2.6.0-test8 and HIGMEM = segfaults and panics?

I'm running a dual-athalon system. When I compiled the 2.6.0-test8 kernel I enabled HIGHMEM for 4 Gigs as I'm at 1.5G now and planning on purchasing an additional 512Meg DIMM next weekend (yeah, should have with the 1.5Gig).

At any rate the box comes up just fine and runs for a while but once the memory is in use for a few hours and seems to exceed 220+ Megs about any command I execute will Segfault and the kernel has panic'd twice (couldn't read the whole oops).

I backed out the HIGHMEM and left it at 1.5Gigs and it seems to be running fine again.

I poked around the archives and couldn't find anything in particular related to this. Anyone seen anything similar, known bug I've missed?

Robert

:wq!
---------------------------------------------------------------------------
Robert L. Harris | GPG Key ID: E344DA3B

                                         @ x-hkp://pgp.mit.edu

DISCLAIMER:
These are MY OPINIONS ALONE. I speak for no-one else.

Life is not a destination, it's a journey.
Microsoft produces 15 car pileups on the highway.
Don't stop traffic to stand and gawk at the tragedy.

I run a dual athlon system with 2 GB, and it has never done anything like that.

Have you run memtest86 on your system to verify it isn't a physical memory problem?

Are you using registered RAM chips? I seem to remember from my motherboard documentation that registered is required to use more than 2 memory slots.

Zan Lynx <[email protected]>

Yup, I've played musical DIMMS as well. It's currently up running with 1.5G install without the HIGMEM and I've taxed it pretty hard today without issue.

What kernel are you running on?

They are registered.

Thus spake Zan Lynx ([email protected]):

:wq!
---------------------------------------------------------------------------
Robert L. Harris | GPG Key ID: E344DA3B

                                         @ x-hkp://pgp.mit.edu

DISCLAIMER:
These are MY OPINIONS ALONE. I speak for no-one else.

Life is not a destination, it's a journey.
Microsoft produces 15 car pileups on the highway.
Don't stop traffic to stand and gawk at the tragedy.

Kernel 2.6.0-test8.
It's a Tyan MPX board with 4 512MB sticks of ECC registered 266 DDR.

Here's the highmem bit of my config file:
# CONFIG_NOHIGHMEM is not set
CONFIG_HIGHMEM4G=y
# CONFIG_HIGHMEM64G is not set
CONFIG_HIGHMEM=y
CONFIG_HIGHPTE=y

So, I don't think it's a general bug.

Zan Lynx <[email protected]>

I can recompile it and run it for a bit. What debug options would be useful for this?

When I had it enabled I had this:

# CONFIG_NOHIGHMEM is not set
CONFIG_HIGHMEM4G=y
# CONFIG_HIGHMEM64G is not set
CONFIG_HIGHMEM=y
CONFIG_HIGHPTE=y

Thus spake Zan Lynx ([email protected]):

:wq!
---------------------------------------------------------------------------
Robert L. Harris | GPG Key ID: E344DA3B

                                         @ x-hkp://pgp.mit.edu

DISCLAIMER:
These are MY OPINIONS ALONE. I speak for no-one else.

Life is not a destination, it's a journey.
Microsoft produces 15 car pileups on the highway.
Don't stop traffic to stand and gawk at the tragedy.