Re: MCM Crashes

[wretch <the.wretch@xxxxxxxxx>]

Hi Dave

I finally captured the serial port output (see below).
I've repeated the test a couple of times during the past few days all with the same result.
Even when a run a script that only executes 'ps' every second (thus not accessing the flash)
crashes the board.

Hope you can point me in the right direction.

Regards,
Peter


------ Serial port output begin -----
Unable to handle kernel access at virtual address 00d4c000
Oops: 0002
IRP: c0009a24 SRP: c0008d32 DCCR: 00000480 USP: 9ffffae0 MOF: 00000000
 r0: c1c18000  r1: c1c58000   r2: 00000000  r3: 00000000
 r4: 00005ad9  r5: c1c59f84   r6: 9ffffb10  r7: 00000000
 r8: 00000000  r9: 00d4d400  r10: c1c18000 r11: c00d9c80
r12: c1c1809d r13: 00000000 oR10: c1c18000
R_MMU_CAUSE: 00d4d139
Process myscript (pid: 253, stackpage=c1c58000)
                                
Stack from 9ffffae0:
       000d1d68 00000001 00089202 00000000 00000000 000c4150 00000000 000ad328
       9fffffbc 000000fd 000d1d68 000d1d7c 00000000 9ffffb10 000890ee 9fffffbc
       00000001 000d1d68 00000000 0008553a 00000000 00000000 000c4150 00000000
Call Trace:
Stack from c1c59de4:
       c00085a6 c1c59f2c c005bf0a c005c068 c00d9c80 00000000 c1c59f2c c1c58000
       c00cc340 00000002 c1c59ee8 c005c12e 00d4c000 c1c59ee8 c00085a6 c005f136
       00000000 00000000 9ffffb10 c1c59f84 00005ad9 00000000 c1c90000 c1c59ee8
Call Trace: [<c00085a6>] [<c005bf0a>] [<c005c068>] [<c005c12e>] [<c00085a6>] [<c005f136>] [<c0012066>]
       [<c0012614>] [<c005ef10>] [<c005bd22>] [<c0008d32>] [<c0009a24>] [<dbed0092>] [<c0008d32>] [<c0009920>]
                      
       [<c005bc26>]
Code: 69 96 0c 30 0f 05 5f 9d a1 00 ed db (10) e0 6a 96 5f ad 95 00 69 9a 5f 9d
Oops: bitten by watchdog
IRP: c000e544 SRP: c0009348 DCCR: 00000400 USP: 9ffffae0 MOF: 00000000
 r0: c1c18000  r1: 000000fd   r2: c1c58099  r3: c1c58000
 r4: c00f4000  r5: c000e41e   r6: c1c5802c  r7: c1c58064
 r8: c1c58010  r9: 0000000e  r10: 00000011 r11: 00000011
r12: c1c18095 r13: 00000000 oR10: 00000011
R_MMU_CAUSE: 35583010
Process myscript (pid: 253, stackpage=c1c58000)
                                
Stack from 9ffffae0:
       00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
       00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
       00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
Call Trace:
Stack from c1c59ca8:
       c00085a6 c1c59d24 c005bf0a c005c068 00000011 00000000 c00f4000 c1c58000
       c1c58099 000000fd c1c59ce0 c005c0e8 c1c18000 c005bdac 00000000 00000011
       00000000 c1c18095 00000011 00000011 0000000e c1c58010 c1c58064 c1c5802c
Call Trace: [<c00085a6>] [<c005bf0a>] [<c005c068>] [<c005c0e8>] [<c005bdac>] [<c000e41e>] [<c0009348>]
       [<c000e544>] [<c000e5bc>] [<c0020cce>] [<c0009348>] [<c000963e>] [<c005c13c>] [<c00085a6>] [<c005f136>]
                      
       [<c0012066>] [<c0012614>] [<c005ef10>] [<c005bd22>] [<c0008d32>] [<c0009a24>] [<dbed0092>] [<c0008d32>]
                      
       [<c0009920>] [<c005bc26>]
Code: 69 9a 14 e1 e9 9b 5f 0d f1 00 69 9a (1c) e1 e9 9b 5f 0d f5 00 69 9a 20 e1

------ Serial port output end -----

On 3/24/06, wretch <the.wretch@xxxxxxx.com> wrote:

The board requires a reset to get out of this state. No reflash required.
Capturing the serial port output is a bit difficult at the moment.
I will try to post the serial output on Monday.


Peter

On 3/24/06, Dave Whittaker <dwhittaker@xxxxxxx.com > wrote:

Could you elaborate on what happens when the board dies. Can it be rebooted or does it require a reflash? Any output from the serial port?

 

Dave

---

From: owner-dev-etrax@xxxxxxx.com [mailto:owner-dev-etrax@xxxxxxx.com] On Behalf Of wretch
Sent: Friday, March 24, 2006 3:54 AM
To: dev-etrax@xxxxxxx.com
Subject: MCM Crashes

Hi group,

We have a number of custom boards (based upon the dev server 83+ design) and we are having a problem with one particular board.
The problem is that the MCM (4+16) on this board crashes after a couple of minutes/hours uptime.

At first I thought it was the custom SW that caused this, so I ran a test without the SW and it ran (just idle) for a more than a day.
It must be the SW you would think (me too at first), but this morning I ran a test with a simple shell script (while true do; find /; done;)
and after a couple of minutes the MCM crashed, so that rules out the custom SW.

We also experimented with different environments to test if the problem was heat related but no change;

We have replace one or two MCM in the past because those refused to program.
Boards got x-rayed and no short we found :(
However this board shows a different behaviour.

Any ideas or similar problems ???

Regards,
Peter