port-i386/42041: kernel assert ("(boundary & (boundary - 1)) == 0") triggers with i915drm in recent -5 kernel

[jasper%pointless.net@localhost]

>Number:         42041
>Category:       port-i386
>Synopsis:       kernel assert ("(boundary & (boundary - 1)) == 0") triggers 
>with i915drm in recent -5 kernel
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    port-i386-maintainer
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Thu Sep 10 17:10:00 +0000 2009
>Originator:     Jasper Wallace
>Release:        5.0-STABLE
>Organization:
>Environment:
NetBSD lemonade 5.0_STABLE NetBSD 5.0_STABLE (LEMONADE) #1: Mon May 25 16:27:45 
BST 2009  jasper@meep-meep:/usr/obj/sys/arch/i386/compile/LEMONADE i386
>Description:
kernels built form an up-to-date netbsd-5 source tree panic with (hand copied) 
when booted:

i915drm0 at vga0: Intel i915GM
panic: kernel diagnostic assertion "(boundary & (boundary - 1)) == 0" failed: 
file "/usr/src/sys/uvm/uvm_pglist.c", line 403
db{0}> bt
breakpoint(
panic(
__kernassert(
uvm_pglistalloc(
_bus_dmamem_alloc_range(
_bus_dmamem_alloc(
drm_pci_alloc(cad9d9a4, 1000, 1000, ffffffff, 88, c15b3008, c15b3000, 2, 
cad9d9d0,0) at netbsd:drm_pci_alloc+0xcc
i915_driver_load([blah]) at netbsd:i915_driver_load + 0x1fe
[stuff]

P.S. 'sync' form the ddb prompt dosn't produce a core dump, do i somehow need 
to tell the kernel to dump on wd0b (It might not of probed the hard drive at 
this point in the boot sequence).

P.P.S. I'm cleaning out my source tree and rebuilding tools and kernel, will 
try a GENERIC kernel as well
>How-To-Repeat:
build a recent -5 kernel and boot it on a Samsung q30 plus

>Fix: