Florian reported the following kernel NULL pointer dereference issue on
a BCM7250 board:
[ 2.829744] Unable to handle kernel NULL pointer dereference at virtual address
0000000c when read
[ 2.838740] [
0000000c] *pgd=
80000000004003, *pmd=
00000000
[ 2.844178] Internal error: Oops: 206 [#1] SMP ARM
[ 2.848990] Modules linked in:
[ 2.852061] CPU: 0 PID: 1 Comm: swapper/0 Not tainted
6.8.0-next-20240305-gd95fcdf4961d #66
[ 2.860436] Hardware name: Broadcom STB (Flattened Device Tree)
[ 2.866371] PC is at brcmnand_read_by_pio+0x180/0x278
[ 2.871449] LR is at __wait_for_common+0x9c/0x1b0
[ 2.876178] pc : [<
c094b6cc>] lr : [<
c0e66310>] psr:
60000053
[ 2.882460] sp :
f0811a80 ip :
00000012 fp :
00000000
[ 2.887699] r10:
00000000 r9 :
00000000 r8 :
c3790000
[ 2.892936] r7 :
00000000 r6 :
00000000 r5 :
c35db440 r4 :
ffe00000
[ 2.899479] r3 :
f15cb814 r2 :
00000000 r1 :
00000000 r0 :
00000000
The issue only happens when dma mode is disabled or not supported on STB
chip. The pio mode transfer calls brcmnand_read_data_bus function which
dereferences ctrl->soc->read_data_bus. But the soc member in STB chip is
NULL hence triggers the access violation. The function needs to check
the soc pointer first.
Fixes: 546e42599120 ("mtd: rawnand: brcmnand: Add BCMBCA read data bus interface")
Reported-by: Florian Fainelli <florian.fainelli@broadcom.com>
Tested-by: Florian Fainelli <florian.fainelli@broadcom.com>
Signed-off-by: William Zhang <william.zhang@broadcom.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Link: https://lore.kernel.org/linux-mtd/20240320222623.35604-1-william.zhang@broadcom.com
struct brcmnand_soc *soc = ctrl->soc;
int i;
- if (soc->read_data_bus) {
+ if (soc && soc->read_data_bus) {
soc->read_data_bus(soc, flash_cache, buffer, fc_words);
} else {
for (i = 0; i < fc_words; i++)