Happy new year Arm-netbook !
Thank you Felix to make me recall this point of the project.
I'd read, well I tried to... the "EOMA68-A20 2.7.4 preproduction board" page (thank's to Pablo I guess) and some other pages linked to.
I'am sad to see that I'am under skilled to play with. In my old times I knew how to compile a linux kernel... but here it is a big step for me.
Anyway, if some discussions are coming here about this topic, maybe I could see how to help, and learn more about this step.
Does somebody has work in progress making this boot image and would like to share his experience here (or in an other way) ?
Franck
El sáb, 17 dic 2022 a las 18:25, lkcl (luke.leighton@gmail.com) escribió:
next step is a boot image to be able to help test the Cards.
i will not be able to assist with this.
if someone else would like to volunteer to create a boot test image that would be very helpful. i can arrange for a Card in advance (and one of the precious few micro-desktops to be sent on loan).
parabolath was the last person to work on this.
https://rhombus-tech.net/allwinner/a20/EOMA68-A20_2-7-4_preproduction/
On Thu, Jan 12, 2023 at 8:26 AM Franck Sinimalé franck@sinimale.fr wrote:
I'am sad to see that I'am under skilled to play with. In my old times I knew how to compile a linux kernel... but here it is a big step for me.
yes this one is a cross-compile. there are extra prefix ENV vars to define to get it to work.
l.
I'm curious -- does the old parabola image no longer work?
unfortunately, I'm swamped with work at the moment, otherwise I would take a look myself.
-A
On Thu, Jan 12, 2023 at 12:54:58PM +0000, Luke Kenneth Casson Leighton wrote:
On Thu, Jan 12, 2023 at 8:26 AM Franck Sinimalé franck@sinimale.fr wrote:
I'am sad to see that I'am under skilled to play with. In my old times I knew how to compile a linux kernel... but here it is a big step for me.
yes this one is a cross-compile. there are extra prefix ENV vars to define to get it to work.
l.
arm-netbook mailing list arm-netbook@lists.phcomp.co.uk Configuration, options, subscribe and unsubscribe at: http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook Send large attachments to: arm-netbook@files.phcomp.co.uk
On Thursday, January 12, 2023, Andreas Grapentin andreas@grapentin.org wrote:
I'm curious -- does the old parabola image no longer work??
from 2016? unfortunately the MicroSD card with the only copy destroyed itself.
i do have a debian/9 image left but it is not in a state to be used for micro-desktop testing.
unfortunately, I'm swamped with work at the moment, otherwise I would take a look myself.
appreciated.
l.
On 12. Jan 2023, at 15:44, Luke Kenneth Casson Leighton lkcl@lkcl.net wrote:
On Thursday, January 12, 2023, Andreas Grapentin andreas@grapentin.org wrote:
I'm curious -- does the old parabola image no longer work??
from 2016? unfortunately the MicroSD card with the only copy destroyed itself.
I still have a copy, I think. I’ll go looking for it. maybe someone can give it a spin.
On Thursday, January 12, 2023, Andreas Grapentin andreas@grapentin.org wrote:
On 12. Jan 2023, at 15:44, Luke Kenneth Casson Leighton lkcl@lkcl.net
wrote:
On Thursday, January 12, 2023, Andreas Grapentin andreas@grapentin.org wrote:
I'm curious -- does the old parabola image no longer work??
from 2016? unfortunately the MicroSD card with the only copy destroyed itself.
I still have a copy, I think. I’ll go looking for it.
yes please. do you use bittorrent, can you publish a magnet link? i will grab and seed it if you do?
l.
On 13. Jan 2023, at 04:12, Luke Kenneth Casson Leighton lkcl@lkcl.net wrote:
On Thursday, January 12, 2023, Andreas Grapentin <andreas@grapentin.org mailto:andreas@grapentin.org> wrote:
On 12. Jan 2023, at 15:44, Luke Kenneth Casson Leighton lkcl@lkcl.net
wrote:
On Thursday, January 12, 2023, Andreas Grapentin andreas@grapentin.org wrote:
I'm curious -- does the old parabola image no longer work??
from 2016? unfortunately the MicroSD card with the only copy destroyed itself.
I still have a copy, I think. I’ll go looking for it.
yes please. do you use bittorrent, can you publish a magnet link? i will grab and seed it if you do?
I’ve found the parabola image I used back in 2021, and tried to boot it on an EOMA card revision 2.7.4 and a microdesktop revision 1.7
In the process of attaching peripheries, I must have blown something on the microdesktop, the power LED no longer lights up, tested multiple chargers. I’m fairly certain that the LED did light up green when I started working on the board today. Is it possible that I blew a fuse somewhere?
Apart from that, I currently can’t get the system to boot. I have output from the boot loader, but it’s not much:
U-Boot SPL 2017.01-2 (Jan 18 2017 - 21:30:38) DRAM: 2048 MiB CPU: 912000000Hz, AXI/AHB/APB: 3/2/2 Trying to boot from MMC2MMC Device 1 not found spl: could not find mmc device. error: -19 SPL: failed to boot from all boot devices ### ERROR ### Please RESET the board ###
This is when I insert the sd card in the EOMA card directly. When I insert the sd card into the microdesktop, I don’t get anything. Still working on it though.
I’m confident that this is salvageable, considering that I do get boot loader output through UART. I’ll keep looking into it. If you have any hints, I’d appreciate it.
in the meantime, I’ll prepare a magnet link / http direct download for the image I’m using, in case it’s useful for someone else. I’ll send those links to the list ASAP.
Best, Andreas
On Mon, Jan 16, 2023 at 6:56 PM Andreas Grapentin andreas@grapentin.org wrote:
I’ve found the parabola image I used back in 2021, and tried to boot it on an EOMA card revision 2.7.4 and a microdesktop revision 1.7
brilliant
In the process of attaching peripheries, I must have blown something on the microdesktop, the power LED no longer lights up, tested multiple chargers. I’m fairly certain that the LED did light up green when I started working on the board today. Is it possible that I blew a fuse somewhere?
yes it should. no there's no fuse. you'll need a multimeter, be *careful*. i suggest taking the Card out. if it's really borked you can power it with a 5V USB "charger" cable (into the 2 standard USB sockets).
also, check you inserted the Card correctly: not one row along or one pin along, and make sure no contact between the two PCBs.
Apart from that, I currently can’t get the system to boot. I have output from the boot loader, but it’s not much:
U-Boot SPL 2017.01-2 (Jan 18 2017 - 21:30:38) DRAM: 2048 MiB CPU: 912000000Hz, AXI/AHB/APB: 3/2/2 Trying to boot from MMC2MMC Device 1 not found spl: could not find mmc device. error: -19 SPL: failed to boot from all boot devices ### ERROR ### Please RESET the board ###
ok if you've no power on the MD then the Level Shifter IC will not be powered up and the MicroSD card socket obviously will not be recognised.
I’m confident that this is salvageable, considering that I do get boot loader output through UART. I’ll keep looking into it. If you have any hints, I’d appreciate it.
it's possible to reconfigure u-boot (etc) to have the OS on the Card's micro-sd.
in the meantime, I’ll prepare a magnet link / http direct download for the image I’m using, in case it’s useful for someone else. I’ll send those links to the list ASAP.
awesome.
l.
On 16. Jan 2023, at 20:01, Luke Kenneth Casson Leighton lkcl@lkcl.net wrote:
On Mon, Jan 16, 2023 at 6:56 PM Andreas Grapentin andreas@grapentin.org wrote:
In the process of attaching peripheries, I must have blown something on the microdesktop, the power LED no longer lights up, tested multiple chargers. I’m fairly certain that the LED did light up green when I started working on the board today. Is it possible that I blew a fuse somewhere?
yes it should. no there's no fuse. you'll need a multimeter, be *careful*. i suggest taking the Card out. if it's really borked you can power it with a 5V USB "charger" cable (into the 2 standard USB sockets).
also, check you inserted the Card correctly: not one row along or one pin along, and make sure no contact between the two PCBs.
forgot to mention, I’m using the expansion header on the micro desktop for the UART connection. Power to the EOMA card directly through the microUSB works, that’s what I’m doing right now. I haven’t tried the USB sockets on the micro desktop though, that might be worth a shot.
I’m confident that this is salvageable, considering that I do get boot loader output through UART. I’ll keep looking into it. If you have any hints, I’d appreciate it.
it's possible to reconfigure u-boot (etc) to have the OS on the Card's micro-sd.
I believe I’m using a quite old u-boot boot script. Included below. I’m not too familiar with u-boot, any tips what needs to be changed to switch the microSD slot, or maybe include a fallback option to try booting from the internal microSD if the external one is unavailable?
if test -n ${distro_bootpart}; then setenv bootpart ${distro_bootpart}; else setenv bootpart 1; fi part uuid ${devtype} ${devnum}:${bootpart} uuid
setenv bootargs console=${console} root=PARTUUID=${uuid} rw rootwait
if load ${devtype} ${devnum}:${bootpart} ${kernel_addr_r} /boot/zImage; then if load ${devtype} ${devnum}:${bootpart} ${fdt_addr_r} /boot/dtbs/${fdtfile}; then if load ${devtype} ${devnum}:${bootpart} ${ramdisk_addr_r} /boot/initramfs-linux.img; then bootz ${kernel_addr_r} ${ramdisk_addr_r}:${filesize} ${fdt_addr_r}; else bootz ${kernel_addr_r} - ${fdt_addr_r}; fi; fi; fi
On Monday, January 16, 2023, Andreas Grapentin andreas@grapentin.org wrote:
On 16. Jan 2023, at 20:01, Luke Kenneth Casson Leighton lkcl@lkcl.net
wrote:
On Mon, Jan 16, 2023 at 6:56 PM Andreas Grapentin andreas@grapentin.org
wrote:
In the process of attaching peripheries, I must have blown something on
the microdesktop, the power LED no longer lights up, tested multiple chargers. I’m fairly certain that the LED did light up green when I started working on the board today. Is it possible that I blew a fuse somewhere?
yes it should. no there's no fuse. you'll need a multimeter, be
*careful*.
i suggest taking the Card out. if it's really borked you can power it with a 5V USB "charger" cable (into the 2 standard USB sockets).
also, check you inserted the Card correctly: not one row along or one pin along, and make sure no contact between the two PCBs.
forgot to mention, I’m using the expansion header on the micro desktop
for the UART connection.
yep ido that.
Power to the EOMA card directly through the microUSB works, that’s what
I’m doing right now.
the SY6280 is *only* powered by the 5v MD Power rail. you *cannot* power the MD through the USB-OTG and if it was possible it would damage everything as the power drain would be exceeded.
the 5v rail connects to 3.3v which then powers the Level Shifter and the microsd socket.
you abolutely cannot expect the VREF of the Card to power the microsd socket.
it is imperative that you provide independent 5v power.
I haven’t tried the USB sockets on the micro desktop though, that might
be worth a shot.
if 12v is not working it will be your only option.
it's possible to reconfigure u-boot (etc) to have the OS on the Card's
micro-sd.
I believe I’m using a quite old u-boot boot script.
that is irrelevant. u-boot-spl.bin is what needs compiling and that is done with compile options. "make config".
i think.
you need to establish which phase has been achieved. there are MULTIPLE phases. EVERY ONE OF THEM needs properly configuring.
l.
Hello,
it's possible to reconfigure u-boot (etc) to have the OS on the Card's
micro-sd.
I believe I’m using a quite old u-boot boot script.
that is irrelevant. u-boot-spl.bin is what needs compiling and that is done with compile options. "make config".
i think.
you need to establish which phase has been achieved. there are MULTIPLE phases. EVERY ONE OF THEM needs properly configuring.
I should have a working u-boot configuration and binary somewhere on my hard drive. Let me check my files. Hopefully I will also find some comments and notes that are not yet in the rhombus-tech wiki. My search and answer on the mailing list may take me a few days.
Pablo
On Mon, Jan 16, 2023 at 07:55:47PM +0100, Andreas Grapentin wrote:
I’ve found the parabola image I used back in 2021, and tried to boot it on an EOMA card revision 2.7.4 and a microdesktop revision 1.7
In the process of attaching peripheries, I must have blown something on the microdesktop, the power LED no longer lights up, tested multiple chargers. I’m fairly certain that the LED did light up green when I started working on the board today. Is it possible that I blew a fuse somewhere?
Apart from that, I currently can’t get the system to boot. I have output from the boot loader, but it’s not much:
U-Boot SPL 2017.01-2 (Jan 18 2017 - 21:30:38) DRAM: 2048 MiB CPU: 912000000Hz, AXI/AHB/APB: 3/2/2 Trying to boot from MMC2MMC Device 1 not found spl: could not find mmc device. error: -19 SPL: failed to boot from all boot devices ### ERROR ### Please RESET the board ###
This is when I insert the sd card in the EOMA card directly. When I insert the sd card into the microdesktop, I don’t get anything. Still working on it though.
Could you please clarify your setup. Is this sunxi-u-boot (legacy u-boot) via fel or written to the sd card?
Are there changes software or hardwire wise? I don't understand why it worked 2021 and fails to boot now.
Pablo
On Tue, Jan 17, 2023 at 08:46:34AM +0100, Pablo Rath wrote:
Could you please clarify your setup. Is this sunxi-u-boot (legacy u-boot) via fel or written to the sd card?
Unfortunately, I don't remember what bootloader I installed on that image. I'm currently in the process of making a toolchain cabable of building the legacy sunxi u-boot.
Are there changes software or hardwire wise? I don't understand why it worked 2021 and fails to boot now.
IIRC I had only uart output working back in 2021, and while setting up the uart connection again, I miswired the microdesktop headers and must have shorted something that destroyed some active component on the microdesktop. That's my reading of the situation.
I'll upload the 2021 image later today, so other people can take a look. But I've already seen that the installation is a bit broken and there's no clear upgrade path. I went through the motions and manually updated the installation in a chroot. Once I have boot back up working I'll give that version a spin and also upload it.
-A
On Tue, Jan 17, 2023 at 10:05:25AM +0100, Andreas Grapentin wrote:
On Tue, Jan 17, 2023 at 08:46:34AM +0100, Pablo Rath wrote:
Could you please clarify your setup. Is this sunxi-u-boot (legacy u-boot) via fel or written to the sd card?
Unfortunately, I don't remember what bootloader I installed on that image. I'm currently in the process of making a toolchain cabable of building the legacy sunxi u-boot.
Ok. You are aware of that wiki page, are you?: https://rhombus-tech.net/allwinner_a10/Buildroot_Toolchain/
Once you have the toolchain running and a copy of legacy sunxi u-boot make sure to checkout this tag: git checkout v2013.07-sunxi.4
make sure to change the boards.cfg (MMC_SUNXI_SLOT=2), and write sunxi U-Boot to SD-Card with the commands described after v2013.07 or earlier. This were the steps I had to take to boot a sunxi 3.4 kernel and a Debian rootfs in 2020. I have documented this here: https://rhombus-tech.net/allwinner/a20/EOMA68-A20_2-7-4_preproduction/ (Section Sunxi U-Boot (legacy))
If it helps, i could upload a copy of sunxi-spl.bin and u-boot.bin for you.
I'll upload the 2021 image later today, so other people can take a look. But I've already seen that the installation is a bit broken and there's no clear upgrade path. I went through the motions and manually updated the installation in a chroot. Once I have boot back up working I'll give that version a spin and also upload it.
Ok, I'll have a look at your image.
Pablo
On 17. Jan 2023, at 11:10, Pablo Rath pablo@parobalth.org wrote:
Ok. You are aware of that wiki page, are you?: https://rhombus-tech.net/allwinner_a10/Buildroot_Toolchain/
Once you have the toolchain running and a copy of legacy sunxi u-boot make sure to checkout this tag: git checkout v2013.07-sunxi.4
make sure to change the boards.cfg (MMC_SUNXI_SLOT=2), and write sunxi U-Boot to SD-Card with the commands described after v2013.07 or earlier. This were the steps I had to take to boot a sunxi 3.4 kernel and a Debian rootfs in 2020. I have documented this here: https://rhombus-tech.net/allwinner/a20/EOMA68-A20_2-7-4_preproduction/ (Section Sunxi U-Boot (legacy))
If it helps, i could upload a copy of sunxi-spl.bin and u-boot.bin for you.
Thank you for those instructions, they were immensely helpful. I compiled u-boot from the v2013.07-sunxi.4 tag, made the changes to the boards.cfg you suggested, and deployed to SDcard, and the behaviour has changed. The system now boot loops, repeating this output:
U-Boot SPL 2013.07-07841-g569c37da7d-dirty (Jan 17 2023 - 19:37:10) Board: EOMA68_A20 DRAM: 1024 MiB CPU: 960000000Hz, AXI/AHB/APB: 3/2/2 SUNXI SD/MMC: 0
for reference, this was the previous output:
U-Boot SPL 2017.01-2 (Jan 18 2017 - 21:30:38) DRAM: 2048 MiB CPU: 912000000Hz, AXI/AHB/APB: 3/2/2 Trying to boot from MMC2MMC Device 1 not found spl: could not find mmc device. error: -19 SPL: failed to boot from all boot devices ### ERROR ### Please RESET the board ###
first thing I notice is that the reported CPU frequency and available RAM is different. I wish that u-boot would present me with a command line instead of resetting, but I haven’t figured out how to do that yet.
-A
On Tue, Jan 17, 2023 at 07:48:40PM +0100, Andreas Grapentin wrote:
On 17. Jan 2023, at 11:10, Pablo Rath pablo@parobalth.org wrote:
Ok. You are aware of that wiki page, are you?: https://rhombus-tech.net/allwinner_a10/Buildroot_Toolchain/
Once you have the toolchain running and a copy of legacy sunxi u-boot make sure to checkout this tag: git checkout v2013.07-sunxi.4
make sure to change the boards.cfg (MMC_SUNXI_SLOT=2), and write sunxi U-Boot to SD-Card with the commands described after v2013.07 or earlier. This were the steps I had to take to boot a sunxi 3.4 kernel and a Debian rootfs in 2020. I have documented this here: https://rhombus-tech.net/allwinner/a20/EOMA68-A20_2-7-4_preproduction/ (Section Sunxi U-Boot (legacy))
If it helps, i could upload a copy of sunxi-spl.bin and u-boot.bin for you.
Thank you for those instructions, they were immensely helpful. I compiled u-boot from the v2013.07-sunxi.4 tag, made the changes to the boards.cfg you suggested, and deployed to SDcard, and the behaviour has changed. The system now boot loops, repeating this output:
U-Boot SPL 2013.07-07841-g569c37da7d-dirty (Jan 17 2023 - 19:37:10) Board: EOMA68_A20 DRAM: 1024 MiB CPU: 960000000Hz, AXI/AHB/APB: 3/2/2 SUNXI SD/MMC: 0
Good. I think you must be almost there. Sorry for the inconvenience but I double checked today and I made changes to boards.cfg myself. So I used SLOT=2 for both lines (EOMA68_A20 and EOMA68_A20_SPL) Here is my git diff: commit 3b4b766cee85be6bac873b5cf797d0def979d88d Author: Pablo Rath pablo@parobalth.org Date: Sat Jul 25 13:42:53 2020 +0200
Changed to slot 2
diff --git a/boards.cfg b/boards.cfg index 5b61dcefa5..5a4e293e1a 100644 --- a/boards.cfg +++ b/boards.cfg @@ -361,8 +361,8 @@ Cubieboard2_FEL arm armv7 sunxi - DNS_M82 arm armv7 sunxi - sunxi sun4i:DNS_M82,SPL EOMA68_A10 arm armv7 sunxi - sunxi sun4i:EOMA68_A10,SPL,MMC_SUNXI_SLOT=3,SUNXI_EMAC EOMA68_A10_SPL arm armv7 sunxi - sunxi sun4i:EOMA68_A10,SPL_FEL,MMC_SUNXI_SLOT=3,SUNXI_EMAC -EOMA68_A20 arm armv7 sunxi - sunxi sun7i:EOMA68_A20,SPL,MMC_SUNXI_SLOT=3,SUNXI_EMAC -EOMA68_A20_SPL arm armv7 sunxi - sunxi sun7i:EOMA68_A20,SPL_FEL,MMC_SUNXI_SLOT=3,SUNXI_EMAC +EOMA68_A20 arm armv7 sunxi - sunxi sun7i:EOMA68_A20,SPL,MMC_SUNXI_SLOT=2 +EOMA68_A20_SPL arm armv7 sunxi - sunxi sun7i:EOMA68_A20,SPL_FEL,MMC_SUNXI_SLOT=2 Gooseberry_A721 arm armv7 sunxi - sunxi sun4i:GOOSEBERRY_A721,SPL H6 arm armv7 sunxi - sunxi sun4i:H6,SPL Hackberry arm armv7 sunxi - sunxi sun4i:HACKBERRY,SPL
Now to your boot loop. Obviously something is not right yet and so you don't get to a U-boot prompt because U-Boot is not (fully) loaded. My suggestions:
1. Try an empty, formatted sd card and write sunxi-spl.bin and u-boot.bin with the right offsets:
dd if=spl/sunxi-spl.bin of=${card} bs=1024 seek=8 dd if=u-boot.bin of=${card} bs=1024 seek=32
Boot this card and see if you get to the U-Boot prompt.
2. Try "u-boot-sunxi-with-spl.bin" from your U-Boot build via Fel
sunxi-fel uboot u-boot-sunxi-with-spl.bin
See https://rhombus-tech.net/allwinner/a20/EOMA68-A20_2-7-4_preproduction/#bootv... and https://rhombus-tech.net/allwinner/a20/EOMA68-A20_2-7-4_preproduction/#notes if you have never used Fel before.
3. I could send you my sunxi-spl.bin, u-boot.bin and u-boot-sunxi-with-spl.bin to test.
4. Your u-boot build went fine with no (critical) errors?
Pablo
On Tuesday, January 17, 2023, Andreas Grapentin andreas@grapentin.org wrote:
On 17. Jan 2023, at 11:10, Pablo Rath pablo@parobalth.org wrote:
Ok. You are aware of that wiki page, are you?: https://rhombus-tech.net/allwinner_a10/Buildroot_Toolchain/
Once you have the toolchain running and a copy of legacy sunxi u-boot make sure to checkout this tag: git checkout v2013.07-sunxi.4
make sure to change the boards.cfg (MMC_SUNXI_SLOT=2), and write sunxi
U-Boot to SD-Card with the commands described after v2013.07 or earlier. This were the steps I had to take to boot a sunxi 3.4 kernel and a Debian rootfs in 2020.
I have documented this here: https://rhombus-tech.net/allwinner/a20/EOMA68-A20_2-7-4_preproduction/ (Section Sunxi U-Boot (legacy))
If it helps, i could upload a copy of sunxi-spl.bin and u-boot.bin for you.
Thank you for those instructions, they were immensely helpful. I compiled
u-boot from the v2013.07-sunxi.4 tag, made the changes to the boards.cfg you suggested, and deployed to SDcard, and the behaviour has changed. The system now boot loops, repeating this output:
U-Boot SPL 2013.07-07841-g569c37da7d-dirty (Jan 17 2023 - 19:37:10) Board: EOMA68_A20 DRAM: 1024 MiB CPU: 960000000Hz, AXI/AHB/APB: 3/2/2 SUNXI SD/MMC: 0
okaay the SPL loader will now be trying to look on sd/mmc 0 for a u-boot image. clearly it is not finding it.
and that's because sd/mmc 0 is routed to the microdesktop. sd/mmc *3* is on the EOMA68-A20 Card.
which socket is the micro-sd card inserted in? (and have you remembered to dd u-boot.bin onto it?) and, did you fix the power issues which would allow the MD SDcard Level Shifter to activate?
for reference, this was the previous output:
U-Boot SPL 2017.01-2 (Jan 18 2017 - 21:30:38) DRAM: 2048 MiB CPU: 912000000Hz, AXI/AHB/APB: 3/2/2 Trying to boot from MMC2MMC Device 1 not found
yup there *is* no SD/MMC 1, u-boot-spl was compiled incorrectly in this case.
spl: could not find mmc device. error: -19 SPL: failed to boot from all boot devices ### ERROR ### Please RESET the board ###
first thing I notice is that the reported CPU frequency and available RAM
is different.
not important.
I wish that u-boot
... you are not yet *running* u-boot. it is so big (1mb) it cannot fit into 16k SRAM. the SPL loader initialises DDR3 memory so that u-boot *can* be loaded into DRAM.
but you are not there yet.
would present me with a command line instead of resetting, but I haven’t
figured out how to do that yet.
the boot process is complex and multi-stage. what the SPL loader is going into is a mere 16k SRAM.
there is absolutely flat-out zero chance of getting a "prompt" in the *SPL* loader.
in 2nd stage - actual u-boot - not a problem at all.
l.
On 18. Jan 2023, at 14:54, Luke Kenneth Casson Leighton lkcl@lkcl.net wrote:
On Tuesday, January 17, 2023, Andreas Grapentin <andreas@grapentin.org mailto:andreas@grapentin.org> wrote:
U-Boot SPL 2013.07-07841-g569c37da7d-dirty (Jan 17 2023 - 19:37:10) Board: EOMA68_A20 DRAM: 1024 MiB CPU: 960000000Hz, AXI/AHB/APB: 3/2/2 SUNXI SD/MMC: 0
okaay the SPL loader will now be trying to look on sd/mmc 0 for a u-boot image. clearly it is not finding it.
From looking at the sunxi spl code, this is the enumeration of all detected mmc slots, not the configured one. It would be good to compare the output to a known good board.
which socket is the micro-sd card inserted in? (and have you remembered to dd u-boot.bin onto it?) and, did you fix the power issues which would allow the MD SDcard Level Shifter to activate?
The card is inserted in the socket on the EOMA card, I figured since the micro desktop is not getting power, the slot would probably be unusable.
I started looking at the micro desktop board with my multimeter and did some checks, but it’s difficult without reference to a known good board. But what is maybe noteworthy is that of the three larger capacitors on the board just behind the power socket, both sides of all of them are connected to ground. I assume that part of the board is responsible for the power control? If so, then the small chip in the middle might be blown. Again, hard to verify without a known good board.
for reference, this was the previous output:
U-Boot SPL 2017.01-2 (Jan 18 2017 - 21:30:38) DRAM: 2048 MiB CPU: 912000000Hz, AXI/AHB/APB: 3/2/2 Trying to boot from MMC2MMC Device 1 not found
yup there *is* no SD/MMC 1, u-boot-spl was compiled incorrectly in this case.
I think you’re right, let’s discard that previous output as garbage.
the boot process is complex and multi-stage. what the SPL loader is going into is a mere 16k SRAM.
Understood. I’ll keep trying to get u-boot proper up and running.
I’ve noticed that the boot loops above are pretty random — it might be possible I’m not supplying enough power to the microsd slot, I’ll try to use a stronger power supply.
Otherwise, I’m pretty certain I’m using the correct microsd slot, because configuring any other slot in boards.cfg results in timeout errors in the MMC code, which leads to the reported error -1
-A
On 19. Jan 2023, at 10:06, Andreas Grapentin andreas@grapentin.org wrote:
I started looking at the micro desktop board with my multimeter and did some checks, but it’s difficult without reference to a known good board. But what is maybe noteworthy is that of the three larger capacitors on the board just behind the power socket, both sides of all of them are connected to ground. I assume that part of the board is responsible for the power control? If so, then the small chip in the middle might be blown. Again, hard to verify without a known good board.
I found the schematics for the micro desktop, and was able to take a closer look. I tested around the RT8288A, and noticed that pins 1, 2, 3 and 8 have basically zero resistance to ground. That’s not good.
I’ll get a hot-air solder station and a microscope, and get that chip off the board, then do some more tests to see if the other components around that area are okay. It could also be that one of the capacitors nearby is blown.
I’ll try to get this micro desktop back to working order, because I have the suspicion that the power issues of the micro desktop are causing the boot loops, and I don’t want to break anything else :)
-A
On Thu, Jan 19, 2023 at 10:44:09AM +0100, Andreas Grapentin wrote:
On 19. Jan 2023, at 10:06, Andreas Grapentin andreas@grapentin.org wrote:
I started looking at the micro desktop board with my multimeter and did some checks, but it’s difficult without reference to a known good board. But what is maybe noteworthy is that of the three larger capacitors on the board just behind the power socket, both sides of all of them are connected to ground. I assume that part of the board is responsible for the power control? If so, then the small chip in the middle might be blown. Again, hard to verify without a known good board.
I found the schematics for the micro desktop, and was able to take a closer look. I tested around the RT8288A, and noticed that pins 1, 2, 3 and 8 have basically zero resistance to ground. That’s not good.
I’ll get a hot-air solder station and a microscope, and get that chip off the board, then do some more tests to see if the other components around that area are okay. It could also be that one of the capacitors nearby is blown.
I’ll try to get this micro desktop back to working order, because I have the suspicion that the power issues of the micro desktop are causing the boot loops, and I don’t want to break anything else :)
So we have two "active" testers with a broken micro desktop each. This is the thread where we discussed my power problems and the blown RT8288 in 2020 (see attachment and Lukes replies in the thread [a redesign of this area for micro desktop 1.8] and a schematic as attachment):
https://lists.phcomp.co.uk/pipermail/arm-netbook/2020-July/016353.html
Andreas, Do you want to accept Lukes offer for one of the five precious micro desktops? If you are done with your testing, you could send it to me. I am curious if my Computer Card is still working.
Pablo
--- crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
On Thu, Jan 19, 2023 at 9:06 AM Andreas Grapentin andreas@grapentin.org wrote:
On 18. Jan 2023, at 14:54, Luke Kenneth Casson Leighton lkcl@lkcl.net wrote:
On Tuesday, January 17, 2023, Andreas Grapentin <andreas@grapentin.org mailto:andreas@grapentin.org> wrote:
U-Boot SPL 2013.07-07841-g569c37da7d-dirty (Jan 17 2023 - 19:37:10) Board: EOMA68_A20 DRAM: 1024 MiB CPU: 960000000Hz, AXI/AHB/APB: 3/2/2 SUNXI SD/MMC: 0
okaay the SPL loader will now be trying to look on sd/mmc 0 for a u-boot image. clearly it is not finding it.
From looking at the sunxi spl code, this is the enumeration of all detected mmc slots, not the configured one. It would be good to compare the output to a known good board.
ah - right. then you may have an earlier revision of the EOMA68-A20 which does indeed have sd/mmc 0 attached to the micro-sd socket.
in which case the advised build instructions - which are for the *newer* EOMA68-A20 - will obviously not work.
I started looking at the micro desktop board with my multimeter and did some checks, but it’s difficult without reference to a known good board. But what is maybe noteworthy is that of the three larger capacitors on the board just behind the power socket, both sides of all of them are connected to ground.
that's a serious short, yes, if the resistance between them is zero (or close to it). check the 5v rail.
I assume that part of the board is responsible for the power control?
7-22v down to 5v, yes.
I’ve noticed that the boot loops above are pretty random — it might be possible I’m not supplying enough power to the microsd slot, I’ll try to use a stronger power supply.
that would be unwise as it will cause further burnouts due to the short.
Otherwise, I’m pretty certain I’m using the correct microsd slot, because configuring any other slot in boards.cfg results in timeout errors in the MMC code, which leads to the reported error -1
you are better off powering from USB-OTG entirely bypassing the MD or removing the RT8288 entirely from the PCB.
the problem with not using the MD is you have to insert wires directly into the EOMA68-A20 tx/rx and GND holes. done that before, it was a pain-in-the-ass but it worked. you do NOT need power because they are open-drain.
l.
On 19. Jan 2023, at 11:44, Luke Kenneth Casson Leighton lkcl@lkcl.net wrote:
that's a serious short, yes, if the resistance between them is zero (or close to it). check the 5v rail.
I removed the RT8288, and now the 5V rail is no longer short to ground. However, I still can’t power the board over USB, I only get around 13mV on the 5V rail. I’ll check around some more and see what I can find. maybe a resistor is faulty.
-A
--- crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
On Sun, Jan 22, 2023 at 7:46 PM Andreas Grapentin andreas@grapentin.org wrote:
On 19. Jan 2023, at 11:44, Luke Kenneth Casson Leighton lkcl@lkcl.net wrote:
that's a serious short, yes, if the resistance between them is zero (or close to it). check the 5v rail.
I removed the RT8288, and now the 5V rail is no longer short to ground. However, I still can’t power the board over USB, I only get around 13mV on the 5V rail. I’ll check around some more and see what I can find. maybe a resistor is faulty.
next thing to check will be the SY6280 which is a current-limiter IC. datasheets and schematics: https://hands.com/~lkcl/eoma/microdesktop/
l.
On 22. Jan 2023, at 23:15, Luke Kenneth Casson Leighton lkcl@lkcl.net wrote: On Sun, Jan 22, 2023 at 7:46 PM Andreas Grapentin andreas@grapentin.org wrote:
I removed the RT8288, and now the 5V rail is no longer short to ground. However, I still can’t power the board over USB, I only get around 13mV on the 5V rail. I’ll check around some more and see what I can find. maybe a resistor is faulty.
next thing to check will be the SY6280 which is a current-limiter IC. datasheets and schematics: https://hands.com/~lkcl/eoma/microdesktop/
I’m looking at the SY6280, but I’m not sure what I’m looking at. I see 5V at the OUT pin going to the USB OTG, and I see no voltage an the IN line to the 5V rail. Enable is pulled high, and both resistors are good at 10k and 6.8k respectively. I have no idea how to check C9 or C8, but I assume those are good.
In my limited understanding, this does seem to indicate that the chip is faulty, right? I’ll try to get a replacement and see if swapping it out changes anything. I’ll also try to get a replacement for the RT8288 just for kicks.
-A
On 23. Jan 2023, at 13:49, Andreas Grapentin andreas@grapentin.org wrote:
On 22. Jan 2023, at 23:15, Luke Kenneth Casson Leighton lkcl@lkcl.net wrote: On Sun, Jan 22, 2023 at 7:46 PM Andreas Grapentin andreas@grapentin.org wrote:
I removed the RT8288, and now the 5V rail is no longer short to ground. However, I still can’t power the board over USB, I only get around 13mV on the 5V rail. I’ll check around some more and see what I can find. maybe a resistor is faulty.
next thing to check will be the SY6280 which is a current-limiter IC. datasheets and schematics: https://hands.com/~lkcl/eoma/microdesktop/
I’m looking at the SY6280, but I’m not sure what I’m looking at. I see 5V at the OUT pin going to the USB OTG, and I see no voltage an the IN line to the 5V rail. Enable is pulled high, and both resistors are good at 10k and 6.8k respectively. I have no idea how to check C9 or C8, but I assume those are good.
In my limited understanding, this does seem to indicate that the chip is faulty, right? I’ll try to get a replacement and see if swapping it out changes anything. I’ll also try to get a replacement for the RT8288 just for kicks.
Seems like the actual SY6280 is tricky to get — Can I use this one as a drop-in replacement?
https://www.digikey.de/en/products/detail/glf-integrated-power/GLF2311A-T1G7... https://www.digikey.de/en/products/detail/glf-integrated-power/GLF2311A-T1G7/16681132
-A
Luke Kenneth Casson Leighton lkcl@lkcl.net writes:
On Thursday, January 12, 2023, Andreas Grapentin andreas@grapentin.org wrote:
I'm curious -- does the old parabola image no longer work??
from 2016? unfortunately the MicroSD card with the only copy destroyed itself.
i do have a debian/9 image left but it is not in a state to be used for micro-desktop testing.
This seems to have useful tips:
https://wiki.debian.org/InstallingDebianOn/Allwinner
I guess we ought to have a page on https://rhombus-tech.net/ for tips on how to get things installed, and either link to that from the debian wiki, or hopefully make the procedure simple enough that it can just be added to one of the lists of boards on the debian page (probably with a link to the rhombus-tech page).
While wondering about this, I noticed that the search on rhombus-tech.net was not working, apparently because 'omega' was missing, which turns out to be provided by xapian-omega (I guess that got lost in a upgrade) so I've reinstalled that, which allows one to search the wiki again:
https://rhombus-tech.net/ikiwiki.cgi?P=a20
I also notice this page:
https://rhombus-tech.net/allwinner/a20/archlinux/
which seems only to be linked from the Sitemap:
https://rhombus-tech.net/sitemap/
I'd have thought we want a top-level "Installing" page, with a sub-page for the EOMA68-A20, which could be populated with links and tips, but I'm not sure because the wiki as it is has some structure, but I don't understand it -- for instance, I'd not use a page called "allwinner_a10" for information that turns out to mostly be about the EOMA68-A20, but I know that's for hysterical raisins, so is presumably not easy to change -- I think the wiki needs something to say "Don't worry about the A10 vs. A20 distinction, we started out using the A10, but it's all about A20 now" somewhere, but I'm not sure where.
BTW If a mass-renaming is in order, I can do that trivially via git, but I suspect we don't want to break links, so I won't do that unilaterally.
Cheers, Phil.
On Friday, January 13, 2023, Philip Hands phil@hands.com wrote:
Luke Kenneth Casson Leighton lkcl@lkcl.net writes:
On Thursday, January 12, 2023, Andreas Grapentin andreas@grapentin.org wrote:
I'm curious -- does the old parabola image no longer work??
from 2016? unfortunately the MicroSD card with the only copy destroyed itself.
i do have a debian/9 image left but it is not in a state to be used for micro-desktop testing.
This seems to have useful tips:
yyeah except EOMA68 critically relies (under the Certiication Mark) for having "devicetree overlays", which the debian kernel developers have refused to enable (even despite repeated requests from raspberry pi "shield" users, and also the fact that 96boards systems require dtb overlays)
EOMA68 is the third system i know that needs overlays but in this case (a first) *requires* them.
I guess we ought to have a page on https://rhombus-tech.net/ for tips on how to get things installed,
found it.
https://rhombus-tech.net/allwinner/a20/EOMA68-A20_2-7-4_preproduction/
more later have to run ty for instlling xapian really appreciated
l.
On Jan 12, 2023, at 05:55, Luke Kenneth Casson Leighton lkcl@lkcl.net wrote:
yes this one is a cross-compile. there are extra prefix ENV vars to define to get it to work.
Is there a page describing/documenting how to set up the cross-compile tools and environment (or at least which target architecture to use)?
Is there a page discussing which kernel source (+patches) supports this board. Sounds like we might want a page covering the status of support for different hardware/CPU features (if it doesn’t already exist).
On Thursday, January 12, 2023, Richard Wilbur richard.wilbur@gmail.com wrote:
On Jan 12, 2023, at 05:55, Luke Kenneth Casson Leighton lkcl@lkcl.net
wrote:
yes this one is a cross-compile. there are extra prefix ENV vars to define to get it to work.
Is there a page describing/documenting how to set up the cross-compile
tools and environment (or at least which target architecture to use)?
yes. check previous messages and rhombus-tech.net wiki changelog.
Is there a page discussing which kernel source (+patches) supports this
board. Sounds like we might want a page covering the status of support for different hardware/CPU features (if it doesn’t already exist).
it's an A20. that's enough.
l.
On Thu, Jan 12, 2023 at 09:26:04AM +0100, Franck Sinimalé wrote:
I'd read, well I tried to... the "EOMA68-A20 2.7.4 preproduction board" page (thank's to Pablo I guess) and some other pages linked to.
I'am sad to see that I'am under skilled to play with. In my old times I knew how to compile a linux kernel... but here it is a big step for me.
Anyway, if some discussions are coming here about this topic, maybe I could see how to help, and learn more about this step.
Does somebody has work in progress making this boot image and would like to share his experience here (or in an other way) ?
Most of what I know and did is in the rhomubus-tech wiki but feel free to ask me a *specific* question. https://linux-sunxi.org/ is an invaluable source of knowledge and many topics have been discussed on this list (see the archives).
To learn more try to set up a cross-compiler, learn some git basics, build Legacy U-Boot (sunxi), mainline U-Boot, a sunxi Kernel and a mainline Kernel, learn how to debootstrap a rootfs, work with sd cards (dd) and common command-line tools. Learn how to use UART and how to solder. It will get boring if you don't have actual hardware to work with. For learning purposes most armhf boards will do. Do you own one?
kind regards Pablo
arm-netbook@lists.phcomp.co.uk
-
Andreas Grapentin -
Franck Sinimalé -
Luke Kenneth Casson Leighton -
Pablo Rath -
Philip Hands -
Richard Wilbur