20:37 <Jasper[m]> Hi! I'm currently trying to get a Spotify Car Thing to check out, apparently it's running a G12A. Are there any good non-windows programs that I can use to pull the thing apart software wise? I've seen meson64-tools on the site, but I'm wondering if there's any other neat tidbits.

21:36 <xdarklight> Jasper[m]: what do you mean with pulling it apart? I typically dump the .dtb from the vendor update file (or from the running system) - that contains most bits to create a .dts for mainline

21:40 <Jasper[m]> xdarklight: Dump emmc, extract firmware, extract dts like you mention. Update file is going to be hard without bt mitm'ing since they're not up on the website.

21:41 <Jasper[m]> Somebody on the github repo people are using for documentation have listed all of that stuff, but not how to reproduce actually grabbing all of that

21:43 <Jasper[m]> s/have/has/

21:43 <xdarklight> Jasper[m]: the .dtb should be at /sys/firmware/fdt, I typically use dd to dump eMMC (assuming I have either read-access on the vendor firmware or I already got far enough with a mainline kernel to gain read-access)

21:44 <Jasper[m]> I'm fairly sure whatever's running on the thing is so far removed from having any form of access that the only thing you can do is run the flash tools

21:44 <Jasper[m]> Looks like Android though, at least it's using a/b partitioning

21:46 <xdarklight> ah I think I understand. the typical board I have has Android running that I can access via serial console (typically even root access)

21:47 <Jasper[m]> Is there a default baudrate on the serial out? I think someone mentioned it having 2 sets of uart testpads: one of them being for uboot and the other for Android, but he only got garbage out of it.

21:48 <xdarklight> 115200 is the one used by the boot ROM on all Amlogic SoCs that I have seen

21:48 <xdarklight> (vendor u-boot or Linux keep 115200 - at least anything that I've seen so far)

21:56 <Jasper[m]> Huh, interestinf

21:56 <Jasper[m]> *interesting

22:01 <Jasper[m]> @xdarklight last question before I leave you alone for the night. Someone on that github repo is trying to build u-boot for the car thing using https://github.com/spsgsb/uboot (superbird_production) and the fip files from u200. Are these fip files board independent or do you actually need the ones included with the board?

22:02 <Jasper[m]> The guy is having trouble getting it to boot. Before I start asking him what error he got I figured I should ask the stupid questions first :p

22:03 <xdarklight> Jasper[m]: my understanding is that the FIP files can be board dependent. I think they contain some DDR calibration settings - so mixing with other boards is likely possible, but it depends on how "close" the two boards are. I'm not an expert in the boot chain though, so take this with a grain of salt

22:04 <Jasper[m]> Okay, will add a word of caution and a reference to the chat. Maybe he can ask you some direct questions.

22:04 <Jasper[m]> Would you suggest using linux-meson's u-boot tree instead of spsgsb's?

22:05 <xdarklight> I think it's best if he asks here or on the linux-amlogic (or u-boot) mailing list so others can add their 2cents where needed

22:05 <Jasper[m]> I understand

22:06 <xdarklight> depends on what the goal is: I'm often too busy to get mainline Linux running, so I end up with vendor u-boot often. but where possible I rip out vendor u-boot and replace it with the mainline version ;)

22:06 <xdarklight> https://github.com/LibreELEC/amlogic-boot-fip has a collection of FIP files in case the u200 ones don't do the job for you. maybe there's another board that works

22:09 <xdarklight> https://github.com/superna9999/pyamlboot/issues/9#issuecomment-720959729 - Neil's pyamlboot seems to have some u-boot chainloading support in case u-boot is in *update* mode (this is different from the SoC's "boot from USB" mode). that probably doesn't require any DDR calibration and other tricks (as the u-boot on the device will set that up). IIRC others have successfully dumped the FIP from an u-boot binary. so once you can dump eMMC you're

22:09 <xdarklight> probably also able to retrieve the FIP

22:09 <Jasper[m]> I made a last reply for tonight, hopefully he joins in. https://github.com/err4o4/spotify-car-thing-reverse-engineering/issues/7 here's the issue link btw. I might create a slightly better repo/hub for it since this one (apart from the issues page) is pretty much dead

22:10 <Jasper[m]> Thank you for that link! Will take a look tomorrow.