Date: Mon, 18 May 2020 14:34:35 +0000 From: Dan Kotowski <dan.kotowski@a9development.com> To: myfreeweb <greg@unrelenting.technology> Cc: John-Mark Gurney <jmg@funkthat.com>, "freebsd-arm@freebsd.org" <freebsd-arm@freebsd.org> Subject: Re: FreeBSD on Layerscape/QorIQ LX2160X Message-ID: <_lvahpuNQE69s4KpHud6ANL6yzL3RCVI-MTyB0_J_ULyW-3UWsqAXnm5gfoFcOyvfRQRabZk4Z4bQgyp15a001kA-WcvsvnWcjqgMBkgGTY=@a9development.com> In-Reply-To: <7AFDA7E0-82EC-4CD2-BB03-B7E33D019EDA@unrelenting.technology> References: <seVuHXPQIJwJasZ3wb2s2t61PcRQY5fyhT6njAvxEJ47Sj02pDtHbh-6EyHOf5M22BZdd27q4wK-zMPfCkJ2FIhFNnhILGxOQPebbQ19-0M=@a9development.com> <72D47662-240E-4851-844A-A4F92F18B3F2@unrelenting.technology> <20200516051207.GT4213@funkthat.com> <3nsQUg1Gm4VFYfpHVELk6PWaHyYNb3CoyoKnLV55_3VR48tr90bhaseG3sJg007L8czZ4mXUmR_YMQvYVdMbUs1bsoqGtZp5d17FqYT6b-o=@a9development.com> <7AFDA7E0-82EC-4CD2-BB03-B7E33D019EDA@unrelenting.technology>
next in thread | previous in thread | raw e-mail | index | archive | help
> > After a solid 16hrs this weekend, I can confirm that the pre-built imag= es do NOT "just work", at least not with SolidRun's HoneyComb version rev 1= .4. > > The 12.1 arm7 GENERICSD image fails to boot. The included EFI/BOOT/boot= arm.efi fails to load. I dropped in a copy of EFI/BOOT/bootarm64.efi from t= he aarch64 image, but that failed to handoff the process properly. Tested w= ith both u-boot and UEFI. > > Why would you even try 32-bit? Just throwing spaghetti at the wall to see what would stick. > > The 12.1 aarch64 memstick installer iamge fails out trying to mount the= UFS partition and drops into the mountroot> shell with error 19. Listing G= EOM managed disk devices returned an empty list. > > What devices does the boot log show? No xhci I guess, how about pci, sdhc= i? Does this answer your question? https://gist.github.com/agrajag9/ce1107a196= 2283b64205bc8f35123a5c If not, I'll happily run what's needed to get you something better. > > My silicon is rev 1.4 so I have to use the SD slot for the firmware lay= er, which is a mild inconvenience because I can't use that slot for an OS i= amge (https://github.com/SolidRun/lx2160a_build/issues/36#issuecomment-6298= 65674). > > The comment there says silicon rev 2 starts with board rev 1.5, but linux= 4kix on Twitter said the 1.4 board is rev 2 silicon! I did flash a known-working firmware build img to the eMMC and set the DIP = switches but it did nothing, although I won't rule out Layer 8 failures as = I'm still pretty fresh to playing this far down in a system. > > Unfortunately that's where I ran out of time. I spent most of the time = tracing and tweaking SolidRun's build scripts, testing various combinations= of firmware builds and boot images, and learning my way through both the u= -boot and UEFI implementations. > > Please don't waste time on u-boot. Any particular reason? Most of SolidRun's documentation is based on u-boot,= the online documentation from u-boot and the in-console help lines have be= en much easier to follow, and I'm finding it has a lot more capability to i= nteract with the underlying systems - e.g. reading/writing arbitrary blobs = from/to memory/disk. > > Barring other simple ideas, my next step is to make buildworld for TARG= ET=3Darm64 TARGET_ARCH=3Daarch64, drop that on an NVMe gumstick, and see if= I can get that booted. > > You can just dd the memstick to an nvme or sata drive as well. Yep, I have a nice fresh NVMe drive now on the way to play with later this = week. Dan Kotowski
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?_lvahpuNQE69s4KpHud6ANL6yzL3RCVI-MTyB0_J_ULyW-3UWsqAXnm5gfoFcOyvfRQRabZk4Z4bQgyp15a001kA-WcvsvnWcjqgMBkgGTY=>