I have added the spidev module to my 2.6.29 kernel and it seems to initialize correctly but I get a crash when I try to use the device.

I added the following to board_snapper9260.c:

static struct spi_board_info spi0_board_info[] __initdata = {
        {
                .modalias       = "spidev",
                .max_speed_hz   = 15 * 1000 * 1000,
                .chip_select    = 0,
                .bus_num        = 1,
        }
};

static void __init sn9260_board_init(void)
{
    at91_add_device_spi(spi0_board_info,ARRAY_SIZE(spi0_board_info));
    .........
}

On boot dmesg shows:

atmel_spi atmel_spi.1: Atmel SPI Controller at 0xfffcc000 (irq 13)
atmel_spi atmel_spi.1: registered master spi1
spi spi1.0: setup: 12861888 Hz bpw 8 mode 0x0 -> csr0 00000702
atmel_spi atmel_spi.1: registered child spi1.0

....and after using mdev -s I have a /dev/spidev1.0 device.

If I then try and use one of the spidev test tools (spidev_test or spidev_fdx) I get the following crash:

root@snapper:~$ ./spidev_fdx -v -m 3 /dev/spidev1.0 
/dekernel BUG at include/linux/spi/spi.h:609!
Unable to handle kernel NULL pointer dereference at virtual address 00000000
pgd = c39b8000
[00000000] *pgd=2396d031, *pte=00000000, *ppte=00000000
Internal error: Oops: 817 [#1]
Modules linked in:
CPU: 0    Not tainted  (2.6.29 #24)
PC is at __bug+0x20/0x2c
LR is at release_console_sem+0x198/0x1b0
pc : [<c002a840>]    lr : [<c003ddfc>]    psr: 60000093
sp : c398fe48  ip : c398fd80  fp : c398fe54
r10: 00000000  r9 : 00000002  r8 : c398e000
r7 : c397a004  r6 : c3937e00  r5 : c398fea4  r4 : c3937e00
r3 : 00000000  r2 : c02d2118  r1 : 00001916  r0 : 0000002e
Flags: nZCv  IRQs off  FIQs on  Mode SVC_32  ISA ARM  Segment user
Control: 0005317f  Table: 239b8000  DAC: 00000015
Process spidev_fdx (pid: 883, stack limit = 0xc398e268)
Stack: (0xc398fe48 to 0xc3990000)
fe40:                   c398fe84 c398fe58 c017f5cc c002a830 00000000 c398fe68 
fe60: c0081e90 00000000 c398fe68 c398fe68 c3997ee0 c3919848 c398fefc c398fe88 
fe80: c017feb4 c017f544 c0155c04 c38c9270 c3997ea0 c38c9260 00000004 c3919800 
fea0: 00000000 c391981c c3919840 00000000 00000000 c017f814 c398fe64 00000000 
fec0: 00000000 00000000 00000000 00000000 c398e000 bed7aca0 40406b00 00000003 
fee0: c3870240 bed7aca0 c398e000 00000000 c398ff14 c398ff00 c0090c44 c017f938 
ff00: c3870240 bed7aca0 c398ff7c c398ff18 c0091188 c0090c20 00000000 c398e000 
ff20: c398ff3c c398ff30 c0156860 c0141830 c398ff54 c398ff40 c0042500 c015685c 
ff40: 00000100 c398e000 c398ff84 c398ff58 c0042354 c00421d8 00000003 bed7aca0 
ff60: 40406b00 c3870240 c0026f28 00000000 c398ffa4 c398ff80 c009121c c0090cc0 
ff80: c00425a4 00000000 ffffffff 00000000 00000000 00000036 00000000 c398ffa8 
ffa0: c0026d80 c00911ec ffffffff 00000000 00000003 40406b00 bed7aca0 bed7aca0 
ffc0: ffffffff 00000000 00000000 00000036 00000000 00000000 40024000 bed7acf4 
ffe0: 00000000 bed7ac78 00008824 400deb5c 80000010 00000003 00000000 00000000 
Backtrace: 
[<c002a820>] (__bug+0x0/0x2c) from [<c017f5cc>] (spidev_sync+0x98/0xf0)
[<c017f534>] (spidev_sync+0x0/0xf0) from [<c017feb4>] (spidev_ioctl+0x58c/0x650)
 r5:c3919848 r4:c3997ee0
[<c017f928>] (spidev_ioctl+0x0/0x650) from [<c0090c44>] (vfs_ioctl+0x34/0x78)
[<c0090c10>] (vfs_ioctl+0x0/0x78) from [<c0091188>] (do_vfs_ioctl+0x4d8/0x52c)
 r5:bed7aca0 r4:c3870240
[<c0090cb0>] (do_vfs_ioctl+0x0/0x52c) from [<c009121c>] (sys_ioctl+0x40/0x64)
[<c00911dc>] (sys_ioctl+0x0/0x64) from [<c0026d80>] (ret_fast_syscall+0x0/0x2c)
 r7:00000036 r6:00000000 r5:00000000 r4:ffffffff
Code: e1a01000 e59f000c eb004ed0 e3a03000 (e5833000) 
v/---[ end trace 5ea79f764524e3a9 ]---
spidev1.0: spi mode 0, 0 bits per word, 15000000 Hz max
Segmentation fault

Did I miss something? is there something else I need to set up before using this module?

Any help would be appreciated.

Regards,

Pete

This looks like a genuine kernel bug. I will try and find some time over the next few days to attempt to replicate this problem.

~Ryan

Ryan,

I have taken the drivers/spi/spi.c and include/linux/spi/spi.h code from the OE Angstrom
2.6.29 kernel and they seem to fix the problem. I have not gone in to the code in any great
depth but it looks like the crash was related to spi bus locking.

Regards,

Pete