15:19 <clarity> Did anyone get spl usb gadget / sdp support working for imx8m?

15:46 <Harm> It just works on my imx8mn

15:48 <clarity> Hmm

15:48 <clarity> Which version of u-boot? Which board config?

15:51 <Harm> Private board config, v2021.10-rc2. I see I have some patches that are not in that release yet

15:53 <Harm> clarity: https://pastebin.com/cJUYkQvv

15:53 <Harm> got the code from the mailing list, apparently it's not in master yet

15:54 <Harm> This should also work on v2021.07

15:55 <clarity> Ok.. so I need that patch but I'll also need to figure out a way to make space

15:55 <Harm> look for 'usb: ehci-mx6: add IMX8MM OTG support' in the mailing list

15:55 <clarity> If I start with imx8mm_evk defconfig and enable usb gadget, ehci hcd & co, I seem to bump into the sram limit

15:55 <Harm> ouch

15:57 <Harm> If it helps, this is my defconfig: https://pastebin.com/uCN5saV3

16:01 <clarity> Hmm you don't have usb gadget support in SPL?

16:02 <clarity> SPL_USB_GADGET, SPL_USB_SDP_SUPPORT etc

16:23 <Harm> clarity: As far as I know the imx8mn does that through the ROM api. I'm using the uuu tool from NXP to load u-boot and that is using the SDP

16:29 <mrnuke> sjg1: Are you working on #ifdef stuff that touches or relocates padding_pkcs_15_verify(). I'm seein gsome undefined reference to padding_pkcs_15_verify in v2021.10-rc2, and I think it's a combination of previous ECDSA work, and our improvements to the #ifdef stack

16:39 <clarity> Harm: Yea that's what I'm after. What happens is that it loads SPL through uuu, but then SPL doesn't know how to load the next part.. I assume SPL needs sdp support too

16:41 <Harm> And that's happening just fine on my board using that config. Can't that be done through the ROM API on your board?

16:42 <clarity> What would the code for that be called?

16:42 <Harm> I think this is the code: /arch/arm/mach-imx/spl_imx_romapi.c

16:42 <clarity> Aha

16:43 <clarity> I'll look into it, never saw that before

16:43 <Harm> Try enabling CONFIG_SPL_BOOTROM_SUPPORT

16:46 <clarity> depends on IMX8MN || IMX8MP || IMX8ULP

16:47 <Harm> :\

16:52 <marex> MX8M/MQ = bootrom-the-old-version , MX8MN/MP = bootrom-v2

16:53 <Harm> Right, so you'll need the SDP code inside SPL then

16:53 <marex> apparently different MX8Ms have different bootroms implemented by different engineering teams, so they have different features and different bugs

16:53 <Harm> wow

16:53 <marex> yes, it sucks

16:54 <marex> MX8MN/MP for example doesn't support persist-secondary-boot (reboot into B-copy of bootloader with the option of falling back to A-copy using watchdog cold boot if B-copy is broken, e.g. in case broken update got installed)

16:54 <marex> MX8MM still supported that ...

16:56 <marex> yes, on MM/MQ , you will need UDC support in SPL ...

16:57 <marex> you can hack out the dependency on ATF (which really is mostly useless on MX8M) to free up some space in the OCRAM

16:57 <marex> on mx8mm, the ATF is at 0x960000 onward, so that could be freed up

17:00 <Harm> I know I can control from which eMMC boot partition I'm booting by setting a flag in the eMMC. Also, when I zero out the first boot partition it automatically tries starting from the second on my 8MN

17:01 <Harm> Won't you need ATF for enabling other CPU cores and power saving modes/cpu frequncy scaling?

17:01 <marex> sure, it will, except for recovery and u-boot you dont really care

17:02 <Harm> yeah

19:31 <sjg1> mrnuke: No not me

19:49 <mrnuke> sjg1: Okay. I'll go ahead and write the fix

22:09 <mrnuke> I'm trying to run the RSA tests in test/lib/rsa.c using " NO_SDL=1 ./test/py/test.py --bd sandbox --setup-show -krsa"

22:10 <mrnuke> but I'm seeing "840/840 deselected" and no test actually runs

22:10 <mrnuke> grep UT_LIB_RSA build-sandbox/.config shows that "CONFIG_UT_LIB_RSA=y" is indeed selected

23:27 <rfs613> mrnuke: just a wild guess here, perhaps you also need CONFIG_RSA or similar options set?

23:28 <rfs613> test/lib/rsa.c seems to be conditional on CONFIG_RSA_VERIFY_WITH_PKEY