[LinuxBIOS] Latest on via epia-m for LinuxBIOS v2?

[Matthew Bloch]

Hi there,

I am trying to build a LinuxBIOS setup to turn an Epia-M board into a
NAS server and am stuck on the configuration.

FILO is built and tested (with grub).

I've now checked out LinuxBIOSv2, and am trying to use this configuration:

mnas:~/LinuxBIOSv2/targets/via/epia-m/epia-m# cat ../Config.lb
target epia-m

mainboard via/epia-m

option  MAXIMUM_CONSOLE_LOGLEVEL=8
option  DEFAULT_CONSOLE_LOGLEVEL=8
option  CONFIG_CONSOLE_SERIAL8250=1

option ROM_SIZE=256*1024
option HAVE_OPTION_TABLE=1
option CONFIG_ROM_PAYLOAD=1
option HAVE_FALLBACK_BOOT=0

option _RAMBASE=0x00004000

romimage "normal"
        option USE_FALLBACK_IMAGE=0
        option ROM_IMAGE_SIZE=0x30000
        option LINUXBIOS_EXTRA_VERSION=".0-Normal"
        option PAYLOAD_SIZE=0xc000
        payload /root/filo-0.5/filo.elf
end

buildrom ./linuxbios.rom ROM_SIZE "normal"

However trying to build it gives me:

./romcc -mcpu=c3 ... (snip lots of -Is and -Ds)
/root/LinuxBIOSv2/src/mainboard/via/epia-m/auto.c -o auto.inc
earlymtrr.c:23.0:
 "XIP_ROM_BASE is not a multiple of XIP_ROM_SIZE"
make[1]: *** [auto.inc] Error 1
make[1]: Leaving directory
`/root/LinuxBIOSv2/targets/via/epia-m/epia-m/normal'

I'm assuming something elementary is missing from the config but I'm not
sure where to look next.

Can someone point me in the right direction?  I will happily update the
HOWTO in exchange :)

cheers,

[Corey Osgood]

Matthew Bloch wrote:
> Hi there,
>
> I am trying to build a LinuxBIOS setup to turn an Epia-M board into a
> NAS server and am stuck on the configuration.
>
> FILO is built and tested (with grub).
>
> I've now checked out LinuxBIOSv2, and am trying to use this configuration:
>
> mnas:~/LinuxBIOSv2/targets/via/epia-m/epia-m# cat ../Config.lb
> target epia-m
>
> mainboard via/epia-m
>
> option  MAXIMUM_CONSOLE_LOGLEVEL=8
> option  DEFAULT_CONSOLE_LOGLEVEL=8
> option  CONFIG_CONSOLE_SERIAL8250=1
>
> option ROM_SIZE=256*1024
> option HAVE_OPTION_TABLE=1
> option CONFIG_ROM_PAYLOAD=1
> option HAVE_FALLBACK_BOOT=0
>
> option _RAMBASE=0x00004000
>
> romimage "normal"
>         option USE_FALLBACK_IMAGE=0
>         option ROM_IMAGE_SIZE=0x30000
>         option LINUXBIOS_EXTRA_VERSION=".0-Normal"
>         option PAYLOAD_SIZE=0xc000
>         payload /root/filo-0.5/filo.elf
> end
>
> buildrom ./linuxbios.rom ROM_SIZE "normal"

First off, can't do this. You must have a fallback image, normal is 
optional.
> However trying to build it gives me:
>
> ./romcc -mcpu=c3 ... (snip lots of -Is and -Ds)
> /root/LinuxBIOSv2/src/mainboard/via/epia-m/auto.c -o auto.inc
> earlymtrr.c:23.0:
>  "XIP_ROM_BASE is not a multiple of XIP_ROM_SIZE"

This is exactly as it says. In src/mainboard/via/epia-m/Config.lb, there 
should be an XIP_ROM_SIZE and XIP_ROM_BASE, the base must be a multiple 
of the size. I had a bit of trouble working around this myself at one 
point, I think I determined that I just couldn't have a VGA rom and only 
the fallback boot, having both seemed to straighten things out.

-Corey
> make[1]: *** [auto.inc] Error 1
> make[1]: Leaving directory
> `/root/LinuxBIOSv2/targets/via/epia-m/epia-m/normal'
>
> I'm assuming something elementary is missing from the config but I'm not
> sure where to look next.
>
> Can someone point me in the right direction?  I will happily update the
> HOWTO in exchange :)
>
> cheers,

[Matthew Bloch]

Corey Osgood wrote:
[snip]
> This is exactly as it says. In src/mainboard/via/epia-m/Config.lb, there
> should be an XIP_ROM_SIZE and XIP_ROM_BASE, the base must be a multiple
> of the size. I had a bit of trouble working around this myself at one
> point, I think I determined that I just couldn't have a VGA rom and only
> the fallback boot, having both seemed to straighten things out.

So just to be clear we officially have no idea how to make a VGA BIOS as
detailed in the HOWTO any more :)  Using the Config.vga.filo gives the
same error, so I'll give up on VGA as well.

Now the Config.filo.lb template gives me a working linuxbios.rom file -
however I had to rebuild filo to get it under about 65K.  (General
question) Is it just a quirk of the build system that I have to put two
copies of the same BIOS image onto the flash?  Under what failure
condition is the "fallback" used?  Surely the most useful fallback would
be the vendor BIOS, which of course wouldn't fit ..?

Anyhow I'm happy that the ROM ought to work as I've tested the new FILO
from my GRUB, so I'm now trying to flash it:

Here's me verifying the vendor BIOS:

mnas:~/LinuxBIOSv2/targets/via/epia-m/epia-m# flashrom -v /root/oldbios
Calibrating delay loop... ok
No LinuxBIOS table found.
Found chipset "VT8235": Enabling flash write... OK.
Found board "VIA EPIA M/MII/...": Enabling flash write... OK.
SST39SF020A found at physical address: 0xfffc0000
Flash part is SST39SF020A (256 KB)
Flash image seems to be a legacy BIOS. Disabling checks.
Verifying flash - VERIFIED

Okay so now I'll try to flash the new one:

mnas:~/LinuxBIOSv2/targets/via/epia-m/epia-m# flashrom linuxbios.rom
Calibrating delay loop... ok
No LinuxBIOS table found.
Found chipset "VT8235": Enabling flash write... OK.
Found board "VIA EPIA M/MII/...": Enabling flash write... OK.
SST39SF020A found at physical address: 0xfffc0000
Flash part is SST39SF020A (256 KB)
Note: If the following flash access fails, you might need to specify -m
<vendor>:<mainboard>

Hmmm, no signs of success or failure, will try to verify:

mnas:~/LinuxBIOSv2/targets/via/epia-m/epia-m# flashrom -v linuxbios.rom
Calibrating delay loop... ok
No LinuxBIOS table found.
Found chipset "VT8235": Enabling flash write... OK.
Found board "VIA EPIA M/MII/...": Enabling flash write... OK.
SST39SF020A found at physical address: 0xfffc0000
Flash part is SST39SF020A (256 KB)
Note: If the following flash access fails, you might need to specify -m
<vendor>:<mainboard>
Verifying flash - FAILED

Looks like something bad happened, or nothing, let's see...

mnas:~/LinuxBIOSv2/targets/via/epia-m/epia-m# flashrom -v /root/oldbios
Calibrating delay loop... ok
No LinuxBIOS table found.
Found chipset "VT8235": Enabling flash write... OK.
Found board "VIA EPIA M/MII/...": Enabling flash write... OK.
SST39SF020A found at physical address: 0xfffc0000
Flash part is SST39SF020A (256 KB)
Flash image seems to be a legacy BIOS. Disabling checks.
Verifying flash - VERIFIED

Ah, so nothing happened!  Any idea why this didn't work?

Thanks,

[Corey Osgood]

Matthew Bloch wrote:
> Corey Osgood wrote:
> [snip]
> > This is exactly as it says. In src/mainboard/via/epia-m/Config.lb, there
> > should be an XIP_ROM_SIZE and XIP_ROM_BASE, the base must be a multiple
> > of the size. I had a bit of trouble working around this myself at one
> > point, I think I determined that I just couldn't have a VGA rom and only
> > the fallback boot, having both seemed to straighten things out.
>
> So just to be clear we officially have no idea how to make a VGA BIOS as
> detailed in the HOWTO any more :)  Using the Config.vga.filo gives the
> same error, so I'll give up on VGA as well.

What I meant was I needed normal+fallback+vga, that did the trick for me 
at the time, but I can't recall why. For some reason having a fallback 
image size of 256k - vgarom size screwed up the XIP size calculation, 
perhaps because the vga rom wasn't exactly 64k?
> Now the Config.filo.lb template gives me a working linuxbios.rom file -
> however I had to rebuild filo to get it under about 65K.  (General
> question) Is it just a quirk of the build system that I have to put two
> copies of the same BIOS image onto the flash?  Under what failure
> condition is the "fallback" used?  Surely the most useful fallback would
> be the vendor BIOS, which of course wouldn't fit ..?

Fallback can be for any variety of things. For example, you could set it 
up so that there's no or reduced serial output during a normal boot, but 
a full spew during fallback, or else you could use a different payload 
for fallback.
> Anyhow I'm happy that the ROM ought to work as I've tested the new FILO
> from my GRUB, so I'm now trying to flash it:
>
> Here's me verifying the vendor BIOS:
>
> mnas:~/LinuxBIOSv2/targets/via/epia-m/epia-m# flashrom -v /root/oldbios
> Calibrating delay loop... ok
> No LinuxBIOS table found.
> Found chipset "VT8235": Enabling flash write... OK.
> Found board "VIA EPIA M/MII/...": Enabling flash write... OK.
> SST39SF020A found at physical address: 0xfffc0000
> Flash part is SST39SF020A (256 KB)
> Flash image seems to be a legacy BIOS. Disabling checks.
> Verifying flash - VERIFIED
>
> Okay so now I'll try to flash the new one:
>
> mnas:~/LinuxBIOSv2/targets/via/epia-m/epia-m# flashrom linuxbios.rom
> Calibrating delay loop... ok
> No LinuxBIOS table found.
> Found chipset "VT8235": Enabling flash write... OK.
> Found board "VIA EPIA M/MII/...": Enabling flash write... OK.
> SST39SF020A found at physical address: 0xfffc0000
> Flash part is SST39SF020A (256 KB)
> Note: If the following flash access fails, you might need to specify -m
> <vendor>:<mainboard>
>
> Hmmm, no signs of success or failure, will try to verify:
>
> mnas:~/LinuxBIOSv2/targets/via/epia-m/epia-m# flashrom -v linuxbios.rom
> Calibrating delay loop... ok
> No LinuxBIOS table found.
> Found chipset "VT8235": Enabling flash write... OK.
> Found board "VIA EPIA M/MII/...": Enabling flash write... OK.
> SST39SF020A found at physical address: 0xfffc0000
> Flash part is SST39SF020A (256 KB)
> Note: If the following flash access fails, you might need to specify -m
> <vendor>:<mainboard>
> Verifying flash - FAILED
>
> Looks like something bad happened, or nothing, let's see...
>
> mnas:~/LinuxBIOSv2/targets/via/epia-m/epia-m# flashrom -v /root/oldbios
> Calibrating delay loop... ok
> No LinuxBIOS table found.
> Found chipset "VT8235": Enabling flash write... OK.
> Found board "VIA EPIA M/MII/...": Enabling flash write... OK.
> SST39SF020A found at physical address: 0xfffc0000
> Flash part is SST39SF020A (256 KB)
> Flash image seems to be a legacy BIOS. Disabling checks.
> Verifying flash - VERIFIED
>
> Ah, so nothing happened!  Any idea why this didn't work?
>
> Thanks

There may be some sort of flash protection in place, check your vendor 
bios for a "Flash protect" option. I have this on my jetway and it gives 
the same effect, but via's a whole different beast. Does the vendor bios 
still verify? Peter Stuge should know more, hopefully he'll jump in soon.

-Corey

[Matthew Bloch]

Corey Osgood wrote:
> Matthew Bloch wrote:
> > Corey Osgood wrote:
> > [snip]
> > > This is exactly as it says. In src/mainboard/via/epia-m/Config.lb, there
> > > should be an XIP_ROM_SIZE and XIP_ROM_BASE, the base must be a multiple
> > > of the size. I had a bit of trouble working around this myself at one
> > > point, I think I determined that I just couldn't have a VGA rom and only
> > > the fallback boot, having both seemed to straighten things out.
> >
> > So just to be clear we officially have no idea how to make a VGA BIOS as
> > detailed in the HOWTO any more :)  Using the Config.vga.filo gives the
> > same error, so I'll give up on VGA as well.
>
> What I meant was I needed normal+fallback+vga, that did the trick for me
> at the time, but I can't recall why. For some reason having a fallback
> image size of 256k - vgarom size screwed up the XIP size calculation,
> perhaps because the vga rom wasn't exactly 64k?

Do you have a working Config.lb I could look at?

[snip]
> Fallback can be for any variety of things. For example, you could set it
> up so that there's no or reduced serial output during a normal boot, but
> a full spew during fallback, or else you could use a different payload
> for fallback.

Ah okay, under what circumstances does the fallback run?

[snip]
> There may be some sort of flash protection in place, check your vendor
> bios for a "Flash protect" option. I have this on my jetway and it gives
> the same effect, but via's a whole different beast. Does the vendor bios
> still verify? Peter Stuge should know more, hopefully he'll jump in soon.

I cut & pasted this program:
  http://linuxbios.org/pipermail/linuxbios/2004-March/007257.html
and it seems to have flashed without trouble (both the program itself
and flashrom said the new BIOS was verified, hooray).

Despite my earlier test I'm hesitant to reboot as I've no backstop in
case I brick it - will review my options for recover, possibly purchase
a backup chip and pop back in a few days :)

cheers,

[Corey Osgood]

Matthew Bloch wrote:
> Corey Osgood wrote:
> > Matthew Bloch wrote:
> > > Corey Osgood wrote:
> > > [snip]
> > > > This is exactly as it says. In src/mainboard/via/epia-m/Config.lb, there
> > > > should be an XIP_ROM_SIZE and XIP_ROM_BASE, the base must be a multiple
> > > > of the size. I had a bit of trouble working around this myself at one
> > > > point, I think I determined that I just couldn't have a VGA rom and only
> > > > the fallback boot, having both seemed to straighten things out.
> > >
> > > So just to be clear we officially have no idea how to make a VGA BIOS as
> > > detailed in the HOWTO any more :)  Using the Config.vga.filo gives the
> > > same error, so I'll give up on VGA as well.
> >
> > What I meant was I needed normal+fallback+vga, that did the trick for me
> > at the time, but I can't recall why. For some reason having a fallback
> > image size of 256k - vgarom size screwed up the XIP size calculation,
> > perhaps because the vga rom wasn't exactly 64k?
>
> Do you have a working Config.lb I could look at?

Sorry, no. It was for a pcchips board with the same chipset that I never 
got going. I don't think I have the code any more.
> [snip]
> > Fallback can be for any variety of things. For example, you could set it
> > up so that there's no or reduced serial output during a normal boot, but
> > a full spew during fallback, or else you could use a different payload
> > for fallback.
>
> Ah okay, under what circumstances does the fallback run?

IIRC, 3 sequential reboots or one failed boot by the normal image.
> [snip]
> > There may be some sort of flash protection in place, check your vendor
> > bios for a "Flash protect" option. I have this on my jetway and it gives
> > the same effect, but via's a whole different beast. Does the vendor bios
> > still verify? Peter Stuge should know more, hopefully he'll jump in soon.
>
> I cut & pasted this program:
>   http://linuxbios.org/pipermail/linuxbios/2004-March/007257.html
> and it seems to have flashed without trouble (both the program itself
> and flashrom said the new BIOS was verified, hooray).

Good find! This should get into flashrom.
> Despite my earlier test I'm hesitant to reboot as I've no backstop in
> case I brick it - will review my options for recover, possibly purchase
> a backup chip and pop back in a few days :)
>
> cheers,

Yeah, that's definitely the best practice. There's really no option for 
recovery if something goes wrong, except to hotswap and flash the chip 
in another system.

-Corey

[Matthew Bloch]

Corey Osgood wrote:
> Yeah, that's definitely the best practice. There's really no option for
> recovery if something goes wrong, except to hotswap and flash the chip
> in another system.

Hm, I do have another epia-m in a soon-to-be-redundant server in London
and this gave me hope:
  http://www.viaarena.com/default.aspx?PageID=5&ArticleID=37
But I think I will buy a BIOS Savior so I can try a few botch jobs with
impunity :)

[Corey Osgood]

Matthew Bloch wrote:
> Corey Osgood wrote:
> > Yeah, that's definitely the best practice. There's really no option for
> > recovery if something goes wrong, except to hotswap and flash the chip
> > in another system.
>
> Hm, I do have another epia-m in a soon-to-be-redundant server in London
> and this gave me hope:
>   http://www.viaarena.com/default.aspx?PageID=5&ArticleID=37
> But I think I will buy a BIOS Savior so I can try a few botch jobs with
> impunity :)

You'd need an RD1-PL IIRC, and they're sold out every place I know of, 
and IOSS has stopped manufacturing them. You can check the compatibility 
list at ioss.com.tw to be sure. This might be helpful as well: 
http://www.linuxbios.org/pipermail/linuxbios/2007-April/019809.html

-Corey

[Matthew Bloch]

Corey Osgood wrote:
> You'd need an RD1-PL IIRC, and they're sold out every place I know of,
> and IOSS has stopped manufacturing them. You can check the compatibility
> list at ioss.com.tw to be sure. This might be helpful as well:
> http://www.linuxbios.org/pipermail/linuxbios/2007-April/019809.html

Oh, that's a shame, I'd just sent an order.  I like the coloured
handles.  Well that does leave only one option for recovery which is the
spare board & hot plug.  Went ahead and flashed without success and was
all ready to pack up with a glum face for the evening (hadn't even
bothered with a serial cable) ... but on a tip from the linuxbios/EPIA
wiki page, replaced my 256MB stick with 512MB and after a few seconds of
hard drive clicking it came up on the network!  Hooray!  So from a cold
boot, the board is now available on the network in about 30s - lame of
course, I need to dig out a flash drive to fix that :)  I'm sure there's
room for improvement but I should at least dig out a serial cable before
I try to optimise this too much more.

Thanks a lot,

[Peter Stuge]

On Mon, Aug 27, 2007 at 11:34:21PM +0100, Matthew Bloch wrote:
> Okay so now I'll try to flash the new one:
>
> mnas:~/LinuxBIOSv2/targets/via/epia-m/epia-m# flashrom linuxbios.rom

Add -w to actually write.

We should add a message to flashrom in case no action option is
specified. (-v, -r or -w)

On Mon, Aug 27, 2007 at 07:31:12PM -0500, Corey Osgood wrote:
> > Do you have a working Config.lb I could look at?
>
> Sorry, no. It was for a pcchips board with the same chipset that I
> never got going. I don't think I have the code any more.

I'm afraid I've lost my working configs too.
> > I cut & pasted this program:
> >   http://linuxbios.org/pipermail/linuxbios/2004-March/007257.html
>
> Good find! This should get into flashrom.

flashrom does support the board, just needs -w. :)

On Tue, Aug 28, 2007 at 01:11:36AM +0100, Matthew Bloch wrote:
> I like the coloured handles.  Well that does leave only one option
> for recovery which is the spare board & hot plug.

You can hot swap flash chips on your one EPIA-M board too - you don't
really need a second board, just more flash chips, as long as you
have a backup of the factory BIOS..
> Went ahead and flashed

..well, ok. :)
> wiki page, replaced my 256MB stick with 512MB and after a few
> seconds of hard drive clicking it came up on the network!  Hooray!
> So from a cold boot, the board is now available on the network in
> about 30s - lame of course, I need to dig out a flash drive to fix
> that :)  I'm sure there's room for improvement

Yep, I've gotten a 600MHz EPIA-MII to 10-or-so seconds with a fairly
fat kernel. (A bunch of extra drivers that need time to init.)
> but I should at least dig out a serial cable before I try to
> optimise this too much more.

Yes, definately. And if you're going to profile the boot by
timestamping serial output - make sure you don't use a buffering
terminal emulator such as minicom.

Instead I suggest xc or picocom.

//Peter