Skip to contentSkip to conversation

config: imx: Define CFG_SYS_UBOOT_BASE for Phytec's imx8mm SoM QSPI boot

PATCH In review Lukasz Majewski 3 participants 5 messages Aug 20, 2024 – Aug 30, 2024 Arm i.MX and Layerscape

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 config: imx: Define CFG_SYS_UBOOT_BASE for Phytec's imx8mm SoM QSPI boot

The image offset when booting from SPI-NOR (QSPI, FSPI driver) is different than one for SD card / eMMC as extra space (0x1000) is consumed by FSPI configuration header (CONFIG_FSPI_CONF_HEADER).

---
Changes for v2:
- Avoid line breaks with defines

---
 include/configs/phycore_imx8mm.h | 7 +++++++
 1 file changed, 7 insertions(+)
include/configs/phycore_imx8mm.h +7 -0
index 0910ae2d87..a445d842ac 100644
@@ -11,8 +11,15 @@
 #include <linux/stringify.h>
 #include <asm/arch/imx-regs.h>
 
+#define UBOOT_ITB_OFFSET			0x57C00
+#define FSPI_CONF_BLOCK_SIZE		0x1000
+#define UBOOT_ITB_OFFSET_FSPI (UBOOT_ITB_OFFSET + FSPI_CONF_BLOCK_SIZE)
+#ifdef CONFIG_FSPI_CONF_HEADER
+#define CFG_SYS_UBOOT_BASE (QSPI0_AMBA_BASE + UBOOT_ITB_OFFSET_FSPI)
+#else
 #define CFG_SYS_UBOOT_BASE \
 		(QSPI0_AMBA_BASE + CONFIG_SYS_MMCSD_RAW_MODE_U_BOOT_SECTOR * 512)
+#endif
 
 #ifdef CONFIG_SPL_BUILD
 /* malloc f used before GD_FLG_FULL_MALLOC_INIT set */
Signature · 1 line
2.39.2
reviewer
#1 lore raw

Hi Lukasz,

Quoted 9 lines · Lukasz Majewski

The image offset when booting from SPI-NOR (QSPI, FSPI driver) is different than one for SD card / eMMC as extra space (0x1000) is consumed by FSPI configuration header (CONFIG_FSPI_CONF_HEADER).

---
Changes for v2:
- Avoid line breaks with defines

From the other thread, my understanding is that QSPI boot is still not working with this patch applied.

Please resend this patch when the QSPI boot problem is resolved.

Thanks

#2 lore raw

Hi Fabio,

Quoted 9 lines

Hi Lukasz,

Quoted 5 lines · Lukasz Majewski

The image offset when booting from SPI-NOR (QSPI, FSPI driver) is different than one for SD card / eMMC as extra space (0x1000) is consumed by FSPI configuration header (CONFIG_FSPI_CONF_HEADER).

Quoted 8 lines
Quoted 4 lines

---
Changes for v2:
- Avoid line breaks with defines

From the other thread, my understanding is that QSPI boot is still not working with this patch applied.

But this is an orthogonal problem - the problem is with binman image generation.

This patch follows the code already present in e.g. imx8mm_beacon* boards.

Quoted 2 lines

Please resend this patch when the QSPI boot problem is resolved.

As I've written above - problem is solely with binman code generation - after reverting 3 last commits from imx8mm-u-boot.dtsi the image is correctly generated.

Quoted 1 line

Thanks

Best regards,

Lukasz Majewski

--

DENX Software Engineering GmbH, Managing Director: Erika Unter
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-59 Fax: (+49)-8142-66989-80 Email: [email protected]

#3 lore raw

Hi Lukasz,

Quoted 20 lines · Lukasz Majewski

Hi Fabio,

Quoted 8 lines

Hi Lukasz,

Quoted 4 lines · Lukasz Majewski

The image offset when booting from SPI-NOR (QSPI, FSPI driver) is different than one for SD card / eMMC as extra space (0x1000) is consumed by FSPI configuration header (CONFIG_FSPI_CONF_HEADER).

Quoted 7 lines
Quoted 4 lines

---
Changes for v2:
- Avoid line breaks with defines

From the other thread, my understanding is that QSPI boot is still not
working with this patch applied.

But this is an orthogonal problem - the problem is with binman image generation.

Quoted 8 lines

This patch follows the code already present in e.g. imx8mm_beacon* boards.

Quoted 2 lines

Please resend this patch when the QSPI boot problem is resolved.

As I've written above - problem is solely with binman code generation - after reverting 3 last commits from imx8mm-u-boot.dtsi the image is correctly generated.

I can also see this. When I apply all your patches and revert those 3 commits, SDP boot via uuu and flashing SPI NOR flash with the qspi Image works. And then also SPI boot works.
When I don't revert the 3 commit, it does not work. (SDP boot via uuu does not work and the Image does not boot when flashed into SPI NOR flash).

I haven't investigated what the problem is or how it can be fixed though. So I guess further investigation is needed here.

Benjamin

Quoted 14 lines
Quoted 2 lines

Thanks

Best regards,

Lukasz Majewski

--

DENX Software Engineering GmbH, Managing Director: Erika Unter
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-59 Fax: (+49)-8142-66989-80 Email: [email protected]

#4 lore raw

Hi Benjamin,

Quoted 39 lines

Hi Lukasz,

Quoted 20 lines · Lukasz Majewski

Hi Fabio,

Quoted 8 lines

Hi Lukasz,

Quoted 4 lines · Lukasz Majewski

The image offset when booting from SPI-NOR (QSPI, FSPI driver) is different than one for SD card / eMMC as extra space (0x1000) is consumed by FSPI configuration header (CONFIG_FSPI_CONF_HEADER).

Quoted 7 lines
Quoted 4 lines

---
Changes for v2:
- Avoid line breaks with defines

From the other thread, my understanding is that QSPI boot is
still not working with this patch applied.

But this is an orthogonal problem - the problem is with binman image generation.

Quoted 8 lines

This patch follows the code already present in e.g. imx8mm_beacon* boards.

Quoted 2 lines

Please resend this patch when the QSPI boot problem is resolved.

As I've written above - problem is solely with binman code generation - after reverting 3 last commits from imx8mm-u-boot.dtsi the image is correctly generated.

I can also see this. When I apply all your patches and revert those 3 commits, SDP boot via uuu and flashing SPI NOR flash with the qspi Image works. And then also SPI boot works.
When I don't revert the 3 commit, it does not work. (SDP boot via uuu does not work and the Image does not boot when flashed into SPI NOR flash).

Quoted 2 lines

I haven't investigated what the problem is or how it can be fixed though.

The "Hackish" solution is to add:

@@ -85,6 +85,10 @@ void ddr_load_train_firmware(enum fw_type type)

                }
        }

+ imem_start -= 0x1000;
+ dmem_start -= 0x1000;
+

        pr_from32 = imem_start;
        pr_to32 = IMEM_OFFSET_ADDR;
        for (i = 0x0; i < imem_len; ) {

(The -= 0x1000 is the size of QSPI header added to the beginning of the image.)

Simon, has recently prepared set of binman related patches, so maybe they provided offset to fix it.

Quoted 21 lines

So I guess further investigation is needed here.

Benjamin

Quoted 15 lines
Quoted 2 lines

Thanks

Best regards,

Lukasz Majewski

--

DENX Software Engineering GmbH, Managing Director: Erika Unter
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany Phone: (+49)-8142-66989-59 Fax: (+49)-8142-66989-80 Email:
[email protected]

Best regards,

Lukasz Majewski

--

DENX Software Engineering GmbH, Managing Director: Erika Unter
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-59 Fax: (+49)-8142-66989-80 Email: [email protected]