c +2 -0
@@ -23,6 +23,8 @@
  #include <fdtdec.h>
  #include <watchdog.h>
  #include <asm/arch/pinmux.h>
+#include <asm/arch/fpga_manager.h>
+#include <mmc.h>


 DECLARE_GLOBAL_DATA_PTR;

arch/arm/mach-socfpga/spl_a10.c +0 -0
@@ -68,11 +70,48 @@ u32 spl_boot_mode(const u32 boot_device)


 void spl_board_init(void)
 {

arch/arm/mach-socfpga/spl_a10.c u32 spl_boot_mode(const u32 boot_device) · line 73

Quoted 1 line
arch/arm/mach-socfpga/spl_a10.c +1 -0
+	char buf[16 * 1024] __aligned(ARCH_DMA_MINALIGN);

ALLOC_CACHE_ALIGN_BUFFER()

Quoted 5 lines
Quoted 2 lines

#define ARCH_DMA_MINALIGN CONFIG_SYS_CACHELINE_SIZE
They are not same thing?

See include/memalign.h and other drivers, the macro is preferred as it
hides the details.

Okay.

Quoted 37 lines
Quoted 37 lines
Quoted 37 lines
Quoted 35 lines

+
  /* enable console uart printing */
  preloader_console_init();
  WATCHDOG_RESET();

  arch_early_init_r();

arch/arm/mach-socfpga/spl_a10.c +2 -0
+
+	/* If the full FPGA is already loaded, ie.from EPCQ,

config fpga pins */

arch/arm/mach-socfpga/spl_a10.c +11 -0
+	if (is_fpgamgr_user_mode()) {
+		int ret = config_pins(gd->fdt_blob, "shared");
+
+		if (ret)
+			return;
+
+		ret = config_pins(gd->fdt_blob, "fpga");
+		if (ret)
+			return;
+	} else if (!is_fpgamgr_early_user_mode()) {
+		/* Program IOSSM(early IO release) or full

FPGA */

arch/arm/mach-socfpga/spl_a10.c +4 -0
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename = get_fpga_filename(gd-
Quoted 2 lines

fdt_blob, &len);

arch/arm/mach-socfpga/spl_a10.c +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf,
Quoted 1 line

sizeof(buf), 0);

Why is this code here twice ? The same code seems to be below ...

Quoted 4 lines
Quoted 3 lines

The 1st calling for periph program, then running ddr calibration, then
2nd calling for core program.

Then maybe the code can be deduplicated ?

Hmm...seems cannot, because
1. DDR calibration is not part of fpga code.
2. fpga driver can only be used to process one bistream at a time, because different mode has different handling.

Quoted 29 lines
Quoted 28 lines
Quoted 28 lines
Quoted 28 lines
arch/arm/mach-socfpga/spl_a10.c +3 -0
+	}
+
+	/* If the IOSSM/full FPGA is already loaded, start DDR

*/

arch/arm/mach-socfpga/spl_a10.c +1 -0
+	if (is_fpgamgr_early_user_mode() ||

is_fpgamgr_user_mode())

arch/arm/mach-socfpga/spl_a10.c +7 -0
+		ddr_calibration_sequence();
+
+	if (!is_fpgamgr_user_mode()) {
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename = get_fpga_filename(gd-
Quoted 2 lines

fdt_blob, &len);

arch/arm/mach-socfpga/spl_a10.c +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf,

sizeof(buf), 0);

arch/arm/mach-socfpga/spl_a10.c +1 -0
+	}

 }

 void board_init_f(ulong dummy)

#43 lore raw

PATCH v8 6/8 spl : socfpga: Implement fpga bitstream loading with socfpga loadfs

Quoted 106 lines · Tien Fong
Quoted 62 lines · Marek Vasut
Quoted 61 lines · Tien Fong
Quoted 59 lines · Marek Vasut
Quoted 55 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add support for loading FPGA bitstream to get DDR up running before
U-Boot is loaded into DDR. Boot device initialization, generic firmware
loader and SPL FAT support are required for this whole mechanism to
work.

---

changes for v7
- Removed casting for get_fpga_filename
- Removed hard coding DDR address for loading core bistream, using
loadable
  property from FIT.
- Added checking for config_pins, return if error.
---
 arch/arm/mach-socfpga/spl_a10.c | 41
++++++++++++++++++++++++++++++++++++++++-

 1 file changed, 40 insertions(+), 1 deletion(-)
arch/arm/mach- renamed from arch/arm/mach-socfpga/spl_a10.c +0 -0

socfpga/spl_a10.c
index c97eacb..36003b1 100644

arch/arm/mach-socfpga/spl_a10.c +1 -1
@@ -1,6 +1,6 @@
  // SPDX-License-Identifier: GPL-2.0+
  /*
- *  Copyright (C) 2012 Altera Corporation <www.altera.com>
+ *  Copyright (C) 2012-2019 Altera Corporation <www.altera.com
Quoted 1 line

  */

 #include <common.h>

arch/arm/mach-socfpga/spl_a10.c +2 -0
@@ -23,6 +23,8 @@
  #include <fdtdec.h>
  #include <watchdog.h>
  #include <asm/arch/pinmux.h>
+#include <asm/arch/fpga_manager.h>
+#include <mmc.h>


 DECLARE_GLOBAL_DATA_PTR;

arch/arm/mach-socfpga/spl_a10.c +0 -0
@@ -68,11 +70,48 @@ u32 spl_boot_mode(const u32 boot_device)


 void spl_board_init(void)
 {

arch/arm/mach-socfpga/spl_a10.c u32 spl_boot_mode(const u32 boot_device) · line 73

Quoted 1 line
arch/arm/mach-socfpga/spl_a10.c +1 -0
+	char buf[16 * 1024] __aligned(ARCH_DMA_MINALIGN);

ALLOC_CACHE_ALIGN_BUFFER()

Quoted 5 lines
Quoted 2 lines

#define ARCH_DMA_MINALIGN CONFIG_SYS_CACHELINE_SIZE
They are not same thing?

See include/memalign.h and other drivers, the macro is preferred as it
hides the details.

Okay.

Quoted 37 lines
Quoted 37 lines
Quoted 37 lines
Quoted 35 lines

+
  /* enable console uart printing */
  preloader_console_init();
  WATCHDOG_RESET();

  arch_early_init_r();

arch/arm/mach-socfpga/spl_a10.c +2 -0
+
+	/* If the full FPGA is already loaded, ie.from EPCQ,

config fpga pins */

arch/arm/mach-socfpga/spl_a10.c +11 -0
+	if (is_fpgamgr_user_mode()) {
+		int ret = config_pins(gd->fdt_blob, "shared");
+
+		if (ret)
+			return;
+
+		ret = config_pins(gd->fdt_blob, "fpga");
+		if (ret)
+			return;
+	} else if (!is_fpgamgr_early_user_mode()) {
+		/* Program IOSSM(early IO release) or full

FPGA */

arch/arm/mach-socfpga/spl_a10.c +4 -0
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename = get_fpga_filename(gd-
Quoted 2 lines

fdt_blob, &len);

arch/arm/mach-socfpga/spl_a10.c +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf,
Quoted 1 line

sizeof(buf), 0);

Why is this code here twice ? The same code seems to be below ...

Quoted 8 lines
Quoted 4 lines
Quoted 3 lines

The 1st calling for periph program, then running ddr calibration, then
2nd calling for core program.

Then maybe the code can be deduplicated ?

Hmm...seems cannot, because
1. DDR calibration is not part of fpga code.
2. fpga driver can only be used to process one bistream at a time, because different mode has different handling.

+ } else if (!is_fpgamgr_early_user_mode()) {
+ /* Program IOSSM(early IO release) or full FPGA */
+ fpga_fs_info fpga_fsinfo;
+ int len;
+
+ fpga_fsinfo.filename = get_fpga_filename(gd->fdt_blob, &len);
+
+ if (fpga_fsinfo.filename)
+ socfpga_loadfs(&fpga_fsinfo, buf, sizeof(buf), 0);
...
+ if (!is_fpgamgr_user_mode()) {
+ fpga_fs_info fpga_fsinfo;
+ int len;
+
+ fpga_fsinfo.filename = get_fpga_filename(gd->fdt_blob, &len);
+
+ if (fpga_fsinfo.filename)
+ socfpga_loadfs(&fpga_fsinfo, buf, sizeof(buf), 0);

These two chunks look the same to me , no ?

Signature · 2 lines
Best regards,
Marek Vasut
#44 lore raw

PATCH v8 6/8 spl : socfpga: Implement fpga bitstream loading with socfpga loadfs

Quoted 161 lines · Marek Vasut
Quoted 137 lines · Tien Fong
Quoted 73 lines · Marek Vasut
Quoted 71 lines · Tien Fong
Quoted 68 lines · Marek Vasut
Quoted 63 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add support for loading FPGA bitstream to get DDR up running
before
U-Boot is loaded into DDR. Boot device initialization, generic
firmware
loader and SPL FAT support are required for this whole mechanism to
work.

---

changes for v7
- Removed casting for get_fpga_filename
- Removed hard coding DDR address for loading core bistream,
using
loadable
  property from FIT.
- Added checking for config_pins, return if error.
---
 arch/arm/mach-socfpga/spl_a10.c | 41
++++++++++++++++++++++++++++++++++++++++-

 1 file changed, 40 insertions(+), 1 deletion(-)
arch/arm/mach- renamed from arch/arm/mach-socfpga/spl_a10.c +0 -0

socfpga/spl_a10.c
index c97eacb..36003b1 100644

arch/arm/mach-socfpga/spl_a10.c +1 -1
@@ -1,6 +1,6 @@
  // SPDX-License-Identifier: GPL-2.0+
  /*
- *  Copyright (C) 2012 Altera Corporation <www.altera.com>
+ *  Copyright (C) 2012-2019 Altera Corporation <www.altera

.com

Quoted 2 lines

  */

 #include <common.h>

arch/arm/mach-socfpga/spl_a10.c +2 -0
@@ -23,6 +23,8 @@
  #include <fdtdec.h>
  #include <watchdog.h>
  #include <asm/arch/pinmux.h>
+#include <asm/arch/fpga_manager.h>
+#include <mmc.h>


 DECLARE_GLOBAL_DATA_PTR;

arch/arm/mach-socfpga/spl_a10.c +0 -0
@@ -68,11 +70,48 @@ u32 spl_boot_mode(const u32
 boot_device)


 void spl_board_init(void)
 {

arch/arm/mach-socfpga/spl_a10.c u32 spl_boot_mode(const u32 boot_device) · line 73

Quoted 1 line
arch/arm/mach-socfpga/spl_a10.c +1 -0
+	char buf[16 * 1024] __aligned(ARCH_DMA_MINALIGN);

ALLOC_CACHE_ALIGN_BUFFER()

Quoted 6 lines
Quoted 2 lines

#define ARCH_DMA_MINALIGN CONFIG_SYS_CACHELINE_SIZE
They are not same thing?

See include/memalign.h and other drivers, the macro is preferred as
it
hides the details.

Okay.

Quoted 46 lines
Quoted 46 lines
Quoted 46 lines
Quoted 43 lines

+
  /* enable console uart printing */
  preloader_console_init();
  WATCHDOG_RESET();

  arch_early_init_r();

arch/arm/mach-socfpga/spl_a10.c +2 -0
+
+	/* If the full FPGA is already loaded, ie.from

EPCQ,
config fpga pins */

arch/arm/mach-socfpga/spl_a10.c +2 -0
+	if (is_fpgamgr_user_mode()) {
+		int ret = config_pins(gd->fdt_blob,

"shared");

arch/arm/mach-socfpga/spl_a10.c +9 -0
+
+		if (ret)
+			return;
+
+		ret = config_pins(gd->fdt_blob, "fpga");
+		if (ret)
+			return;
+	} else if (!is_fpgamgr_early_user_mode()) {
+		/* Program IOSSM(early IO release) or full

FPGA */

arch/arm/mach-socfpga/spl_a10.c +4 -0
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename =

get_fpga_filename(gd-

Quoted 3 lines

fdt_blob, &len);

arch/arm/mach-socfpga/spl_a10.c +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf,
Quoted 1 line

sizeof(buf), 0);

Why is this code here twice ? The same code seems to be below
...

Quoted 5 lines
Quoted 4 lines

The 1st calling for periph program, then running ddr calibration,
then
2nd calling for core program.

Then maybe the code can be deduplicated ?

Hmm...seems cannot, because
1. DDR calibration is not part of fpga code.
2. fpga driver can only be used to process one bistream at a time, because different mode has different handling.

arch/arm/mach-socfpga/spl_a10.c +6 -0
+	} else if (!is_fpgamgr_early_user_mode()) {
+		/* Program IOSSM(early IO release) or full FPGA */
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename = get_fpga_filename(gd-
Quoted 1 line

fdt_blob, &len);

arch/arm/mach-socfpga/spl_a10.c +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf,

sizeof(buf), 0);
...

arch/arm/mach-socfpga/spl_a10.c +5 -0
+	if (!is_fpgamgr_user_mode()) {
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename = get_fpga_filename(gd-
Quoted 1 line

fdt_blob, &len);

arch/arm/mach-socfpga/spl_a10.c +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf,

sizeof(buf), 0);

Quoted 1 line

These two chunks look the same to me , no ?

Yes, they are being called twice at different fpga mode, and different sequence, before and after DDR calibration.

Quoted 1 line
#45 lore raw

PATCH v8 6/8 spl : socfpga: Implement fpga bitstream loading with socfpga loadfs

Quoted 157 lines · Tien Fong
Quoted 156 lines · Marek Vasut
Quoted 137 lines · Tien Fong
Quoted 73 lines · Marek Vasut
Quoted 71 lines · Tien Fong
Quoted 68 lines · Marek Vasut
Quoted 63 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add support for loading FPGA bitstream to get DDR up running
before
U-Boot is loaded into DDR. Boot device initialization, generic
firmware
loader and SPL FAT support are required for this whole mechanism to
work.

---

changes for v7
- Removed casting for get_fpga_filename
- Removed hard coding DDR address for loading core bistream,
using
loadable
  property from FIT.
- Added checking for config_pins, return if error.
---
 arch/arm/mach-socfpga/spl_a10.c | 41
++++++++++++++++++++++++++++++++++++++++-

 1 file changed, 40 insertions(+), 1 deletion(-)
arch/arm/mach- renamed from arch/arm/mach-socfpga/spl_a10.c +0 -0

socfpga/spl_a10.c
index c97eacb..36003b1 100644

arch/arm/mach-socfpga/spl_a10.c +1 -1
@@ -1,6 +1,6 @@
  // SPDX-License-Identifier: GPL-2.0+
  /*
- *  Copyright (C) 2012 Altera Corporation <www.altera.com>
+ *  Copyright (C) 2012-2019 Altera Corporation <www.altera

.com

Quoted 2 lines

  */

 #include <common.h>

arch/arm/mach-socfpga/spl_a10.c +2 -0
@@ -23,6 +23,8 @@
  #include <fdtdec.h>
  #include <watchdog.h>
  #include <asm/arch/pinmux.h>
+#include <asm/arch/fpga_manager.h>
+#include <mmc.h>


 DECLARE_GLOBAL_DATA_PTR;

arch/arm/mach-socfpga/spl_a10.c +0 -0
@@ -68,11 +70,48 @@ u32 spl_boot_mode(const u32
 boot_device)


 void spl_board_init(void)
 {

arch/arm/mach-socfpga/spl_a10.c

Quoted 1 line
arch/arm/mach-socfpga/spl_a10.c +1 -0
+	char buf[16 * 1024] __aligned(ARCH_DMA_MINALIGN);

ALLOC_CACHE_ALIGN_BUFFER()

Quoted 6 lines
Quoted 2 lines

#define ARCH_DMA_MINALIGN CONFIG_SYS_CACHELINE_SIZE
They are not same thing?

See include/memalign.h and other drivers, the macro is preferred as
it
hides the details.

Okay.

Quoted 46 lines
Quoted 46 lines
Quoted 46 lines
Quoted 43 lines

+
  /* enable console uart printing */
  preloader_console_init();
  WATCHDOG_RESET();

  arch_early_init_r();

arch/arm/mach-socfpga/spl_a10.c +2 -0
+
+	/* If the full FPGA is already loaded, ie.from

EPCQ,
config fpga pins */

arch/arm/mach-socfpga/spl_a10.c +2 -0
+	if (is_fpgamgr_user_mode()) {
+		int ret = config_pins(gd->fdt_blob,

"shared");

arch/arm/mach-socfpga/spl_a10.c +9 -0
+
+		if (ret)
+			return;
+
+		ret = config_pins(gd->fdt_blob, "fpga");
+		if (ret)
+			return;
+	} else if (!is_fpgamgr_early_user_mode()) {
+		/* Program IOSSM(early IO release) or full

FPGA */

arch/arm/mach-socfpga/spl_a10.c +4 -0
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename =

get_fpga_filename(gd-

Quoted 3 lines

fdt_blob, &len);

arch/arm/mach-socfpga/spl_a10.c +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf,
Quoted 1 line

sizeof(buf), 0);

Why is this code here twice ? The same code seems to be below
...

Quoted 5 lines
Quoted 4 lines

The 1st calling for periph program, then running ddr calibration,
then
2nd calling for core program.

Then maybe the code can be deduplicated ?

Hmm...seems cannot, because
1. DDR calibration is not part of fpga code.
2. fpga driver can only be used to process one bistream at a time, because different mode has different handling.

(quoted diff) +6 -0
+	} else if (!is_fpgamgr_early_user_mode()) {
+		/* Program IOSSM(early IO release) or full FPGA */
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename = get_fpga_filename(gd-
Quoted 1 line

fdt_blob, &len);

(quoted diff) +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf,

sizeof(buf), 0);
...

(quoted diff) +5 -0
+	if (!is_fpgamgr_user_mode()) {
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename = get_fpga_filename(gd-
Quoted 1 line

fdt_blob, &len);

Quoted 8 lines
Quoted 6 lines
(quoted diff) +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf,

sizeof(buf), 0);

These two chunks look the same to me , no ?

Yes, they are being called twice at different fpga mode, and different sequence, before and after DDR calibration.

So why can't these two chunks of code be de-duplicated ?

Signature · 2 lines
Best regards,
Marek Vasut
#46 lore raw

PATCH v8 6/8 spl : socfpga: Implement fpga bitstream loading with socfpga loadfs

Quoted 189 lines · Marek Vasut
Quoted 188 lines · Tien Fong
Quoted 186 lines · Marek Vasut
Quoted 158 lines · Tien Fong
Quoted 82 lines · Marek Vasut
Quoted 79 lines · Tien Fong
Quoted 75 lines · Marek Vasut
Quoted 69 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add support for loading FPGA bitstream to get DDR up running
before
U-Boot is loaded into DDR. Boot device initialization, generic
firmware
loader and SPL FAT support are required for this whole mechanism to
work.

Quoted 1 line

---

changes for v7
- Removed casting for get_fpga_filename
- Removed hard coding DDR address for loading core bistream,
using
loadable
  property from FIT.
- Added checking for config_pins, return if error.
---
 arch/arm/mach-socfpga/spl_a10.c | 41
++++++++++++++++++++++++++++++++++++++++-

 1 file changed, 40 insertions(+), 1 deletion(-)
arch/arm/mach- renamed from arch/arm/mach-socfpga/spl_a10.c +0 -0

socfpga/spl_a10.c
index c97eacb..36003b1 100644

arch/arm/mach-socfpga/spl_a10.c +0 -1
@@ -1,6 +1,6 @@
  // SPDX-License-Identifier: GPL-2.0+
  /*
- *  Copyright (C) 2012 Altera Corporation <www.altera.

com>
+ *  Copyright (C) 2012-2019 Altera Corporation <www.al tera
.com

Quoted 3 lines

  */

 #include <common.h>

arch/arm/mach-socfpga/spl_a10.c +2 -0
@@ -23,6 +23,8 @@
  #include <fdtdec.h>
  #include <watchdog.h>
  #include <asm/arch/pinmux.h>
+#include <asm/arch/fpga_manager.h>
+#include <mmc.h>


 DECLARE_GLOBAL_DATA_PTR;

arch/arm/mach-socfpga/spl_a10.c +0 -0
@@ -68,11 +70,48 @@ u32 spl_boot_mode(const u32
 boot_device)


 void spl_board_init(void)
 {

arch/arm/mach-socfpga/spl_a10.c +1 -0
+	char buf[16 * 1024]
Quoted 1 line

__aligned(ARCH_DMA_MINALIGN);

ALLOC_CACHE_ALIGN_BUFFER()

Quoted 7 lines
Quoted 2 lines

#define ARCH_DMA_MINALIGN CONFIG_SYS_CACHELINE_SIZE
They are not same thing?

See include/memalign.h and other drivers, the macro is preferred
as
it
hides the details.

Okay.

Quoted 55 lines
Quoted 55 lines
Quoted 55 lines
Quoted 50 lines

+
  /* enable console uart printing */
  preloader_console_init();
  WATCHDOG_RESET();

  arch_early_init_r();

arch/arm/mach-socfpga/spl_a10.c +2 -0
+
+	/* If the full FPGA is already loaded, ie.from

EPCQ,
config fpga pins */

arch/arm/mach-socfpga/spl_a10.c +2 -0
+	if (is_fpgamgr_user_mode()) {
+		int ret = config_pins(gd->fdt_blob,

"shared");

arch/arm/mach-socfpga/spl_a10.c +5 -0
+
+		if (ret)
+			return;
+
+		ret = config_pins(gd->fdt_blob,

"fpga");

arch/arm/mach-socfpga/spl_a10.c +4 -0
+		if (ret)
+			return;
+	} else if (!is_fpgamgr_early_user_mode()) {
+		/* Program IOSSM(early IO release) or

full
FPGA */

arch/arm/mach-socfpga/spl_a10.c +4 -0
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename =

get_fpga_filename(gd-

Quoted 4 lines

fdt_blob, &len);

arch/arm/mach-socfpga/spl_a10.c +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo,
Quoted 2 lines

buf,
sizeof(buf), 0);

Why is this code here twice ? The same code seems to be below
...

Quoted 5 lines
Quoted 4 lines

The 1st calling for periph program, then running ddr calibration,
then
2nd calling for core program.

Then maybe the code can be deduplicated ?

Hmm...seems cannot, because
1. DDR calibration is not part of fpga code.
2. fpga driver can only be used to process one bistream at a time,
because different mode has different handling.

(quoted diff) +2 -0
+	} else if (!is_fpgamgr_early_user_mode()) {
+		/* Program IOSSM(early IO release) or full FPGA

*/

(quoted diff) +4 -0
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename = get_fpga_filename(gd-
Quoted 2 lines

fdt_blob, &len);

(quoted diff) +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf,

sizeof(buf), 0);
...

(quoted diff) +5 -0
+	if (!is_fpgamgr_user_mode()) {
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename = get_fpga_filename(gd-
Quoted 2 lines

fdt_blob, &len);

(quoted diff) +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf,

sizeof(buf), 0);

Quoted 5 lines
Quoted 4 lines
Quoted 1 line

These two chunks look the same to me , no ?

Yes, they are being called twice at different fpga mode, and different
sequence, before and after DDR calibration.

So why can't these two chunks of code be de-duplicated ?

de-duplicated means calling for one time only?
If that's the case, how de-deplicated? One call for processing one bistream, and this driver has different handling for different FPGA mode and different sequence(before and after DDR calibration).

Quoted 1 line
#47 lore raw

PATCH v8 6/8 spl : socfpga: Implement fpga bitstream loading with socfpga loadfs

Quoted 194 lines · Tien Fong
Quoted 193 lines · Marek Vasut
Quoted 192 lines · Tien Fong
Quoted 186 lines · Marek Vasut
Quoted 158 lines · Tien Fong
Quoted 82 lines · Marek Vasut
Quoted 79 lines · Tien Fong
Quoted 75 lines · Marek Vasut
Quoted 69 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add support for loading FPGA bitstream to get DDR up running
before
U-Boot is loaded into DDR. Boot device initialization, generic
firmware
loader and SPL FAT support are required for this whole mechanism to
work.

Quoted 1 line

---

changes for v7
- Removed casting for get_fpga_filename
- Removed hard coding DDR address for loading core bistream,
using
loadable
  property from FIT.
- Added checking for config_pins, return if error.
---
 arch/arm/mach-socfpga/spl_a10.c | 41
++++++++++++++++++++++++++++++++++++++++-

 1 file changed, 40 insertions(+), 1 deletion(-)
arch/arm/mach- renamed from arch/arm/mach-socfpga/spl_a10.c +0 -0

socfpga/spl_a10.c
index c97eacb..36003b1 100644

arch/arm/mach-socfpga/spl_a10.c +0 -1
@@ -1,6 +1,6 @@
  // SPDX-License-Identifier: GPL-2.0+
  /*
- *  Copyright (C) 2012 Altera Corporation <www.altera.

com>
+ *  Copyright (C) 2012-2019 Altera Corporation <www.al tera
.com

Quoted 3 lines

  */

 #include <common.h>

arch/arm/mach-socfpga/spl_a10.c +2 -0
@@ -23,6 +23,8 @@
  #include <fdtdec.h>
  #include <watchdog.h>
  #include <asm/arch/pinmux.h>
+#include <asm/arch/fpga_manager.h>
+#include <mmc.h>


 DECLARE_GLOBAL_DATA_PTR;

arch/arm/mach-socfpga/spl_a10.c +0 -0
@@ -68,11 +70,48 @@ u32 spl_boot_mode(const u32
 boot_device)


 void spl_board_init(void)
 {

arch/arm/mach-socfpga/spl_a10.c +1 -0
+	char buf[16 * 1024]
Quoted 1 line

__aligned(ARCH_DMA_MINALIGN);

ALLOC_CACHE_ALIGN_BUFFER()

Quoted 7 lines
Quoted 2 lines

#define ARCH_DMA_MINALIGN CONFIG_SYS_CACHELINE_SIZE
They are not same thing?

See include/memalign.h and other drivers, the macro is preferred
as
it
hides the details.

Okay.

Quoted 55 lines
Quoted 55 lines
Quoted 55 lines
Quoted 50 lines

+
  /* enable console uart printing */
  preloader_console_init();
  WATCHDOG_RESET();

  arch_early_init_r();

arch/arm/mach-socfpga/spl_a10.c +2 -0
+
+	/* If the full FPGA is already loaded, ie.from

EPCQ,
config fpga pins */

arch/arm/mach-socfpga/spl_a10.c +2 -0
+	if (is_fpgamgr_user_mode()) {
+		int ret = config_pins(gd->fdt_blob,

"shared");

arch/arm/mach-socfpga/spl_a10.c +5 -0
+
+		if (ret)
+			return;
+
+		ret = config_pins(gd->fdt_blob,

"fpga");

arch/arm/mach-socfpga/spl_a10.c +4 -0
+		if (ret)
+			return;
+	} else if (!is_fpgamgr_early_user_mode()) {
+		/* Program IOSSM(early IO release) or

full
FPGA */

arch/arm/mach-socfpga/spl_a10.c +4 -0
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename =

get_fpga_filename(gd-

Quoted 4 lines

fdt_blob, &len);

arch/arm/mach-socfpga/spl_a10.c +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo,
Quoted 2 lines

buf,
sizeof(buf), 0);

Why is this code here twice ? The same code seems to be below
...

Quoted 5 lines
Quoted 4 lines

The 1st calling for periph program, then running ddr calibration,
then
2nd calling for core program.

Then maybe the code can be deduplicated ?

Hmm...seems cannot, because
1. DDR calibration is not part of fpga code.
2. fpga driver can only be used to process one bistream at a time,
because different mode has different handling.

(quoted diff) +2 -0
+	} else if (!is_fpgamgr_early_user_mode()) {
+		/* Program IOSSM(early IO release) or full FPGA

*/

(quoted diff) +4 -0
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename = get_fpga_filename(gd-
Quoted 2 lines

fdt_blob, &len);

(quoted diff) +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf,

sizeof(buf), 0);
...

(quoted diff) +5 -0
+	if (!is_fpgamgr_user_mode()) {
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename = get_fpga_filename(gd-
Quoted 2 lines

fdt_blob, &len);

(quoted diff) +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf,

sizeof(buf), 0);

Quoted 1 line

These two chunks look the same to me , no ?

Yes, they are being called twice at different fpga mode, and different
sequence, before and after DDR calibration.

Quoted 5 lines
Quoted 1 line

So why can't these two chunks of code be de-duplicated ?

de-duplicated means calling for one time only?
If that's the case, how de-deplicated? One call for processing one bistream, and this driver has different handling for different FPGA mode and different sequence(before and after DDR calibration).

I mean, make the two copies of code a single function and call that function twice.

Signature · 2 lines
Best regards,
Marek Vasut
#48 lore raw

PATCH v8 6/8 spl : socfpga: Implement fpga bitstream loading with socfpga loadfs

Quoted 227 lines · Marek Vasut
Quoted 226 lines · Tien Fong
Quoted 224 lines · Marek Vasut
Quoted 222 lines · Tien Fong
Quoted 215 lines · Marek Vasut
Quoted 183 lines · Tien Fong
Quoted 93 lines · Marek Vasut
Quoted 89 lines · Tien Fong
Quoted 84 lines · Marek Vasut
Quoted 77 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add support for loading FPGA bitstream to get DDR
up
running
before
U-Boot is loaded into DDR. Boot device
initialization,
generic
firmware
loader and SPL FAT support are required for this
whole
mechanism to
work.

.com

Quoted 2 lines

---

changes for v7
- Removed casting for get_fpga_filename
- Removed hard coding DDR address for loading core bistream,
using
loadable
  property from FIT.
- Added checking for config_pins, return if error.
---
 arch/arm/mach-socfpga/spl_a10.c | 41
++++++++++++++++++++++++++++++++++++++++-

 1 file changed, 40 insertions(+), 1 deletion(-)
arch/arm/mach- renamed from arch/arm/mach-socfpga/spl_a10.c +0 -0

socfpga/spl_a10.c
index c97eacb..36003b1 100644

arch/arm/mach-socfpga/spl_a10.c +0 -1
@@ -1,6 +1,6 @@
  // SPDX-License-Identifier: GPL-2.0+
  /*
- *  Copyright (C) 2012 Altera Corporation <www.alt

era.
com>
+ *  Copyright (C) 2012-2019 Altera Corporation <ww w.al
tera
.com

Quoted 4 lines

  */

 #include <common.h>

arch/arm/mach-socfpga/spl_a10.c +2 -0
@@ -23,6 +23,8 @@
  #include <fdtdec.h>
  #include <watchdog.h>
  #include <asm/arch/pinmux.h>
+#include <asm/arch/fpga_manager.h>
+#include <mmc.h>


 DECLARE_GLOBAL_DATA_PTR;

arch/arm/mach-socfpga/spl_a10.c +0 -0
@@ -68,11 +70,48 @@ u32 spl_boot_mode(const u32
 boot_device)


 void spl_board_init(void)
 {

arch/arm/mach-socfpga/spl_a10.c +1 -0
+	char buf[16 * 1024]
Quoted 1 line

__aligned(ARCH_DMA_MINALIGN);

ALLOC_CACHE_ALIGN_BUFFER()

Quoted 8 lines
Quoted 3 lines

#define ARCH_DMA_MINALIGN CONFIG_SYS_CACHELINE_S
IZE
They are not same thing?

See include/memalign.h and other drivers, the macro is preferred
as
it
hides the details.

Okay.

Quoted 66 lines
Quoted 66 lines
Quoted 66 lines
Quoted 59 lines

+
  /* enable console uart printing */
  preloader_console_init();
  WATCHDOG_RESET();

  arch_early_init_r();

arch/arm/mach-socfpga/spl_a10.c +2 -0
+
+	/* If the full FPGA is already loaded,

ie.from
EPCQ,
config fpga pins */

arch/arm/mach-socfpga/spl_a10.c +2 -0
+	if (is_fpgamgr_user_mode()) {
+		int ret = config_pins(gd-
Quoted 1 line

fdt_blob,

"shared");

arch/arm/mach-socfpga/spl_a10.c +5 -0
+
+		if (ret)
+			return;
+
+		ret = config_pins(gd->fdt_blob,

"fpga");

arch/arm/mach-socfpga/spl_a10.c +3 -0
+		if (ret)
+			return;
+	} else if (!is_fpgamgr_early_user_mode())

{

arch/arm/mach-socfpga/spl_a10.c +1 -0
+		/* Program IOSSM(early IO release)

or
full
FPGA */

arch/arm/mach-socfpga/spl_a10.c +4 -0
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename =

get_fpga_filename(gd-

Quoted 5 lines

fdt_blob, &len);

arch/arm/mach-socfpga/spl_a10.c +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinf
Quoted 3 lines

o,
buf,
sizeof(buf), 0);

Why is this code here twice ? The same code seems to be
below
...

Quoted 5 lines
Quoted 4 lines

The 1st calling for periph program, then running ddr calibration,
then
2nd calling for core program.

Then maybe the code can be deduplicated ?

Hmm...seems cannot, because
1. DDR calibration is not part of fpga code.
2. fpga driver can only be used to process one bistream at a
time,
because different mode has different handling.

(quoted diff) +2 -0
+	} else if (!is_fpgamgr_early_user_mode()) {
+		/* Program IOSSM(early IO release) or full

FPGA
*/

(quoted diff) +4 -0
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename = get_fpga_filename(gd-
Quoted 3 lines

fdt_blob, &len);

(quoted diff) +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf,

sizeof(buf), 0);
...

(quoted diff) +5 -0
+	if (!is_fpgamgr_user_mode()) {
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename = get_fpga_filename(gd-
Quoted 3 lines

fdt_blob, &len);

(quoted diff) +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf,

sizeof(buf), 0);

Quoted 1 line

These two chunks look the same to me , no ?

Yes, they are being called twice at different fpga mode, and different
sequence, before and after DDR calibration.

Quoted 7 lines
Quoted 5 lines
Quoted 1 line

So why can't these two chunks of code be de-duplicated ?

de-duplicated means calling for one time only?
If that's the case, how de-deplicated? One call for processing one bistream, and this driver has different handling for different FPGA mode and different sequence(before and after DDR calibration).

I mean, make the two copies of code a single function and call that function twice.

Okay, i got you.

Quoted 1 line
#49 lore raw

PATCH v8 6/8 spl : socfpga: Implement fpga bitstream loading with socfpga loadfs

In reply to Chee
Quoted 146 lines · Tien Fong
Quoted 145 lines · Marek Vasut
Quoted 126 lines · Tien Fong
Quoted 68 lines · Marek Vasut
Quoted 67 lines · Tien Fong
Quoted 65 lines · Marek Vasut
Quoted 57 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add support for loading FPGA bitstream to get DDR up running
before
U-Boot is loaded into DDR. Boot device initialization, generic
firmware
loader and SPL FAT support are required for this whole mechanism to
work.

---

changes for v7
- Removed casting for get_fpga_filename
- Removed hard coding DDR address for loading core bistream,
using
loadable
property from FIT.
- Added checking for config_pins, return if error.

---
 arch/arm/mach-socfpga/spl_a10.c | 41
++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 40 insertions(+), 1 deletion(-)
arch/arm/mach- renamed from arch/arm/mach-socfpga/spl_a10.c +0 -0

socfpga/spl_a10.c
index c97eacb..36003b1 100644

arch/arm/mach-socfpga/spl_a10.c +1 -1
@@ -1,6 +1,6 @@
 // SPDX-License-Identifier: GPL-2.0+
 /*
- *  Copyright (C) 2012 Altera Corporation <www.altera.com>
+ *  Copyright (C) 2012-2019 Altera Corporation <www.altera

.com

Quoted 1 line

*/

#include <common.h>

arch/arm/mach-socfpga/spl_a10.c +2 -0
@@ -23,6 +23,8 @@
 #include <fdtdec.h>
 #include <watchdog.h>
 #include <asm/arch/pinmux.h>
+#include <asm/arch/fpga_manager.h>
+#include <mmc.h>
 
 DECLARE_GLOBAL_DATA_PTR;
 

arch/arm/mach-socfpga/spl_a10.c u32 spl_boot_mode(const u32 · line 74

Quoted 5 lines
arch/arm/mach-socfpga/spl_a10.c +1 -0
@@ -68,11 +70,48 @@ u32 spl_boot_mode(const u32
 boot_device)
 
 void spl_board_init(void)
 {
+	char buf[16 * 1024] __aligned(ARCH_DMA_MINALIGN);

ALLOC_CACHE_ALIGN_BUFFER()

Quoted 6 lines
Quoted 2 lines

#define ARCH_DMA_MINALIGN CONFIG_SYS_CACHELINE_SIZE
They are not same thing?

See include/memalign.h and other drivers, the macro is preferred as
it
hides the details.

Okay.

Quoted 41 lines
Quoted 41 lines
Quoted 41 lines
Quoted 38 lines
arch/arm/mach-socfpga/spl_a10.c +3 -0
+
 	/* enable console uart printing */
 	preloader_console_init();
 	WATCHDOG_RESET();
 
 	arch_early_init_r();
+
+	/* If the full FPGA is already loaded, ie.from
 EPCQ,

config fpga pins */

arch/arm/mach-socfpga/spl_a10.c +2 -0
+	if (is_fpgamgr_user_mode()) {
+		int ret = config_pins(gd->fdt_blob,

"shared");

arch/arm/mach-socfpga/spl_a10.c +9 -0
+
+		if (ret)
+			return;
+
+		ret = config_pins(gd->fdt_blob, "fpga");
+		if (ret)
+			return;
+	} else if (!is_fpgamgr_early_user_mode()) {
+		/* Program IOSSM(early IO release) or full

FPGA */

arch/arm/mach-socfpga/spl_a10.c +4 -0
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename =

get_fpga_filename(gd-

Quoted 2 lines

fdt_blob, &len);

arch/arm/mach-socfpga/spl_a10.c +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf,
Quoted 1 line

sizeof(buf), 0);

Why is this code here twice ? The same code seems to be below
...

Quoted 5 lines
Quoted 4 lines

The 1st calling for periph program, then running ddr calibration,
then
2nd calling for core program.

Then maybe the code can be deduplicated ?

Hmm...seems cannot, because
1. DDR calibration is not part of fpga code.
2. fpga driver can only be used to process one bistream at a time, because different mode has different handling.

(quoted diff) +6 -0
+	} else if (!is_fpgamgr_early_user_mode()) {
+		/* Program IOSSM(early IO release) or full FPGA */
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename = get_fpga_filename(gd-
Quoted 1 line

fdt_blob, &len);

(quoted diff) +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf,

sizeof(buf), 0);
...

(quoted diff) +5 -0
+	if (!is_fpgamgr_user_mode()) {
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename = get_fpga_filename(gd-
Quoted 1 line

fdt_blob, &len);

Quoted 8 lines
Quoted 6 lines
(quoted diff) +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf,

sizeof(buf), 0);

These two chunks look the same to me , no ?

Yes, they are being called twice at different fpga mode, and different sequence, before and after DDR calibration.

I believe i understand the issue here. The code is a little more convoluted then it should be. The issue actually stems from first_loading_rbf_to_buffer which behaves differently depending on the state of the fpga.
...
+ uname = fit_get_name(buffer_p, images_noffset, NULL);
+ if (uname) {
+ debug("FPGA: %s\n", uname);
+
+ if (strstr(uname, "fpga-periph") &&

+                               (!is_fpgamgr_early_user_mode() ||
+                               is_fpgamgr_user_mode())) {
+                               fpga_node_name = uname;
+                               printf("FPGA: Start to program ");
+                               printf("peripheral/full bitstream ...\n");
+                               break;
+                       } else if (strstr(uname, "fpga-core") &&
+                                       (is_fpgamgr_early_user_mode() &&
+                                       !is_fpgamgr_user_mode())) {
+                               fpga_node_name = uname;
+                               printf("FPGA: Start to program core ");
+                               printf("bitstream ...\n");
+                               break;

+ }
+ }
...
so when called with an unprogrammed fpga or a fully programmed fpga, it will program the fpga-periph. When called with a programmed periph image and unprogrammed core is will program the the core.

so socfpga_loadfs needs to be called twice to fully program the fpga.

My question is, for a configuration where only the periph image is programmed, would this not result in an erroneous error message?

+ debug("FPGA: No suitable bitstream was found, count: %d.\n", i);

I wounder if it would not be better to move the fpga image fit parsing out of the socfpga_loadfs function so

1) you could only call the fpga-core programming if needed
2) socfpga_loadfs is explicitly called to program either the core or the periph image

alternately, perhaps a call to socfpga_loadfs should program both images if present, and initialize the ddr if required?

I am not so keen on socfpga_loadfs behaving differently like this with the same function call.

--dalon

Quoted 4 lines

_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
https://lists.denx.de/listinfo/u-boot

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 3282 bytes
Desc: not available
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20190214/96593726/attachment.bin>

#50 lore raw

PATCH v8 6/8 spl : socfpga: Implement fpga bitstream loading with socfpga loadfs

Quoted 217 lines · Dalon
Quoted 177 lines · Tien Fong
Quoted 175 lines · Marek Vasut
Quoted 147 lines · Tien Fong
Quoted 77 lines · Marek Vasut
Quoted 75 lines · Tien Fong
Quoted 72 lines · Marek Vasut
Quoted 67 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add support for loading FPGA bitstream to get DDR up running
before
U-Boot is loaded into DDR. Boot device initialization, generic
firmware
loader and SPL FAT support are required for this whole mechanism to
work.

Quoted 1 line

---

changes for v7
- Removed casting for get_fpga_filename
- Removed hard coding DDR address for loading core bistream,
using
loadable
  property from FIT.
- Added checking for config_pins, return if error.
---
 arch/arm/mach-socfpga/spl_a10.c | 41
++++++++++++++++++++++++++++++++++++++++-

 1 file changed, 40 insertions(+), 1 deletion(-)
arch/arm/mach- renamed from arch/arm/mach-socfpga/spl_a10.c +0 -0

socfpga/spl_a10.c
index c97eacb..36003b1 100644

arch/arm/mach-socfpga/spl_a10.c +0 -1
@@ -1,6 +1,6 @@
  // SPDX-License-Identifier: GPL-2.0+
  /*
- *  Copyright (C) 2012 Altera Corporation <www.altera.

com>
+ *  Copyright (C) 2012-2019 Altera Corporation <www.al tera
.com

Quoted 2 lines

  */

 #include <common.h>

arch/arm/mach-socfpga/spl_a10.c +2 -0
@@ -23,6 +23,8 @@
  #include <fdtdec.h>
  #include <watchdog.h>
  #include <asm/arch/pinmux.h>
+#include <asm/arch/fpga_manager.h>
+#include <mmc.h>


 DECLARE_GLOBAL_DATA_PTR;

arch/arm/mach-socfpga/spl_a10.c +0 -0
@@ -68,11 +70,48 @@ u32 spl_boot_mode(const u32
 boot_device)


 void spl_board_init(void)
 {

arch/arm/mach-socfpga/spl_a10.c +1 -0
+	char buf[16 * 1024]
Quoted 1 line

__aligned(ARCH_DMA_MINALIGN);

ALLOC_CACHE_ALIGN_BUFFER()

Quoted 7 lines
Quoted 2 lines

#define ARCH_DMA_MINALIGN CONFIG_SYS_CACHELINE_SIZE
They are not same thing?

See include/memalign.h and other drivers, the macro is preferred
as
it
hides the details.

Okay.

Quoted 50 lines
Quoted 50 lines
Quoted 50 lines
Quoted 45 lines

+
  /* enable console uart printing */
  preloader_console_init();
  WATCHDOG_RESET();

  arch_early_init_r();

arch/arm/mach-socfpga/spl_a10.c +2 -0
+
+	/* If the full FPGA is already loaded, ie.from

EPCQ,
config fpga pins */

arch/arm/mach-socfpga/spl_a10.c +2 -0
+	if (is_fpgamgr_user_mode()) {
+		int ret = config_pins(gd->fdt_blob,

"shared");

arch/arm/mach-socfpga/spl_a10.c +5 -0
+
+		if (ret)
+			return;
+
+		ret = config_pins(gd->fdt_blob,

"fpga");

arch/arm/mach-socfpga/spl_a10.c +4 -0
+		if (ret)
+			return;
+	} else if (!is_fpgamgr_early_user_mode()) {
+		/* Program IOSSM(early IO release) or

full
FPGA */

arch/arm/mach-socfpga/spl_a10.c +4 -0
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename =

get_fpga_filename(gd-

Quoted 3 lines

fdt_blob, &len);

arch/arm/mach-socfpga/spl_a10.c +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo,
Quoted 2 lines

buf,
sizeof(buf), 0);

Why is this code here twice ? The same code seems to be below
...

Quoted 5 lines
Quoted 4 lines

The 1st calling for periph program, then running ddr calibration,
then
2nd calling for core program.

Then maybe the code can be deduplicated ?

Hmm...seems cannot, because
1. DDR calibration is not part of fpga code.
2. fpga driver can only be used to process one bistream at a time,
because different mode has different handling.

(quoted diff) +2 -0
+	} else if (!is_fpgamgr_early_user_mode()) {
+		/* Program IOSSM(early IO release) or full FPGA

*/

(quoted diff) +4 -0
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename = get_fpga_filename(gd-
Quoted 2 lines

fdt_blob, &len);

(quoted diff) +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf,

sizeof(buf), 0);
...

(quoted diff) +5 -0
+	if (!is_fpgamgr_user_mode()) {
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename = get_fpga_filename(gd-
Quoted 2 lines

fdt_blob, &len);

(quoted diff) +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf,

sizeof(buf), 0);

Quoted 4 lines
Quoted 1 line

These two chunks look the same to me , no ?

Yes, they are being called twice at different fpga mode, and different
sequence, before and after DDR calibration.

I believe i understand the issue here.  The code is a little more convoluted then it should be.  The issue actually stems from first_loading_rbf_to_buffer which behaves differently depending on the state of the fpga.
...
+               uname = fit_get_name(buffer_p, images_noffset, NULL); +               if (uname) {
+                       debug("FPGA: %s\n", uname);
+
+                       if (strstr(uname, "fpga-periph") && +                               (!is_fpgamgr_early_user_mode() || +                               is_fpgamgr_user_mode())) { +                               fpga_node_name = uname;
+                               printf("FPGA: Start to program "); +                               printf("peripheral/full bitstream ...\n");
+                               break;
+                       } else if (strstr(uname, "fpga-core") && +                                       (is_fpgamgr_early_user_mode() &&
+                                       !is_fpgamgr_user_mode())) { +                               fpga_node_name = uname;
+                               printf("FPGA: Start to program core ");
+                               printf("bitstream ...\n"); +                               break;
+                       }
+               }
...
so when called with an unprogrammed fpga or a fully programmed fpga, it will
program the fpga-periph.  When called with a programmed periph image and
unprogrammed core is will program the the core.

Quoted 8 lines

so socfpga_loadfs needs to be called twice to fully program the fpga.

My question is, for a configuration where only the periph image is programmed,
would this not result in an erroneous error message?

+               debug("FPGA: No suitable bitstream was found, count:
%d.\n", i);

This is not error, value 1 would be returned instead -ve and 0. socfpga_loadfs would skip the core configuration when 1 is detected. You would see the message print out "FPGA: Skipping configuration ..." after "FPGA: Checking FPGA configuration setting ..."

Quoted 8 lines

I wounder if it would not be better to move the fpga image fit parsing out
of the socfpga_loadfs function so

1) you could only call the fpga-core programming if needed
2) socfpga_loadfs is explicitly called to program either the core or the periph image

I don't know would this cause more messy, parsing fpga image fit job supposely done inside fpga driver.

Quoted 15 lines

alternately, perhaps a call to socfpga_loadfs should program both images if present, and initialize the ddr if required?

I am not so keen on socfpga_loadfs behaving differently like this with
the same function call.

--dalon

Quoted 5 lines

_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
https://lists.denx.de/listinfo/u-boot

#51 lore raw

PATCH v8 6/8 spl : socfpga: Implement fpga bitstream loading with socfpga loadfs

Quoted 218 lines · Tien Fong
Quoted 201 lines · Dalon
Quoted 161 lines · Tien Fong
Quoted 160 lines · Marek Vasut
Quoted 135 lines · Tien Fong
Quoted 71 lines · Marek Vasut
Quoted 70 lines · Tien Fong
Quoted 68 lines · Marek Vasut
Quoted 59 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add support for loading FPGA bitstream to get DDR up running
before
U-Boot is loaded into DDR. Boot device initialization, generic
firmware
loader and SPL FAT support are required for this whole mechanism to
work.

---

changes for v7
- Removed casting for get_fpga_filename
- Removed hard coding DDR address for loading core bistream,
using
loadable
property from FIT.
- Added checking for config_pins, return if error.

---
 arch/arm/mach-socfpga/spl_a10.c | 41
++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 40 insertions(+), 1 deletion(-)
arch/arm/mach- renamed from arch/arm/mach-socfpga/spl_a10.c +0 -0

socfpga/spl_a10.c
index c97eacb..36003b1 100644

arch/arm/mach-socfpga/spl_a10.c +0 -1
@@ -1,6 +1,6 @@
 // SPDX-License-Identifier: GPL-2.0+
 /*
- *  Copyright (C) 2012 Altera Corporation <www.altera.

com>
+ * Copyright (C) 2012-2019 Altera Corporation <www.al tera
.com

Quoted 1 line

*/

#include <common.h>

arch/arm/mach-socfpga/spl_a10.c +2 -0
@@ -23,6 +23,8 @@
 #include <fdtdec.h>
 #include <watchdog.h>
 #include <asm/arch/pinmux.h>
+#include <asm/arch/fpga_manager.h>
+#include <mmc.h>
 
 DECLARE_GLOBAL_DATA_PTR;
 

arch/arm/mach-socfpga/spl_a10.c u32 spl_boot_mode(const u32 · line 75

Quoted 6 lines
arch/arm/mach-socfpga/spl_a10.c +1 -0
@@ -68,11 +70,48 @@ u32 spl_boot_mode(const u32
 boot_device)
 
 void spl_board_init(void)
 {
+	char buf[16 * 1024]
 __aligned(ARCH_DMA_MINALIGN);

ALLOC_CACHE_ALIGN_BUFFER()

Quoted 7 lines
Quoted 2 lines

#define ARCH_DMA_MINALIGN CONFIG_SYS_CACHELINE_SIZE
They are not same thing?

See include/memalign.h and other drivers, the macro is preferred
as
it
hides the details.

Okay.

Quoted 45 lines
Quoted 45 lines
Quoted 45 lines
Quoted 40 lines
arch/arm/mach-socfpga/spl_a10.c +3 -0
+
 	/* enable console uart printing */
 	preloader_console_init();
 	WATCHDOG_RESET();
 
 	arch_early_init_r();
+
+	/* If the full FPGA is already loaded, ie.from
 EPCQ,

config fpga pins */

arch/arm/mach-socfpga/spl_a10.c +2 -0
+	if (is_fpgamgr_user_mode()) {
+		int ret = config_pins(gd->fdt_blob,

"shared");

arch/arm/mach-socfpga/spl_a10.c +5 -0
+
+		if (ret)
+			return;
+
+		ret = config_pins(gd->fdt_blob,

"fpga");

arch/arm/mach-socfpga/spl_a10.c +4 -0
+		if (ret)
+			return;
+	} else if (!is_fpgamgr_early_user_mode()) {
+		/* Program IOSSM(early IO release) or

full
FPGA */

arch/arm/mach-socfpga/spl_a10.c +4 -0
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename =

get_fpga_filename(gd-

Quoted 2 lines

fdt_blob, &len);

arch/arm/mach-socfpga/spl_a10.c +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo,
Quoted 2 lines

buf,
sizeof(buf), 0);

Why is this code here twice ? The same code seems to be below
...

Quoted 5 lines
Quoted 4 lines

The 1st calling for periph program, then running ddr calibration,
then
2nd calling for core program.

Then maybe the code can be deduplicated ?

Hmm...seems cannot, because
1. DDR calibration is not part of fpga code.
2. fpga driver can only be used to process one bistream at a time,
because different mode has different handling.

(quoted diff) +2 -0
+	} else if (!is_fpgamgr_early_user_mode()) {
+		/* Program IOSSM(early IO release) or full FPGA

*/

(quoted diff) +4 -0
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename = get_fpga_filename(gd-
Quoted 1 line

fdt_blob, &len);

(quoted diff) +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf,

sizeof(buf), 0);
...

(quoted diff) +5 -0
+	if (!is_fpgamgr_user_mode()) {
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename = get_fpga_filename(gd-
Quoted 1 line

fdt_blob, &len);

(quoted diff) +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf,

sizeof(buf), 0);

Quoted 4 lines
Quoted 1 line

These two chunks look the same to me , no ?

Yes, they are being called twice at different fpga mode, and different
sequence, before and after DDR calibration.

I believe i understand the issue here. The code is a little more convoluted then it should be. The issue actually stems from first_loading_rbf_to_buffer which behaves differently depending on the state of the fpga.
...

(quoted diff) +10 -0
+               uname = fit_get_name(buffer_p, images_noffset, NULL);
+               if (uname) {
+                       debug("FPGA: %s\n", uname);
+
+                       if (strstr(uname, "fpga-periph") &&
+                               (!is_fpgamgr_early_user_mode() ||
+                               is_fpgamgr_user_mode())) {
+                               fpga_node_name = uname;
+                               printf("FPGA: Start to program ");
+                               printf("peripheral/full bitstream

...\n");

(quoted diff) +3 -0
+                               break;
+                       } else if (strstr(uname, "fpga-core") &&
+                                       (is_fpgamgr_early_user_mode()

&&

(quoted diff) +3 -0
+                                       !is_fpgamgr_user_mode())) {
+                               fpga_node_name = uname;
+                               printf("FPGA: Start to program core

");

(quoted diff) +4 -0
+                               printf("bitstream ...\n");
+                               break;
+                       }
+               }
...

so when called with an unprogrammed fpga or a fully programmed fpga, it will
program the fpga-periph. When called with a programmed periph image and
unprogrammed core is will program the the core.

Quoted 8 lines

so socfpga_loadfs needs to be called twice to fully program the fpga.

My question is, for a configuration where only the periph image is programmed,
would this not result in an erroneous error message?

+ debug("FPGA: No suitable bitstream was found, count:
%d.\n", i);

This is not error, value 1 would be returned instead -ve and 0. socfpga_loadfs would skip the core configuration when 1 is detected. You would see the message print out "FPGA: Skipping configuration ..." after "FPGA: Checking FPGA configuration setting ..."

Quoted 4 lines

I wounder if it would not be better to move the fpga image fit parsing out
of the socfpga_loadfs function so

Quoted 5 lines
Quoted 3 lines

1) you could only call the fpga-core programming if needed
2) socfpga_loadfs is explicitly called to program either the core or the periph image

I don't know would this cause more messy, parsing fpga image fit job supposely done inside fpga driver.

i agree, but perhaps the ddr calibration call should be in socfpga_loadfs? and perhaps a single call to socfpga_loadfs should program both the periph and core images if present?

Quoted 9 lines
Quoted 9 lines

alternately, perhaps a call to socfpga_loadfs should program both images if present, and initialize the ddr if required?

I am not so keen on socfpga_loadfs behaving differently like this with
the same function call.

--dalon

Quoted 8 lines
Quoted 4 lines
Quoted 4 lines

_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
https://lists.denx.de/listinfo/u-boot

_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
https://lists.denx.de/listinfo/u-boot

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 3282 bytes
Desc: not available
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20190214/dd0f3ef7/attachment.bin>

#52 lore raw

PATCH v8 6/8 spl : socfpga: Implement fpga bitstream loading with socfpga loadfs

In reply to Westergreen
Quoted 230 lines · Dalon
Quoted 177 lines · Tien Fong
Quoted 175 lines · Marek Vasut
Quoted 147 lines · Tien Fong
Quoted 77 lines · Marek Vasut
Quoted 75 lines · Tien Fong
Quoted 72 lines · Marek Vasut
Quoted 67 lines · tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Add support for loading FPGA bitstream to get DDR up running
before
U-Boot is loaded into DDR. Boot device initialization, generic
firmware
loader and SPL FAT support are required for this whole mechanism to
work.

Quoted 1 line

---

changes for v7
- Removed casting for get_fpga_filename
- Removed hard coding DDR address for loading core bistream,
using
loadable
  property from FIT.
- Added checking for config_pins, return if error.
---
 arch/arm/mach-socfpga/spl_a10.c | 41
++++++++++++++++++++++++++++++++++++++++-

 1 file changed, 40 insertions(+), 1 deletion(-)
arch/arm/mach- renamed from arch/arm/mach-socfpga/spl_a10.c +0 -0

socfpga/spl_a10.c
index c97eacb..36003b1 100644

arch/arm/mach-socfpga/spl_a10.c +0 -1
@@ -1,6 +1,6 @@
  // SPDX-License-Identifier: GPL-2.0+
  /*
- *  Copyright (C) 2012 Altera Corporation <www.altera.

com>
+ *  Copyright (C) 2012-2019 Altera Corporation <www.al tera
.com

Quoted 2 lines

  */

 #include <common.h>

arch/arm/mach-socfpga/spl_a10.c +2 -0
@@ -23,6 +23,8 @@
  #include <fdtdec.h>
  #include <watchdog.h>
  #include <asm/arch/pinmux.h>
+#include <asm/arch/fpga_manager.h>
+#include <mmc.h>


 DECLARE_GLOBAL_DATA_PTR;

arch/arm/mach-socfpga/spl_a10.c +0 -0
@@ -68,11 +70,48 @@ u32 spl_boot_mode(const u32
 boot_device)


 void spl_board_init(void)
 {

arch/arm/mach-socfpga/spl_a10.c +1 -0
+	char buf[16 * 1024]
Quoted 1 line

__aligned(ARCH_DMA_MINALIGN);

ALLOC_CACHE_ALIGN_BUFFER()

Quoted 7 lines
Quoted 2 lines

#define ARCH_DMA_MINALIGN CONFIG_SYS_CACHELINE_SIZE
They are not same thing?

See include/memalign.h and other drivers, the macro is preferred
as
it
hides the details.

Okay.

Quoted 50 lines
Quoted 50 lines
Quoted 50 lines
Quoted 45 lines

+
  /* enable console uart printing */
  preloader_console_init();
  WATCHDOG_RESET();

  arch_early_init_r();

arch/arm/mach-socfpga/spl_a10.c +2 -0
+
+	/* If the full FPGA is already loaded, ie.from

EPCQ,
config fpga pins */

arch/arm/mach-socfpga/spl_a10.c +2 -0
+	if (is_fpgamgr_user_mode()) {
+		int ret = config_pins(gd->fdt_blob,

"shared");

arch/arm/mach-socfpga/spl_a10.c +5 -0
+
+		if (ret)
+			return;
+
+		ret = config_pins(gd->fdt_blob,

"fpga");

arch/arm/mach-socfpga/spl_a10.c +4 -0
+		if (ret)
+			return;
+	} else if (!is_fpgamgr_early_user_mode()) {
+		/* Program IOSSM(early IO release) or

full
FPGA */

arch/arm/mach-socfpga/spl_a10.c +4 -0
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename =

get_fpga_filename(gd-

Quoted 3 lines

fdt_blob, &len);

arch/arm/mach-socfpga/spl_a10.c +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo,
Quoted 2 lines

buf,
sizeof(buf), 0);

Why is this code here twice ? The same code seems to be below
...

Quoted 5 lines
Quoted 4 lines

The 1st calling for periph program, then running ddr calibration,
then
2nd calling for core program.

Then maybe the code can be deduplicated ?

Hmm...seems cannot, because
1. DDR calibration is not part of fpga code.
2. fpga driver can only be used to process one bistream at a time,
because different mode has different handling.

(quoted diff) +2 -0
+	} else if (!is_fpgamgr_early_user_mode()) {
+		/* Program IOSSM(early IO release) or full FPGA

*/

(quoted diff) +4 -0
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename = get_fpga_filename(gd-
Quoted 2 lines

fdt_blob, &len);

(quoted diff) +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf,

sizeof(buf), 0);
...

(quoted diff) +5 -0
+	if (!is_fpgamgr_user_mode()) {
+		fpga_fs_info fpga_fsinfo;
+		int len;
+
+		fpga_fsinfo.filename = get_fpga_filename(gd-
Quoted 2 lines

fdt_blob, &len);

(quoted diff) +3 -0
+
+		if (fpga_fsinfo.filename)
+			socfpga_loadfs(&fpga_fsinfo, buf,

sizeof(buf), 0);

Quoted 4 lines
Quoted 1 line

These two chunks look the same to me , no ?

Yes, they are being called twice at different fpga mode, and different
sequence, before and after DDR calibration.

I believe i understand the issue here.  The code is a little more convoluted then it should be.  The issue actually stems from first_loading_rbf_to_buffer which behaves differently depending on the state of the fpga.
...
+               uname = fit_get_name(buffer_p, images_noffset, NULL); +               if (uname) {
+                       debug("FPGA: %s\n", uname);
+
+                       if (strstr(uname, "fpga-periph") && +                               (!is_fpgamgr_early_user_mode() || +                               is_fpgamgr_user_mode())) { +                               fpga_node_name = uname;
+                               printf("FPGA: Start to program "); +                               printf("peripheral/full bitstream ...\n");
+                               break;
+                       } else if (strstr(uname, "fpga-core") && +                                       (is_fpgamgr_early_user_mode() &&
+                                       !is_fpgamgr_user_mode())) { +                               fpga_node_name = uname;
+                               printf("FPGA: Start to program core ");
+                               printf("bitstream ...\n"); +                               break;
+                       }
+               }
...
so when called with an unprogrammed fpga or a fully programmed fpga, it will
program the fpga-periph.  When called with a programmed periph image and
unprogrammed core is will program the the core.

so socfpga_loadfs needs to be called twice to fully program the fpga.

My question is, for a configuration where only the periph image is programmed,
would this not result in an erroneous error message?

+               debug("FPGA: No suitable bitstream was found, count:
%d.\n", i);

I wounder if it would not be better to move the fpga image fit parsing out
of the socfpga_loadfs function so

Quoted 6 lines

1) you could only call the fpga-core programming if needed
2) socfpga_loadfs is explicitly called to program either the core or the periph image

alternately, perhaps a call to socfpga_loadfs should program both images if present, and initialize the ddr if required?

DDR calibration is not a FPGA stuff, so i think it should not be part of FPGA driver code.
I believe a call to socfpga_loadfs would be more complicated for both images. What about for the case only periph program is required?

Quoted 12 lines

I am not so keen on socfpga_loadfs behaving differently like this with
the same function call.

--dalon

Quoted 5 lines

_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
https://lists.denx.de/listinfo/u-boot

#53 lore raw

PATCH v8 7/8 ARM: socfpga: Synchronize the configuration for A10 SoCDK

In reply to tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

Update the default configuration file to enable the necessary functionality the get the kit working.

---

changes for v8
- Moved the FIT related configs to the patch of configuration for FPGA SoCFPGA A10 SoCDK.

changes for v7
- Keep minimal configs.

---
 configs/socfpga_arria10_defconfig | 15 +++++++++------
 1 file changed, 9 insertions(+), 6 deletions(-)
configs/socfpga_arria10_defconfig +9 -6
index c870543..bdbf90e 100644
@@ -10,10 +10,13 @@ CONFIG_NR_DRAM_BANKS=1
 CONFIG_USE_BOOTARGS=y
 CONFIG_BOOTARGS="console=ttyS0,115200"
 # CONFIG_USE_BOOTCOMMAND is not set
+CONFIG_SYS_CONSOLE_IS_IN_ENV=y
+CONFIG_SYS_CONSOLE_OVERWRITE_ROUTINE=y
+CONFIG_SYS_CONSOLE_ENV_OVERWRITE=y
 CONFIG_DEFAULT_FDT_FILE="socfpga_arria10_socdk_sdmmc.dtb"
+CONFIG_VERSION_VARIABLE=y
 CONFIG_DISPLAY_BOARDINFO_LATE=y
 CONFIG_SPL_FPGA_SUPPORT=y
-CONFIG_SPL_SPI_LOAD=y
 CONFIG_CMD_ASKENV=y
 CONFIG_CMD_GREPENV=y
 # CONFIG_CMD_FLASH is not set
@@ -22,9 +25,7 @@ CONFIG_CMD_MMC=y
 CONFIG_CMD_CACHE=y
 CONFIG_CMD_EXT4_WRITE=y
 CONFIG_MTDIDS_DEFAULT="nor0=ff705000.spi.0"
-# CONFIG_SPL_DOS_PARTITION is not set
-# CONFIG_ISO_PARTITION is not set
-# CONFIG_EFI_PARTITION is not set
+CONFIG_OF_SPL_REMOVE_PROPS="interrupts interrupt-parent dmas dma-names"
 CONFIG_DEFAULT_DEVICE_TREE="socfpga_arria10_socdk_sdmmc"
 CONFIG_ENV_IS_IN_MMC=y
 CONFIG_SPL_ENV_SUPPORT=y
@@ -32,7 +33,6 @@ CONFIG_SPL_DM=y
 CONFIG_SPL_DM_SEQ_ALIAS=y
 CONFIG_SPL_DM_MMC=y
 CONFIG_SPL_MMC_SUPPORT=y
-CONFIG_SPL_EXT_SUPPORT=y
 CONFIG_SPL_FS_FAT=y
 CONFIG_SPL_DRIVERS_MISC_SUPPORT=y
 CONFIG_FS_LOADER=y
@@ -43,11 +43,14 @@ CONFIG_DM_GPIO=y
 CONFIG_DWAPB_GPIO=y
 CONFIG_DM_MMC=y
 CONFIG_MTD_DEVICE=y
+CONFIG_MMC_DW=y
+CONFIG_PHY_MICREL=y
+CONFIG_PHY_MICREL_KSZ90X1=y
 CONFIG_DM_ETH=y
 CONFIG_ETH_DESIGNWARE=y
 CONFIG_MII=y
+CONFIG_SYS_NS16550=y
 CONFIG_SPI=y
 CONFIG_TIMER=y
 CONFIG_SPL_TIMER=y
 CONFIG_DESIGNWARE_APB_TIMER=y
-CONFIG_USE_TINY_PRINTF=y
Signature · 1 line
2.2.0
#54 lore raw

PATCH v8 8/8 ARM: socfpga: Increase Malloc pool size to support FAT filesystem in SPL

In reply to tien.fong.chee at intel.com

From: Tien Fong Chee <[email protected]>

After some series of patches to maximise reusable of memory pool, here come to result of reasonable size required for whole SDMMC boot working on A10 SoCDK. Size required come from default max cluster(0x10000) + others(0x2000) + additional memory for headroom(0x3000).

---

changes for v7
- Added 0x3000 for memory headroom.

---
 include/configs/socfpga_common.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
include/configs/socfpga_common.h +2 -2
index 4551cb2..548b458 100644
@@ -1,6 +1,6 @@
 /* SPDX-License-Identifier: GPL-2.0+ */
 /*
- * Copyright (C) 2012 Altera Corporation <www.altera.com>
+ * Copyright (C) 2012-2019 Altera Corporation <www.altera.com>
  */
 #ifndef __CONFIG_SOCFPGA_COMMON_H__
 #define __CONFIG_SOCFPGA_COMMON_H__
@@ -258,7 +258,7 @@ unsigned int cm_get_qspi_controller_clk_hz(void);
 #if defined(CONFIG_TARGET_SOCFPGA_ARRIA10)
 /* SPL memory allocation configuration, this is for FAT implementation */
 #ifndef CONFIG_SYS_SPL_MALLOC_START
-#define CONFIG_SYS_SPL_MALLOC_SIZE	0x00010000
+#define CONFIG_SYS_SPL_MALLOC_SIZE	0x00015000
 #define CONFIG_SYS_SPL_MALLOC_START	(CONFIG_SYS_INIT_RAM_SIZE - \
 					 CONFIG_SYS_SPL_MALLOC_SIZE + \
 					 CONFIG_SYS_INIT_RAM_ADDR)
Signature · 1 line
2.2.0