[ipxe-devel] Problem chainloading iPXE from some legacy Intel PXE

[Jedrzej Kalinowski]

 On Fri, 29 Apr 2011 02:02:18 +0100, Michael Brown wrote:
> ines: 38
>
> On Thursday 28 Apr 2011 22:22:50 Jedrzej Kalinowski wrote:
>>  I've got a problem chainloading iPXE from the legacy PXE ROM 
>> (Intel's)
>>  on some of my machines and would be very grateful for your help.
>>
>>  I was using undionly.kpxe and undionly.kkpxe images, both with no
>>  result. Both images cannot locate undi network devices. I've 
>> recompiled
>>  the images with DEBUG=undinet,undionly,pxeparent option and the 
>> only
>>  additional notice to "no more network devices" I get is "no 
>> preloaded
>>  undi device found".
>>
>>  Here are the screenshots of both cases:
>>  http://www.iem.pw.edu.pl/~kalinoj1/public/undionly.kkpxe.jpg
>>  http://www.iem.pw.edu.pl/~kalinoj1/public/undionly.kpxe.jpg
>>
>>  The images work fine on VirtualBox and my other Intel-NIC equipped 
>> PCs.
>>
>>  Can having two NICs on one machine (it is the case, here) interfere
>>  somehow? Can I provide you with some more precise debug?
>
> Two NICs shouldn't be a problem here.  Could you try the following 
> patch? :
>
> diff --git a/src/arch/i386/drivers/net/undionly.c
> b/src/arch/i386/drivers/net/undionly.c
> index c38b574..abb839b 100644
> --- a/src/arch/i386/drivers/net/undionly.c
> +++ b/src/arch/i386/drivers/net/undionly.c
> @@ -58,6 +58,7 @@ static int undibus_probe ( struct root_device 
> *rootdev ) {
>  	/* Check for a valie preloaded UNDI device */
>  	if ( ! undi->entry.segment ) {
>  		DBG ( "No preloaded UNDI device found!\n" );
> +		DBG_HDA ( 0, undi, sizeof ( *undi ) );
>  		return -ENODEV;
>  	}
>
> Thanks,

 Hi,
 I've applied the patch and basically I got zeros everywhere:
 http://www.iem.pw.edu.pl/~kalinoj1/public/DEBUG-undionly.kkpxe.jpg
 http://www.iem.pw.edu.pl/~kalinoj1/public/DEBUG-undionly.kpxe.jpg


 BTW. I also tried removing the second NIC, and you were right - it 
 didn't change anything.

 BR,
 Jedrzej Kalinowski