[ipxe-devel] Windows having problems parsing iBFT from recent iPXE versions?

[Floris Bos]

Hi,

On 10/29/2014 02:56 PM, Michael Brown wrote:
> On 28/10/14 22:43, Floris Bos wrote:
>> I am having problems getting a Windows 2012 iSCSI installation to work
>> with recent iPXE versions,
>> while it does work with an older iPXE version.
>
> Have you tried performing a bisection (http://ipxe.org/howto/bisect) 
> to identify the commit causing the problem?
>

Problems seem to start after this commit:

==
[tcp] Do not send RST for unrecognised connections
author    Michael Brown <mcb30 at ipxe.org>
     Fri, 12 Jul 2013 02:10:03 +0100 (03:10 +0200)
committer    Michael Brown <mcb30 at ipxe.org>
     Fri, 12 Jul 2013 02:20:05 +0100 (03:20 +0200)
commit    18d0818f94c6750c680cba582932ae16659c0c38
tree    d2083540b2755e44a2d0a06534e4c9f557386ea1    tree | snapshot 
(tar.gz tar.bz2 zip)
parent    936134ed460618e18cc05d677a442d43d5e739a1    commit | diff
[tcp] Do not send RST for unrecognised connections

On large networks with substantial numbers of monitoring agents,
unwanted TCP connection attempts may end up flooding iPXE's ARP cache.

Fix by silently dropping packets received for unrecognised TCP
connections.  This should not cause problems, since many firewalls
will also silently drop any such packets.

Reported-by: Jarrod Johnson <jarrod.b.johnson at gmail.com>
Signed-off-by: Michael Brown <mcb30 at ipxe.org>
==

Parent 936134ed460618e18cc05d677a442d43d5e739a1 does work.
However if I try with HEAD with just this one commit reversed it does 
NOT work.
Are there any related commits?


 >For what it's worth: in the current iPXE and wimboot builds this 
script should probably be:

If it doesn't hurt I would prefer to keep using old syntax, so it is 
easier to test with all versions of iPXE without changing scripts.

>> With undionly.kkpxe compiled today from git sources:
>
> Another possibility is that a small increase in code or data size has 
> pushed the iBFT below the 512kB boundary.  Try with undionly.kpxe 
> instead (which should reclaim more base memory); this may work.
>

Does not work with .kpxe either.

And perhaps we should test for location of iBFT in iPXE, and print out a 
big fat warning -if not fatal error- when it is under 512 KB?

BTW, I noticed the Linux code -in addition to scanning memory- also 
seems to look in the RSDT ACPI table for a reference to the iBFT table.
Is there any way for us to add a reference to our iBFT there?

-- 
Yours sincerely,

Floris Bos