Random kernel Oops from fget_light on 3.10.17 kernel

[sebimohan]

Hi,

We are making a custom board based on imx6 sabresd. Initial revision was running on 3.0.35 kernel.

After upgrading to Freescale 3.10.17 kernel we are getting random kernel panic from system calls(PC is at fget_light on all crash)

(Some of our applications calls select and ioctls to driver)

Is this issue related to any kernel configuration changes or is it a general issue? Any clue on the reason behind this issue??

Below is the log of kernel Oops caused by a watchdog keep alive application which just calls an ioctl to keep the watchdog alive

Unable to handle kernel NULL pointer dereference at virtual address 00000000

pgd = ac53c000

[00000000] *pgd=3c4fa831, *pte=00000000, *ppte=00000000

Internal error: Oops: 17 [#1] SMP ARM

CPU: 0 PID: 889 Comm: wdog_keep_alive Tainted: P           O 3.10.17-1.0.0_ga+g232293e #1

task: ac9bcf00 ti: ac492000 task.ti: ac492000

PC is at fget_light+0x30/0xf0

LR is at SyS_ioctl+0x1c/0x64

pc : [<800da970>]    lr : [<800d1a5c>]    psr: 80030013

sp : ac493f80  ip : 00000000  fp : 7efd3cac

r10: 00000000  r9 : ac492000  r8 : 8000e044

r7 : 00000000  r6 : 000083b8  r5 : 00000003  r4 : 80045705

r3 : 00000000  r2 : 00000000  r1 : ac493f8c  r0 : 00000003

Flags: Nzcv  IRQs on  FIQs on  Mode SVC_32  ISA ARM  Segment user

Control: 10c53c7d  Table: 3c53c04a  DAC: 00000015

Process wdog_keep_alive (pid: 889, stack limit = 0xac492238)

Stack: (0xac493f80 to 0xac494000)

3f80: 80045705 00000003 8000e044 00000000 00000000 00000000 000083b8 00000036

3fa0: 8000e044 8000dec0 00000000 00000000 00000003 80045705 00000000 00000000

3fc0: 00000000 00000000 000083b8 00000036 00000000 00000000 76fe8000 7efd3cac

3fe0: 00010734 7efd3c94 0000853c 76f51ccc 60030010 00000003 00000000 00000000

[<800da970>] (fget_light+0x30/0xf0) from [<8000e044>] (sys_call_table+0x0/0x5f0)

Code: e593c000 e35c0001 0a00001f e5933004 (e5932000)

---[ end trace 91fceda01df377fa ]---

Thanks

SM

[igorpadykov]

Hi S M

these errors are caused by custom module, may be some wrong calls e.t.c.

One can try to debug it using attached document or

AN4553 Using Open Source Debugging Tools for Linux on i.MX Processors

Another option try latest BSP

L3.10.53_1.1.0_iMX6QDLS_Bundle : i.MX 6Quad, i.MX 6Dual, i.MX 6DualLite,

i.MX 6Solo Linux Binary Demo Files and Linux BSP Documentation

Size (K): 1392363 Format: gz Rev #: L3.10.53_1.1.0 Modified: 1/23/2015

Wdog usage can be found in imx-test package:

Index of /buildsources/i/imx-test/imx-test-3.10.17-1.0.0

Best regards

igor

[sebimohan]

Hi Igor

The panic is at "include/linux/fdtable.h" in function fcheck_files, fdt->max_fds where fdt obtained is NULL.

At that point files->count is 0. But there is no possibility that the close of that fd happened in between because the device nodes causing the panic is opened by single application only and that application have a single main(), no extra threads.

  these errors are caused by custom module, may be some wrong calls e.t.c.

What could be the wrong calls?. A custom module is there, but it just wait in poll_wait and read and write threads wake it up. Then read or write initiate from application.

Thanks

SM

[igorpadykov]

Hi S M

had you tried Demo images, without custom changes ?

L3.10.53_1.1.0_iMX6QDLS_Bundle : i.MX 6Quad, i.MX 6Dual, i.MX 6DualLite,

i.MX 6Solo Linux Binary Demo Files

~igor

[sebimohan]

Hi Igor,

We are using custom board based on SabreSD for our development. It is not possible to use the Demo Images in that.

I was not having the kernel panic issue with 3.0.35 Kernel.

Thanks,

SM