Using mem=4G fixed the problem. I thought this system only had 4G, so I
didn't think the mem boot parameter would be necessary. Guess there's
nothing else to do but wait for the patches to be picked up.
Thanks,
Mike
Vivek Goyal wrote:
On Thu, May 18, 2006 at 04:56:57PM -0700, Mike Mason wrote:
I don't see the warning when running crash on a live system. On a live
system, the "list modules" command shows the entire linked list. When
viewing the vmcore, I get this:
crash> list modules
c02fc354
f8c6f684
list: read error: kernel virtual address: f8c6f684 type: "list entry"
Also, if you do a "vtop f8c6f680", it will give you the physical address
that would be passed to read_netdump() to access. You can then
check that physical address against the ranges of physical memory
stored in the vmcore by doing a "help -n".
"vtop f8c6f680" shows:
VIRTUAL PHYSICAL
f8c6f680 101ede680
Hi Mike,
This physical address is more than 4G and that's a problem in SLES10 kernels.
I had posted the 64bit resources patches so that memory more than 4G is
exported through /proc/iomem and kdump can capture it. These patches are
still in Greg's tree and I am hoping that these will get in 2.6.18, if
community likes the patches.
Patchset is big and risky hence SuSE is not picking it for SLES10. Might
get into SLES10 SP1. That's why you see the problem. Try booting your
first kernel with mem=4G parameter and capture the dump, things should
be fine.
Maneesh had posted a hack patch which could be considered for inclusion
in SLES10. I guess SuSE is yet to look into it.
Thanks
Vivek
