[PATCH 00/17] Armada 370/XP NAND driver

[Sascha Hauer]

On Tue, Oct 13, 2015 at 12:38:51PM +0200, Robert Jarzmik wrote:
> Sascha Hauer <s.hauer at pengutronix.de> writes:
> 
> > In your temporary patch you have:
> >
> >> @@ -67,6 +71,9 @@ static int zylonite_devices_init(void)
> >>  			   &smsc91x_pdata);
> >>  	add_generic_device("mrvl_nand", DEVICE_ID_DYNAMIC, NULL,
> >>  			   0x43100000, 0x1000, IORESOURCE_MEM, &nand_pdata);
> >> +	clk = clk_get_sys("nand", NULL);
> >> +	if (!IS_ERR(clk))
> >> +		clkdev_add_physbase(clk, 0x43100000, NULL);
> >
> > Could it be that you register the clk lookup after the nand driver is probed? This
> > hunk should be moved before the nand device registration.
> That's what I had thought, but changing the order gave me a panic, so I didn't
> push forward.
> 
> This is what I had, decoded in [1].
> 
> Cheers.
> 
> --
> Robert
> 
> [1] My backtrace with the inversion :
> 
> barebox at Zylonite:/ bootm /mnt/tftp/barebox.bin
> bootm /mnt/tftp/barebox.bin
> netconsole: netconsole initialized with 255.255.255.255:6662
> eth0: 100Mbps full duplex link detected
> DHCP client bound to address 192.168.1.232
> 
> Loading ARM barebox image '/mnt/tftp/barebox.bin'
> commandline: ram=64M console=ttyS0,115200   mtdparts=pxa3xx_nand-0:128k at 0(TIMH)ro,128k at 128k(OBMI)ro,768k at 256k(barebox),256k at 1024k(barebox-env),12M at 1280k(kernel),38016k at 13568k(root)
> arch_number: 1233
> netconsole: port not set
> netconsole: registered as netconsole-1
> smc91c111 smc91c1110: chip is revision= 9, version= 2
> mdio_bus: miibus0: probed
> eth0: got preset MAC address: 00:0e:0c:a7:26:f7
> zylonite_devices_init(): clk = 83703328
> alloc_nand_resource(): clk_get() -> 83703328
> unable to handle NULL pointer dereference at address 0x00000004
> pc : [<83f15b9c>]    lr : [<83f15b90>]
> sp : 83fffed0  ip : 000001b0  fp : 00000000
> r10: 00000002  r9 : 83706888  r8 : 83f8c88c
> r7 : 00000000  r6 : 00000000  r5 : 83706948  r4 : 00000000
> r3 : 83706888  r2 : 43100000  r1 : 83f60783  r0 : 00000000
> Flags: nZCv  IRQs off  FIQs off  Mode SVC_32
> (mrvl_nand_probe+0x1ac/0x650) from device_probe (/drivers/base/driver.c:93)
> (device_probe+0x30/0xb8) from match.part.3 (/drivers/base/driver.c:153)
> (match.part.3+0x44/0x54) from register_device (/drivers/base/driver.c:192)
> (register_device+0x120/0x15c) from add_generic_device (/drivers/base/resource.c:84)
> (add_generic_device+0x3c/0x48) from zylonite_devices_init (/arch/arm/boards/zylonite/board.c:78)
> (zylonite_devices_init+0x9c/0x108) from start_barebox (/common/startup.c:101)
> (start_barebox+0x30/0xe8) from __start (/arch/arm/cpu/start.c:142)
> (__start+0x124/0x13c) from 0x8386c290
> 
> (unwind_backtrace+0x0/0x90) from panic (/common/misc.c:209)
> (panic+0x28/0x38) from do_exception (/arch/arm/cpu/interrupts.c:85)
> (do_exception+0x10/0x14) from do_data_abort (/arch/arm/cpu/interrupts.c:139)
> (do_data_abort+0x2c/0x38) from do_abort_6 (:?)
> ### ERROR ### Please RESET the board ###
> 
> And a special gift :
> rj at belgarion:~/mio_linux/barebox$ addr2line -i -e barebox 0x83f15b9c
> /home/rj/mio_linux/barebox/drivers/mtd/nand/nand_mrvl_nfc.c:1172
> /home/rj/mio_linux/barebox/drivers/mtd/nand/nand_mrvl_nfc.c:1202

This seems to be in mrvl_nand_probe_dt() and we are probing without
device tree here, right?

We have:

	struct device_node *np = host->dev->device_node;
	const struct of_device_id *match =
		of_match_node(mrvl_nand_dt_ids, np);
	const struct mrvl_nand_variant *variant = match->data;

of_match_node() is a static inline wrapper returning NULL, so next line
we dereference a NULL pointer.

Sascha

-- 
Pengutronix e.K.                           |                             |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |