@Seiya

Not sure about the Sam 460, it should have but the X1000 does have 2d/3d acceleration in the Linux kernels.

@scabit, Hypex

Sorry, I'm a little bit busy at the moment. I'll try to reply on the weekend.

Thanks for you patience!

_________________
A1SE: G3@600MHz, 2GB, 1GBit network card
A1XE: G4@933MHz, 2GB, refitted AC'97 codec
microA1: G3@800MHz, 1GB

- A1 Linux support -

@Geri

Quote:

Hmm, can you try to boot only the bare kernel image

tried with the 3.2.0..after two minutes of kernel loading without messages, an error similar to my last picture compare and immediately the a1 reboot...

_________________
--------------------------------------
Search and Destroy
---------------------------------------

@Geri

*Bump*

_________________
AmigaOne uA1-c 512M RAM - Only Amiga Makes It Possible!
Check my blog AmigaOne Computing

@scabit

Well, looks like he tried.

@Hypex

Quote:

Hypex wrote: @Geri Quote: After a whole day of debugging (and copying around whole partitions) it turned out that the inode size needs to be fixed to 128 bytes to make the partition readable by the slb_v2 bootloader. Wow that was a lot of work! Okay so what would cause this to be another size? Does it relate to block size?

No, it doesn't relate to the block size. But read more about this here under "AmigaOS4 slb_v2".

Quote:

Hypex wrote: @Geri Regarding compiling, I forgot until I found this! Infact I was looking for the SyncRead error on amigaworld.net but nothing came up amd only five threads are listed in the Linux section. So googled it to locate it. Should submit a bug report on that one.

The original MoL developer doesn't seem to be active anymore. Thus I wouldn't hold my breath that the bug report is ever dealt with.

Quote:

Regarding kernel source files. So, how about a poor mans source, include header files? On that note do we need specific A1 header files? And do you have just headers for these newer kernels which I will migrate too soon?

I only have kernel headers for the 3.2.0 kernel. However I'm not sure anymore, if they can be used on a PPC machine, as I compiled this kernel on a x86 machine as well. I guess I should make more use of distcc and ccache to speed up the compilation on my A1s.

Quote:

Quote: Thanks for the info! But I think I have enough addon-cards now. It would only save me a PCI slot. As long as you don't need it your system will be fine.

Well, my A1XE already has a Sii3114 SATA controller. Still I could need a free PCI slot for a 1GBit ethernet card, but then I could also remove the sound card, as the onboard sound is working well.

My A1SE on the other side needs a sound card, but suffers from the buggy PCI bus 0, where any DMA operation fails. In this case the card wouldn't help, except for saving me a PCI slot.

Quote:

You were looking for one. I think a friend would like one as well. I see NEC chip on there but is that USB? Seems unlikely if VIA can do all chipsets.

Yes, it looks like the NEC chip handles the USB ports. The VIA chip on the card probably is for the Firewire ports. And the TQFP chip nearby the SATA ports could be a SiliconImage controller chip.

Quote:

I suspect the problem is at my end. I don't think the initrd is setup correctly. And/or modules. I'll get it working and see if an specific issues show up.

Okay! Let me know, if something shows up!

Quote:

I guess not. I just so what I do with a new image. Remove previous modules from lib dir and replace with A1 modules. Infact it's slightly more complicated, as I now take them from the initrd for a stripped copy. Or can just copy across the dir list in the initrd. Then just a matter of replacing the kernel in the installer files. UBoot image. Scripts...

Sounds easy.

Quote:

Quote: What do you mean exactly with "broken"? IIRC it panics at boot. I'll fix it.

Okay.

Quote:

Quote: Where there any error messages from the 3c59x driver (transmit timeout) or something like this? I just went to check the logs and they are cut off. Linux froze recently as well. I don' remember booting into it since.

I wonder why Linux freezes that often on my A1XE, but not on my A1SE. It would be good to know about the root cause of this behavior.

Quote:

Quote: Thus I wonder why it doesn't work the same way on my A1SE with a Etch and Lenny-chroot environment!? As do I. Perhaps related to the problems I encountered. Remember your advice? Have you checked all source leading to MoL kernel modules?

The build of the MoL kernel module stopped due to a problem with the tools provided by the kernel sources. Maybe it has something todo with the Squeeze base systems that the chroot environments where running on, because IIRC the Linux kernel tools reported an error with the libc (even if rebuild for the powerpc architecture), when running manually.

Quote:

Quote: How should that work at all? Do you have all Linux/POSIX includes available on OS4? I started work on this project a long time ago in the OS4.0 days IIRC. When I started I actually downloaded some Linux include files and put them in their own sys folder in the includes. So I had all that I needed but it would always break as it conflicted with OS4 ones. So then I would remove files and the OS4 includes would lack features...

Okay, so you tried to get the MoL kernel module to build under OS4 (but still for a Linux kernel)?

Quote:

Quote: Hmm, isn't kconfig only the "kernel configuration" tool that generates the .config files? Yes I guess it is. But it felt good when about eight sources compile and one process completes!

Quote:

Quote: Did you work already on a AmigaOS4 version of the MoL kernel module? No. Didn't get faar did I? When I got to that I was intending to substitute Linux MMU functions for Exec ones through some translation code. Or I'd hoped.

At least it sounds like a lot of work!

Quote:

Quote: I recall Ross V told me MoL would be impossible on OS4, to paraphrase, and so Project Moami begun. But, 68K Amiga could run MacOS. And a PowerPC Amiga could run MacOS PPC through WarpUP and Fusion. So it's certainly not impossible. I side project was to patch the OS4 WarpUP implmentation so get Fusion to work but I neber got around to researching it too much.

IMHO, patching Fusion sounds like a better approach than getting MoL to work under OS4.

Quote:

Quote: Well, they shouldn't conflict with each other. You could disable DMA for both the IDE and LIBATA drivers. I did it on purpose. I forget why. If DMA was disabled on libata then I wouldn't include ide_core setttings as they would be redundant. That is, assuming libata can control all DMA? In which case my logical is really flawed if it doesnt!

ide_core and libata provide drivers for the same hardware, but are different subsystems and so need their own kernel parameters. However libata already largely replaced the ide_core drivers (except for Debian (Squeeze), which is a little bit behind with this transition). But if you boot with the 3.2.0 kernel for example, then a pata_via driver will show up as replacement for the via82cxxx driver. In this case the ide_core parameters wouldn't be necessary anymore.

Quote:

I recall this was the solution to specifying dma off settings on a number of ide controlers where sata would still be missed. As it controlls all bases?

For libata drivers you can only disable DMA for all harddisks and/or CDROM drives on all controllers, but not for a single device or driver.

Quote:

But I should include bus disabling on ide. It is restricted to bus and not devices. As I thought it more important to stop a full bus scan that a device scan. However this isn't UBoot, but Linux, so what was I thinking!?

For the via82cxxx driver I had to "noprobe" all channels of the VIA IDE controller on my A1SE to avoid a long delay. The new libata based pata_via driver however quickly runs over the empty IDE channels.

Quote:

Quote: That's why I don't understand, why the boot image wasn't copied to the boot partition. Well at least it is partially working. I'll do some tests.

Good!

Quote:

Quote: Did you try out my proposal with using the "/target/"'s sort command to sort the list-devices output. I can't remember! Was that one in an email? I tried one you suggested but it didn't work and I forget which one. But good point, it has to work in the installer enviroment. I'm not worried as I intend to rewrite that section. So it DOES work!

I think it was this command line: "sort -k 1.1,1.8 -k 1.9n"

But you have to use the sort command installed in the /target system (/target/usr/bin/sort), because busybox' version is buggy.

Quote:

As this point, with Wheezy breating down our necks.

Yes, rumors say it will be released in two weeks.

Quote:

Should we forfeit Squeeze? It may be hard as it needs 3.2 kernel which on A1 is still WIP?

I wouldn't do so. I'm not that happy with Wheezy yet. Especially because GNOME 3.4 is a pain and it is unlikey that it works fine on any A1.

@scabit

Quote:

scabit wrote: @Geri @Hypex So I figured I would try to build the MOL kernel files under Debian Lenny just to see what happened. I downloaded the mol-kernel sources, and then downloaded the 2.6.18 sources you pointed to in a previous post. I assume I do not need to apply any A1 specific patches to this kernel source before building it? i.e. it is already the patched version, yes?

Yes, it is a patched version. But as I mentioned before, I compiled the 2.6.18 kernel on a x86 machine, which breaks the build of the MoL kernel modules. Thus this kernel source won't help you much without recompiling the kernel on a PPC machine.

Quote:

I installed zlib1g-dev and autoconf and gcc..all is well there.

Well, under Debian Lenny you could also simply do "apt-get build-dep mol".

Quote:

A big problem I seem to have is that when I try to extract the 2.6.18 kernel sources, I cannot do it using the filemanager because it won;t allow me permission to install the files in /usr/src/linux. When I do it from a command line prompt, it installs where ever I happen to be and creates a directory named linux-sources-2.6.18, with LOTS of files and directories inside.

That's probably because the filemanager is not running with root priviledges (and that is a good thing!). Best is to do such thing in a shell and "cd" to the directory where you want to extract the kernel source.

Quote:

I cd to the mol kernel main directory (in my home directory) and type ./autogen.sh That works great. I then type make

It's much easier to compile MoL, if you run "make rpms", which will generate two separate RPM packages, one for the main MoL program and one for the MoL kernel modules. These packages can be converted to Debian packages using the "alien" tool later on.

Quote:

It makes the mol build, then goes to linux and tries to compile the kernel, but can't find the right environment variable. I tried copying the linux-sources-2.6.18 directory to /usr/src/linux and I get an error "Unconfigured kernel source!" How do I fix that? I'm sure it is some very simple and highly cryptic 3 character command that will fix everything.....

There are two PowerPC architectures available in kernel versions 2.6.16 till 2.6.26. The default since 2.6.16 is the "powerpc" architecture, which is a merge of the "ppc" and "ppc64" architecture code. The AmigaOne platform code for this old kernel versions is based on the "ppc" architecture, which you would have to force with "make ARCH=ppc ...".

Quote:

scabit wrote: @Geri Quote: --- The kernel source directory '/usr/src/linux' does not exist. --- Set the KERNEL_SOURCE environment variable to the appropriate directory. Okay, I installed the 2.6.18 kernel sources you pointed to and did this: ln -s /usr/src/kernel-source-2.6.18 /usr/src/linux and I still get the above error. How do I set the KERNEL_SOURCE environment variable to the proper directory? I'd like to give building the MOL kernel modules a try. All the MOL kernel stuff builds, but it breaks at the above error trying to build the kernel 2.6.18 modules. OH, this is on LENNY using kernel 2.6.18

Just append it to the make command or export the variable like this:

Quote:

export KERNEL_SOURCE=/usr/src/linux/

If all goes well, I'll recompile the 2.6.18 kernel this week and try to rebuild MoL on my A1XE. However I can't promise it at the moment.

@damned

Quote:

damned wrote: @Geri Quote: Hmm, can you try to boot only the bare kernel image tried with the 3.2.0..after two minutes of kernel loading without messages, an error similar to my last picture compare and immediately the a1 reboot...

That sounds like the kernel boots, but can't find any root device (which is expected). But then I wonder why there are no kernel messages on the screen.

Still no luck with a null modem cable?

@Hypex

Quote:

Hypex wrote: @scabit Well, looks like he tried.

The weekend ends at Monday 12:00 am CE[S]T here!

Last edited by Geri on 23-Apr-2013 at 07:47 AM.

_________________
A1SE: G3@600MHz, 2GB, 1GBit network card
A1XE: G4@933MHz, 2GB, refitted AC'97 codec
microA1: G3@800MHz, 1GB

- A1 Linux support -

@Geri

Well, I tried apt-get build-dep mol and downloaded another 9 megabytes of stuff.
Then I did the export KERNEL_SOURCE=/usr/src/linux
Then make rpms

Still got all the same errors - i.e. can't find directory /usr/src/linux

I tried to build the kernel (2.6.18) by going to /usr/src/linux/linux-source-2.6.18 and did a "make ARCH=ppc" and it built a lot of stuff, but I got this error:

from arch/powerpc/kernel/irq.c:34:
include/asm/system.h: In function ‘xchg_ptr’:
include/asm/system.h:193: warning: ‘xchg_u32’ is static but used in inline function ‘xchg_ptr’ which is not static
arch/powerpc/kernel/irq.c: In function ‘irq_alloc_host’:
arch/powerpc/kernel/irq.c:345: error: ‘mem_init_done’ undeclared (first use in this function)
arch/powerpc/kernel/irq.c:345: error: (Each undeclared identifier is reported only once
arch/powerpc/kernel/irq.c:345: error: for each function it appears in.)
arch/powerpc/kernel/irq.c: In function ‘irq_create_of_mapping’:
arch/powerpc/kernel/irq.c:586: error: dereferencing pointer to incomplete type
arch/powerpc/kernel/irq.c: In function ‘irq_of_parse_and_map’:
arch/powerpc/kernel/irq.c:614: error: storage size of ‘oirq’ isn’t known
arch/powerpc/kernel/irq.c:616: warning: implicit declaration of function ‘of_irq_map_one’
arch/powerpc/kernel/irq.c:614: warning: unused variable ‘oirq’
make[1]: *** [arch/powerpc/kernel/irq.o] Error 1
make: *** [arch/powerpc/kernel] Error 2

So, undeclared variables in irq.c. What's up with that?

Scott

Last edited by scabit on 22-Apr-2013 at 02:55 AM.

_________________
AmigaOne uA1-c 512M RAM - Only Amiga Makes It Possible!
Check my blog AmigaOne Computing

@scabit

Quote:

scabit wrote: @Geri Well, I tried apt-get build-dep mol and downloaded another 9 megabytes of stuff. Then I did the export KERNEL_SOURCE=/usr/src/linux Then make rpms Still got all the same errors - i.e. can't find directory /usr/src/linux

Did you link /usr/src/linux to the directory containing the kernel source? For example:
Quote:

/usr/src/linux-source-2.6.18 -> /usr/src/linux

Otherwise you can also point the KERNEL_SOURCE variable directly to the corresponding directory.

Quote:

I tried to build the kernel (2.6.18) by going to /usr/src/linux/linux-source-2.6.18 and did a "make ARCH=ppc" and it built a lot of stuff, but I got this error: from arch/powerpc/kernel/irq.c:34: include/asm/system.h: In function ‘xchg_ptr’: include/asm/system.h:193: warning: ‘xchg_u32’ is static but used in inline function ‘xchg_ptr’ which is not static

I stumbled over this warning, but...

Quote:

arch/powerpc/kernel/irq.c: In function ‘irq_alloc_host’: arch/powerpc/kernel/irq.c:345: error: ‘mem_init_done’ undeclared (first use in this function) arch/powerpc/kernel/irq.c:345: error: (Each undeclared identifier is reported only once arch/powerpc/kernel/irq.c:345: error: for each function it appears in.) arch/powerpc/kernel/irq.c: In function ‘irq_create_of_mapping’: arch/powerpc/kernel/irq.c:586: error: dereferencing pointer to incomplete type arch/powerpc/kernel/irq.c: In function ‘irq_of_parse_and_map’: arch/powerpc/kernel/irq.c:614: error: storage size of ‘oirq’ isn’t known arch/powerpc/kernel/irq.c:616: warning: implicit declaration of function ‘of_irq_map_one’ arch/powerpc/kernel/irq.c:614: warning: unused variable ‘oirq’ make[1]: *** [arch/powerpc/kernel/irq.o] Error 1 make: *** [arch/powerpc/kernel] Error 2 So, undeclared variables in irq.c. What's up with that?

...I never saw this error up to now. You're compiling this under Debian Lenny, right?

_________________
A1SE: G3@600MHz, 2GB, 1GBit network card
A1XE: G4@933MHz, 2GB, refitted AC'97 codec
microA1: G3@800MHz, 1GB

- A1 Linux support -

@Geri

As is typical with Debian, I am discovering that installing programs or building images requires multiple iterations, and that each iteration appears to go a little further than the previous one.
By rebooting and executing the exact same "make ARCH=ppc" command from within usr/src/linux/linux-source-2.6.18, I next got a different error after about 30 minutes of building. This error was fatal and stopped the compilation. It showed an undefined symbol.
So I tried again, and this time got through the entire compile process and apparently began linking, because after about 30 minutes of compiling it went to this:

building modules, step 2
MODPOST

then my micro A1 locked up solid.
So I am trying it again. Note - is there any way to get the compiler to NOT start all over again, that is, every time I run "make" it begins recompiling and creating all the object files from the beginning. Why doesn't it default to only building the ones that have not been built yet? Is there a command line argument that can make that happen? man make only shows a way to force it to compile everything from scratch, not to skip what is already compiled.

I'll let you know if I am ever able to successfully build a 2.6.18 kernel. Oh yes, this is under LENNY using the 2.6.18 kernel sources from sourceforge you pointed me to previously.

Scott

Last edited by scabit on 23-Apr-2013 at 06:12 PM.

_________________
AmigaOne uA1-c 512M RAM - Only Amiga Makes It Possible!
Check my blog AmigaOne Computing

@Geri

Okay, third times the charm! I was able to build a vmlinux image from the kernel 2.6.18 sources. It is 4.6 Megabytes in size. I tried to identify the start address and entry point using mkimage -l but it gave me an error saying "bad magic number...not a valid image file".
so I forged ahead and did this:

mkimage -A ppc -O linux -T multi -C none -a 0x000000 -e 0x000000 -n "Linux-2.6.18" -d vmlinux uImage.new.2.6.18

It build a uImage.new.2.6.18 of size 4.6 Megabytes.
I set up an entry in my a1boot.conf to boot this image, identical to using the uImage.2.6.18 provided by Hypex from ages ago, but using the new file name. I copied this new kernel image into my booting linux partition /boot directory where the a1boot.conf file resides.
Rebooted, it loaded the kernel image, with my name "Linux-2.6.18" with load of 000000 and entry of 000000 and said "checksum ok". Then it just sat there and did nothing. Forever.
(okay, I didn't time it for exactly that long...just guessing).

One question, should the -T multi in my mkimage be a -T kernel ?

Where did i go astray in making this kernel image?

Thanks,

Scott

_________________
AmigaOne uA1-c 512M RAM - Only Amiga Makes It Possible!
Check my blog AmigaOne Computing

@scabit

How does your UBoot image look compared with the orginal 2.6.18 one? As a rule of thumb if you see different longwords from offset 64 (IIRC where the code starts) then you have a problem. For an umcompressed image you should see three "60000000" words in a row at this point.

Looks like three giant 68K BRAs. 68K BRA was 6000.

BTW an ELF image won't work so if you are trying to create a boot image from an ELF kernel that wil crash it. It needs raw code. And with only kernel and no initrd set type to kernel.

In any case, apart from the kernel, have the working headers enabled you to compile mol kernel modules yet?!

Last edited by Hypex on 26-Apr-2013 at 04:38 PM.
Last edited by Hypex on 26-Apr-2013 at 04:34 PM.

@Hypex

Okay, the original image I built showed "ELF" in the first part of it. Obviously an ELF image. I tried using mkimage on the vmlinux I created with the -T kernel option and the start of the output image file looks like this:

00000000 27 05 19 56 14 D3 79 E1 51 7C 09 F6 00 49 02 CA 00 00 00 00 00 00 00 00 52 22 0E B8 '..V..y.Q|...I..........R"..
0000001C 05 07 02 00 44 65 62 69 61 6E 20 4C 69 6E 75 78 20 35 2E 30 20 4E 65 77 00 00 00 00 ....Debian Linux 5.0 New....
00000038 00 00 00 00 00 00 00 00 7F 45 4C 46 01 02 01 00 00 00 00 00 00 00 00 00 00 02 00 14 .........ELF................
00000054 00 00 00 01 C0 00 00 00 00 00 00 34 00 39 7F 3C 00 00 00 00 00 34 00 20 00 02 00 28 ...........4.9.

_________________
AmigaOne uA1-c 512M RAM - Only Amiga Makes It Possible!
Check my blog AmigaOne Computing

@scabit

It should look like this, as ripped from the prefab uImageHD-2.6.18 file:

0000: 27051956 5E269F3E 4BABCF8A 003A4086 '..V^&.>K«Ï..:@.
0010: 00000000 00000000 318C88F6 05070200 ........1..ö....
0020: 4C696E75 782D322E 362E3138 00000000 Linux-2.6.18....
0030: 00000000 00000000 00000000 00000000 ................
0040: 60000000 60000000 60000000 7C7F1B78 `...`...`...|..x
0050: 7C9E2378 7CBD2B78 7CDC3378 7CFB3B78 |.#x|½+x|Ü3x|û;x
0060: 3B000000 483887D9 480037C5 48003711 ;...H8.ÙH.7ÅH.7.

@Hypex

So, any idea what I am doing wrong in building the 2.6.18 kernel image under Lenny? i think I have all the proper tools installed, and it builds via make command with a lot of warnings but no errors. The only file it produces that I can tell is a vmlinux file, which I thought could be made into a kernel image file using mkimage. But obviously, the image I am creating is not valid. Is this because I am supposed to use a file other than vmlinux from which to generate the kernel image, or is vmlinux the correct file for me to use and it is not being generated correctly?

Thanks!

_________________
AmigaOne uA1-c 512M RAM - Only Amiga Makes It Possible!
Check my blog AmigaOne Computing

@scabit

Sorry for the late answer. I got sidetracked by other things.

Quote:

scabit wrote: @Geri As is typical with Debian, I am discovering that installing programs or building images requires multiple iterations, and that each iteration appears to go a little further than the previous one. By rebooting and executing the exact same "make ARCH=ppc" command from within usr/src/linux/linux-source-2.6.18, I next got a different error after about 30 minutes of building. This error was fatal and stopped the compilation. It showed an undefined symbol. So I tried again, and this time got through the entire compile process and apparently began linking, because after about 30 minutes of compiling it went to this: building modules, step 2 MODPOST then my micro A1 locked up solid.

A well known problem.

Quote:

So I am trying it again. Note - is there any way to get the compiler to NOT start all over again, that is, every time I run "make" it begins recompiling and creating all the object files from the beginning. Why doesn't it default to only building the ones that have not been built yet? Is there a command line argument that can make that happen? man make only shows a way to force it to compile everything from scratch, not to skip what is already compiled.

"make" only builds or updates the objects here that have not been compiled yet or have been modified at the source code level (given that the kernel config didn't change fundamentally in the meantime).

It could be a problem with the timestamp of the Linux kernel source files!? In this case it may also be a good idea to install "ntp" to synchronize the time with some NTP servers on the Internet.

Quote:

scabit wrote: @Geri Okay, third times the charm! I was able to build a vmlinux image from the kernel 2.6.18 sources. It is 4.6 Megabytes in size. I tried to identify the start address and entry point using mkimage -l but it gave me an error saying "bad magic number...not a valid image file".

"vmlinux" is not a U-Boot image file, but an ordinary ELF executable.

Quote:

so I forged ahead and did this: mkimage -A ppc -O linux -T multi -C none -a 0x000000 -e 0x000000 -n "Linux-2.6.18" -d vmlinux uImage.new.2.6.18 It build a uImage.new.2.6.18 of size 4.6 Megabytes. I set up an entry in my a1boot.conf to boot this image, identical to using the uImage.2.6.18 provided by Hypex from ages ago, but using the new file name. I copied this new kernel image into my booting linux partition /boot directory where the a1boot.conf file resides. Rebooted, it loaded the kernel image, with my name "Linux-2.6.18" with load of 000000 and entry of 000000 and said "checksum ok". Then it just sat there and did nothing. Forever. (okay, I didn't time it for exactly that long...just guessing).

You need to use a raw binary image for U-Boot, as it should be created in arch/ppc/boot/images/vmlinux.bin.

Quote:

One question, should the -T multi in my mkimage be a -T kernel ?

Yes, in this case you should use "-T kernel".

However it is much easier to simply run "make ARCH=ppc uImage", which will create the right image upfront! BTW "make ARCH=ppc help" gives you a list of all available build target.

Anyway, I uploaded a proberly build 2.6.18 Debian kernel package to the linux-image-old files section of the amigaone-linux project together with rebuild MoL packages. I had to build the kernel under Etch as well to avoid any modules mismatches. The MoL modules were tested with a complete OSX installation.

_________________
A1SE: G3@600MHz, 2GB, 1GBit network card
A1XE: G4@933MHz, 2GB, refitted AC'97 codec
microA1: G3@800MHz, 1GB

- A1 Linux support -

@Geri

Thanks Geri. I was about to look on my X1000 for answers.

As I remembered I built a Linux kernel there once...

@Hypex

Quote:

Hypex wrote: @Geri Thanks Geri. I was about to look on my X1000 for answers.

Thanks for taking over support in the meantime.

Quote:

As I remembered I built a Linux kernel there once...

I guess the X1000 can make use of a simple vmlinux file, right?

_________________
A1SE: G3@600MHz, 2GB, 1GBit network card
A1XE: G4@933MHz, 2GB, refitted AC'97 codec
microA1: G3@800MHz, 1GB

- A1 Linux support -

@Geri

I'll take a look and see if I got a vmlinux.bin file when I built the kernel.

As for the files you put up on SourceForge....there are 5 files:

mol-kmods...deb
mol...powerpc.deb
unionfs-modules-2.6.18-6....
gspca-modules-2.6.18-6...
linux-image-2.6.18-6...

So, I imagine I know what to do with the .deb files and the linux-image file. But what are the other two files used for?

Scott

_________________
AmigaOne uA1-c 512M RAM - Only Amiga Makes It Possible!
Check my blog AmigaOne Computing

@scabit

Quote:

scabit wrote: @Geri I'll take a look and see if I got a vmlinux.bin file when I built the kernel. As for the files you put up on SourceForge....there are 5 files: mol-kmods...deb mol...powerpc.deb unionfs-modules-2.6.18-6.... gspca-modules-2.6.18-6... linux-image-2.6.18-6... So, I imagine I know what to do with the .deb files and the linux-image file. But what are the other two files used for?

gspca is a driver for a number of USB video devices and unionfs is an overlay filesystem mainly used by some live install CDs. I just found their source code together with the kernel code on the Etch partition and decided to build them, too. So the linux-image and the mol packages are all you really need to get MoL running.

BTW: Please not that the kernel has a longer version string (2.6.18-6-book3s-amigaone). Thus you can install it alongside the existing 2.6.18 kernel.

_________________
A1SE: G3@600MHz, 2GB, 1GBit network card
A1XE: G4@933MHz, 2GB, refitted AC'97 codec
microA1: G3@800MHz, 1GB

- A1 Linux support -

@Geri

Quote:

You need to use a raw binary image for U-Boot, as it should be created in arch/ppc/boot/images/vmlinux.bin.

You are correct. I found the vmlinux.bin file dated 4-24-13 in my arch/ppc/boot/images directory - so I know I created it - and used it to build a proper image file...and it works! Wel,, one small problem, I get a lot of these types of warnings at bootup:

snd_ac97_codec: disagrees about version of symbol struct_module
gameport: disagrees about version of symbol struct_module
snd_via82xx: disagrees about version of symbol struct_module


I am guessing there is a particular version number in a file somewhere which differs from the name I used when I built the file? Or something like that.
Anyway, at least I can rebuild a kernel if I ever wish to do so.

Now I'll go and get your MOL kernel files from SourceForge and see how they work....

Thanks,

Scott

_________________
AmigaOne uA1-c 512M RAM - Only Amiga Makes It Possible!
Check my blog AmigaOne Computing