Hi Andre,
Just FYI, if you want to build bootloader for SATA boot, please use the ATF build instructions like below:
make DEBUG=1 USE_COHERENT_MEM=0 LOG_LEVEL=20 SECURE=0 CLOCKSPRESET=CPU_1000_DDR_800 DDR_TOPOLOGY=2 BOOTDEV=SATA PARTNUM=0 PLAT=a3700 WTP=`pwd`/../A3700-utils-marvell all fip
The A3700 bootROM has partition capability, so if you dd the flash-image.bin to a partition with type 0x4d, and the ESPRESSObin is able to find the bootloader at power up. Of course you can dd to LBA0 to make it simpler.
Thanks, Benjamin
-----Original Message----- From: André Przywara [mailto:andre.przywara@arm.com] Sent: Sunday, July 23, 2017 2:06 PM To: Andreas Färber afaerber@suse.de Cc: Kostya Porotchkin kostap@marvell.com; Thomas Petazzoni thomas.petazzoni@free-electrons.com; Benjamin (Guodan) Huang gdhuang@marvell.com; Marc Zyngier Marc.Zyngier@arm.com; U-Boot u-boot@lists.denx.de; Stefan Roese sr@denx.de Subject: [EXT] Re: How to update Espressobin (U-Boot) firmware?
External Email
---------------------------------------------------------------------- On 19/07/17 18:01, Andre Przywara wrote:
Hi Andreas,
On 19/07/17 17:58, Andreas Färber wrote:
Hi,
Am 19.07.2017 um 18:50 schrieb Andre Przywara:
On 19/07/17 17:37, Kostya Porotchkin wrote:
I probably have to build and try the SATA image as the next step.
Yeah, that was my plan for tonight as well, since this would allow me to experiment safely with the generated image.
Oh, BTW: Chain-loading mainline U-Boot worked quite nicely and I could launch the generic debian-testing installer via EFI boot. The only issue was that I was using the U-Boot DT directly, which isn't enough for the kernel (missing clocks and interrupts). I will try to send some patches to bring the U-Boot DT in sync with the kernel one. If this is in place, one could simply attach some (UEFI enabled) USB pen drive and it should start without further ado (given a newer U-Boot in the SPI flash with EFI support).
Which devices have you successfully enumerated on mainline U-Boot? Does SATA work for you? I had reported an endless reset on another board. USB was not quite working either last time I tried on the other board - either the initial scan or a reset had similar symptoms to SATA.
I was using USB (2.0), that worked with origin/master. I haven't tested SATA yet, but will try tonight and let you know.
Quick update: Indeed SATA does not work with mainline U-Boot, I see it probing, but it apparently runs into a timeout. Also MMC (SD card) didn't work for me, I think there was no device detected at all. USB works. This was with chainloading U-Boot, so it might as well be a problem with those devices not being turned off properly and/or the mainline init code not being able to cope with all corner cases (like already initialized devices).
All of MMC, USB and SATA work with the Marvell U-Boot version on the SPI flash.
Unfortunately I can't do further tests or debugging at the moment (although the build process now seems to work correctly), as I managed to snap off the microUSB socket from my board (while pushing apparently too hard to the SATA cable), so I lost console access and couldn't recover yet. I didn't have success with a 1.8V serial adapter connected to those UART pins on the JTAG header. I will try later to solder a TX and RX wire to the PL2303 chip's pins directly (where I can measure 1.8V in contrast to those JTAG header pins) - but only after my holidays ;-)
Cheers, Andre.