[U-Boot] imx6dl_mamoj broken

Hi Raffaele, hi Simone,
your imx6dl_mamoj is broken on u-boot-imx. Reason is that SPL overrruns and it is bigger as the OCRAM on the DL SOC.
arm: + imx6dl_mamoj +spl/u-boot-spl.bin exceeds file size limit: + limit: 53248 bytes + actual: 58597 bytes + excess: 5349 bytes
+make[1]: *** [spl/u-boot-spl.bin] Error 1 +make[1]: *** Deleting file `spl/u-boot-spl.bin' +make: *** [sub-make] Error 2
Can you check it and fix it, please ? You see the error here:
https://travis-ci.org/sbabic/u-boot-imx/jobs/584953098
Best regards, Stefano Babic

Hi Stefano,
On Mon, Sep 16, 2019 at 1:23 PM Stefano Babic sbabic@denx.de wrote:
Hi Raffaele, hi Simone,
your imx6dl_mamoj is broken on u-boot-imx. Reason is that SPL overrruns and it is bigger as the OCRAM on the DL SOC.
arm: + imx6dl_mamoj
+spl/u-boot-spl.bin exceeds file size limit:
- limit: 53248 bytes
- actual: 58597 bytes
- excess: 5349 bytes
+make[1]: *** [spl/u-boot-spl.bin] Error 1 +make[1]: *** Deleting file `spl/u-boot-spl.bin' +make: *** [sub-make] Error 2
Any idea which commit make this blow-out. we didn't do anything with this board recently. I guess some default CLK framework or so?

On 16/09/19 10:06, Jagan Teki wrote:
Hi Stefano,
On Mon, Sep 16, 2019 at 1:23 PM Stefano Babic sbabic@denx.de wrote:
Hi Raffaele, hi Simone,
your imx6dl_mamoj is broken on u-boot-imx. Reason is that SPL overrruns and it is bigger as the OCRAM on the DL SOC.
arm: + imx6dl_mamoj
+spl/u-boot-spl.bin exceeds file size limit:
- limit: 53248 bytes
- actual: 58597 bytes
- excess: 5349 bytes
+make[1]: *** [spl/u-boot-spl.bin] Error 1 +make[1]: *** Deleting file `spl/u-boot-spl.bin' +make: *** [sub-make] Error 2
Any idea which commit make this blow-out. we didn't do anything with this board recently.
Yes, I took a look at its include file, there is nothing new or obciously big.
I guess some default CLK framework or so?
I suppose the same, yes.
Regards, Stefano

Hi Jagan,
On Mon, 16 Sep 2019 13:36:03 +0530 Jagan Teki jagan@amarulasolutions.com wrote: ...
arm: + imx6dl_mamoj
+spl/u-boot-spl.bin exceeds file size limit:
- limit: 53248 bytes
- actual: 58597 bytes
- excess: 5349 bytes
+make[1]: *** [spl/u-boot-spl.bin] Error 1 +make[1]: *** Deleting file `spl/u-boot-spl.bin' +make: *** [sub-make] Error 2
Any idea which commit make this blow-out. we didn't do anything with this board recently. I guess some default CLK framework or so?
This check fails since the recent changes in SPL size limit check, see commit 25d38f9b34be (imx6: spl: Reduce SPL limit size in case CONFIG_SECURE_BOOT is enabled).
i.MX6DL has 128KiB OCRAM, probably we could increase default SPL_SIZE_LIMIT config in common/spl/Kconfig for i.MX6DL?
-- Anatolij

Hi Anatolji,
On 16/09/19 22:20, Anatolij Gustschin wrote:
Hi Jagan,
On Mon, 16 Sep 2019 13:36:03 +0530 Jagan Teki jagan@amarulasolutions.com wrote: ...
arm: + imx6dl_mamoj
+spl/u-boot-spl.bin exceeds file size limit:
- limit: 53248 bytes
- actual: 58597 bytes
- excess: 5349 bytes
+make[1]: *** [spl/u-boot-spl.bin] Error 1 +make[1]: *** Deleting file `spl/u-boot-spl.bin' +make: *** [sub-make] Error 2
Any idea which commit make this blow-out. we didn't do anything with this board recently. I guess some default CLK framework or so?
This check fails since the recent changes in SPL size limit check, see commit 25d38f9b34be (imx6: spl: Reduce SPL limit size in case CONFIG_SECURE_BOOT is enabled).
i.MX6DL has 128KiB OCRAM, probably we could increase default SPL_SIZE_LIMIT config in common/spl/Kconfig for i.MX6DL?
Well, the current size was computed for the Solo /DL - Dual / Quad has twice OCRAM. According to chapter "8.4 Internal ROM / RAM map", for Solo /Dl there is 68KB available for user (address 0x907000--0x917FFF). I do not remember the details, but then we have to reserve space for gd, stack, etc. This leads to the actual value.
Regards, Stefano

On 2019-09-16 3:08 p.m., Stefano Babic wrote:
Hi Anatolji,
On 16/09/19 22:20, Anatolij Gustschin wrote:
Hi Jagan,
On Mon, 16 Sep 2019 13:36:03 +0530 Jagan Teki jagan@amarulasolutions.com wrote: ...
arm: + imx6dl_mamoj
+spl/u-boot-spl.bin exceeds file size limit:
- limit: 53248 bytes
- actual: 58597 bytes
- excess: 5349 bytes
+make[1]: *** [spl/u-boot-spl.bin] Error 1 +make[1]: *** Deleting file `spl/u-boot-spl.bin' +make: *** [sub-make] Error 2
Any idea which commit make this blow-out. we didn't do anything with this board recently. I guess some default CLK framework or so?
This check fails since the recent changes in SPL size limit check, see commit 25d38f9b34be (imx6: spl: Reduce SPL limit size in case CONFIG_SECURE_BOOT is enabled).
i.MX6DL has 128KiB OCRAM, probably we could increase default SPL_SIZE_LIMIT config in common/spl/Kconfig for i.MX6DL?
Well, the current size was computed for the Solo /DL - Dual / Quad has twice OCRAM. According to chapter "8.4 Internal ROM / RAM map", for Solo /Dl there is 68KB available for user (address 0x907000--0x917FFF). I do not remember the details, but then we have to reserve space for gd, stack, etc. This leads to the actual value.
The patch I posted a while ago ("[PATCH v2] ARM: imx: Support larger SPL size on IMX6DQ") may be helpful for this case. If the platform doesn't need to support the "lite" chip versions with smaller OCRAM size, then the available size for the SPL can be increased.

Hi Robert,
On Mon, Sep 16, 2019 at 8:38 PM Robert Hancock hancock@sedsystems.ca wrote:
The patch I posted a while ago ("[PATCH v2] ARM: imx: Support larger SPL size on IMX6DQ") may be helpful for this case. If the platform doesn't need to support the "lite" chip versions with smaller OCRAM size, then the available size for the SPL can be increased.
I do think your patch is useful and I hope it can be applied, but I think it does not help in this case as the target name is imx6dl_mamoj.
Thanks

Hi Robert,
On 17/09/19 01:37, Robert Hancock wrote:
On 2019-09-16 3:08 p.m., Stefano Babic wrote:
Hi Anatolji,
On 16/09/19 22:20, Anatolij Gustschin wrote:
Hi Jagan,
On Mon, 16 Sep 2019 13:36:03 +0530 Jagan Teki jagan@amarulasolutions.com wrote: ...
arm: + imx6dl_mamoj +spl/u-boot-spl.bin exceeds file size limit: + limit: 53248 bytes + actual: 58597 bytes + excess: 5349 bytes
+make[1]: *** [spl/u-boot-spl.bin] Error 1 +make[1]: *** Deleting file `spl/u-boot-spl.bin' +make: *** [sub-make] Error 2
Any idea which commit make this blow-out. we didn't do anything with this board recently. I guess some default CLK framework or so?
This check fails since the recent changes in SPL size limit check, see commit 25d38f9b34be (imx6: spl: Reduce SPL limit size in case CONFIG_SECURE_BOOT is enabled).
i.MX6DL has 128KiB OCRAM, probably we could increase default SPL_SIZE_LIMIT config in common/spl/Kconfig for i.MX6DL?
Well, the current size was computed for the Solo /DL - Dual / Quad has twice OCRAM. According to chapter "8.4 Internal ROM / RAM map", for Solo /Dl there is 68KB available for user (address 0x907000--0x917FFF). I do not remember the details, but then we have to reserve space for gd, stack, etc. This leads to the actual value.
The patch I posted a while ago ("[PATCH v2] ARM: imx: Support larger SPL size on IMX6DQ") may be helpful for this case.
Your patch is "in" - it is applied to u-boot-imx. However, it cannot help because the current limitation is set for Solo / DL. With your patch, SPL on Dual / Quad can be larger.
If the platform doesn't need to support the "lite" chip versions with smaller OCRAM size,
The SOC is a DL, 128KB
then the available size for the SPL can be increased.
It cannot
Best regards, Stefano Babic

Hi Anatolij/Stefano/Jagan,
On Mon, Sep 16, 2019 at 5:20 PM Anatolij Gustschin agust@denx.de wrote:
This check fails since the recent changes in SPL size limit check, see commit 25d38f9b34be (imx6: spl: Reduce SPL limit size in case CONFIG_SECURE_BOOT is enabled).
If you agree I can send a patch doing:
--- a/configs/imx6dl_mamoj_defconfig +++ b/configs/imx6dl_mamoj_defconfig @@ -1,7 +1,6 @@ CONFIG_ARM=y CONFIG_ARCH_MX6=y CONFIG_SYS_TEXT_BASE=0x17800000 -CONFIG_SECURE_BOOT=y CONFIG_TARGET_MX6DL_MAMOJ=y CONFIG_NR_DRAM_BANKS=1 # CONFIG_CMD_BMODE is not set
which allows the target to build again.
It seems that this target does not boot currently, but if we remove CONFIG_SECURE_BOOT then it can build and boot, I suppose.

On Tue, Sep 17, 2019 at 2:51 AM Fabio Estevam festevam@gmail.com wrote:
Hi Anatolij/Stefano/Jagan,
On Mon, Sep 16, 2019 at 5:20 PM Anatolij Gustschin agust@denx.de wrote:
This check fails since the recent changes in SPL size limit check, see commit 25d38f9b34be (imx6: spl: Reduce SPL limit size in case CONFIG_SECURE_BOOT is enabled).
If you agree I can send a patch doing:
--- a/configs/imx6dl_mamoj_defconfig +++ b/configs/imx6dl_mamoj_defconfig @@ -1,7 +1,6 @@ CONFIG_ARM=y CONFIG_ARCH_MX6=y CONFIG_SYS_TEXT_BASE=0x17800000 -CONFIG_SECURE_BOOT=y CONFIG_TARGET_MX6DL_MAMOJ=y CONFIG_NR_DRAM_BANKS=1 # CONFIG_CMD_BMODE is not set
which allows the target to build again.
It seems that this target does not boot currently, but if we remove CONFIG_SECURE_BOOT then it can build and boot, I suppose.
But this would drop HABv4, which indeed required for this board since all the production boards are secure enabled.

Hi Fabio,
On 16/09/19 23:21, Fabio Estevam wrote:
Hi Anatolij/Stefano/Jagan,
On Mon, Sep 16, 2019 at 5:20 PM Anatolij Gustschin agust@denx.de wrote:
This check fails since the recent changes in SPL size limit check, see commit 25d38f9b34be (imx6: spl: Reduce SPL limit size in case CONFIG_SECURE_BOOT is enabled).
If you agree I can send a patch doing:
--- a/configs/imx6dl_mamoj_defconfig +++ b/configs/imx6dl_mamoj_defconfig @@ -1,7 +1,6 @@ CONFIG_ARM=y CONFIG_ARCH_MX6=y CONFIG_SYS_TEXT_BASE=0x17800000 -CONFIG_SECURE_BOOT=y CONFIG_TARGET_MX6DL_MAMOJ=y CONFIG_NR_DRAM_BANKS=1 # CONFIG_CMD_BMODE is not set
which allows the target to build again.
It seems that this target does not boot currently, but if we remove CONFIG_SECURE_BOOT then it can build and boot, I suppose.
I can't decide for the board maintainer - yes, I have also dropped some feature, and I can build, but I have no idea which feature can be removed (and I could not test at all). The decision should be taken by board maintainer, I hope as soon as possible.
Best regards, Stefano
participants (5)
-
Anatolij Gustschin
-
Fabio Estevam
-
Jagan Teki
-
Robert Hancock
-
Stefano Babic