Hi LinuxBIOS list subscribers,
I've been struggling for a couple of days now trying to get LinuxBIOS working on a DigitalLogic SmartCore-P5 166MHz PC/104 card. I am using the v1 tree, since the v1 tree contained smartcore support, but the v2 tree does not. I had no problems building and flashing the LinuxBIOS, and it starts up just fine and ends up in elfboot as expected. Elfboot reports to find my payload, loads it, and attempts to start executing it. At that point, nothing more happens. Due to the limited "debugability" at this point, it's hard to say exactly what has happened, but a serial output character I have put early in the FILO startup code is not showing up so I guess the FILO code is never reached at all(?) I have verified that the filo elf image I'm trying to load is OK by loading it from GRUB (it worked OK).
I have searched both the archives and other parts of the Internet for some hints about this, but the only interesting thing I was able to find was a FAQ in the FILO tree saying there was a bug in some linuxbios/elfboot checksum calculation routine that would cause FILO to crash. According to that FAQ, using the latest linuxbios patches should fix this problem. But I couldn't find any patches newer than the v1 stable tree, so I assumed this was old info(?)...
Does this sound familiar to anyone? Any suggestions would be highly appreciated. The v2 tree contains support for another DigitalLogic card that I'm not familiar with. Would there be any chance that config would work on my SmartCore-P5 card?
Cheers, /Kjell
the existing digitallogic support in v2 is not going to work on your card.
So your payload is FILO? And filo won't come up.
Are you on serial port console? You didn't try to set filo up for console did you? What baud rate?
ron
I was almost sure the DL support in v2 wouldn't with my card, so I didn't try that. :-) Is there any reason the smartcore:s weren't migrated to v2? Any interest in me giving it a shot after getting it to work with v1?
Yep, my intended payload is FILO, but I have also tried Etherboot with the same result. I'm trying to load Linux from an ext2 fs residing on the IDE compact flash slot in the card. The Linux system is already working well, the BIOS work is an attempt to speed up the boot process. The flash is 256kB, so there's no chance to squeeze the kernel into it, and my understanding from the documentation is that FILO should then be the way to go. Right? Since I wanted to avoid possibly screwing up my original BIOS, I have burned the romimage into another flash chip with a standalone flash programmer in the lab. We didn't have any 256kB flashes in stock, so I've been using 128kB and 512kB flashes instead and adjusted the PAYLOAD_SIZE accordingly. The result is the same regardless whether I'm using 128kB or 512kB flashes - FILO won't come up, but since I haven't actually been able to try the original 256kB size I was thinking maybe there's something more to adjust than the PAYLOAD_SIZE to compensate for another flash size? I have enabled extensive debugging (level 9), and read the outputs carefully (sorry, can't attach the output today since that's on another machine in the lab) but I cnnot find anything suspicious there.
Yes, I'm on a serial console. In fact, my DL board is the variant that is even not populated with the VGA chipset (optional). Unless the FILO config options are very unclear - No, I have disabled console and enabled serial at 115200. I have tried the built FILO image by loading it from GRUB, and my FILO image then worked fine at the serial link.
The note in the FILO FAQ about a checksum error in LinuxBIOS causing a crash, is that something that sounds familiar?
Cheers, /Kjell
On 13 maj 2005, at 23:05, Ronald G. Minnich wrote:
the existing digitallogic support in v2 is not going to work on your card.
So your payload is FILO? And filo won't come up.
Are you on serial port console? You didn't try to set filo up for console did you? What baud rate?
ron
LinuxBIOS mailing list LinuxBIOS@openbios.org http://www.openbios.org/mailman/listinfo/linuxbios