[syslinux] EFI: PXE: "My IP is 0.0.0.0"

[Gene Cumm]

On Mon, Jun 22, 2015 at 6:03 AM, S. Schauenburg <s.schauenburg at gmail.com> wrote:
> Ady2 suggested to me on IRC to add additional information to the ML.
>
> With BIOS P89 v1.32 (03/05/2015) I saw errors from core_udp_sendto:
> Getting cached packet
> My IP is 0.0.0.0
> core_udp_sendto: stalling on configure with no mapping
> core_udp_sendto: stalling on configure with no mapping
> core_udp_sendto: stalling on configure with no mapping
> core_udp_sendto: stalling on configure with no mapping
> core_udp_sendto: stalling on configure with no mapping
>
> With BIOS P89 v1.40 (05/06/2015) I don't see any errors from core_udp_sendto
> anymore.

But you still see the "My IP is 0.0.0.0"?  The firmware version change
is definitely useful.  Without this, I would have considered the
stalling messages as expected.

> Hardware used: HP ProLiant DL380 Gen9
> NIC being used: NIC1 (not NIC0, since that's connected to a different
> network)

This is definitely useful and I suspect the major reason for the issue.


> On Mon, Jun 22, 2015 at 8:54 AM, S. Schauenburg <s.schauenburg at gmail.com>
> wrote:
>>
>> Hello Gene,
>>
>> thanks a lot for your patch! It doesn't work yet, but I have some more
>> information.

Excellent!  I figured that if I didn't have a solution, the temporary
extra Print() calls should at least provide more details, which they
did.  I was surprised to see that your server returns 8 protocol
handles (likely IPv4 and IPv6 for each of the 4 NIC ports).

>> Before the patch, the boot sequence ended with the following:
>>   NBP file download successfully.
>> Getting cached packet
>> My IP is 0.0.0.0
>>
>> After the patch it ended with the following (note that the IP is the same
>> as the Station IP address):
>>   NBP file download successfully.
>> LibLocateHandle returned nr 8
>> Handle [0] gave -3 00000000
>> Handle [1] gave -3 00000000
>> Handle [2] gave 0 0ad402a7
>> Getting cached packet
>> My IP is 10.212.2.167

I'd guess that the ARP packets are being seen on NIC0 and not NIC1.
I'm hoping to have another patch this week.

>> On Sun, Jun 21, 2015 at 8:48 PM, Gene Cumm via Syslinux
>> <syslinux at zytor.com> wrote:
>>>
>>> On Sat, Jun 20, 2015 at 5:09 PM, Gene Cumm <gene.cumm at gmail.com> wrote:
>>> > Starting a new thread for the moment and BCCing the two reporters of
>>> > the issue.
>>> >
>>> > Looking over the code, I think I finally see why we get junk data from
>>> > the PXE base.  If we tell the PXE base that a buffer is 0 bytes long,
>>> > we should expect junk/null data.
>>> >
>>> > I'm hoping to have a patch for testing over the weekend.  If either of
>>> > you would rather start a bug on the bugzilla, feel free.
>>>
>>> Attached are a patch and a limited set of test binaries.  Patch is
>>> relative to 38e861eb.  This should give more indication to the EFI PXE
>>> "My IP is 0.0.0.0" issue.
>>>
>>> --
>>> -Gene
>>>
>>> _______________________________________________
>>> Syslinux mailing list
>>> Submissions to Syslinux at zytor.com
>>> Unsubscribe or set options at:
>>> http://www.zytor.com/mailman/listinfo/syslinux
>>
>>
>



-- 
-Gene