[Pkg-xen-devel] Bug#721946: Bug#721946: xen-hypervisor-4.1-amd64: dom0_mem cannot exceed some value

[Alessandro Vesely]

On Fri 06/Sep/2013 10:12:29 +0200 Ian Campbell wrote:
> On Thu, 2013-09-05 at 20:52 +0200, Alessandro Vesely wrote:
>> 
>> I tried GRUB_CMDLINE_XEN="dom0_mem=8192M":  that delivers 6964868K total, then
>> crashes when used=2837436K free=4127432K.
> 
> On a modern dom0 kernel you need to specify the maximum memory as well,
> i.e. dom0_mem=8192M,max:8192M

Nope, I tried the string you said, and got a crash.  I attach the log
file again, as there are some small differences between this boot and
the two ones in the previous attachment, that I am unable to
interpret.  The outstanding difference I see is:

BUG: unable to handle kernel paging request at ffff88013fc00000
Oops: 0002 [#1] SMP
BUG: scheduling while atomic: gimp-2.8/4922/0x10000001
BUG: unable to handle kernel paging request at ffff88013fe00000
Oops: 0002 [#2] SMP
BUG: scheduling while atomic: postgres/5960/0x10000001

Lines of those kind were present in the first log (dom0_mem=8192M) and
in this one (dom0_mem=8192M,max:8192M), but not in the second boot of
the first log (dom0_mem=2048M).  They are related to the black screen
behavior, correct?

> http://blog.xen.org/index.php/2012/04/30/memory-where-it-has-not-gone/
> has some background discussion.

Thanks for the pointer.  In the logs I find "register mmio base:
0xFEA30000".  Is that a byte address?  It would seem something could
change when passing from the 6G below it to higher values.  I only
tried 2G and 8G thus far.

Ale
-------------- next part --------------
A non-text attachment was scrubbed...
Name: kern2.log
Type: text/x-log
Size: 112002 bytes
Desc: not available
URL: <http://lists.alioth.debian.org/pipermail/pkg-xen-devel/attachments/20130906/35449506/attachment-0001.bin>