Introducing ubiboot N9 (multiboot OS loader) - Page 103 - maemo.org

Garp | # 1021 | 2013-12-29, 11:36 | Report

Originally Posted by juiceme View Post

Well, there are basically 2 kinds of backups you can take off Harmattan.

The first type is the normal backup via the "settings -> backup" menu. This creates backup of your settings, messages, notes, whatever application data you have (or most of it) and the result is a new subdirectoty under MyDocs/.backups/ which contains 7 zipfiles and a metadata file.
This backup does not contain your applications, so to reinstall those you need to either reload from store or save the DEB's when you install applications.

The second type of backup is the type taken with BackupMenu or ubiboot maintanance console, which is a snapshot of the filesystem state at the moment it was taken.
This type saves the state exactly as it is, which is a problem if your state is not correct (like on your N9), as if you restore this kind of backup to your device after a reflash, you will get the device back in the (broken/faulty) state it was before

So, if you have a device in a broken/faulty state, you do not want to use the second type of backup.

Thanks, that's what I thought! How to use maintenance console for Backup, can't see it from README?
OK need to flash back to closed mode or no-preserve BUT that will destroy my 5 partitions plus one extended and reduce it to 3 standard, correct?

EDIT: Reason why can't download from Store in Sailfish OS could that be same error in Harmattan OS as with Microsoft Exchange not working?

Last edited by Garp; 2013-12-29 at 12:10.

peterleinchen | # 1022 | 2013-12-29, 12:11 | Report

I believe you missed this ?

--
what I would try is to fully backup (partitioning wise) my working device, then flash there the rootfs, copy that rootfs (p2) via 'dd' to the other device, test. If not working try once more also with /home (p3). Maybe you will loose partitiong table on working device, but this could be restored easily/manually.
If successful restore the working device.

Last edited by peterleinchen; 2013-12-29 at 12:18.

The Following 4 Users Say Thank You to peterleinchen For This Useful Post:
ck2nb, Garp, hrbani, juiceme

fejd | # 1023 | 2013-12-29, 16:28 | Report

Originally Posted by juiceme View Post

The ubiboot image is very large; about 12M in compressed form, true, that is because it includes the root filesystem of the repair mode in the kernel image.
However I have not ever seen that it would be too big to be flashed to the kernel MTD partition.

Would it be possible that your kernel partition is smaller than on other devices? I believe that even on MTD there is some flash filesystem that removes damaged flash blocks from use, maybe your kernel partition has been reduced in size?

Could be, I suspect the flash is damaged, it kind of makes sense now with the issues I've had previously with lost data and settings. How do I see the size of the kernel MTD partition? df shows me:

~ # df -h
Filesystem Size Used Available Use% Mounted on
/dev/root 3.9G 1.8G 1.9G 49% /
devtmpfs 10.0M 252.0K 9.8M 2% /dev
tmpfs 4.0M 88.0K 3.9M 2% /tmp
tmpfs 512.0K 160.0K 352.0K 31% /var/run
none 10.0M 252.0K 9.8M 2% /dev
tmpfs 64.0M 436.0K 63.6M 1% /dev/shm
/dev/mmcblk0p3 2.0G 1.1G 790.6M 59% /home
/dev/mmcblk0p1 49.8G 14.5G 35.3G 29% /home/user/MyDocs
aegisfs 3.9G 1.8G 1.9G 49% /var/run/applauncherd
aegisfs 3.9G 1.8G 1.9G 49% /home/user/.odnp-fpcd/private
aegisfs 3.9G 1.8G 1.9G 49% /home/user/.odnp/private
aegisfs 3.9G 1.8G 1.9G 49% /home/user/.positioningd/private
aegisfs 3.9G 1.8G 1.9G 49% /home/user/private
aegisfs 3.9G 1.8G 1.9G 49% /var/cache/timed/aegis
aegisfs 3.9G 1.8G 1.9G 49% /home/user/.slpgwd/layer
aegisfs 3.9G 1.8G 1.9G 49% /home/user/.mms/private
aegisfs 3.9G 1.8G 1.9G 49% /home/user/.facebook/private
aegisfs 3.9G 1.8G 1.9G 49% /home/user/.signon/private
aegisfs 3.9G 1.8G 1.9G 49% /etc/ssl/certs
/dev/mapper/signonfs 6.8M 41.0K 6.4M 1% /home/user/.signon/signonfs-mnt

Here's the output from sfdisk:
~ # /sbin/sfdisk -l

Disk /dev/mmcblk0: 1957120 cylinders, 4 heads, 16 sectors/track
Units = cylinders of 32768 bytes, blocks of 1024 bytes, counting from 0

Device Boot Start End #cyls #blocks Id System
/dev/mmcblk0p1 16 1632511 1632496 52239872 c W95 FAT32 (LBA)
end: (c,h,s) expected (1023,3,16) found (1023,63,32)
/dev/mmcblk0p2 1760512 1891583 131072 4194304 83 Linux
/dev/mmcblk0p3 1891584 1957119 65536 2097152 83 Linux
/dev/mmcblk0p4 1632512 1760511 128000 4096000 83 Linux
start: (c,h,s) expected (1023,3,16) found (1023,63,32)
end: (c,h,s) expected (1023,3,16) found (1023,63,32)

Disk /dev/dm-0: 0 cylinders, 0 heads, 0 sectors/track

sfdisk: ERROR: sector 0 does not have an msdos signature
/dev/dm-0: unrecognized partition table type
No partitions found

Should I repartition it? I'd have to say I've never done that but there's a thread on the topic here that looks helpful:
http://talk.maemo.org/showthread.php?t=91914

fejd | # 1024 | 2013-12-30, 21:54 | Report

Originally Posted by juiceme View Post

The ubiboot image is very large; about 12M in compressed form, true, that is because it includes the root filesystem of the repair mode in the kernel image.
However I have not ever seen that it would be too big to be flashed to the kernel MTD partition.

Would it be possible that your kernel partition is smaller than on other devices? I believe that even on MTD there is some flash filesystem that removes damaged flash blocks from use, maybe your kernel partition has been reduced in size?

How do I verify the size of the kernel partition? I've run sfdisk and it shows me:

Device Boot Start End #cyls #blocks Id System
/dev/mmcblk0p1 16 1632511 1632496 52239872 c W95 FAT32 (LBA)
end: (c,h,s) expected (1023,3,16) found (1023,63,32)
/dev/mmcblk0p2 1760512 1891583 131072 4194304 83 Linux
/dev/mmcblk0p3 1891584 1957119 65536 2097152 83 Linux
/dev/mmcblk0p4 1632512 1760511 128000 4096000 83 Linux
start: (c,h,s) expected (1023,3,16) found (1023,63,32)
end: (c,h,s) expected (1023,3,16) found (1023,63,32)

This is right after flashing MOSLO and trying to flash ubiboot. If the flash is damaged, would it help to run a disk check and mark the sectors, then flash again?

p.s. Is there any way to format the output to make it more readable? d.s.

Last edited by fejd; 2013-12-30 at 21:57.

juiceme | # 1025 | 2013-12-30, 22:20 | Report

Originally Posted by fejd View Post

How do I verify the size of the kernel partition? I've run sfdisk and it shows me:

Device Boot Start End #cyls #blocks Id System
/dev/mmcblk0p1 16 1632511 1632496 52239872 c W95 FAT32 (LBA)
end: (c,h,s) expected (1023,3,16) found (1023,63,32)
/dev/mmcblk0p2 1760512 1891583 131072 4194304 83 Linux
/dev/mmcblk0p3 1891584 1957119 65536 2097152 83 Linux
/dev/mmcblk0p4 1632512 1760511 128000 4096000 83 Linux
start: (c,h,s) expected (1023,3,16) found (1023,63,32)
end: (c,h,s) expected (1023,3,16) found (1023,63,32)

This is right after flashing MOSLO and trying to flash ubiboot. If the flash is damaged, would it help to run a disk check and mark the sectors, then flash again?

p.s. Is there any way to format the output to make it more readable? d.s.

Kernel is not flashed on your mmcblk0 flash, there's a differnt flash filesystem on the device that consists of the parts of device which hold for example the factory calibration data.

The location for kernel is the MTD2 partition of the system NAND.
You can get some information with mtdinfo command, for example on my device it says:

~ # /usr/sbin/mtdinfo -m 2
mtd2
Name: kernel
Type: nand
Eraseblock size: 262144 bytes, 256.0 KiB
Amount of eraseblocks: 64 (16777216 bytes, 16.0 MiB)
Minimum input/output unit size: 4096 bytes
Sub-page size: 1024 bytes
OOB size: 128 bytes
Character device major/minor: 90:4
Bad blocks are allowed: true
Device is writable: true

peterleinchen | # 1026 | 2013-12-30, 22:37 | Report

Looks same on my device.
Had only info from
cat /proc/mtd
which did not reveal size of blocks.
But neither commands will tell us if / how many blocks are 'burnt'?

[OT]sfdisk -d /dev/mmcblk0
will show output more readable[/OT]

fejd | # 1027 | 2013-12-30, 22:46 | Report

Originally Posted by juiceme View Post

Kernel is not flashed on your mmcblk0 flash, there's a differnt flash filesystem on the device that consists of the parts of device which hold for example the factory calibration data.

The location for kernel is the MTD2 partition of the system NAND.
You can get some information with mtdinfo command, for example on my device it says:

~ # /usr/sbin/mtdinfo -m 2
mtd2
Name: kernel
Type: nand
Eraseblock size: 262144 bytes, 256.0 KiB
Amount of eraseblocks: 64 (16777216 bytes, 16.0 MiB)
Minimum input/output unit size: 4096 bytes
Sub-page size: 1024 bytes
OOB size: 128 bytes
Character device major/minor: 90:4
Bad blocks are allowed: true
Device is writable: true

mtdinfo for partition 2 shows:

~ # /usr/sbin/mtdinfo -m 2
mtd2
Name: kernel
Type: nand
Eraseblock size: 131072 bytes, 128.0 KiB
Amount of eraseblocks: 64 (8388608 bytes, 8.0 MiB)
Minimum input/output unit size: 2048 bytes
Sub-page size: 512 bytes
OOB size: 64 bytes
Character device major/minor: 90:4
Bad blocks are allowed: true
Device is writable: true

mtd_debug gives me:

~ # /usr/sbin/mtd_debug info /dev/mtd2
mtd.type = MTD_NANDFLASH
mtd.flags = MTD_CAP_NANDFLASH
mtd.size = 8388608 (8M)
mtd.erasesize = 131072 (128K)
mtd.writesize = 2048 (2K)
mtd.oobsize = 64
regions = 0

So it looks like the size is set to 8M. Is this a factory setting or is it set during flashing? Could it be a result of damaged sectors? I've flashed two different variants with the same result:
059H2M6_RM-696 NDT 64GB ROW BLACK
059L7N6_N9 RM-696 Country Variant Sweden SE Black 64GB

If I print the MTD info for Moslo, its size is set to 16M:

~ # /usr/sbin/mtd_debug info /dev/mtd5
mtd.type = MTD_NANDFLASH
mtd.flags = MTD_CAP_NANDFLASH
mtd.size = 16777216 (16M)
mtd.erasesize = 131072 (128K)
mtd.writesize = 2048 (2K)
mtd.oobsize = 64
regions = 0

peterleinchen | # 1028 | 2013-12-30, 22:54 | Report

Hmm, I have exactly double sized mtd2 and mtd5.
So does not look like erase block problems.
Maybe HW revision?
What is yours? (lshal | grep hardware)

Last edited by peterleinchen; 2013-12-30 at 23:09.

taixzo | # 1029 | 2013-12-30, 23:04 | Report

Originally Posted by peterleinchen View Post

@taixzo
Here you go ...

Please have in mind I have working harmattan on p5, so you need to set ot to where ever your harmattan rootfs is. And I have my nit and firefox also on p5, so hopefully this will help and not confuse more.
Please omit all those echo changes

I tried to apply the correct changes, but Nitdroid still does not boot. The last entries in the Ubiboot log say it correctly loaded the kernel; where would things be logged after that?

fejd | # 1030 | 2013-12-30, 23:08 | Report

Originally Posted by peterleinchen View Post

Hmm, I have exactly double sized mtd2 and mtd5.
So does not look like erase block problems.
Maybe HW revision?
What is yours?

When I run flasher, the hardware revision shows 1501:

Device identifier: 357923041815376 (SN: N/A)
Found device RM-696, hardware revision 1501