43 replies to this topic

[ims]

logo

Updating

white display

Reading USB

Creating partition - very long time nothing happens 5mins 30 sec

Making filesystem

Update kernel

Update splash

Updating rootfs - here is reading from USB cca 25 secs

Rebooting in X seconds - countdown

 

and again and again ...

[betacentauri]

I need to know which device is currently used for rootfs. Is „cat /proc/cmdline“ really empty?
Please post „dmesg“ output.

Please try to mount /dev/mmcblk0p4. Does the content look like a rootfs?

[ims]

(none) login: root
# cat /proc/cmdline

#

Dmesg is attached in dmesg and dmesg1 (after next reboot) file.

When I got it yesterday, it is long time on Creating partition and in cca 6minuts cycle it is rebooted and all repeated again.

But it seems, /dev/mmcblk0p4 not exist more...

 

And fdisk not displaying mmcblk0 content more (see txt)

Attached Files

•  dmesg.txt   19.2KB   3 downloads
•  fdisk.txt   1.31KB   2 downloads
•  dmesg1.txt   24.84KB   3 downloads

Edited by ims, 23 January 2020 - 08:05.

[betacentauri]

Hi ims,

I fear box is dead:

<3>[   10.729385] mmcblk0: error -110 sending status command, retrying
<3>[   10.737443] mmcblk0: error -110 sending status command, retrying
<3>[   10.745490] mmcblk0: error -110 sending status command, aborting
<3>[   10.751515] end_request: I/O error, dev mmcblk0, sector 0
<3>[   10.756933] Buffer I/O error on device mmcblk0, logical block 0

So flash memory can’t be read.

[betacentauri]

I have no idea how to fix that.
Only maybe flashing a new kernel, if the user has flashed a totally wrong one.

[ims]

 

Only maybe flashing a new kernel, if the user has flashed a totally wrong one.

but how ? With ofgwrite ?

Edited by ims, 23 January 2020 - 09:47.

[ims]

Btw - has vu some as was there in dm, that was possible via serial port set in "board bios" booting from external flash ?

Then there on dm800se was possible use usb flash card instead internal and to have all on usb ( I used it for PLi 6.0 on dm800se).

 

There could be used serial in RJ11 <-> COM, i think.

[betacentauri]

You can also only flash the kernel with ofgwrite:
ofgwrite -k <path to kernel>
(It will only work when flash is available.
Mhmm which is strange as kernel was loaded from the flash, but there were errors in the dmesg output regarding flash memory. So at least flash memory is not totally dead)

But please wait with that for a moment.

Can you do a:
ls -la /dev/root

I still don’t know which device is used as rootfs. Maybe I overlooked something. Can you see it?

I don’t think it’s possible to change the root device in the bios. You could use own kernel with custom cmdline. In the command line you can specify other device like USB stick. That might work, but you shouldn’t update the kernel anymore.

[ims]

# ls -la /dev/root
ls: /dev/root: No such file or directory

 

for me is very strange, that there exist /etc/fstab with line /dev/sda1 as /

but /dev/sda1 is first pluged device, imho.

 

Do you think, that internal mmc is damaged ?

Weird, that it waiting 5minuts on "Creating partitions". Then "Update rootfs" works cca 25seconds and in this time is 100% readed something from usb stick, On finish is displayed "Update complete".

[betacentauri]

Well, in dmesg there are the error messages regarding mmcblk0 which is the flash memory device. That's why I think something is wrong with the flash memory. There are more messages regarding this:

<6>[   20.999730]  mmcblk0: unable to read partition table

<3>[   31.229943] end_request: I/O error, dev mmcblk0, sector 0
<3>[   31.235360] Buffer I/O error on device mmcblk0, logical block 0
<3>[   31.241302] end_request: I/O error, dev mmcblk0, sector 8
<3>[   31.246732] Buffer I/O error on device mmcblk0, logical block 1
<3>[   31.252668] end_request: I/O error, dev mmcblk0, sector 16
<3>[   31.258165] Buffer I/O error on device mmcblk0, logical block 2
<3>[   31.264102] end_request: I/O error, dev mmcblk0, sector 24
<3>[   31.269609] Buffer I/O error on device mmcblk0, logical block 3

It might be that initramfs is running on your box. That might explain why you can connect to the box and why /proc/cmdline is empty.

But this would also tell us that the kernel cannot be read from the flash memory. So it's maybe also no option to only flash a new kernel.

 

Are the devices /dev/mmcblk0* present on the box?

ls -la /dev/mmcblk0*

If yes, what is the output when you try to mount one of the partitions?

mkdir /tmp/m
mount /dev/mmcblk0p4 /tmp/m

Edited by betacentauri, 23 January 2020 - 17:34.

[WanWizard]

What happens if you blindly try to dump the entire device using dd?

[ims]

# ls -la /dev/mmcblk0*
brw-rw----    1 root     root      179,   0 Jan  1 00:05 /dev/mmcblk0
brw-rw----    1 root     root      179,  16 Jan  1 00:00 /dev/mmcblk0boot0
brw-rw----    1 root     root      179,  32 Jan  1 00:00 /dev/mmcblk0boot1
-rw-rw-rw-    1 root     root       4124128 Jan  1 00:05 /dev/mmcblk0p1
-rw-rw-rw-    1 root     root       1036856 Jan  1 00:05 /dev/mmcblk0p3
brw-rw----    1 root     root      179,  48 Jan  1 00:00 /dev/mmcblk0rpmb

 

# mkdir /tmp/m
# mount /dev/mmcblk0p4 /tmp/m
mount: mounting /dev/mmcblk0p4 on /tmp/m failed: No such file or directory

 

# cat /proc/partitions
major minor  #blocks  name

  31        0       2048 mtdblock0
  31        1         64 mtdblock1
  31        2         64 mtdblock2
  31        3         64 mtdblock3
  31        4       4096 mtdblock4
 179        0    3817472 mmcblk0
   8        0    7909376 sda
   8        1    7909348 sda1
 179       48        512 mmcblk0rpmb
 179       32       4096 mmcblk0boot1
 179       16       4096 mmcblk0boot0

# ls /dev | grep mm
mmcblk0
mmcblk0boot0
mmcblk0boot1
mmcblk0p1
mmcblk0p3
mmcblk0rpmb

I think, mmcblk0 is not accessible more I/O error... but mmcblk0p3 and mmcblk0p1 is readable with dd if=/dev/mmcblk0p3 bs=512k count=1
 

[betacentauri]

According to dmesg output the partition table cannot be read. And therefore rootfs partition(mmcblk0p4) is missing. Strange that p1 and p3 is there.

Never saw something like this before. Maybe try to create partition table by yourself. But there are still I/O errors, which can't be fixed by creating new partition table. And I think USB flashing creates new partition table (at least on the duo4k it does).

 

Sorry, at the moment I have no idea how to fix the box.

[ims]

And where is stored actual rootfs now ?

 

Btw - can I try replace p1 and p3 with zeroes, if it will work ?

[ims]

in p1 is new kernel, imho

 

Weird, that into mmc is not possible read/write and into p1 is possible write and p2/p4 not exist.

Edited by ims, 23 January 2020 - 23:41.

[betacentauri]

I’m not sure, but I think a initramfs is running. And so rootfs is in RAM. Or how much size does root have?
df -h

What happens when you boot the box without attached usb stick? Or is sda a hdd?

You want to write zeros into p1 and p3? Why?

[betacentauri]

Weird, that into mmc is not possible read/write and into p1 is possible write and p2/p4 not exist.

Yes, indeed.
So you flashed new kernel? Did something change?

[ims]

# df -h
Filesystem                Size      Used Available Use% Mounted on
rootfs                  105.9M    105.9M         0 100% /
none                    105.9M         0    105.9M   0% /dev
tmpfs                   115.9M     12.0K    115.9M   0% /tmp
shm                     115.9M         0    115.9M   0% /dev/shm
/dev/sda1                 7.5G      2.8G      4.8G  37% /mnt/usb/sda1

sda is usb. Box has not hdd (owner not sent me it with hdd). Without pluged usb is on LCD logo with "Vu+ SOLO 4k" and box is not accessible via telnet.

[ims]

 

Weird, that into mmc is not possible read/write and into p1 is possible write and p2/p4 not exist.

Yes, indeed.
So you flashed new kernel? Did something change?

 

It seems, each run with usb flashed kernel - tried original vu, and pli - size is changed and is equal with used kernel_auto.bin  (tried dd if=/dev/mmcblk0p1 to file on usb and content seems is equal)

Edited by ims, 23 January 2020 - 23:57.

[betacentauri]

If it only “boots” with attached usb stick, then I’m quite sure that a initramfs is running.

Only solution I can think of is to flash own build kernel with custom cmdline which uses rootfs which is on an attached usb stick/hdd.