Compiling the Linux Kernel for the Kindle - Page 5

knc1 · 05-18-2014, 08:47 AM

Quote:

Originally Posted by Jackie Im

hi~
I bought kindle paper white 2nd.
I would like to compile this model's kernel.
I typed "make mario_mx_defconfig" command in kernel folder.
but error was happened.
error was "No rule to make target 'mario_mx_defconfig'. Stop."
which folder do i type "make mario_mx_defconfig " in?
how can I decide 'mario_mx_defconfig'?

Start us off at the start of what you are doing -

Which kernel source are you using?
Where did you get it (with a link please)?

newman · 08-29-2015, 06:37 AM

Budalb (or anyone else) -

Why do you think KF8 (MOBI8) rendering is dependent on the 2.6.26 kernel (i.e. the kernel from K3's 3.4 FW)?

Three years ago you wrote:

Quote:

Originally Posted by budalb

But I think it was mentioned that Yifan's approach does not update the linux kernel. The Firmware 3.4 requires kernel updates (eink_fb) to have MOBI8 support. And it was mentioned that the reason of not updating the kernel is the kernel used in Kindle 3 does not backward support Kindle 2(including DX and DXG). Is it possible to recompile Kindle 3 kernel with selecting Kindle 2 CPU support?

And similarly here:

Quote:

Originally Posted by budalb

2) The KF8 issue is more complicated. htmlreader is used to process mobi8 files. htmlreader sends requests to a process "webreader". But the "webreader" crashes, probably due to incompatible rink_fb modules. The 3.4 was built on top of different kernel from DXG, so all the kernel module updates are not effective in DXG.

Attached is the crash messages.

I recently upgraded DXi to 3.4.2 for the purpose of enabling of TLS in Experimental Browser. Having now 3.4.x FW I was wondering if KF8/MOBI8 files could be enabled too.

I was wondering why do you think that KF8 support relies on K3's kernel?

I read your web reader crash log and I couldn't find there any reference to kernel...

Also I did a diff of 3.2.1 to 3.4.2 of eInk kernel code (http://pastebin.com/36HBdpfg) and I doubt there's anything related to KF8.

This is what I get when I attempt to open a KF8 book:

Spoiler:

Code:

150828:095707 cvm[3444]: I BookletManager:SwitchingBooklets:from=Home,to=Bookworm:
150828:095707 cvm[3444]: I Reader:BOOK INFO:book asin=unknown,file size=9609280,file last mod date=2015-08-24 16.27.58 +0200,content type=ebook,length=HTMLPosition_ 366299,access=2015-08-24 16.33.46 +0200,last read position=HTMLPosition_ 0,isEncrypted=false,isSample=false,isNew=false,isTTSMetdataPresent=false,isTTSMetadataAllowed=true,fileExtn=mobi:
150828:095707 cvm[3444]: I lipc:gip:prop=userstoreFreeSpace, source=com.lab126.volumd:Get int property
150828:095707 cvm[3444]: E HTMLBook:Error::Failed to get Start Reading Location from Metadata
150828:095707 cvm[3444]: I Reader:Information::Reader.setUpGUIForBook()
150828:095707 cvm[3444]: I SidecarPageNumberProvider:Information::Page number sidecar unavailable
150828:095707 cvm[3444]: E HTMLRenderer:HTML_WEBREADER_RESPONSE_ERR:command=/command/open_uri:Connection refused
150828:095707 cvm[3444]: E HTMLRenderer:HTTP_CONNECTION_FATAL:reason=processCommand_ Failed command:failied to get response
150828:095707 cvm[3444]: E HTMLBookViewer:OPEN BOOK FAILURE:identifier=1918561039,exception=processCommand_ Failed command:
150828:095707 cvm[3444]: I ReaderGUI:SWITCH VIEW:name=com.amazon.ebook.booklet.reader.gui.view.pagecontentview.qb:
150828:095708 cvm[3444]: I lipc:gip:prop=userstoreFreeSpace, source=com.lab126.volumd:Get int property
150828:095708 cvm[3444]: W SystemBarImpl:SPIN:spinCount=0:count <=0 when stopSpinner() called
150828:095708 cvm[3444]: I BookletManager:SwitchingBooklets:from=Bookworm,to=Home:
150828:095708 cvm[3444]: W SystemBarImpl:SPIN:spinCount=0:count <=0 when stopSpinner() called
150828:095708 cvm[3444]: W SystemBarImpl:SPIN:spinCount=0:count <=0 when stopSpinner() called
150828:095708 cvm[3444]: I HTMLBookViewer:Information::HTMLBookViewer closeBook save last read position 0
150828:095709 cvm[3444]: E HTMLRenderer:HTTP_CONNECTION_FATAL:reason=Connection refused:failied to get response
150828:095709 cvm[3444]: E HTMLBookViewer:CLOSE_BOOK_FAILED:message=Connection refused:
150828:095709 cvm[3444]: java.io.IOException: Connection refused
150828:095709 cvm[3444]: 	at com.amazon.ebook.booklet.htmlreader.impl.m.IeC(whe:336)
150828:095709 cvm[3444]: 	at com.amazon.ebook.booklet.htmlreader.impl.lb.Lo(ume:1705)
150828:095709 cvm[3444]: 	at com.amazon.ebook.booklet.reader.vc.ZYc(aj:4087)
150828:095709 cvm[3444]: 	at com.amazon.ebook.booklet.reader.vc.Jp(aj:4168)
150828:095709 cvm[3444]: 	at com.amazon.ebook.booklet.reader.vc.ryc(aj:711)
150828:095709 cvm[3444]: 	at com.amazon.ebook.booklet.reader.vc.haC(aj:1509)
150828:095709 cvm[3444]: 	at com.amazon.ebook.booklet.reader.vc.stop(aj:1000)
150828:095709 cvm[3444]: 	at com.amazon.ebook.framework.impl.j.uGB(sn:199)
150828:095709 cvm[3444]: 	at com.amazon.ebook.framework.impl.j.jhB(sn:392)
150828:095709 cvm[3444]: 	at com.amazon.ebook.framework.impl.o.iEB(sbd:23)
150828:095709 cvm[3444]: 	at com.amazon.ebook.framework.impl.b.run(sbd:160)
150828:095709 cvm[3444]: 	at java.awt.event.InvocationEvent.dispatch(InvocationEvent.java:152)
150828:095709 cvm[3444]: 	at java.awt.EventQueue.dispatchEvent(EventQueue.java:371)
150828:095709 cvm[3444]: 	at java.awt.EventQueueProxy.dispatchEvent(EventDispatchThread.java:353)
150828:095709 cvm[3444]: 	at java.awt.EventDispatchThread.pumpOneEvent(EventDispatchThread.java:164)
150828:095709 cvm[3444]: 	at java.awt.EventDispatchThread.pumpEvents(EventDispatchThread.java:154)
150828:095709 cvm[3444]: 	at java.awt.EventDispatchThread.pumpEvents(EventDispatchThread.java:149)
150828:095709 cvm[3444]: 	at java.awt.EventDispatchThread.run(EventDispatchThread.java:134)
150828:095709 cvm[3444]: 	at java.lang.Thread.startup(Thread.java:1065)
150828:095709 cvm[3444]: 
150828:095709 cvm[3444]: W HTMLRendererMonitor:Warning::WebReader restart needed - restarting now
150828:095709 cvm[3444]: W SystemBarImpl:SPIN:spinCount=0:count <=0 when stopSpinner() called
150828:095709 cvm[3444]: E Reader:Error::
150828:095709 cvm[3444]: com.amazon.ebook.booklet.reader.sdk.BookException: gui.alert.invalid_book
150828:095709 cvm[3444]: 	at com.amazon.ebook.booklet.htmlreader.impl.lb.pgC(ume:1680)
150828:095709 cvm[3444]: 	at com.amazon.ebook.booklet.htmlreader.impl.hb.run(qs:81)
150828:095709 cvm[3444]: 	at com.amazon.ebook.booklet.htmlreader.impl.fb.run(qs:366)
150828:095709 cvm[3444]: 	at java.lang.Thread.startup(Thread.java:1065)
150828:095709 cvm[3444]: 
150828:095709 cvm[3444]: I AlertImpl:PostingAlert:msg=:Posting an Alert
150828:095709 cvm[3444]: I MainView:Information::The device has WAN module so skip popping up the WiFi dialog.
150828:095709 cvm[3444]: I HTMLJsonSidecarFile:Information::Delayed Write to sidecar file

Thanks!

knc1 · 08-29-2015, 08:02 AM

Because it couldn't contact Amazon with only 3G available?

But that doesn't make much sense - it should work in Airplane mode.

- - - -

Check the dates on those posts - because the yifan.lu procedure idoit was following **did** replace the kernel (both copies).

knc1 · 08-29-2015, 09:27 AM

Quote:

Originally Posted by knc1

Because it couldn't contact Amazon with only 3G available?

But that doesn't make much sense - it should work in Airplane mode.

- - - -

Check the dates on those posts - because the yifan.lu procedure idoit was following **did** replace the kernel (both copies).

In fact, his old blog post mentions that **both** kernels must be replaced.

So something quoted above must be out-of-date.

newman · 08-29-2015, 10:09 AM

Quote:

Originally Posted by knc1

Check the dates on those posts - because the yifan.lu procedure idoit was following **did** replace the kernel (both copies).

Idiot? You mean Budalb?

Quote:

Originally Posted by knc1

In fact, his old blog post mentions that **both** kernels must be replaced.

So something quoted above must be out-of-date.

I couldn't find that old blog post, would you post a link for me, please?

knc1 · 08-29-2015, 11:24 AM

Next to last line, first post of this thread.

Also:
http://igorsk.blogspot.com/2007/12/h...oader-and.html

Which is what yifan.lu was working from.

= = = =

I mean this person:
https://www.mobileread.com/forums/member.php?u=150585

Who has replaced his kernels (but I don't think he has replaced his modules).
In fact, he has chickened out (DXG too expensive, too easy to brick) and ordered a K3-3G to experiment with. (I don't blame him, my DXG devices are still factory stock for the same reason).

thatworkshop · 08-29-2015, 12:04 PM

Quote:

Originally Posted by newman

Idiot? You mean Budalb?

I couldn't find that old blog post, would you post a link for me, please?

Both kernels are replaced in step 1 when kernel-flash.sh is hit.

Code:

flash_kernel()

Oh come on! I'm knot an 'idiot'... I am knot, I am knot

newman · 08-29-2015, 05:57 PM

Quote:

Originally Posted by idoit

Both kernels are replaced in step 1 when kernel-flash.sh is hit.

Code:

flash_kernel()

Oh come on! I'm knot an 'idiot'... I am knot, I am knot :D

Oh, I am sorry! And I was wondering why such a strange name.. :) It was just idiocy of mine after all :).

Quote:

Originally Posted by knc1

In fact, his old blog post mentions that **both** kernels must be replaced.

Right, both kernels need to be flashed. It's actually in /test/flashtools/ as well as in official kernel update scripts done this way. (See my next post.)

Quote:

Originally Posted by knc1

So something quoted above must be out-of-date.

What you mean? Did I quote someone claiming that flash of regular kernel on 2/DX with 3.x is enough? (I am sorry for my ignorance.)

What I am really wondering is figuring out if 3.4 kernel is needed for KF8 support. From what I've seen so far I doubt it: 3.4 FW was released at the time when KF8 standard was released, so it probably had something to do with KF8 but looking at all 3.3→3.4 kernel changes from GPL release (http://pastebin.com/DNg9DA4v) I see no reason to thing the support went into kernel. If KF8 got added to K3 in 3.4 FW it got enabled via the proprietary Java apps, which perhaps somehow does not communicate well on frankenfirmware (perhaps similar issue to the one with Active Content/Kindles?).

knc1 · 08-29-2015, 06:07 PM

There was a problem with Active Documents on DX(G) devices that had been flashed to 3.2.1 -
There is a note in some readme that I read today, along with a link to fix.

Some of these relationships are stored in "database" (or maybe better: "data files") on the Kindle.

Since the hardware will still identify itself as a 2.something, just running 3.firmware is not always enough, you have to tweak the data entry so your 3.firmware knows its ok to run on a 2.something device.

or something like that.

I seem to recall that yifan.lu did make some of those data updates, but maybe he missed something related to the kf8.

= = = =

I was referring to the **content** of some of the quotes as being now out-of-date.

newman · 08-29-2015, 06:08 PM

Am I right that

2/DX uses MTD for U-Boot (regular and backup), kernel (regular and backup), ... but
K3 uses just the first sectors of MMC (moviNAND) for one bootloader and one kernel
so 2/DX with 3.x FW has to use some smart combination of both (e.g. booting from MTD but letting official update scripts to flash the 3.x kernel to a place where it can't anything bad, i.e. beginning of MMC)?

(See the rest of the post why I think so.)

FW 3.4 [moviNAND]

Luigi-based moviNAND device partition scheme:

Code:

MOVINAND_DEVICE=/dev/mmcblk0
MOVINAND_OFFSET_PARTITION=0     #0x00000000 Partition Map
MOVINAND_OFFSET_HEADER=1024     #0x00000400 DCD Header
MOVINAND_OFFSET_U_BOOT=3072     #0x00000C00 Boot Loader                       
MOVINAND_OFFSET_KERNEL=266240   #0x00041000 Linux Kernel
MOVINAND_OFFSET_ISISWF=3674112  #0x00381000 ISIS Waveform
MOVINAND_OFFSET_ROOTFS=3936256  #0x003C1000 Root filesystem

Code:

BLOCK_SIZE=1024
KERNEL_OFFSET=266,240
KERNEL_SEEK=$(( ${KERNEL_OFFSET} / ${BLOCK_SIZE} )) = 260 blocks
$ dd if=${KERNEL_FILE} of=/dev/mmcblk0 seek=${KERNEL_SEEK} bs=${BLOCK_SIZE}
`--------------------------------------^^^^--skip KERNEL_SEEK blocks (260 blocks * 512 B = 133,120 B)

3.{3,4} uImage kernel size = 2,1 MB (64 B header)

Looking at mine mmcblk0, there's no U-Boot, just an empty place to be filled eventually (see the partition scheme)

FW 2.5.8 [MTD]

what I have on DXi in /dev/mtdblock{1,4} is a YifanaLu's kernel (verified by dd-ing it out by md5sum)
in /dev/mtblock{0,3} is U-Boot from 2.5.8 (i.e. the original one) (verified by dd-ing it out by md5sum)

/proc/mtd:

Code:

dev:    size   erasesize  name
mtd0: 00020000 00010000 "Bootloader1"
mtd1: 00380000 00010000 "Kernel1"
mtd2: 00008000 00002000 "BoardId"
mtd3: 00020000 00010000 "Bootloader2"
mtd4: 00380000 00010000 "Kernel2"
mtd5: 00010000 00010000 "BootEnv"
mtd6: 00020000 00010000 "Diags"
mtd7: 00006000 00002000 "Vectors"
mtd8: 00010000 00010000 "unused0"

MTD needs to be unlocked before erase→flash

Code:

flash_unlock /dev/mtd/0..1..3..4

From 2.5.8 FW update; how to flash U-Boot and kernel, both regular and backup ones:

size of u-boot.bin = 131,072 B

Code:

$ dd if=imx31_mario/u-boot.bin of=/dev/mtdblock/0 bs=131072
$ dd if=imx31_mario/u-boot.bin of=/dev/mtdblock/3 bs=131072

size of uImage = 1,875,952 B

Code:

$ dd if=uImage of=/dev/mtdblock/1 bs=131072
$ dd if=uImage of=/dev/mtdblock/4 bs=131072

newman · 08-29-2015, 06:31 PM

Quote:

Originally Posted by knc1

There was a problem with Active Documents on DX(G) devices that had been flashed to 3.2.1 -
There is a note in some readme that I read today, along with a link to fix.

Right, I fixed it with https://www.mobileread.com/forums/sho...31#post1747631 (needs some changes which are post #354 and some subsequent I can't find right now - it was just a zip with a shell updated script).

Quote:

Originally Posted by knc1

Some of these relationships are stored in "database" (or maybe better: "data files") on the Kindle.

Since the hardware will still identify itself as a 2.something, just running 3.firmware is not always enough, you have to tweak the data entry so your 3.firmware knows its ok to run on a 2.something device.

or something like that.

I seem to recall that yifan.lu did make some of those data updates, but maybe he missed something related to the kf8.

Ah-ha! Any hint where those "data files" may be placed?

NiLuJe · 08-29-2015, 07:11 PM

@newman: I may be mixing my timelines here, but wasn't KF8 support introduced in FW 3.3?

EDIT: Yuuuup, brainfart.

knc1 · 08-29-2015, 07:41 PM

Quote:

Originally Posted by newman

- - - -
Looking at mine mmcblk0, there's no U-Boot, just an empty place to be filled eventually (see the partition scheme)
- - - -

Looking "at mine" what?
If your looking at your K3 - then you can't see the bootloader storage (its in 'hidden' storage of the eMMC device).

Check the teardown's that are indexed, at least one of them gives the part number, then web search the datasheet - - it tells all.

= = = =

If all of this is about being able to read certain book formats with the native reader - - -
I would just forget about that, there is a keyboard fork of KOReader that will handle your book display needs.
I have forgotten the name - idoit can tell you - he uses it.

newman · 08-29-2015, 07:43 PM

Quote:

Originally Posted by NiLuJe

@newman: I may be mixing my timelines here, but wasn't KF8 support introduced in FW 3.3?

EDIT: Yuuuup, brainfart.

It's 3.4 (https://www.mobileread.com/forums/sho...d.php?t=190723) not that I was able to find it on amazon.com...

thatworkshop · 08-29-2015, 07:56 PM

Quote:

Originally Posted by NiLuJe

@newman: I may be mixing my timelines here, but wasn't KF8 support introduced in FW 3.3?

EDIT: Yuuuup, brainfart.

Well, v3.3 introduced viewing and retrieving archived items.
A bit mixing but definitely up to a point of brainfart, yet

it takes lots of willpower to produce one.

08-29-2015, 11:24 AM	#66
knc1 Going Viral Posts: 17,212 Karma: 18210809 Join Date: Feb 2012 Location: Central Texas Device: No K1, PW2, KV, KOA	Next to last line, first post of this thread. Also: http://igorsk.blogspot.com/2007/12/h...oader-and.html Which is what yifan.lu was working from. = = = = I mean this person: https://www.mobileread.com/forums/member.php?u=150585 Who has replaced his kernels (but I don't think he has replaced his modules). In fact, he has chickened out (DXG too expensive, too easy to brick) and ordered a K3-3G to experiment with. (I don't blame him, my DXG devices are still factory stock for the same reason). Last edited by knc1; 08-29-2015 at 11:28 AM.

08-29-2015, 06:08 PM	#70
newman homo viator Posts: 78 Karma: 252594 Join Date: May 2010 Device: Kindle DXi with 3.4.2 firmware, Kobo Glo HD	Am I right that 2/DX uses MTD for U-Boot (regular and backup), kernel (regular and backup), ... but K3 uses just the first sectors of MMC (moviNAND) for one bootloader and one kernel so 2/DX with 3.x FW has to use some smart combination of both (e.g. booting from MTD but letting official update scripts to flash the 3.x kernel to a place where it can't anything bad, i.e. beginning of MMC)? (See the rest of the post why I think so.) FW 3.4 [moviNAND] Luigi-based moviNAND device partition scheme: Code: MOVINAND_DEVICE=/dev/mmcblk0 MOVINAND_OFFSET_PARTITION=0 #0x00000000 Partition Map MOVINAND_OFFSET_HEADER=1024 #0x00000400 DCD Header MOVINAND_OFFSET_U_BOOT=3072 #0x00000C00 Boot Loader MOVINAND_OFFSET_KERNEL=266240 #0x00041000 Linux Kernel MOVINAND_OFFSET_ISISWF=3674112 #0x00381000 ISIS Waveform MOVINAND_OFFSET_ROOTFS=3936256 #0x003C1000 Root filesystem Code: BLOCK_SIZE=1024 KERNEL_OFFSET=266,240 KERNEL_SEEK=$(( ${KERNEL_OFFSET} / ${BLOCK_SIZE} )) = 260 blocks $ dd if=${KERNEL_FILE} of=/dev/mmcblk0 seek=${KERNEL_SEEK} bs=${BLOCK_SIZE} `--------------------------------------^^^^--skip KERNEL_SEEK blocks (260 blocks * 512 B = 133,120 B) 3.{3,4} uImage kernel size = 2,1 MB (64 B header) Looking at mine mmcblk0, there's no U-Boot, just an empty place to be filled eventually (see the partition scheme) FW 2.5.8 [MTD] what I have on DXi in /dev/mtdblock{1,4} is a YifanaLu's kernel (verified by dd-ing it out by md5sum) in /dev/mtblock{0,3} is U-Boot from 2.5.8 (i.e. the original one) (verified by dd-ing it out by md5sum) /proc/mtd: Code: dev: size erasesize name mtd0: 00020000 00010000 "Bootloader1" mtd1: 00380000 00010000 "Kernel1" mtd2: 00008000 00002000 "BoardId" mtd3: 00020000 00010000 "Bootloader2" mtd4: 00380000 00010000 "Kernel2" mtd5: 00010000 00010000 "BootEnv" mtd6: 00020000 00010000 "Diags" mtd7: 00006000 00002000 "Vectors" mtd8: 00010000 00010000 "unused0" MTD needs to be unlocked before erase→flash Code: flash_unlock /dev/mtd/0..1..3..4 From 2.5.8 FW update; how to flash U-Boot and kernel, both regular and backup ones: size of u-boot.bin = 131,072 B Code: $ dd if=imx31_mario/u-boot.bin of=/dev/mtdblock/0 bs=131072 $ dd if=imx31_mario/u-boot.bin of=/dev/mtdblock/3 bs=131072 size of uImage = 1,875,952 B Code: $ dd if=uImage of=/dev/mtdblock/1 bs=131072 $ dd if=uImage of=/dev/mtdblock/4 bs=131072

08-29-2015, 07:11 PM	#72
NiLuJe BLAM! Posts: 13,477 Karma: 26012492 Join Date: Jun 2010 Location: Paris, France Device: Kindle 2i, 3g, 4, 5w, PW, PW2, PW5; Kobo H2O, Forma, Elipsa, Sage, C2E	@newman: I may be mixing my timelines here, but wasn't KF8 support introduced in FW 3.3? EDIT: Yuuuup, brainfart. Last edited by NiLuJe; 08-29-2015 at 07:14 PM.

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Downgrade Kindle 3.x Kernel to 3.0.2 (on linux k3flasher) to provide forensic 3.3	twobob	Kindle Developer's Corner	45	02-26-2013 08:32 PM
PRS-T1 How-to: Compiling the kernel and kernel modules	altruizine	Sony Reader Dev Corner	9	09-03-2012 09:12 PM
iLiad Linux Kernel flashing to iliad	atmel	iRex Developer's Corner	1	11-20-2007 09:55 AM
iLiad Linux kernel update!	atmel	iRex Developer's Corner	4	11-12-2007 11:06 AM
iLiad Linux 2.6 Kernel Port for the Iliad	CommanderROR	iRex Developer's Corner	6	08-21-2007 08:33 AM

08-29-2015, 08:02 AM	#63
knc1 Going Viral Posts: 17,212 Karma: 18210809 Join Date: Feb 2012 Location: Central Texas Device: No K1, PW2, KV, KOA	Because it couldn't contact Amazon with only 3G available? But that doesn't make much sense - it should work in Airplane mode. - - - - Check the dates on those posts - because the yifan.lu procedure idoit was following did replace the kernel (both copies).

08-29-2015, 06:07 PM	#69
knc1 Going Viral Posts: 17,212 Karma: 18210809 Join Date: Feb 2012 Location: Central Texas Device: No K1, PW2, KV, KOA	There was a problem with Active Documents on DX(G) devices that had been flashed to 3.2.1 - There is a note in some readme that I read today, along with a link to fix. Some of these relationships are stored in "database" (or maybe better: "data files") on the Kindle. Since the hardware will still identify itself as a 2.something, just running 3.firmware is not always enough, you have to tweak the data entry so your 3.firmware knows its ok to run on a 2.something device. or something like that. I seem to recall that yifan.lu did make some of those data updates, but maybe he missed something related to the kf8. = = = = I was referring to the content of some of the quotes as being now out-of-date.

Advert

Advert