More info about split-second message. RE: [syslinux] run FreeBSD "pxeboot" from pxelinux.0?

McMahon, Chris CMcMahon at
Tue Nov 18 15:09:25 PST 2003

	In the split-second before the reboot, I can see that the pxeboot.0
process is indeed started-- pxeboot gets out just over 2 lines of text
before the reboot occurs.   
	So it's not a "ready" message.  
	The text in question is (without *'s): 
PXE Loader 1.00 

Building the boot loader arguments 
Relocating the loader and the BTX 

	and it's gone.  
	BTW, I altered the display colors (yellow on blue) and was able to
read the split-second message, although I had to re-run a couple of times.
(nice feature, those colors.) 
	I hope that helps... 

-----Original Message-----
From: H. Peter Anvin [mailto:hpa at] 
Sent: Friday, November 14, 2003 2:22 PM
To: McMahon, Chris
Cc: 'syslinux at'; 'mike at'
Subject: Re: [syslinux] run FreeBSD "pxeboot" from pxelinux.0?

McMahon, Chris wrote:
> 	Doing ls -al on pxeboot.0 shows 157696. This is the generic
> "pxeboot" program from FreeBSD 4.8.   
> 	pxeboot's operation is very similar to pxelinux, but more simple:
> pxeboot simply finds and loads a FreeBSD kernel, then a FreeBSD filesystem
> called "mfsroot".  Access to the kernel and mfsroot is (in my case) over
> NFS-- but I'm not even getting that far. pxeboot relies on the "option
> root-path" DHCP parameter for access to the NFS filesystem.  
> 	pxeboot may access the kernel/mfsroot with TFTP, but that option
> must be enabled at compile time, and I have not done that.  I will if I
> to. But again, I'm not getting that far. 
> 	One additional note: upon closer examination, there seems to be some
> sort of message on the screen in the split-second before the first reboot,
> but it disappears too fast to read.  I'll try to capture it, but don't
> expect too much... 
> 	-Chris     

It's probably just "Ready", but who knows... if you want to try to
capture it I'd try hooking up a serial console.

A pxeboot NBP of 157,696 bytes is not compatible with the strict letter
of the spec (which says the NBP should be 32K or less), but it works
with all network stacks so it's certainly supported by the "de facto" spec.

I suspect this is a PXELINUX bug, and that it can be fixed.


More information about the Syslinux mailing list