[syslinux] isolinux problem since 3.74

[H. Peter Anvin]

On 10/13/2009 03:28 PM, Stillwell, Bryan wrote:
> On Tue, 2009-10-13 at 21:39 +0000, H. Peter Anvin wrote:
>> On 10/13/2009 02:15 PM, Stillwell, Bryan wrote:
>>>
>>> Removing these changes in the 3.83 release allows the prototype machines
>>> to boot the installer, but I'm not sure why these changes are needed
>>> since it works on other (released) machines I have access to.  I'm
>>> writing the list because I would like to determine if this is a firmware
>>> bug that we can fix before we release, or a bug in isolinux that this
>>> machine aggrevates?
>>>
>>
>> The other possibility is that you have a broken memory map.  Could you
>> read out the output of meminfo.c32?
> 
> I've attached a screenshot from pxe booting meminfo.c32 on the machine.
> The machine has 8GiB of memory installed right now.
> 

You're using gPXE aren't you... it might be worthwhile to try without
gPXE if you can, since gPXE modifies the memory map.

Either way, the mostly likely problem isn't gPXE but rather memory line
number 6, which is marked "usable" (type 1) but has the flag bits [9].
The value 9 is AddressRangeSlowAccess which was only defined in the ACPI
4.0 specification -- it's likely that something is just severely *weird*
with your system.

Syslinux, like most operating systems, ignore these flag values as they
simply aren't reliable enough to do anything with.  It is certainly
possible that this is causing problems with a bizarre memory map like
this, though.

	-hpa

-- 
H. Peter Anvin, Intel Open Source Technology Center
I work for Intel.  I don't speak on their behalf.