[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [microblaze-uclinux] Bad page state in process



Hi Giulio,

Giulio Mazzoleni wrote:

I was using a Spartan3E 1600E evaluation board with no problem.
Then we have developed a custom board and I migrated our application to
that target.

I started by using the project Xilinx-Spartan3E1600-RevA-edk91.

I'm facing a system crash that shows up as soon as I start multiple
applications on the bord, for example the snmp daemon while an ftp
transfer is already in progress or vice versa.
I can easily reproduce the error by starting the same sequence of
applications. Please note that each of them is working fine if running
alone.

I am receiving these kind of error messages on the console:

Bad page state in process 'ftpd'
page:4440b180 flags:0x79740a61 mapping:6867666a mapcount:1818715753
count:1684433515
Trying to fix it up, but a reboot is needed

[snip]

 > Bad page state in process 'sh'
page:44411280 flags:0x73606706 mapping:00000000 mapcount:0
count:-1207427631

[snip]

Looks like a memory corruption issue, something is trashing the page tables and the kernel is detecting it.

A few things to check:

Do you have any special memory override parameters set in your MSS file, like

MAIN_MEMORY_SIZE
MAIN_MEMORY_START
MAIN_MEMORY_OFFSET

?

Check that the CONFIG_XILINX_ERAM_START and _ERAM_SIZE settings in linux-2.6.x/include/linux/autoconf.h are correct for your board.

Don't edit them there - they are autogenerated via the PetaLinux BSP -> petalinux-copy-autoconfig -> Kconfig.auto, but autoconf.h is where they end up.

Also make sure that CONFIG_XILINX_UNCACHED_SHADOW is turned off, it needs to be used with caution.

I started inspecting the RAM and tried with a simple extension to the
Xilinx RAM test. Filled the RAM with a random pattern and checked it by
reading data back. No problem arose.

A more stressful memory test may uncover something.


I'm using EDK 9.2i and OPB bus. Tried with both mch_opb_sdram and
opb_sdram.
OPB bus is running at 50 MHz.
I'm not currently using the clock feedback.
Tried with cache disabled.
Microblaze processor is version 6.00.b
Tried with bot ethernet_lite and ethernet full.

Always got the same results.

I'll try to crash the system without generating network activity.

I would be very glad if you could give me some hints or pointers.
I already read the thread "Bad page state in process 'swapper'" on this
mailing list.

I just re-read that thread, there was no clear resolution but it seemed likely a hardware problem - changing to the lower clock frequency seemed to fix it.

Regards,

John

___________________________
microblaze-uclinux mailing list
microblaze-uclinux@xxxxxxxxxxxxxx
Project Home Page : http://www.itee.uq.edu.au/~jwilliams/mblaze-uclinux
Mailing List Archive : http://www.itee.uq.edu.au/~listarch/microblaze-uclinux/