On Wed, Jan 28, 2009 at 05:43:31PM +0200, Mart Raudsepp wrote:
Ühel kenal päeval, K, 2009-01-28 kell 10:28, kirjutas Kevin O'Connor:
On Wed, Jan 28, 2009 at 04:25:56AM +0200, Mart Raudsepp wrote:
On T, 2009-01-27 at 20:02 -0500, Kevin O'Connor wrote:
The log output at default loglevel of 1 is attached. At higher log levels the last dump seems to come from a different place. Should I gather such logs with a larger loglevel or does this say something revealing already? Anything else I can provide here?
Those messages aren't reporting a problem - I see them on my epia-cn and on qemu. You'll need to increase the debugging level to see if something else is seen before a failure.
I'm CC'ing the coreboot list again.
I think loglevel 20 went into an infinite loop (just this time maybe?), and I see a lot of these blocks, with either three or four entries:
handle_1c
That's the user timer irq handler.
enter handle_16: a=00000100 b=00000049 c=00008e26 d=00000000 si=00000010 di=00000044 ds=00000000 es=00000000 ip=00008e55 cs=00000000 f=00000246 r=00001fd2
That's grub asking if there is keyboard input.
handle_08
That's the timer irq handler.
So, you're getting interrupts, and it appears grub is running (or at least responding to irqs).
BTW, what are you using for a vgabios? I thought geode didn't have a vga bios available.
-Kevin
On K, 2009-01-28 at 13:36 -0500, Kevin O'Connor wrote:
handle_08
That's the timer irq handler.
So, you're getting interrupts, and it appears grub is running (or at least responding to irqs).
duh, I should configure autoboot after some seconds to be there, to actually see something.
BTW, what are you using for a vgabios? I thought geode didn't have a vga bios available.
I am not using a VGA BIOS at all. Now I realize I didn't have automatic boot of a secletion on timeout in my grub.conf, and I need to test if perhaps linux boot actually works and I can't just continue with that without a PS/2 keyboard and timeout to automatic choice. I'll report back on that after I can test it in the office.
XPe boot definitely didn't work then though. It probably does rely on VGA BIOS at start, but after that the native driver takes over. However it might also have been something else not being suitable for it compared to legacy BIOS, e.g ACPI vs non-ACPI installation (our legacy BIOS doesn't implement ACPI though, iirc). I'll re-test that after GRUB too, and gather logs as necessary.
Regarding a note in the other reply of yours - no, I don't have any SuperIO, and therefore no PS/2 keyboard or the like. I only have a USB keyboard and that I believe is not supported yet by SeaBIOS. I have OHCI not UHCI, as a non-Intel chipset (unless I'm mixing the two again).
Regards, Mart Raudsepp
On Wed, Jan 28, 2009 at 11:50:43PM +0200, Mart Raudsepp wrote:
On K, 2009-01-28 at 13:36 -0500, Kevin O'Connor wrote:
BTW, what are you using for a vgabios? I thought geode didn't have a vga bios available.
I am not using a VGA BIOS at all.
Without a vga bios, you're probably not going to be able to make much use of SeaBIOS. I wouldn't assume boot loaders will just "work" without one.
There has been discussion of porting the open source vga bios to geode. Also, google announced a vga bios to serial converter that you could try.
XPe boot definitely didn't work then though. It probably does rely on VGA BIOS at start, but after that the native driver takes over.
You'll need the vga bios for xp. One can verify this with qemu - without the vga bios it wont start.
Regarding a note in the other reply of yours - no, I don't have any SuperIO, and therefore no PS/2 keyboard or the like. I only have a USB keyboard and that I believe is not supported yet by SeaBIOS. I have OHCI not UHCI, as a non-Intel chipset (unless I'm mixing the two again).
Right - SeaBIOS doesn't have uhci/ohci support. It would be nice to add though.
-Kevin
On Wed, Jan 28, 2009 at 07:52:22PM -0500, Kevin O'Connor wrote:
There has been discussion of porting the open source vga bios to geode. Also, google announced a vga bios to serial converter that you could try.
Do you have a link for that? Sounds interesting.
Thanks, Ward.
Ward Vandewege wrote:
google announced a vga bios to serial converter
Do you have a link for that? Sounds interesting.
http://code.google.com/p/sgabios/
//Peter
On Wed, Jan 28, 2009 at 09:39:59PM -0500, Ward Vandewege wrote:
On Wed, Jan 28, 2009 at 07:52:22PM -0500, Kevin O'Connor wrote:
There has been discussion of porting the open source vga bios to geode. Also, google announced a vga bios to serial converter that you could try.
Do you have a link for that? Sounds interesting.
The email conversation I was thinking of is at:
http://www.coreboot.org/pipermail/coreboot/2008-October/039657.html
The google code is at:
http://code.google.com/p/sgabios/
-Kevin
Ühel kenal päeval, K, 2009-01-28 kell 19:52, kirjutas Kevin O'Connor:
On Wed, Jan 28, 2009 at 11:50:43PM +0200, Mart Raudsepp wrote:
On K, 2009-01-28 at 13:36 -0500, Kevin O'Connor wrote:
BTW, what are you using for a vgabios? I thought geode didn't have a vga bios available.
I am not using a VGA BIOS at all.
Without a vga bios, you're probably not going to be able to make much use of SeaBIOS. I wouldn't assume boot loaders will just "work" without one.
Just one step would be to get SeaBIOS booting something. Then I can present that and we are one step closer in shipping more devices with coreboot, and more likely to get more time resources on getting the rest to work. All the other bootloaders used as payloads work just fine without a VGA BIOS - FILO, etherboot, etc. Under not assuming boot loaders will just "work" without a VGA BIOS, do you mean the boot loaders found in MBRs? Kind of comes back to the currently missing possibility to optionally just use SeaBIOS for interrupt services.
There has been discussion of porting the open source vga bios to geode.
I've seen some vague mentions of it, but nothing concrete besides plans without details, and nothing even close to actual code :( I hope I've just missed something. While on that topic, some more detailed planning might be useful - is there something this is good to be based on? is there a good idea how it should get hooked it? etc
Also, google announced a vga bios to serial converter that you could try.
Yeah, I could see what sgabios does at some point.
XPe boot definitely didn't work then though. It probably does rely on VGA BIOS at start, but after that the native driver takes over.
You'll need the vga bios for xp. One can verify this with qemu - without the vga bios it wont start.
Most unfortunate. What is it wanting from it besides just drawing stuff. Is it expecting some kind of replies or return values back unlike some other things?
Regarding a note in the other reply of yours - no, I don't have any SuperIO, and therefore no PS/2 keyboard or the like. I only have a USB keyboard and that I believe is not supported yet by SeaBIOS. I have OHCI not UHCI, as a non-Intel chipset (unless I'm mixing the two again).
Right - SeaBIOS doesn't have uhci/ohci support. It would be nice to add though.
Indeed :)
Regards, Mart Raudsepp
On Thu, Jan 29, 2009 at 12:33:04PM +0200, Mart Raudsepp wrote:
All the other bootloaders used as payloads work just fine without a VGA BIOS - FILO, etherboot, etc. Under not assuming boot loaders will just "work" without a VGA BIOS, do you mean the boot loaders found in MBRs?
Yes.
There has been discussion of porting the open source vga bios to geode.
I've seen some vague mentions of it, but nothing concrete besides plans without details, and nothing even close to actual code :( I hope I've just missed something. While on that topic, some more detailed planning might be useful - is there something this is good to be based on? is there a good idea how it should get hooked it? etc
There was a thread on this starting at:
http://www.coreboot.org/pipermail/coreboot/2008-October/039657.html
There exists an open source vga bios today ( http://www.nongnu.org/vgabios/ ). So, in theory one could port it to the geode.
I'd recommend porting the current open source vga bios to gcc (much like SeaBIOS is a port of bochs bios to gcc). Then, one would need to either change all the ioport calls in the bios to use the geode equivalents, or implement some vsa/smm wrapper that caught the vga ioport accesses and emulated their functionality.
You'll need the vga bios for xp. One can verify this with qemu - without the vga bios it wont start.
Most unfortunate. What is it wanting from it besides just drawing stuff. Is it expecting some kind of replies or return values back unlike some other things?
I don't know. A qemu test shows grub wants a vga bios too. Lilo might still work.
-Kevin
-
Kevin O'Connor -
Mart Raudsepp -
Peter Stuge -
Ward Vandewege