RK3588 optee question

RK3588 optee question

[Alexander Shiyan]

Hello All.

BL31 = rockchip bl31 from rkbin
BL32 = tee.bin builded from https://github.com/OP-TEE/optee_os.git

DDR cb12b99cc23 hcy 25/10/17-18:57:13,fwver: v1.21
...
NOTICE:  BL31: v2.3():v2.3-964-g8319968f5:derrick.huang, fwver: v1.54
NOTICE:  BL31: Built : 17:20:23, Dec 26 2025
...
INFO:    BL31: Initialising Exception Handling Framework
INFO:    BL31: Initializing runtime services
INFO:    BL31: Initializing BL32
INFO:    BL31: Preparing for EL3 exit to normal world
INFO:    Entry point address = 0xb00000
INFO:    SPSR = 0x3c9

barebox 2026.07.0-00839-g50ae8812b950-dirty #144 Tue Aug 11 14:39:36 MSK 2026

Board: Diasom DS-RK3588-BTB-EVB
rockchip-dmc memory-controller.of: Detected memory size: 0x100000000
netconsole: registered as netconsole-1
fan53555-regulator rk86020: FAN53555 Option[10] Rev[1] Detected!
vdd2_ddr: Bringing 500000uV into 1100000-1100000uV
vddq_ddr: Bringing 500000uV into 600000-600000uV
rockchip_saradc fec10000.adc@fec10000.of: registered as aiodev0
BTB revision: 3
EVB revision: 2
EVB HAT Addon detected.
arm-scmi firmware:scmi.of: SCMI Protocol v2.0 'rockchip:' Firmware version 0x0
optee: smc_abi: probing for conduit method.
optee: smc_abi: revision 4.10 (71196403f6f9f642)
ERROR: optee: smc_abi: driver requires OP-TEE dynamic shared memory support
ERROR: optee firmware:optee.of: probe failed: Function not implemented
psci psci.of: detected version 1.1
xHCI xHCI0: USB XHCI 1.10
ehci fc800000.usb@fc800000.of: USB EHCI 1.00
ehci fc880000.usb@fc880000.of: USB EHCI 1.00
xHCI xHCI1: USB XHCI 1.10
fan53555-regulator rk86021: FAN53555 Option[10] Rev[1] Detected!
fan53555-regulator rk86030: FAN53555 Option[10] Rev[1] Detected!
rtl8169_eth pci-10ec:8125.0: found RTL8125B (base=0x0000000a00000000)
mdio_bus: miibus0: probed
dw_mmc fe2c0000.mmc@fe2c0000.of: registered as mmc1
mmc1: detected SD card version 3.0
WARNING: mmc1: GPT:Primary header thinks Alt. header is not at the end
of the disk.
WARNING: mmc1: GPT:2976871 != 62333951
WARNING: mmc1: GPT:Alternate GPT header not at the end of the disk.
WARNING: mmc1: GPT:2976871 != 62333951
mmc1: GPT: will repair later if global.system.gpt_refresh=1
mmc1: registered mmc1
rk3568-dwcmshc-sdhci fe2e0000.mmc@fe2e0000.of: registered as mmc0
mmc0: detected MMC card version 5.1
mmc0: registered mmc0.boot0
mmc0: registered mmc0.boot1
mmc0: registered mmc0
i2c3: Added multiplexed i2c bus 9
i2c3: Added multiplexed i2c bus 10
i2c3: Added multiplexed i2c bus 11
i2c3: Added multiplexed i2c bus 12
pca954x pca95460: registered 4 multiplexed busses for I2C switch
xHCI xHCI2: USB XHCI 1.10
rockchip-dw-pcie a40000000.pcie@fe150000.of: Phy link never came up
rtl8169_eth pci-10ec:8125.1: found RTL8125B (base=0x0000000980000000)
mdio_bus: miibus1: probed
Setup Machine ID from EMMC serial: 1815144414
nv variable modified, will save nv variables on shutdown
malloc space: 0xb0000000 -> 0xedd0ffff (size 989.1 MiB)
CAM0: No cameras detected.
CAM1: Camera AR0234 detected.
envfs: no envfs (magic mismatch) - envfs never written?
eth1: Got preset MAC address from NVMEM: 80:34:28:75:d4:d4
eth0: Got preset MAC address from NVMEM: 80:34:28:75:cf:e7

Hit any to stop autoboot:    2
barebox@Diasom DS-RK3588-BTB-EVB-HAT:/ rksecure -i
Failed to read secure boot info
rksecure: No such device
barebox@Diasom DS-RK3588-BTB-EVB-HAT:/

Am I missing something?
The following lines confuse me, what is these line mean?:
ERROR: optee: smc_abi: driver requires OP-TEE dynamic shared memory support
ERROR: optee firmware:optee.of: probe failed: Function not implemented

Thanks!

Re: RK3588 optee question

[Ahmad Fatoum]

Hi,

On 8/11/26 1:51 PM, Alexander Shiyan wrote:
> Am I missing something?
> The following lines confuse me, what is these line mean?:
> ERROR: optee: smc_abi: driver requires OP-TEE dynamic shared memory support
> ERROR: optee firmware:optee.of: probe failed: Function not implemented

Try enabling CFG_CORE_DYN_SHM in OP-TEE.

Cheers,
Ahmad


-- 
Pengutronix e.K.                  |                             |
Steuerwalder Str. 21              | http://www.pengutronix.de/  |
31137 Hildesheim, Germany         | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686  | Fax:   +49-5121-206917-5555 |

Re: RK3588 optee question

[Alexander Shiyan]

Hello.

As far as I can see, this option (CFG_CORE_DYN_SHM) is enabled
by default in out/arm-plat-rockchip/conf.mk

Experiments with the open (BL31) TF-A version give a different error:
...
NOTICE:  BL31: v2.15.0(release):v2.15.0-274-gc64fe42d7
NOTICE:  BL31: Built : 14:41:34, Jul  6 2026
...
arm-scmi firmware:scmi.of: SCMI Protocol v2.0 'rockchip:' Firmware version 0x0
optee: smc_abi: probing for conduit method.
WARNING: optee: smc_abi: api uid mismatch
ERROR: optee firmware:optee.of: probe failed: Invalid argument

вт, 11 авг. 2026 г. в 15:19, Ahmad Fatoum <a.fatoum@pengutronix.de>:
>
> Hi,
>
> On 8/11/26 1:51 PM, Alexander Shiyan wrote:
> > Am I missing something?
> > The following lines confuse me, what is these line mean?:
> > ERROR: optee: smc_abi: driver requires OP-TEE dynamic shared memory support
> > ERROR: optee firmware:optee.of: probe failed: Function not implemented
>
> Try enabling CFG_CORE_DYN_SHM in OP-TEE.
>
> Cheers,
> Ahmad
>
>
> --
> Pengutronix e.K.                  |                             |
> Steuerwalder Str. 21              | http://www.pengutronix.de/  |
> 31137 Hildesheim, Germany         | Phone: +49-5121-206917-0    |
> Amtsgericht Hildesheim, HRA 2686  | Fax:   +49-5121-206917-5555 |
>

Re: RK3588 optee question

[Ahmad Fatoum]

Hi,

On 8/11/26 2:54 PM, Alexander Shiyan wrote:
> Hello.
> 
> As far as I can see, this option (CFG_CORE_DYN_SHM) is enabled
> by default in out/arm-plat-rockchip/conf.mk
> 
> Experiments with the open (BL31) TF-A version give a different error:
> ...
> NOTICE:  BL31: v2.15.0(release):v2.15.0-274-gc64fe42d7
> NOTICE:  BL31: Built : 14:41:34, Jul  6 2026
> ...
> arm-scmi firmware:scmi.of: SCMI Protocol v2.0 'rockchip:' Firmware version 0x0
> optee: smc_abi: probing for conduit method.
> WARNING: optee: smc_abi: api uid mismatch
> ERROR: optee firmware:optee.of: probe failed: Invalid argument

Huh, no idea off the top my head what is still missing in the OP-TEE or
TF-A configuration.

Cheers,
Ahmad

> 
> вт, 11 авг. 2026 г. в 15:19, Ahmad Fatoum <a.fatoum@pengutronix.de>:
>>
>> Hi,
>>
>> On 8/11/26 1:51 PM, Alexander Shiyan wrote:
>>> Am I missing something?
>>> The following lines confuse me, what is these line mean?:
>>> ERROR: optee: smc_abi: driver requires OP-TEE dynamic shared memory support
>>> ERROR: optee firmware:optee.of: probe failed: Function not implemented
>>
>> Try enabling CFG_CORE_DYN_SHM in OP-TEE.
>>
>> Cheers,
>> Ahmad
>>
>>
>> --
>> Pengutronix e.K.                  |                             |
>> Steuerwalder Str. 21              | http://www.pengutronix.de/  |
>> 31137 Hildesheim, Germany         | Phone: +49-5121-206917-0    |
>> Amtsgericht Hildesheim, HRA 2686  | Fax:   +49-5121-206917-5555 |
>>
> 

-- 
Pengutronix e.K.                  |                             |
Steuerwalder Str. 21              | http://www.pengutronix.de/  |
31137 Hildesheim, Germany         | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686  | Fax:   +49-5121-206917-5555 |

Re: RK3588 optee question

[Alexander Shiyan]

Could someone share the correct RK3588 tee.bin binary for the test?
Perhaps I compiled something incorrectly...

Thanks!

вт, 11 авг. 2026 г. в 17:07, Ahmad Fatoum <a.fatoum@pengutronix.de>:
>
> Hi,
>
> On 8/11/26 2:54 PM, Alexander Shiyan wrote:
> > Hello.
> >
> > As far as I can see, this option (CFG_CORE_DYN_SHM) is enabled
> > by default in out/arm-plat-rockchip/conf.mk
> >
> > Experiments with the open (BL31) TF-A version give a different error:
> > ...
> > NOTICE:  BL31: v2.15.0(release):v2.15.0-274-gc64fe42d7
> > NOTICE:  BL31: Built : 14:41:34, Jul  6 2026
> > ...
> > arm-scmi firmware:scmi.of: SCMI Protocol v2.0 'rockchip:' Firmware version 0x0
> > optee: smc_abi: probing for conduit method.
> > WARNING: optee: smc_abi: api uid mismatch
> > ERROR: optee firmware:optee.of: probe failed: Invalid argument
>
> Huh, no idea off the top my head what is still missing in the OP-TEE or
> TF-A configuration.
>
> Cheers,
> Ahmad
>
> >
> > вт, 11 авг. 2026 г. в 15:19, Ahmad Fatoum <a.fatoum@pengutronix.de>:
> >>
> >> Hi,
> >>
> >> On 8/11/26 1:51 PM, Alexander Shiyan wrote:
> >>> Am I missing something?
> >>> The following lines confuse me, what is these line mean?:
> >>> ERROR: optee: smc_abi: driver requires OP-TEE dynamic shared memory support
> >>> ERROR: optee firmware:optee.of: probe failed: Function not implemented
> >>
> >> Try enabling CFG_CORE_DYN_SHM in OP-TEE.
> >>
> >> Cheers,
> >> Ahmad
> >>
> >>
> >> --
> >> Pengutronix e.K.                  |                             |
> >> Steuerwalder Str. 21              | http://www.pengutronix.de/  |
> >> 31137 Hildesheim, Germany         | Phone: +49-5121-206917-0    |
> >> Amtsgericht Hildesheim, HRA 2686  | Fax:   +49-5121-206917-5555 |
> >>
> >
>
> --
> Pengutronix e.K.                  |                             |
> Steuerwalder Str. 21              | http://www.pengutronix.de/  |
> 31137 Hildesheim, Germany         | Phone: +49-5121-206917-0    |
> Amtsgericht Hildesheim, HRA 2686  | Fax:   +49-5121-206917-5555 |
>

Re: RK3588 optee question

[Sascha Hauer]

On 2026-08-11 14:51, Alexander Shiyan wrote:
> Hello All.
> 
> BL31 = rockchip bl31 from rkbin
> BL32 = tee.bin builded from https://github.com/OP-TEE/optee_os.git
> 
> DDR cb12b99cc23 hcy 25/10/17-18:57:13,fwver: v1.21
> ...
> NOTICE:  BL31: v2.3():v2.3-964-g8319968f5:derrick.huang, fwver: v1.54
> NOTICE:  BL31: Built : 17:20:23, Dec 26 2025
> ...
> INFO:    BL31: Initialising Exception Handling Framework
> INFO:    BL31: Initializing runtime services
> INFO:    BL31: Initializing BL32
> INFO:    BL31: Preparing for EL3 exit to normal world
> INFO:    Entry point address = 0xb00000
> INFO:    SPSR = 0x3c9
> 
> barebox 2026.07.0-00839-g50ae8812b950-dirty #144 Tue Aug 11 14:39:36 MSK 2026
> 
> Board: Diasom DS-RK3588-BTB-EVB
> rockchip-dmc memory-controller.of: Detected memory size: 0x100000000
> netconsole: registered as netconsole-1
> fan53555-regulator rk86020: FAN53555 Option[10] Rev[1] Detected!
> vdd2_ddr: Bringing 500000uV into 1100000-1100000uV
> vddq_ddr: Bringing 500000uV into 600000-600000uV
> rockchip_saradc fec10000.adc@fec10000.of: registered as aiodev0
> BTB revision: 3
> EVB revision: 2
> EVB HAT Addon detected.
> arm-scmi firmware:scmi.of: SCMI Protocol v2.0 'rockchip:' Firmware version 0x0
> optee: smc_abi: probing for conduit method.
> optee: smc_abi: revision 4.10 (71196403f6f9f642)
> ERROR: optee: smc_abi: driver requires OP-TEE dynamic shared memory support

In OP-TEE this goes down to:

#if defined(CFG_CORE_DYN_SHM)
          dyn_shm_en = core_mmu_nsec_ddr_is_defined();
          if (dyn_shm_en)
                  args->a1 |= OPTEE_SMC_SEC_CAP_DYNAMIC_SHM;
#endif

So either you have compiled OP-TEE without CFG_CORE_DYN_SHM enabled
(unlikely, it's the default), or core_mmu_nsec_ddr_is_defined() fails
which means OP-TEE doesn't know where your memory is. OP-TEE gets this
information from the device tree. Do you pass a device tree to OP-TEE?
barebox only does it when CONFIG_ARCH_ROCKCHIP_ATF_PASS_FDT is enabled
which currently is not the default as there are some Rockchip TF-A
binaries shipped by Rockchip which do not handle a fdt being passed.

Looking at your TF-A version you might have one of these, so you likely
have to update your TF-A to an upstream one.

Side note: Look at bl31_entry(). if it gets a fdt_addr that will be
passed to both TF-A and OP-TEE. You *could* try passing NULL as fdt to
atf_entry() to pass the fdt only to OP-TEE and not to TF-A, but I
wouldn't encourage that.

Sascha

--
Pengutronix e.K.                           |                             |
Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

Re: RK3588 optee question

[Alexander Shiyan]

Hello Sascha.

After enabling CONFIG_ARCH_ROCKCHIP_ATF_PASS_FDT=y the following build
errors occur:

CC [P]  arch/arm/mach-rockchip/rk3588.pbl.o
In file included from include/linux/array_size.h:6,
from include/linux/kernel.h:6,
from /home/ARM64/barebox/arch/arm/include/asm/barebox-arm.h:17,
from include/mach/rockchip/bootrom.h:9,
from arch/arm/mach-rockchip/bootrom.c:2:
include/linux/build_bug.h:100:41: error: static assertion failed:
"sizeof(structrockchip_scratch_space) <= CONFIG_SCRATCH_SIZE"
100 | #define __static_assert(expr, msg, ...) _Static_assert(expr, msg)
|                                         ^~~~~~~~~~~~~~
include/linux/build_bug.h:99:34: note: in expansion of macro ‘__static_assert’
99 | #define static_assert(expr, ...) __static_assert(expr,
##__VA_ARGS__, #expr)
|                                  ^~~~~~~~~~~~~~~
include/mach/rockchip/bootrom.h:25:1: note: in expansion of macro
‘static_assert’
25 | static_assert(sizeof(struct rockchip_scratch_space) <=
CONFIG_SCRATCH_SIZE);
| ^~~~~~~~~~~~~
In file included from include/linux/container_of.h:5,
from include/linux/list.h:5,
from include/param.h:7,
from include/console.h:10,
from include/stdio.h:6,
from include/common.h:10,
from arch/arm/mach-rockchip/rockchip.c:2:
include/linux/build_bug.h:100:41: error: static assertion failed:
"sizeof(structrockchip_scratch_space) <= CONFIG_SCRATCH_SIZE"
100 | #define __static_assert(expr, msg, ...) _Static_assert(expr, msg)
|                                         ^~~~~~~~~~~~~~
include/linux/build_bug.h:99:34: note: in expansion of macro ‘__static_assert’
99 | #define static_assert(expr, ...) __static_assert(expr,
##__VA_ARGS__, #expr)
|                                  ^~~~~~~~~~~~~~~
include/mach/rockchip/bootrom.h:25:1: note: in expansion of macro
‘static_assert’
25 | static_assert(sizeof(struct rockchip_scratch_space) <=
CONFIG_SCRATCH_SIZE);
| ^~~~~~~~~~~~~
In file included from include/linux/container_of.h:5,
from include/linux/list.h:5,
from include/param.h:7,
from include/console.h:10,
from include/stdio.h:6,
from include/common.h:10,
from arch/arm/mach-rockchip/rk3568.c:2:
include/linux/build_bug.h:100:41: error: static assertion failed:
"sizeof(structrockchip_scratch_space) <= CONFIG_SCRATCH_SIZE"
100 | #define __static_assert(expr, msg, ...) _Static_assert(expr, msg)
|                                         ^~~~~~~~~~~~~~
include/linux/build_bug.h:99:34: note: in expansion of macro ‘__static_assert’
99 | #define static_assert(expr, ...) __static_assert(expr,
##__VA_ARGS__, #expr)
|                                  ^~~~~~~~~~~~~~~
include/mach/rockchip/bootrom.h:25:1: note: in expansion of macro
‘static_assert’
25 | static_assert(sizeof(struct rockchip_scratch_space) <=
CONFIG_SCRATCH_SIZE);
| ^~~~~~~~~~~~~
make[1]: *** [/home/ARM64/barebox/scripts/Makefile.build:170:
arch/arm/mach-rockchip/rockchip.o] Error 1
make[1]: *** Waiting for unfinished jobs....
make[1]: *** [/home/ARM64/barebox/scripts/Makefile.build:170:
arch/arm/mach-rockchip/bootrom.o] Error 1
make[1]: *** [/home/ARM64/barebox/scripts/Makefile.build:170:
arch/arm/mach-rockchip/rk3568.o] Error 1
In file included from include/linux/container_of.h:5,
from include/linux/list.h:5,
from include/param.h:7,
from include/console.h:10,
from include/stdio.h:6,
from include/common.h:10,
from arch/arm/mach-rockchip/rk3588.c:2:
include/linux/build_bug.h:100:41: error: static assertion failed:
"sizeof(structrockchip_scratch_space) <= CONFIG_SCRATCH_SIZE"
100 | #define __static_assert(expr, msg, ...) _Static_assert(expr, msg)
|                                         ^~~~~~~~~~~~~~
include/linux/build_bug.h:99:34: note: in expansion of macro ‘__static_assert’
99 | #define static_assert(expr, ...) __static_assert(expr,
##__VA_ARGS__, #expr)
|                                  ^~~~~~~~~~~~~~~
include/mach/rockchip/bootrom.h:25:1: note: in expansion of macro
‘static_assert’
25 | static_assert(sizeof(struct rockchip_scratch_space) <=
CONFIG_SCRATCH_SIZE);
| ^~~~~~~~~~~~~
In file included from include/linux/container_of.h:5,
from include/linux/list.h:5,
from include/param.h:7,
from include/console.h:10,
from include/stdio.h:6,
from include/common.h:10,
from arch/arm/mach-rockchip/bbu.c:2:
include/linux/build_bug.h:100:41: error: static assertion failed:
"sizeof(structrockchip_scratch_space) <= CONFIG_SCRATCH_SIZE"
100 | #define __static_assert(expr, msg, ...) _Static_assert(expr, msg)
|                                         ^~~~~~~~~~~~~~
include/linux/build_bug.h:99:34: note: in expansion of macro ‘__static_assert’
99 | #define static_assert(expr, ...) __static_assert(expr,
##__VA_ARGS__, #expr)
|                                  ^~~~~~~~~~~~~~~
include/mach/rockchip/bootrom.h:25:1: note: in expansion of macro
‘static_assert’
25 | static_assert(sizeof(struct rockchip_scratch_space) <=
CONFIG_SCRATCH_SIZE);
| ^~~~~~~~~~~~~
make[1]: *** [/home/ARM64/barebox/scripts/Makefile.build:170:
arch/arm/mach-rockchip/rk3588.o] Error 1
make[1]: *** [/home/ARM64/barebox/scripts/Makefile.build:170:
arch/arm/mach-rockchip/bbu.o] Error 1
In file included from include/linux/container_of.h:5,
from include/linux/list.h:5,
from include/param.h:7,
from include/console.h:10,
from include/stdio.h:6,
from include/common.h:10,
from arch/arm/mach-rockchip/atf.c:2:
include/linux/build_bug.h:100:41: error: static assertion failed:
"sizeof(structrockchip_scratch_space) <= CONFIG_SCRATCH_SIZE"
100 | #define __static_assert(expr, msg, ...) _Static_assert(expr, msg)
|                                         ^~~~~~~~~~~~~~
include/linux/build_bug.h:99:34: note: in expansion of macro ‘__static_assert’
99 | #define static_assert(expr, ...) __static_assert(expr,
##__VA_ARGS__, #expr)
|                                  ^~~~~~~~~~~~~~~
include/mach/rockchip/bootrom.h:25:1: note: in expansion of macro
‘static_assert’
25 | static_assert(sizeof(struct rockchip_scratch_space) <=
CONFIG_SCRATCH_SIZE);
| ^~~~~~~~~~~~~
make[1]: *** [/home/ARM64/barebox/scripts/Makefile.build:167:
arch/arm/mach-rockchip/atf.pbl.o] Error 1
In file included from include/linux/container_of.h:5,
from include/linux/list.h:5,
from include/param.h:7,
from include/console.h:10,
from include/stdio.h:6,
from include/common.h:10,
from arch/arm/mach-rockchip/rk3568.c:2:
include/linux/build_bug.h:100:41: error: static assertion failed:
"sizeof(structrockchip_scratch_space) <= CONFIG_SCRATCH_SIZE"
100 | #define __static_assert(expr, msg, ...) _Static_assert(expr, msg)
|                                         ^~~~~~~~~~~~~~
include/linux/build_bug.h:99:34: note: in expansion of macro ‘__static_assert’
99 | #define static_assert(expr, ...) __static_assert(expr,
##__VA_ARGS__, #expr)
|                                  ^~~~~~~~~~~~~~~
include/mach/rockchip/bootrom.h:25:1: note: in expansion of macro
‘static_assert’
25 | static_assert(sizeof(struct rockchip_scratch_space) <=
CONFIG_SCRATCH_SIZE);
| ^~~~~~~~~~~~~
In file included from include/linux/container_of.h:5,
from include/linux/list.h:5,
from include/param.h:7,
from include/console.h:10,
from include/stdio.h:6,
from include/common.h:10,
from arch/arm/mach-rockchip/rk3588.c:2:
include/linux/build_bug.h:100:41: error: static assertion failed:
"sizeof(structrockchip_scratch_space) <= CONFIG_SCRATCH_SIZE"
100 | #define __static_assert(expr, msg, ...) _Static_assert(expr, msg)
|                                         ^~~~~~~~~~~~~~
include/linux/build_bug.h:99:34: note: in expansion of macro ‘__static_assert’
99 | #define static_assert(expr, ...) __static_assert(expr,
##__VA_ARGS__, #expr)
|                                  ^~~~~~~~~~~~~~~
include/mach/rockchip/bootrom.h:25:1: note: in expansion of macro
‘static_assert’
25 | static_assert(sizeof(struct rockchip_scratch_space) <=
CONFIG_SCRATCH_SIZE);
| ^~~~~~~~~~~~~
make[1]: *** [/home/ARM64/barebox/scripts/Makefile.build:167:
arch/arm/mach-rockchip/rk3568.pbl.o] Error 1
make[1]: *** [/home/ARM64/barebox/scripts/Makefile.build:167:
arch/arm/mach-rockchip/rk3588.pbl.o] Error 1
make: *** [Makefile:1142: arch/arm/mach-rockchip] Error 2
make: *** Waiting for unfinished jobs....
AR      drivers/of/built-in.a
AR      drivers/built-in.a

So, to avoid this, can we define default for CONFIG_SCRATCH_SIZE as
CONFIG_ARCH_ROCKCHIP_ATF_FDT_SIZE if this option is used?

Finally, at least for RK3568 (cannot test RK3588 right now), the init
errors are gone:
...
arm-scmi firmware:scmi.of: SCMI Protocol v2.0 'rockchip:' Firmware version 0x0
optee: smc_abi: probing for conduit method.
optee: smc_abi: revision 4.10 (5c827bf3bc358e9a)
psci psci.of: detected version 1.1
...

The rksecure now works like this:
barebox@Diasom DS-RK3568-SOM-EVB:/ rksecure -i
NOTICE: optee: optee: No supplicant or RPC handler for command 0x0
ERROR: rk-secure: TA 5cfa57f6-1a4c-407f-94a7-a56c8c47019d: error 0xffff0009
Failed to read secure boot info
rksecure: Invalid argument

ср, 12 авг. 2026 г. в 09:25, Sascha Hauer <s.hauer@pengutronix.de>:
>
> On 2026-08-11 14:51, Alexander Shiyan wrote:
> > Hello All.
> >
> > BL31 = rockchip bl31 from rkbin
> > BL32 = tee.bin builded from https://github.com/OP-TEE/optee_os.git
> >
> > DDR cb12b99cc23 hcy 25/10/17-18:57:13,fwver: v1.21
> > ...
> > NOTICE:  BL31: v2.3():v2.3-964-g8319968f5:derrick.huang, fwver: v1.54
> > NOTICE:  BL31: Built : 17:20:23, Dec 26 2025
> > ...
> > INFO:    BL31: Initialising Exception Handling Framework
> > INFO:    BL31: Initializing runtime services
> > INFO:    BL31: Initializing BL32
> > INFO:    BL31: Preparing for EL3 exit to normal world
> > INFO:    Entry point address = 0xb00000
> > INFO:    SPSR = 0x3c9
> >
> > barebox 2026.07.0-00839-g50ae8812b950-dirty #144 Tue Aug 11 14:39:36 MSK 2026
> >
> > Board: Diasom DS-RK3588-BTB-EVB
> > rockchip-dmc memory-controller.of: Detected memory size: 0x100000000
> > netconsole: registered as netconsole-1
> > fan53555-regulator rk86020: FAN53555 Option[10] Rev[1] Detected!
> > vdd2_ddr: Bringing 500000uV into 1100000-1100000uV
> > vddq_ddr: Bringing 500000uV into 600000-600000uV
> > rockchip_saradc fec10000.adc@fec10000.of: registered as aiodev0
> > BTB revision: 3
> > EVB revision: 2
> > EVB HAT Addon detected.
> > arm-scmi firmware:scmi.of: SCMI Protocol v2.0 'rockchip:' Firmware version 0x0
> > optee: smc_abi: probing for conduit method.
> > optee: smc_abi: revision 4.10 (71196403f6f9f642)
> > ERROR: optee: smc_abi: driver requires OP-TEE dynamic shared memory support
>
> In OP-TEE this goes down to:
>
> #if defined(CFG_CORE_DYN_SHM)
>           dyn_shm_en = core_mmu_nsec_ddr_is_defined();
>           if (dyn_shm_en)
>                   args->a1 |= OPTEE_SMC_SEC_CAP_DYNAMIC_SHM;
> #endif
>
> So either you have compiled OP-TEE without CFG_CORE_DYN_SHM enabled
> (unlikely, it's the default), or core_mmu_nsec_ddr_is_defined() fails
> which means OP-TEE doesn't know where your memory is. OP-TEE gets this
> information from the device tree. Do you pass a device tree to OP-TEE?
> barebox only does it when CONFIG_ARCH_ROCKCHIP_ATF_PASS_FDT is enabled
> which currently is not the default as there are some Rockchip TF-A
> binaries shipped by Rockchip which do not handle a fdt being passed.
>
> Looking at your TF-A version you might have one of these, so you likely
> have to update your TF-A to an upstream one.
>
> Side note: Look at bl31_entry(). if it gets a fdt_addr that will be
> passed to both TF-A and OP-TEE. You *could* try passing NULL as fdt to
> atf_entry() to pass the fdt only to OP-TEE and not to TF-A, but I
> wouldn't encourage that.
>
> Sascha
>
> --
> Pengutronix e.K.                           |                             |
> Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
> 31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
> Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |
>

Re: RK3588 optee question

[Sascha Hauer]

On 2026-08-12 17:44, Alexander Shiyan wrote:
> Hello Sascha.
> 
> After enabling CONFIG_ARCH_ROCKCHIP_ATF_PASS_FDT=y the following build
> errors occur:
> 
> CC [P]  arch/arm/mach-rockchip/rk3588.pbl.o
> In file included from include/linux/array_size.h:6,
> from include/linux/kernel.h:6,
> from /home/ARM64/barebox/arch/arm/include/asm/barebox-arm.h:17,
> from include/mach/rockchip/bootrom.h:9,
> from arch/arm/mach-rockchip/bootrom.c:2:
> include/linux/build_bug.h:100:41: error: static assertion failed:
> "sizeof(structrockchip_scratch_space) <= CONFIG_SCRATCH_SIZE"

Set CONFIG_SCRATCH_SIZE=0x100000

Seems we have to look into it to get a sane default.

Sascha

--
Pengutronix e.K.                           |                             |
Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

Re: RK3588 optee question

[Alexander Shiyan]

Hello, Sascha.

With CONFIG_SCRATCH_SIZE=0x100000 compiles fine.

Now about the current errors, one build is used for two images for
RK3568 and RK3588.

RK3568, Rockchip BL31, tee.bin from OPTEE_OS:
arm-scmi firmware:scmi.of: SCMI Protocol v2.0 'rockchip:' Firmware version 0x0
optee: smc_abi: probing for conduit method.
optee: smc_abi: revision 4.10 (5c827bf3bc358e9a)
psci psci.of: detected version 1.1

RK3588, open source TF-A, tee.bin from OPTEE_OS:
arm-scmi firmware:scmi.of: SCMI Protocol v2.0 'rockchip:' Firmware version 0x0
optee: smc_abi: probing for conduit method.
WARNING: optee: smc_abi: api uid mismatch
ERROR: optee firmware:optee.of: probe failed: Invalid argument

Switching to a configuration using Rockchip BL31 with the TEE/OPTEE options
currently enabled results in a constant restart of DDR init code, meaning the
code never reaches BL31...

What else can I watch?

Thanks!


чт, 13 авг. 2026 г. в 09:42, Sascha Hauer <s.hauer@pengutronix.de>:
>
> On 2026-08-12 17:44, Alexander Shiyan wrote:
> > Hello Sascha.
> >
> > After enabling CONFIG_ARCH_ROCKCHIP_ATF_PASS_FDT=y the following build
> > errors occur:
> >
> > CC [P]  arch/arm/mach-rockchip/rk3588.pbl.o
> > In file included from include/linux/array_size.h:6,
> > from include/linux/kernel.h:6,
> > from /home/ARM64/barebox/arch/arm/include/asm/barebox-arm.h:17,
> > from include/mach/rockchip/bootrom.h:9,
> > from arch/arm/mach-rockchip/bootrom.c:2:
> > include/linux/build_bug.h:100:41: error: static assertion failed:
> > "sizeof(structrockchip_scratch_space) <= CONFIG_SCRATCH_SIZE"
>
> Set CONFIG_SCRATCH_SIZE=0x100000
>
> Seems we have to look into it to get a sane default.
>
> Sascha
>
> --
> Pengutronix e.K.                           |                             |
> Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
> 31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
> Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |
>
>