On Wednesday 26 August 2020 15:22:37 Andre Heider wrote: > Hi, > > On 26/08/2020 13:40, Pali Rohár wrote: > > On Wednesday 26 August 2020 12:51:18 Pali Rohár wrote: > > > On Wednesday 26 August 2020 11:57:01 Andre Heider wrote: > > > > Hi, > > > > > > > > On 26/08/2020 10:31, Pali Rohár wrote: > > > > > Hello! > > > > > > > > > > On Wednesday 26 August 2020 09:30:11 Andre Heider wrote: > > > > > > Pali, > > > > > > > > > > > > thank you for all these espressobin fixes, much appreciated! > > > > > > > > > > > > This isn't directly related to the patch I'm replying to, but since > > > > > > you seem > > > > > > to be syncing down- with upstream: > > > > > > > > > > > > Upstream u-boot doesn't detect my sata ssd ("SATA link 0 timeout"), > > > > > > nor is > > > > > > > > > > Could you try following Marvell SATA/AHCI patch? > > > > > https://github.com/MarvellEmbeddedProcessors/u-boot-marvell/commit/2c1571432e4fd298be0236b53511f415bbdd21ac > > > > > > > > will do that later and report my findings, thanks! > > That one didn't help unfortunately. u-boot still prints "SATA link 0 > timeout". Booting kernel 5.7.0-0.bpo.2-arm64 with that and sata works, but > booting 4.19.0-10-arm64 and it doesn't: "ata1: SATA link down (SStatus 131 > SControl 300)" is printed all over again.
I guess it was broken also prior applying all those patches. So for now I do not have any fix for this. Somebody else would need to look at it. > > I looked at Debian and booting Debian kernel with Debian initrd failed: > > > > => booti ${kernel_addr_r} ${ramdisk_addr_r}:{filesize} ${fdt_addr_r} > > Moving Image from 0x7000000 to 0x7080000, end=82f2000 > > ERROR: RD image overlaps OS image (OS=0x7080000..0x82f2000) > > > > Debian's kernel is big, so it is needed to move also ramdisk_addr_r env > > variable. Following patch helped: > > > > diff --git a/include/configs/mvebu_armada-37xx.h > > b/include/configs/mvebu_armada-37xx.h > > index 6f72a13301..8929fcf390 100644 > > --- a/include/configs/mvebu_armada-37xx.h > > +++ b/include/configs/mvebu_armada-37xx.h > > @@ -99,7 +99,7 @@ > > "fdt_addr_r=0x6f00000\0" \ > > "kernel_addr=0x7000000\0" \ > > "kernel_addr_r=0x7000000\0" \ > > - "ramdisk_addr_r=0x8000000\0" \ > > + "ramdisk_addr_r=0xa000000\0" \ > > "console=" CONFIG_DEFAULT_CONSOLE "\0" \ > > BOOTENV > > > > And after that Debian kernel with Debian initrd and FDT was booted: > > Heh, nice. That indeed works! Ok, I have sent a new version V2 of this patch series and updated first patch which fix also ramdisk_addr_r. > Well minus one thing, see my reply to "arm: mvebu: Espressobin: Define > console env". For now I have dropped this patch for V2, until it is solved/fixed in some way.