Bug 1556951 - Raspberry Pi 3: unable to boot
Summary: Raspberry Pi 3: unable to boot
Keywords:
Status: CLOSED ERRATA
Alias: None
Product: Fedora
Classification: Fedora
Component: uboot-tools
Version: 28
Hardware: armhfp
OS: Unspecified
unspecified
unspecified
Target Milestone: ---
Assignee: Peter Robinson
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard: AcceptedFreezeException
Depends On:
Blocks: F28BetaBlocker F28BetaFreezeException
TreeView+ depends on / blocked
 
Reported: 2018-03-15 15:33 UTC by Alessio
Modified: 2018-03-21 20:31 UTC (History)
12 users (show)

Fixed In Version: uboot-tools-2018.03-3.fc28
Doc Type: If docs needed, set a value
Doc Text:
Clone Of:
Environment:
Last Closed: 2018-03-21 20:31:53 UTC
Type: Bug
Embargoed:


Attachments (Terms of Use)

Description Alessio 2018-03-15 15:33:08 UTC
Description of problem:

While with 20180310 compose the RPi3 did not automatically boot because of bug 1552130, with 20180314 compose the uboot prompt never appears: the display stays stuck at the "multicolor" screen.


Version-Release number of selected component (if applicable):

Fedora-Minimal-armhfp-28-20180314.n.2-sda.raw
Fedora-Xfce-armhfp-28-20180314.n.2-sda.raw.xz

bcm283x-firmware-20180307-1.7fdcd00.fc28

Comment 1 Peter Robinson 2018-03-15 17:40:04 UTC
I believe this might be an issue with u-boot not firmware. Also looks to be RPi3 only (not RPi2)

Comment 2 Peter Robinson 2018-03-16 05:54:59 UTC
I've not been able to recreate this with Fedora-Minimal-armhfp-28-20180314.n.2-sda.raw.xz on a couple of different RPi3s.

I'm wondering if there's not been a power regression which is causing the device to pull too much power while booting the firmware and that the PSU might not be coping. What is the rating of the PSU you're using?

Comment 3 Alessio 2018-03-16 06:07:08 UTC
(In reply to Peter Robinson from comment #2)

> I'm wondering if there's not been a power regression which is causing the
> device to pull too much power while booting the firmware and that the PSU
> might not be coping. What is the rating of the PSU you're using?

I will check ASAP.
However, as said before, using 20180310 compose at least uboot prompt appears. And F27 boots without issues as well. Obviously using the same PSU and SD card on the same RPi.

Comment 4 Peter Robinson 2018-03-16 06:33:08 UTC
> However, as said before, using 20180310 compose at least uboot prompt

Sure, that doesn't mean there's not regressions in power consumption as stated.

> appears. And F27 boots without issues as well. Obviously using the same PSU
> and SD card on the same RPi.

See above. We usually only see rainbow in two circumstances 1) incompatible RPi (zero, original editions) 2) now a powerful enough PSU. I've tested this across 4 different RPis, includingthe 3+, and not managed to recreate it so I'm trying to ascertain why that is.

Comment 5 Alessio 2018-03-16 06:46:38 UTC
(In reply to Peter Robinson from comment #4)
> See above. We usually only see rainbow in two circumstances 1) incompatible
> RPi (zero, original editions) 2) now a powerful enough PSU. I've tested this
> across 4 different RPis, includingthe 3+, and not managed to recreate it so
> I'm trying to ascertain why that is.

Well. I will try again ASAP. Maybe my fault on copying the image on the SD card, maybe I have to try another PSU.
I'm using arm-image-installer on F27 workstation to copy the image to the SD, do you?
Thanks.

Comment 6 Peter Robinson 2018-03-16 06:50:45 UTC
> I'm using arm-image-installer on F27 workstation to copy the image to the
> SD, do you?

I use a-i-i but on F-28, completely irrelevant though

Comment 7 Alessio 2018-03-16 09:40:22 UTC
(In reply to Peter Robinson from comment #2)
> I'm wondering if there's not been a power regression which is causing the
> device to pull too much power while booting the firmware and that the PSU
> might not be coping. What is the rating of the PSU you're using?

The PSU is 5V - 3000ma

Comment 8 Alessio 2018-03-16 11:01:10 UTC
Update.

I've used another microSD. And... it works!

The previous SD, that works well with F27 and 20180310 compose, is a Samsung EVO+ 32GB U1
This "new" one is a SanDisk 16GB C10

Comment 9 Paul Whalen 2018-03-16 15:46:58 UTC
I reproduced this using a Samsung EVO 16G card, other cards seem to work ok(tested S&P, A-Data). The Samsung card works when I use bcm283x-firmware-20180209-1.b1a7f4a.fc28.

Comment 10 Peter Robinson 2018-03-16 15:58:42 UTC
Do you have another similar rated PSU you could try with the same SD card, I suspect there's been some change with the firmware but we'll need to engage with the vendor for this, and to do that we'll need more data, likely test some more firmwares too.

Comment 11 Paul Whalen 2018-03-16 16:09:30 UTC
I tested using the official 2.5a, Official 2a and a Geekworm 3a PSU with no change using the Samsung card.

Comment 12 Peter Robinson 2018-03-17 05:43:46 UTC
Scratch build here that should fix the problem, please test

https://koji.fedoraproject.org/koji/taskinfo?taskID=25756818

you can install the uboot-images-armv7 noarch rpm on x86 and copy the rpi3_32 u-boot binary on the vfat partition

Comment 13 Alessio 2018-03-19 15:52:23 UTC
(In reply to Peter Robinson from comment #12)
> Scratch build here that should fix the problem, please test
> 
> https://koji.fedoraproject.org/koji/taskinfo?taskID=25756818
> 
> you can install the uboot-images-armv7 noarch rpm on x86 and copy the
> rpi3_32 u-boot binary on the vfat partition

It doesn't seem to solve the issue.

However, I mounted the 20180310 compose.
By process of elimination, I copied the files in the vfat partition from such compose, inside the vfat parition of the 32GB SD card (containing the 20180314 compose).
The file that allows the system to move forward in the boot process, leaving all the other files unchanged, seems to be start_cd.elf from 20180310 compose.
Subsequently there is a kernel panic after unpacking the initramfs.

This make sense?

Comment 14 Alessio 2018-03-19 16:41:03 UTC
Well.
I extracted the content of bcm283x-firmware-20180307-1.7fdcd00.fc28 in the vfat partition, and same issue as this bug.

Using instead the content of bcm283x-firmware-20180209-1.b1a7f4a.fc28 I'm able to reach the initial setup!

Comment 15 Paul Whalen 2018-03-19 16:50:52 UTC
Scratch build seems to be working with bcm283x-firmware-20180307-1.7fdcd00.fc28

U-Boot 2018.03 (Mar 17 2018 - 07:58:37 +0000)

DRAM:  998 MiB
RPI 3 Model B (0xa22082)
MMC:   mmc@7e202000: 0, sdhci@7e300000: 1
Loading Environment from FAT... *** Warning - bad CRC, using default environment

Failed (-5)
In:    serial
Out:   vidconsole
Err:   vidconsole
Net:   No ethernet found.
starting USB...
USB0:   Core Release: 2.80a
scanning bus 0 for devices... 4 USB Device(s) found
       scanning usb for storage devices... 1 Storage Device(s) found
Hit any key to stop autoboot:  0

Comment 16 Peter Robinson 2018-03-20 01:18:58 UTC
> However, I mounted the 20180310 compose.
> By process of elimination, I copied the files in the vfat partition from
> such compose, inside the vfat parition of the 32GB SD card (containing the
> 20180314 compose).
> The file that allows the system to move forward in the boot process, leaving
> all the other files unchanged, seems to be start_cd.elf from 20180310
> compose.
> Subsequently there is a kernel panic after unpacking the initramfs.
> 
> This make sense?

No, not really TBH.

Which files did you copy, from where to where. You mentioned you copied from the 20180310 compose to what compose/image.

Comment 17 Fedora Update System 2018-03-20 05:31:47 UTC
arm-trusted-firmware-1.5-0.4.rc3.fc28 bcm283x-firmware-20180316-1.25cf637.fc28 uboot-tools-2018.03-3.fc28 has been submitted as an update to Fedora 28. https://bodhi.fedoraproject.org/updates/FEDORA-2018-1556e768e2

Comment 18 Alessio 2018-03-20 09:28:41 UTC
(In reply to Peter Robinson from comment #16)
> 
> Which files did you copy, from where to where. You mentioned you copied from
> the 20180310 compose to what compose/image.

Never mind.

However now, with the latest builds, it works even on the 32GB SD card, following these steps:

- using compose Fedora-Minimal-armhfp-28-20180319.n.0-sda.raw.xz (that still doesn't work as it is)
- copy the content of ../bcm283x-firmware-20180316-1.25cf637.fc28.armv7hl.rpm to the vfat partition on the SD card
- copy the rpi_3_32b/u-boot.bin file from uboot-images-armv7-2018.03-3.fc28.noarch.rpm to the vfat partition on the SD card, renaming it to rpi3-u-boot.bin


Thanks.

Comment 19 Fedora Update System 2018-03-20 14:54:28 UTC
arm-trusted-firmware-1.5-0.4.rc3.fc28, bcm283x-firmware-20180316-1.25cf637.fc28, uboot-tools-2018.03-3.fc28 has been pushed to the Fedora 28 testing repository. If problems still persist, please make note of it in this bug report.
See https://fedoraproject.org/wiki/QA:Updates_Testing for
instructions on how to install test updates.
You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-2018-1556e768e2

Comment 20 Fedora Blocker Bugs Application 2018-03-21 03:19:15 UTC
Proposed as a Blocker for 28-beta by Fedora user pbrobinson using the blocker tracking app because:

 Not sure if this is a blocker or a FE, but basically there's some SD cards (primarily seems to be Samsung EVO) that fail to boot that previously worked on the Raspberry Pi. An updated uboot/firmware combination from upstream feedback has been confirmed to fix this issue.

Comment 21 Stephen Gallagher 2018-03-21 14:58:32 UTC
I'd call this a clear +1 FE; I don't think it would pass the "last blocker" test, because it's only a subset of cards[*] that are involved. In other words, if this fix from upstream isn't complete, I wouldn't slip *just* for this. But since we have a suspected fix, let's include it.



[*] Of course, it *would* happen to be the one I just ordered with my new RPi3...

Comment 22 Paul Whalen 2018-03-21 15:32:29 UTC
+1 FE

Comment 23 Adam Williamson 2018-03-21 16:11:43 UTC
Agree with Stephen's rationale, clear +1 FE. That's +3, setting accepted.

Comment 24 Adam Williamson 2018-03-21 16:12:06 UTC
I also endorse the wording 'suspected fix'. All fixes are suspicious. :P

Comment 25 Kevin Fenzi 2018-03-21 16:50:47 UTC
Yeah, +1 FE for sure. I also don't think I would call it a blocker as it's limited in scope.

Comment 26 Mohan Boddu 2018-03-21 17:02:14 UTC
+1 FE since its happening only on certain SD cards.

Comment 27 Fedora Update System 2018-03-21 20:31:53 UTC
arm-trusted-firmware-1.5-0.4.rc3.fc28, bcm283x-firmware-20180316-1.25cf637.fc28, uboot-tools-2018.03-3.fc28 has been pushed to the Fedora 28 stable repository. If problems still persist, please make note of it in this bug report.


Note You need to log in before you can comment on or make changes to this bug.