[syslinux] Pxechn.c32

[Gene Cumm]

On Mon, Jun 4, 2012 at 4:09 PM, H. Peter Anvin <hpa at zytor.com> wrote:
> On 06/04/2012 12:56 PM, Gene Cumm wrote:
>> On Jun 4, 2012 2:06 PM, "H. Peter Anvin" <hpa at zytor.com
>> <mailto:hpa at zytor.com>> wrote:
>>
>>> No, I think the problem is that you overwrite gPXE's copy of the packet,
>>> which is then lost when gPXE is unloaded.
>>
>> I'd expect to be overwriting gPXE's copy but I suspect either
>> 1) wdsnbp.com <http://wdsnbp.com> uses a scanning plan to find the PXE
>> stack (and finds the underlying stack) or
>> 2) gPXE reforms those packets (but not sure how as pxechn.c32 loads/runs
>> wdsnbp.com <http://wdsnbp.com> itself without the PXE restart call)
>>
>> I'm trying to get a COM32 to tell the results of all 5 plans (including
>> not stopping at the first on any scanning plans).
>>
>
> Right now, I think what happens is that it overwrites gPXE's copy, and
> then gPXE is unloaded when pxelinux exits.

Yes, syslinux_shuffle_boot_rm() normally unloads PXELINUX in a way
that it asks gPXE to unload also.  Thanks for the reminder (done so as
unloading gpxelinux.0 should result in a system clean of what's loaded
by it).  So the behavior of pxechn.c32 when gPXE is present should
change.

1) If there is a way, allow for a new pxechn.c32 option to instruct it
to instruct PXELINUX not to unload gPXE or
2) Attempt to find the underlying PXE stack and modify its packets.

-- 
-Gene