MeeGo-Harmattan HE

[jnack95]

At this link we can find The Big Package. (http://www.developer.nokia.com/info/sw.nokia.com/id/db230178-aa63-4c73-ba7f-20930da13cad/Nokia_N950_OneClickFlashers.html)

C'mon TMO, get to opening up the rootfs!

Is there any reason that this image cannot be flashed onto an n900? Basically the developer device is the same? Anybody brave enough to try?

[gorgezilla]

i would try it but in about 8 hours as I get back home from work...if there are other brave ones..please feedback us...

[Hurrian]

Is there any reason that this image cannot be flashed onto an n900? Basically the developer device is the same? Anybody brave enough to try?
The file is an executable, flasher's builtin.
I'm yet to ask MAG how he got rootfs.lzo out of it, I'm assuming he used unshar.

[jnack95]

The file is an executable, flasher's builtin.
I'm yet to ask MAG how he got rootfs.lzo out of it, I'm assuming he used unshar.

Instead of extracting the image, I was talking about actually flashing the whole thing onto the N900 as if it were the N950. I mean what would be different? Same processor architecture, same gpu...it would be interesting to see what worked and what doesn't.........I'll probably not try it though unless someone can extract the image and I can flash it to a micro SD. Anybody got an extra N900 that might become a brick (but probably not)?

[gorgezilla]

Instead of extracting the image, I was talking about actually flashing the whole thing onto the N900 as if it were the N950. I mean what would be different? Same processor architecture, same gpu...it would be interesting to see what worked and what doesn't.........I'll probably not try it though unless someone can extract the image and I can flash it to a micro SD. Anybody got an extra N900 that might become a brick (but probably not)?

bet that MAG&others already are trying everything plausible....hope I read news until I get home to try flash it on my N900

[Hurrian]

Instead of extracting the image, I was talking about actually flashing the whole thing onto the N900 as if it were the N950. I mean what would be different? Same processor architecture, same gpu...it would be interesting to see what worked and what doesn't.........I'll probably not try it though unless someone can extract the image and I can flash it to a micro SD. Anybody got an extra N900 that might become a brick (but probably not)?
Well, MAG trying to backup CAL (MTD1) of the N900 failed and resulted in the cellular radio bugging out (fixed though), so I wouldn't try it. Especially that there are NOLO images inside the image. Not gonna flash directly. I'm going to load it on a uSD, and use the MeeGo Kernel to boot it.

[blackjack4it]

If we extract the files in the self-extractor (I've downloaded the Windows version and extracted with Winrar) we can see that there are the following files:

EMMC.bin (I suppose that this file correspond to RX-51_2009SE_10.2010.13-2.VANILLA_PR_EMMC_MR0_ARM.bin of the Maemo package)

flasher.bat
flasher.exe
msi_package_version.exe
Nokia_Connectivity_Cable_Driver.exe

RX-71.bin (I suppose that this file correspond to RX-51_2009SE_3.2010.02-8_PR_COMBINED_MR0_ARM.bin of the Maemo Package)

These are only my thoughts, I'm not sure about it.

[jnack95]

Well, MAG trying to backup CAL (MTD1) of the N900 failed and resulted in the cellular radio bugging out (fixed though), so I wouldn't try it. Especially that there are NOLO images inside the image. Not gonna flash directly. I'm going to load it on a uSD, and use the MeeGo Kernel to boot it.

That is a better route but I would guess that there is little difference between the kernel in this and the n900 adaption kernel. Keep us posted and good luck=O

[BluesLee]

If we extract the files in the self-extractor (I've downloaded the Windows version and extracted with Winrar) we can see that there are the following files:

EMMC.bin (I suppose that this file correspond to RX-51_2009SE_10.2010.13-2.VANILLA_PR_EMMC_MR0_ARM.bin of the Maemo package)

flasher.bat
flasher.exe
msi_package_version.exe
Nokia_Connectivity_Cable_Driver.exe

RX-71.bin (I suppose that this file correspond to RX-51_2009SE_3.2010.02-8_PR_COMBINED_MR0_ARM.bin of the Maemo Package)

These are only my thoughts, I'm not sure about it.

I get an emmc.bin and an image.bin on Linux,
strings on the latter gives me:


MDFL61_HARMATTAN_1.2011.22-6_PR_RM680
vhash
vdata
kernel
Wrootfs
TAtc
xloader
secondary
cert-sw
RM-680
...

[blackjack4it]

I get an emmc.bin and an image.bin on Linux,
strings on the latter gives me:


MDFL61_HARMATTAN_1.2011.22-6_PR_RM680
vhash
vdata
kernel
Wrootfs
TAtc
xloader
secondary
cert-sw
RM-680
...

Very Interesting...especially Wrootfs...
how can I unpack the bin files?

[MartinK]

BTW, how big is the NAND flash in N9/950 anyway ?

[jflatt]

Very Interesting...especially Wrootfs...
how can I unpack the bin files?
flasher -F <image> -u

[mece]

BTW, how big is the NAND flash in N9/950 anyway ?

512Mb, so I'm guessing we'll have to use something else for rootfs.

[blackjack4it]

flasher -F <image> -u

It says "bad command line" and returns the list of arguments :( I've also tried with the flasher3.5.exe that we use for flashing N900 but it give me the same result :(

[attila77]

512Mb, so I'm guessing we'll have to use something else for rootfs.

The whole partition layout is pretty different (no NAND+/opt trick) so...

[jflatt]

It says "bad command line" and returns the list of arguments :( I've also tried with the flasher3.5.exe that we use for flashing N900 but it give me the same result :(You have to specify a directory to unpack to. Run flasher --help for instructruction

[corsac]

As indicated, you need to provide the folder, so:

flasher --unpack=img -F img.bin will unpack it in img/ folder. Then what we get:

total 769M
28K 1st_RM-680:916
28K 1st_RM-680:922
12K 1st_RX-71:000
28K 1st_RX-71:166
28K 2nd.bin_RM-680:0425
28K 2nd.bin_RM-680:0522
28K 2nd.bin_RM-680:0621,0622,0623,0624
28K 2nd.bin_RM-680:0720,0721,0722,0723,0724,0725,0726
28K 2nd.bin_RM-680:0820,0822,0824,0825,0826
28K 2nd.bin_RM-680:0821,0823
28K 2nd.bin_RM-680:0827,0828,0829,0830,0831,0832
28K 2nd.bin_RM-680:0920,0921,0922,0923,0924,0925
28K 2nd.bin_RM-680:1020,1021,1022,1023,1024,1025,1026,1027,1028
28K 2nd.bin_RM-680:1029
28K 2nd.bin_RM-680:1120,1121
28K 2nd.bin_RM-680:1122,1123
28K 2nd.bin_RM-680:1124,1125
28K 2nd.bin_RX-71:3116
6,9M ape-algo
4,0K cert-sw_RM-680:0425
4,0K cert-sw_RM-680:0522
4,0K cert-sw_RM-680:0621,0622,0623,0624
4,0K cert-sw_RM-680:0720,0721,0722,0723,0724,0725,0726
4,0K cert-sw_RM-680:0820,0822,0824,0825,0826
4,0K cert-sw_RM-680:0821,0823
4,0K cert-sw_RM-680:0827,0828,0829,0830,0831,0832
4,0K cert-sw_RM-680:0920,0921,0922,0923,0924,0925
4,0K cert-sw_RM-680:1020,1021,1022,1023,1024,1025,1026,1027,1028
4,0K cert-sw_RM-680:1029
4,0K cert-sw_RM-680:1120,1121
4,0K cert-sw_RM-680:1122,1123
4,0K cert-sw_RM-680:1124,1125
4,0K cert-sw_RX-71:3116
96K cmt-2nd_RM-660+RM-680
96K cmt-2nd_RM-690
96K cmt-2nd_RM-696
96K cmt-2nd_RM-716
96K cmt-2nd_RX-71
792K cmt-algo_RM-660+RM-680
792K cmt-algo_RM-690
792K cmt-algo_RM-696
792K cmt-algo_RM-716
792K cmt-algo_RX-71
5,9M cmt-mcusw_RM-660+RM-680
6,0M cmt-mcusw_RM-690
6,0M cmt-mcusw_RM-696
6,0M cmt-mcusw_RM-716
6,8M cmt-mcusw_RX-71
361M rootfs.lzo
4,0K rootfs.lzo.layout
76K secondary.bin_RM-680:0425
76K secondary.bin_RM-680:0522
96K secondary.bin_RM-680:0621,0622,0623,0624
96K secondary.bin_RM-680:0720,0721,0722,0723,0724,0725,0726
96K secondary.bin_RM-680:0820,0822,0824,0825,0826
96K secondary.bin_RM-680:0821,0823
96K secondary.bin_RM-680:0827,0828,0829,0830,0831,0832
96K secondary.bin_RM-680:0920,0921,0922,0923,0924,0925
96K secondary.bin_RM-680:1020,1021,1022,1023,1024,1025,1026,1027,1028
96K secondary.bin_RM-680:1029
96K secondary.bin_RM-680:1120,1121
96K secondary.bin_RM-680:1122,1123
92K secondary.bin_RM-680:1124,1125
96K secondary.bin_RX-71:3116
24K xloader.bin_RM-680:0425
24K xloader.bin_RM-680:0522
24K xloader.bin_RM-680:0621,0622,0623,0624
24K xloader.bin_RM-680:0720,0721,0722,0723,0724,0725,0726
24K xloader.bin_RM-680:0820,0822,0824,0825,0826
24K xloader.bin_RM-680:0821,0823
24K xloader.bin_RM-680:0827,0828,0829,0830,0831,0832
24K xloader.bin_RM-680:0920,0921,0922,0923,0924,0925
24K xloader.bin_RM-680:1020,1021,1022,1023,1024,1025,1026,1027,1028
24K xloader.bin_RM-680:1029
24K xloader.bin_RM-680:1120,1121
24K xloader.bin_RM-680:1122,1123
24K xloader.bin_RM-680:1124,1125
24K xloader.bin_RX-71:3116
2,7M zImage

rootfs.lzo is not compressed and it might have an ext4 filesystem inside but I didn't yet found how to mount it (trying some offsets didn't give anything).

[jflatt]

As indicated, you need to provide the folder, so:

flasher --unpack=img -F img.bin will unpack it in img/ folder. Then what we get:

rootfs.lzo is not compressed and it might have an ext4 filesystem inside but I didn't yet found how to mount it (trying some offsets didn't give anything).
See this:
http://wiki.maemo.org/Modifying_the_root_image
and
http://talk.maemo.org/showpost.php?p=403015&postcount=3
I haven't had a chance to try yet though

[corsac]

See this:
http://wiki.maemo.org/Modifying_the_root_image
and
http://talk.maemo.org/showpost.php?p=403015&postcount=3
I haven't had a chance to try yet though

It's not a JFFS2 fs but the mtd stuff might work, at least it deserves a try, thanks!

[Acidspunk]

This thread is really interesting to follow. So basically since the N950 uses the same SoC the OS can be ported to work on the N900 (in theory at least)?

[corsac]

It's not a JFFS2 fs but the mtd stuff might work, at least it deserves a try, thanks!

Ok, it doesn't work at first sight. When adding the loop0 to the block2mtd module, I can see it inside the (just created) mtdblock0 node, but it's exactly the same content as the rootfs.lzo file, so I'm not too sure what I gained, and indeed file replies “data” and I can't mount it using ext4.

[jflatt]

I believe the filesystem still needs mounted, I am thinking the nandsim approach is needed, and the pagesize might need to be set to 4096, but these are all guesses for now

[corsac]

I believe the filesystem still needs mounted, I am thinking the nandsim approach is needed, and the pagesize might need to be set to 4096, but these are all guesses for now

I've tried the nandsim approach (dd-ing the rootfs.lzo image to mtdblock) and the block2mtd approach. In both cases I can read the /dev/mtdblock0 file and it shows the exact same content as the rootfs.lzo file, and I just can't mount it.

[Deaconclgi]

I am very interested in this. My N900 is my spare phone so I may be able to help with testing in the future, after July 3rd, when I return back home.

Keep up the great detective work!

[Hurrian]

This thread is really interesting to follow. So basically since the N950 uses the same SoC the OS can be ported to work on the N900 (in theory at least)?
Not exactly the same SOC, the N9/50 uses an OMAP3 3630+SGX530. When Nokia releases the kernel source, we might be able to adapt it for the N900. But right now, if we wanted to boot the image, we'd probably use the MeeGo Kernel.

[lma]

When Nokia releases the kernel source

Here (http://harmattan-dev.nokia.com/pool/harmattan-beta/free/k/kernel/). Does anyone know if Harmattan is built with thumb instructions? The N900's OMAP can't cope with those.

[corsac]

Here (http://harmattan-dev.nokia.com/pool/harmattan-beta/free/k/kernel/). Does anyone know if Harmattan is built with thumb instructions? The N900's OMAP can't cope with those.

I don't have the config file around but iirc thumb2 wasn't enabled, I'll check tomorrow in case nobody does it first.

[corsac]

I don't have the config file around but iirc thumb2 wasn't enabled, I'll check tomorrow in case nobody does it first.

Ok, it was easy enough to check:


$ grep THUMB arch/arm/configs/rm581_defconfig
CONFIG_ARM_THUMB=y
# CONFIG_ARM_THUMBEE is not set
# CONFIG_THUMB2_KERNEL is not set


Hope that helps.

[Hurrian]

Here (http://harmattan-dev.nokia.com/pool/harmattan-beta/free/k/kernel/). Does anyone know if Harmattan is built with thumb instructions? The N900's OMAP can't cope with those.
Features : swp half thumb fastmult vfp edsp thumbee neon vfp3
IIRC, thumb and thumbee are missing from certain hardware revisions of the N900. Not on mine though.

In other N950 related news, it has a C64x DSP! Everything Maemo 6 should work fine, except probably omap34xxcam.

[lma]

Ok, it was easy enough to check:


$ grep THUMB arch/arm/configs/rm581_defconfig
CONFIG_ARM_THUMB=y
# CONFIG_ARM_THUMBEE is not set
# CONFIG_THUMB2_KERNEL is not set


Hope that helps.

Not much, rx51_defconfig had the same. I notice CONFIG_ARM_ERRATA_430973 (http://cateee.net/lkddb/web-lkddb/ARM_ERRATA_430973.html) is not set, but that doesn't necessarily mean anything either.

Features : swp half thumb fastmult vfp edsp thumbee neon vfp3
IIRC, thumb and thumbee are missing from certain hardware revisions of the N900. Not on mine though.


Not missing, just buggy (see above link).

[lma]

Testing a random Harmattan binary (bc (http://harmattan-dev.nokia.com/pool/harmattan-beta/free/b/bc/bc_1.06.95-2+maemo9+0m6_armel.deb)):

$ arm-linux-gnueabi-readelf -A bc
Attribute Section: aeabi
File Attributes
Tag_CPU_name: "7-A"
Tag_CPU_arch: v7
Tag_CPU_arch_profile: Application
Tag_ARM_ISA_use: Yes
Tag_THUMB_ISA_use: Thumb-2
Tag_FP_arch: VFPv3
Tag_Advanced_SIMD_arch: NEONv1
Tag_ABI_PCS_wchar_t: 4
Tag_ABI_FP_denormal: Needed
Tag_ABI_FP_exceptions: Needed
Tag_ABI_FP_number_model: IEEE 754
Tag_ABI_align_needed: 8-byte
Tag_ABI_align_preserved: 8-byte, except leaf SP
Tag_ABI_enum_size: int
Tag_ABI_HardFP_use: SP and DP
Tag_ABI_VFP_args: VFP registers


I'm not sure if the erratum applies also to thumb2.

[corsac]

Testing a random Harmattan binary (bc (http://harmattan-dev.nokia.com/pool/harmattan-beta/free/b/bc/bc_1.06.95-2+maemo9+0m6_armel.deb)):

$ arm-linux-gnueabi-readelf -A bc
Attribute Section: aeabi
File Attributes
Tag_CPU_name: "7-A"
Tag_CPU_arch: v7
Tag_CPU_arch_profile: Application
Tag_ARM_ISA_use: Yes
Tag_THUMB_ISA_use: Thumb-2
Tag_FP_arch: VFPv3
Tag_Advanced_SIMD_arch: NEONv1
Tag_ABI_PCS_wchar_t: 4
Tag_ABI_FP_denormal: Needed
Tag_ABI_FP_exceptions: Needed
Tag_ABI_FP_number_model: IEEE 754
Tag_ABI_align_needed: 8-byte
Tag_ABI_align_preserved: 8-byte, except leaf SP
Tag_ABI_enum_size: int
Tag_ABI_HardFP_use: SP and DP
Tag_ABI_VFP_args: VFP registers


I'm not sure if the erratum applies also to thumb2.

Is that in the sdk or using a real device?

[lma]

Neither, I just manually fetched a .deb from the harmattan repo and extracted it in /tmp. My arm readelf comes from embedian I think, but even plain readelf (this is on an x86_64 Ubuntu 11.04 box) reports the same.

[Hurrian]

Okay, I just verified that the rootfs.lzo file is NOT a UBIFS dump (compare M6-rootfs.lzo with M5-rootfs.ubifs -- you'll see my point.)

Now, it looks like you'll need to do filesystem reconstruction - the EXT4 header (EF53, or when opened in XVI32 53EF) is found after the filesystem label (rootfs apparently). Why Nokia does this is unclear - but they've done numerous ***-backwards things before, this might be one of them.

The good news is that you can directly mount -t vfat the mmc0 image.
Which leads to my next question, how do we mount rootfs.lzo? It "appears" to be compressed, but we can't be certain, as the magic was stripped out. We should be able to directly mount it just like the mmc0 image. But nothing's working. EXT4, nandsim etc. None.

EDIT: rootfs.lzo is a lzo file. However, Nokia has stripped off the header for no reason at all. I've been trying for the past 3 hours to reconstruct the header from other lzo files, with no dice. If anyone can point me to how flasher-3.5.11 handles this thing, that would be much appreciated.

[corsac]

Now, it looks like you'll need to do filesystem reconstruction - the EXT4 header (EF53, or when opened in XVI32 53EF) is found after the filesystem label (rootfs apparently). Why Nokia does this is unclear - but they've done numerous ***-backwards things before, this might be one of them.

I've found the ext4 magic too, but I'm not too sure it's really an ext4 header (but it might be worth trying identifying other fields)


EDIT: rootfs.lzo is a lzo file. However, Nokia has stripped off the header for no reason at all. I've been trying for the past 3 hours to reconstruct the header from other lzo files, with no dice. If anyone can point me to how flasher-3.5.11 handles this thing, that would be much appreciated.

I really don't think it is compressed. Look at the content in hexdump/xxd, you can see some ascii stuff like filenames and file contents, which you wouldn't if it was compressed. Maybe some parts are compressed (the whole rootfs.lzo file is about ~300M while it'll fit on a 4G partition so at least it's sparse) but not everything.

[Hurrian]

rootfs.lzo is likely a compressed filesystem image - BTW you'll see strings even in other lzo-compressed files ;)

[corsac]

rootfs.lzo is likely a compressed filesystem image - BTW you'll see strings even in other lzo-compressed files ;)

Thanks, I didn't know that (it puzzles me a little though). So yeah, maybe reconstructing an lzo header might be the solution.

[Hurrian]

I think we might need someone who knows about how the lzo file is made. I have absolutely no idea what I'm doing here, and am mostly poking in the dark, hoping with crossed fingers that lzop will decompress it.

So far, the results have been:
lzop: rootfs.lzo: header corrupted (checksum error)
lzop: rootfs.lzo: header corrupted (transmitted in text mode?)
lzop: rootfs.lzo: not a lzop file
lzop: rootfs.lzo: header corrupted

[corsac]

Just to add a little bit more of information, ape-algo looks like a tiny Linux system for flashing.

It's a Linux kernel with a tiny initscript running softupd. The flasher sends it to the device, which boots it and then waits for the flasher connection, which then sends the rest of the firmware.

I don't know how exactly ape-algo system is compacted, it might be lzo to, but I can't extract it.

[Hurrian]

Just to add a little bit more of information, ape-algo looks like a tiny Linux system for flashing.

It's a Linux kernel with a tiny initscript running softupd. The flasher sends it to the device, which boots it and then waits for the flasher connection, which then sends the rest of the firmware.

I don't know how exactly ape-algo system is compacted, it might be lzo to, but I can't extract it.
Look out for the LZO headers. Also, it seems that there are 2 LZO files inside -- or it's just a kludge. Anyways, after extracting the second one, you find a romfs filesystem, but it's mangled. This is quickly turning into a very tedious excercise.

[qgil]

Hi, have you considered asking/discussing at #meego-arm, the meego-handset mailing list of forum.meego.com and work closer to the MeeGo Community Edition team?

[corsac]

Hi, have you considered asking/discussing at #meego-arm, the meego-handset mailing list of forum.meego.com and work closer to the MeeGo Community Edition team?

I'm on the #meego-arm channel but I'm not too sure if discussing this on MeeGo forums is a really good idea. I tried discussing Aegis/MSSF on meego-security-discussion list and I was pointed at the fact that MSSF wasn't in MeeGo and that the N9 wasn't running MeeGo at all and that calling it that way would only add more confusion.

Any advice appreciated :)

[qgil]

This is why I'm suggesting you to ask & discuss in the surrounds of the MeeGo CE project.

My main point however is that I keep seeing here this background tone of 'why Nokia has done this', 'how that is supposed to work', etc, when in fact there might be more than a couple of Nokia engineers interested in your work and happy to give you tips to progress further.

[corsac]

This is why I'm suggesting you to ask & discuss in the surrounds of the MeeGo CE project.

My main point however is that I keep seeing here this background tone of 'why Nokia has done this', 'how that is supposed to work', etc, when in fact there might be more than a couple of Nokia engineers interested in your work and happy to give you tips to progress further.

Yeah, to be fair I had very friendly answers to my questions on meego-security-discussionm, both from Nokia and Intel people. But Intel people are concerned about confusion between MeeGo 1.2 and MeeGo 1.2 Harmattan (which, I guess, is fair). And I'm sure there are a lot of friendly people on the other channels, but I just don't want to bother people by asking those questions on the wrong channels.

[blackjack4it]

Yeah, to be fair I had very friendly answers to my questions on meego-security-discussionm, both from Nokia and Intel people. But Intel people are concerned about confusion between MeeGo 1.2 and MeeGo 1.2 Harmattan (which, I guess, is fair). And I'm sure there are a lot of friendly people on the other channels, but I just don't want to bother people by asking those questions on the wrong channels.

Any good news about this porting :D?

[corsac]

Any good news about this porting :D?

Hmhm, sorry to disappoint you but my efforts aren't at all related to porting Harmattan to N900, it's more to examine and analyze the rootfs.

[blackjack4it]

Hmhm, sorry to disappoint you but my efforts aren't at all related to porting Harmattan to N900, it's more to examine and analyze the rootfs.

Sigh & Sob :( don't worry, your "discovery mission" will help the porting :D

Any news about? :D

[Hurrian]

Sigh & Sob :( don't worry, your "discovery mission" will help the porting :D

Any news about? :D
Still nothing on unpacking the image.
If it does get unpacked, I'll try to use it with the MeeGo 2.6.37 kernel.

[lma]

Still nothing on unpacking the image.
If it does get unpacked, I'll try to use it with the MeeGo 2.6.37 kernel.

Harmattan is 2.6.32 (http://harmattan-dev.nokia.com/pool/harmattan-beta/free/k/kernel/), so I suspect that would be about as succesful as running Fremantle userland on that 2.6.37 :-/

[Hurrian]

Harmattan is 2.6.32 (http://harmattan-dev.nokia.com/pool/harmattan-beta/free/k/kernel/), so I suspect that would be about as succesful as running Fremantle userland on that 2.6.37 :-/
That is, if Nokia... does a Nokia and makes things depend on kernel interfaces removed from 2.6.37 - which there are very few that I know of, and mostly relate to Linux's graphics performance (lol)

[lma]

The diff between Harmattan & kernel.org 2.6.32.39 is 4.1MiB/143K lines, draw your own conclusions.

[Hurrian]

The diff between Harmattan & kernel.org 2.6.32.39 is 4.1MiB/143K lines, draw your own conclusions.
What the...
Who needs to patch the kernel that deeply?

[ali1234]

I have successfully decompressed and mounted the rootfs.lzo. Here is the script to uncompress it. This gives you a regular ext4 image. Should work with the other lzo images, but this is not tested.

Also you need python-lzo installed. It isn't packaged for Ubuntu, but the required liblzo2-2 is in the standard repos.

http://al.robotfuzz.com/~al/maemo/unlzo

[corsac]

Thanks, it does work pretty fine :)

[Hurrian]

To get oberhumer's python-lzo to work on Ubuntu's lzo2, you'll need to apply this patch. (https://www.redhat.com/archives/fedora-extras-commits/2008-April/msg01773.html)

Drop the generates lzo.so in the same folder as the python script, and "python unlzo.py rootfs.lzo".

EDIT: Bad news: The N950's ... unique MMC structure is apparently hardcoded into system files.
Even after setting up the correct mountpoints, Harmattan sputtered and failed, and it clobbered Maemo 5's filesystem.
Used the MeeGo Kernel 2.6.37-29.1. Needed a full reflash to restore everything.

Directions: WARNING THIS WILL CLOBBER MTD5 (aka ROOTFS) -- IT MAY OR MAY NOT CAUSE DAMAGE TO YOUR MTD1 (aka config) and MTD2 (syslog?) PARTITIONS.

(1) Run ali1234's unlzo script.
(2) Make sure a SD card is partitioned into a 2GB ext4 (p1), a 2MB linux-swap (p2) and 32MB vfat (p3)
(3) Mount and copy MeeGo CE's uboot boot files to the vfat partition. Save MeeGo's /lib/modules/{KERNEL} folder (via cp-a) somewhere on your desktop. Unmount MeeGo CE's image.
(4) Copy (cp -a) Harmattan's system to the ext4 (p1) partition on the SD card.
(5) In Harmattan's /lib/modules , remove it's kernel and the "current" symlink.
(6) Copy MeeGo CE's modules folder into modules, and ln -s the folder to "current".
(7) Edit /etc/pmconfig (!)
(8) Edit /etc/fstab (mmcblk0 is the SD card when booting from SD, so edit accordingly, and use mmcblk1p3 as the swap partition.)
(9) Unmount everything, connect SD to N900.
(10) flasher-3.5 -l -k vmlinuz-{KERNEL} -b
(11) ??? Watch dmesg
(12) Break Fremantle

[Tigerite]

You might have more luck using the Harmattan kernel (http://harmattan-dev.nokia.com/pool/harmattan-beta/free/k/kernel/) (but probably not..) You'll have to extract the zImage, of course.

[Tigerite]

What the...
Who needs to patch the kernel that deeply?

I'd be interested as to how these figures were calculated - when I ran a diff (minus the Documentation folder, which adds a fair few thousand lines itself, and the debian folder) I got 521,171 lines and a filesize of 15.2MiB..

As for why they patched it so deeply, the changelog (http://pastebin.com/vqnNKziL) provides quite a bit of info (all 4000+ lines of it!)

[jflatt]

Anybody get anything good from harmattan running on the N900? I got nano copied over, runs fine. It actually doesn't segfault when searching like the maemo5 version does, so it's an improvement there. I tried meego-terminal and linked in all its dependent libs, but seems there's too big of a Qt version mismatch.

[corsac]

And note that Harmattan seems to use hardfp so any binary using floating point will surely die because of illegal instruction or something like that.

[Hurrian]

And note that Harmattan seems to use hardfp so any binary using floating point will surely die because of illegal instruction or something like that.
++

Someone should try chrooting and running apps on MeeGo CE for N900.

[[DarkGUNMAN]]

Guide me through it and I'll try it later today.. should be about on #maemo.

[PartyboyXP]

Its possible to port harmattan to our N900?

[Hurrian]

Its possible to port harmattan to our N900?
How much of Harmattan do you want? IIRC, the N900 can't run thumb-2 instructions, which the closed binaries that Harmattan are compiled for. However, you may get some of the UI working with meegotouch.

[blackjack4it]

If we achieve to "run" Meego Harmattan we can have a lot of advantages...
With "run" I mean like an Hackintosh can run iOS, where with every update we have to fix the libraries that hangs the OS to let it run again..
The advantages could be:
1) A supported OS ("unofficially" because we take N9 updates and port them to our N900)
2) A Nokia tested OS (with these words I'm not destroying the great Meego Community Edition, I'm just comparing two views of the same OS)

If we can achieve to run Harmattan edition, we can put our efforts in "mixing" the two editions (Meego CE and Meego Harmattan) in a new OS that is a Hybrid of the two editions.
(I'm dreaming about a Meego Harmattan Edition with all the closed-source libraries replaced by the Meego CE libraries...a Full Open-Source OS...)

This could bring to a very Powerful OS I think ;)...

[mikecomputing]

Its possible to port harmattan to our N900?

n900CE Is more like Harmattan than Maemo5. Means less porting issues in Meego N900CE than Maemo 5.

But if you mean swipeUI its closed part.

But apps written in qt(quick) and components UI yes (as long as they dont accces n9 related HW):


http://www.developer.nokia.com/Community/Blogs/blog/kate-alholas-forum-nokia-blog/2011/07/08/porting-meego-1.2-harmattan-qt-quick-components

[Hurrian]

If we achieve to "run" Meego Harmattan we can have a lot of advantages...
With "run" I mean like an Hackintosh can run iOS, where with every update we have to fix the libraries that hangs the OS to let it run again..
The advantages could be:
1) A supported OS ("unofficially" because we take N9 updates and port them to our N900)
2) A Nokia tested OS (with these words I'm not destroying the great Meego Community Edition, I'm just comparing two views of the same OS)

If we can achieve to run Harmattan edition, we can put our efforts in "mixing" the two editions (Meego CE and Meego Harmattan) in a new OS that is a Hybrid of the two editions.
(I'm dreaming about a Meego Harmattan Edition with all the closed-source libraries replaced by the Meego CE libraries...a Full Open-Source OS...)

This could bring to a very Powerful OS I think ;)...
Yes, it's like Hackintosh. However, think of the N900 being a SSE2 processor, and the N9/50 being a SSE3-capable processor. We might need to hack up some closed-source pieces of Harmattan, should running the binaries on MeeGo CE be insufficient.

Anyways, I was playing with a friend's Nokia 8800 Sapphire Arte ($1000 S40 brick, I hope I was kidding) - and the screen shows you a clock when locked if you double-tap the screen hard. It also goes into silent profile if you put the phone face down. And it's not a touchscreen phone. Somebody definitely needs to sniff the accelerometer data from the N900.

[SpeedEvil]

Somebody definitely needs to sniff the accelerometer data from the N900.

The n900 accelerometer supports double-tap detection in hardware.
http://wiki.maemo.org/N900_accelerometer
This is not implemented properly in the driver.
Though - run some program using the accelerometer, tap the phone, and you can see the appropriate IRQ (341) increment.

[blackjack4it]

Any good news about porting? :( Is there something that we could do to help the porting?

[Radicalz38]

How much of Harmattan do you want? IIRC, the N900 can't run thumb-2 instructions, which the closed binaries that Harmattan are compiled for. However, you may get some of the UI working with meegotouch.

No N900 does have thumb-2 support and can run thumb-2 instructions. It's just currently too buggy.

[Radicalz38]

The OMAP 3430 consists of:

ARM Cortex A8 processor running at 600Mhz: designed in 65-nm CMOS process
Imagination Technologies PowerVR SGX530 GPU: supports OpenGL ES 1.1/2.0, OpenVG 1.1, OpenGL 2.0/3.0, DirectX 9/10.1 and OpenCL (
TMS320C64x DSP/ISP: Digital/Image Signal Processor running at 430 Mhz. A technical overview is available here. A product bulletin is available here
IVA 2+ accelerator: as described on TI web site, a second generation power-optimized version of the imaging video and audio accelerator used in TI's DaVinci technology. Enables multi-standard (MPEG-4, H.264, WMV) encode and decode at DVD up to HD resolutions. IVA 2+ provides H.264-based video at greater than CIF resolutions.

A 6 pages product bulletin is available here if you are interested by additional details. There is also an interesting presentation of Cortex A8 Processor here

Below is what Linux kernel reports for the processor:

oslo:~# cat /proc/cpuinfo
Processor : ARMv7 Processor rev 3 (v7l)
BogoMIPS : 249.96
Features : swp half thumb fastmult vfp edsp neon vfpv3
CPU implementer : 0x41
CPU architecture: 7
CPU variant : 0x1
CPU part : 0xc08
CPU revision : 3

Hardware : Nokia RX-51 board
Revision : 2101
Serial : 0000000000000000

oslo:~# cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_available_frequencies
600000 550000 500000 250000

The BogoMIPS value reported above corresponds to the one matching the current frequency at which the processor is running. Here, 250MHz. Below is the list of frequencies reported by the device:

arno@oslo:~# cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_available_frequencies
600000 550000 500000 250000

Regarding the features reported by the processor:

swp: XXX
half: XXX
thumb: the processor in fact supports thumb-2, which is an extension of thumb
fastmult: XXX
edsp: XXX
neon: supports the NEON instruction set. There are interesting information on the way it can be used by gcc on http://elinux.org/BeagleBoard
vfp: VFP (Vector Floating Point) technology is a coprocessor extension to the ARM architecture. It provides low-cost single-precision and double-precision floating-point computation.
vfpv3: the ARM 3430 supports VFPv3 floating point instruction set, used for single/double precision scalar operations. This is used by gcc for C floating point operations on float and double.



http://processors.wiki.ti.com/index.php/Cortex-A8_Features

[Radicalz38]

Last post sorry for multiple posts. The above info of cpuinfo was done on a PR1.1 ROM. Sadly thumb support was disabled from PR1.2 up.

Regarding fixes maybe this should shed us some light?
http://66.135.57.166/lists/linux-omap/msg45232.html

[jinnn_1989]

is any1 trying to port Harmattan meego to n900? in other words n9's os to n900?

[blackjack4it]

Bump :D! Any good news? How can we help the project?

[PartyboyXP]

is any1 trying to port harmattan meego to n900? In other words n9's os to n900?

:( :( :( :( :( :( :( :(

[blackjack4it]

:( :( :( :( :( :( :( :(
:( :( :( :( :( :( :( :(

[blackjack4it]

Do anyone know why I receive "403 Forbidden" if I try to see
http://harmattan-dev.nokia.com/pool/harmattan-beta/nokia-binaries/
while I can see
http://harmattan-dev.nokia.com/pool/harmattan-beta/free
http://harmattan-dev.nokia.com/pool/harmattan-beta/non-free?

Thanks for response ;)

EDIT: Another theory came in my mind...now that N9 and N950 (only for developers) came out,
Can we "clone" the entire partitions of N9 or N950 (using a port of softwares like BackupManager) to "unlock" new hopes in bringing Harmattan on the N900?

[qgil]

Do anyone know why I receive "403 Forbidden" if I try to see
http://harmattan-dev.nokia.com/pool/harmattan-beta/nokia-binaries/

Developers have access to these Nokia binaries after accepting the EULA through the SDK installation process.

[blackjack4it]

And about my theory :p?

[blackjack4it]

To get oberhumer's python-lzo to work on Ubuntu's lzo2, you'll need to apply this patch. (https://www.redhat.com/archives/fedora-extras-commits/2008-April/msg01773.html)

Drop the generates lzo.so in the same folder as the python script, and "python unlzo.py rootfs.lzo".

EDIT: Bad news: The N950's ... unique MMC structure is apparently hardcoded into system files.
Even after setting up the correct mountpoints, Harmattan sputtered and failed, and it clobbered Maemo 5's filesystem.
Used the MeeGo Kernel 2.6.37-29.1. Needed a full reflash to restore everything.

Directions: WARNING THIS WILL CLOBBER MTD5 (aka ROOTFS) -- IT MAY OR MAY NOT CAUSE DAMAGE TO YOUR MTD1 (aka config) and MTD2 (syslog?) PARTITIONS.

(1) Run ali1234's unlzo script.
(2) Make sure a SD card is partitioned into a 2GB ext4 (p1), a 2MB linux-swap (p2) and 32MB vfat (p3)
(3) Mount and copy MeeGo CE's uboot boot files to the vfat partition. Save MeeGo's /lib/modules/{KERNEL} folder (via cp-a) somewhere on your desktop. Unmount MeeGo CE's image.
(4) Copy (cp -a) Harmattan's system to the ext4 (p1) partition on the SD card.
(5) In Harmattan's /lib/modules , remove it's kernel and the "current" symlink.
(6) Copy MeeGo CE's modules folder into modules, and ln -s the folder to "current".
(7) Edit /etc/pmconfig (!)
(8) Edit /etc/fstab (mmcblk0 is the SD card when booting from SD, so edit accordingly, and use mmcblk1p3 as the swap partition.)
(9) Unmount everything, connect SD to N900.
(10) flasher-3.5 -l -k vmlinuz-{KERNEL} -b
(11) ??? Watch dmesg
(12) Break Fremantle

Did you also try to use this kernel?
http://harmattan-dev.nokia.com/pool/harmattan-beta/free/k/kernel/

[Hurrian]

Did you also try to use this kernel?
http://harmattan-dev.nokia.com/pool/harmattan-beta/free/k/kernel/
The **** would you think that would work?
As someone said, the whole of Harmattan would need to be rebuilt for the OMAP3430. Better to just go and use MeeGo CE as it aims to be a MeeGo-compliant instance.

[blackjack4it]

The **** would you think that would work?
As someone said, the whole of Harmattan would need to be rebuilt for the OMAP3430. Better to just go and use MeeGo CE as it aims to be a MeeGo-compliant instance.

Hey, calm down and follow the Netiquette ;)

Remember that we are here to share ideas, proposals and try to realize them :p I was just asking if you already tried that way (just to try, if you have another N900 ;) )

[blackjack4it]

Uhm...here we are using the N950 HD libraries in our N900...
we had only to copy and paste the right list of files needed...
[HOWTO]Watch 720p videos on n900 (http://forums.internettablettalk.com/showthread.php?t=77695)
1) The N950's firmware is not Meego, is a Maemo but Meego compliant...
2) The 3530 board is just a 3430 board for smaller distribution.

So, we could try to use the Maemo Kernel to boot Harmattan, not Meego's Kernel (both Community Edition and Harmattan Edition) because of this fact or to find differences between Harmattan Kernel and Kernel-Power, not Meego CE Kernel.
Am I right?

Sadly, I don't have another N900 and the knowledge to try that way and I don't know if it could be done :(

[fahadj2003]

"It’s really about ensuring that you have the best possible experience designed for the features on your Nokia N900 device."

biggest bs i've heard since the "osama bin laden" is dead..
he's banned!! not dead!!

[blackjack4it]

"It’s really about ensuring that you have the best possible experience designed for the features on your Nokia N900 device."


So, let's show to Nokia how powerfull we are by bringing our PROMISED sistem on our device :D


"osama bin laden" is dead..he's banned!! not dead!!

LOL

[fahadj2003]

So, let's show to Nokia how powerfull we are by bringing our PROMISED sistem on our device :D



LOL

no seriously.. osama bin laden was banned from this website..
http://talk.maemo.org/member.php?u=55607

and yea.. meego and flash 10.1 is coming to n900.. we all know how that turned out..

[Hurrian]

Uhm...here we are using the N950 HD libraries in our N900...
we had only to copy and paste the right list of files needed...
[HOWTO]Watch 720p videos on n900 (http://forums.internettablettalk.com/showthread.php?t=77695)
1) The N950's firmware is not Meego, is a Maemo but Meego compliant...
2) The 3530 board is just a 3430 board for smaller distribution.

So, we could try to use the Maemo Kernel to boot Harmattan, not Meego's Kernel (both Community Edition and Harmattan Edition) because of this fact or to find differences between Harmattan Kernel and Kernel-Power, not Meego CE Kernel.
Am I right?

Sadly, I don't have another N900 and the knowledge to try that way and I don't know if it could be done :(
I tried this first on the N900, and ended up blowing mtd5. And aegis was complaining.

[patlak]

2) The 3530 board is just a 3430 board for smaller distribution.

It's a 45nm OMAP 3630 SoC; It has a 1GHz clocked CPU and a 200MHz clocked GPU as opposed to 600MHz and 110MHz. Not sure about DSP, but I think it's the same 430MHz c64x as the 3430.

[ursachicatalin17]

Improving N9 Meego with Maemo 5 features.

http://www.ideasproject.com/ideas/11368

[blackjack4it]

Could someone please upload the content of N950 rootfs/emmc to a uploading site? (like megaupload, rapidshare ecc ecc) I have some ideas in my mind but I cannot extract correctly all the files from the bin files :) thank you very much :)

[Hurrian]

Could someone please upload the content of N950 rootfs/emmc to a uploading site? (like megaupload, rapidshare ecc ecc) I have some ideas in my mind but I cannot extract correctly all the files from the bin files :) thank you very much :)
Geez. On Linux, sudo ./oneclickflasher then go to /tmp, grab the files (flasher and the rootfs file). flasher -u the rootfs.bin, then use unlzo.py (which needs python-lzo, downloadable here (http://www.oberhumer.com/opensource/lzo/download/LZO-v1/python-lzo-1.08.tar.gz), and needs a patch for python2 here (http://sources.gentoo.org/cgi-bin/viewvc.cgi/gentoo-x86/dev-python/python-lzo/files/lzo2compat.patch?annotate=1.2)) on rootfs.lzo.

[blackjack4it]

OMG they released the source code of the Harmattan Kernel O_O!

This is the link to the folder:
http://harmattan-dev.nokia.com/pool/harmattan-beta2/free/k/kernel/

and this is the direct link to the Harmattan Kernel Source:
http://harmattan-dev.nokia.com/pool/harmattan-beta2/free/k/kernel/kernel-source_2.6.32-20112910+0m6_all.deb

By exploring the Harmattan Kernel I saw that:
1) a lot of functions are almost the same, something was removed but a lot of "corrections" and libraries were added
2) Inside the libraries I see a lot of references about 3430 so I think that Harmattan Kernel already have the support for 3430..

OMAP 3430 specs
http://focus.ti.com/general/docs/wtbu/wtbuproductcontent.tsp?templateId=6123&navigationId=12643&contentId=14649&DCMP=WTBU&HQS=Other+OT+omap3430
OMAP 3630 specs
http://focus.ti.com/general/docs/wtbu/wtbuproductcontent.tsp?templateId=6123&navigationId=12837&contentId=52605

This sentence is very interesting (founded in 3630 specs):
"Software-compatible with OMAP 2 and OMAP 3 processors"
This could means that our N900, that is in OMAP 3 processors series, could be already compatible for Harmattan...

"Optification on the N900 was needed mainly because rootfs space was so limited (256 MB with about 70 MB free) so another partition with more space was needed. On the N9/N950 rootfs (i.e. /dev/root) is 4 GB and more than 80% (i.e. over 3 GB ) of it is free, which means that on rootfs itself there is more space than on optfs in N900... Shortly speaking, no need for optification anymore.

The /opt/ folder is still there, but it is just a normal folder on rootfs, not a separate file system / partition symlinked to optfs like on N900. Separate opt partition does not exist anymore. The /opt/ folder remains for compatibility reasons, so that apps ported from N900 can still use it, while actually being installed directly on rootfs.

The /home/ folder is separate partition (mmcblk0p3) and on the N9/N950 it is 2 GB.

MyDocs is another partition (mmcblk0p1) and it is about 10 GB. That's in case of 16 GB N9s and the N950. There will also be a 64 GB version of the N9 where the MyDocs folder will probably be around 58 GB.

To recap, there are the following partitions:
- rootfs 4 GB
- home 2 GB
- MyDocs about 10 GB"

So we know that a possibile cause of "it sputtered and failed" could be a failed try by the new kernel to make a new partition dedicated to the rootfs of 4GB while our is limitated to 256 mb..
Another thing that came in my mind about n950 rootfs is that 80% free of 4 gb is 3,2 gb free so the minimum of the rootfs space required (without changes) would be 812 mb or more...

EDIT: As I supposed, after unpacking the rootfs I noticed a total size of 612 mb...less from what I've thought but still more than 256 mb...
1) A crazy idea is try to "cut" the rootfs from any file that could be not necessary (I've seen a folder that contains a free version of Angry Birds for example, that could be cut away) to fit 256 mb of our rootfs but, if we achieve that step, I don't know how to rebuild the "stripped" rootfs
2) Proceed in Hurrian work, but using a modified Harmattan Kernel that point to the MMC
3) More crazy and useless, swap the rootfs partition and the "swap" partition, so we will have Harmattan installed in Maemo-swap (612 mb in 768 mb) but It will be slower and then use rootfs as SWAP (256 ram + 256 rootfs-RAM)

;)

[blackjack4it]

Cleaned up my messages ;)

[qwazix]

could we possibly use the newer kernel along with fremantle? So that it's easier to port new features over instead of backporing them to the old kernel?

[blackjack4it]

could we possibly use the newer kernel along with fremantle? So that it's easier to port new features over instead of backporing them to the old kernel?

Sorry but I don't have enough knowledge to answer your question :(

I've found that comparison, so linuxtesting.org already tried to deeply compare Harmattan and Maemo...

http://linuxtesting.org/compatibility/Maemo_5.0_to_MeeGo_1.2_Harmattan/arm/abi_compat_report.html

[RDK]

Ok so reading this, Meego for the N900 isn't out yet, but may well be on the way? I love the fact people are still working on the N900, makes me all warm and fuzzy inside. I'll end-up using this phone till I die, or till an upgraded version comes out.

[qwazix]

Meego is available for the N900 and is working quite ok. Not the flavour that is on the N9 of course, that's more like Maemo 6. Search for meego ce 1.3 for how to get Meego on your N900

[qwazix]

hey, would it be possible to run harmattan on the N900 in chroot? just thinking... Easy-Harmattan

[pablocrossa]

hey, would it be possible to run harmattan on the N900 in chroot? just thinking... Easy-Harmattan

I gave it a shot about a week ago I think, here is the relevant thread:
http://talk.maemo.org/showpost.php?p=1210166&postcount=57

I would think that the missing aegis stuff is either from /home partition (as I am missing all of it from an N9, help appreciated) and/or missing instructions in the kernel (seem to recall aegis had some roots in the kernel). So what you said about porting newer features of the new kernel to the N900 kernel (aegis & harmattan required only) and obtaining a /home dump could be enough to make this work :) would be pretty cool :D
EDIT: Oh and the problem was not exactly as on the linked pastebin in the linked post, on the linked pastebin it was non-root user trying to apt-get, in my case it was the restok.conf lines around the 30-something in the pastebin :)
EDIT2: I see the Meego 1.3 CE is kernel 2.5.35 while harmattan is 2.6.32... shame, copying modules from the CE into harmattan would be cool, seems it will not work :(
EDIT3: Yes, an aegis folder inside the kernel, that would help explain why I failed :( would be cool if compiled cleanly against N900 kernel headers as a module, might just check... Or I am probably talking bullsh*t :)