Skip to contentSkip to conversation

i2c: Add msgs is NULL check

PATCH [email protected] 1 participant 1 message Mar 19, 2024 I2C subsystem host drivers

lore
Share & tools
lore
#0 lore raw

Thanks, 5.4 kernel, I'll try

Quoted 357 lines

Hi

Quoted 324 lines · Mika Westerberg

Hi,

Quoted 292 lines · Youwan Wang

fix crash because of null pointers

[  190.538113] kernel NULL pointer dereference at virtual address 0000000000000000

[  190.538115] Mem abort info:

[  190.538116]   ESR = 0x96000004

[  190.538118]   Exception class = DABT (current EL), IL = 32 bits

[  190.538119]   SET = 0, FnV = 0

[  190.538120]   EA = 0, S1PTW = 0

[  190.538121] Data abort info:

[  190.538122]   ISV = 0, ISS = 0x00000004

[  190.538123]   CM = 0, WnR = 0

[  190.538125] user pgtable: 4k pages, 48-bit VAs, pgdp = 00000000cf937cf2

[  190.538126] [0000000000000000] pgd=0000000000000000

[  190.538129] Internal error: Oops: 96000004 [#1] SMP

[  190.538172] Process swapper/0 (pid: 0, stack limit = 0x000000000179ba77)

[  190.538175] CPU: 0 PID: 0 Comm: swapper/0 Kdump: loaded Tainted: G

[  190.538178] pstate: 40000085 (nZcv daIf -PAN -UAO)

[  190.538183] pc : i2c_dw_isr+0x2e4/0x614 [i2c_designware_core]

[  190.538185] lr : i2c_dw_isr+0x9c/0x614 [i2c_designware_core]

[  190.538186] sp : ffff000008003e40

[  190.538187] x29: ffff000008003e40 x28: ffffd9cfad997200

[  190.538189] x27: ffff5f18a4f2c018 x26: 0000000000000000

[  190.538191] x25: ffff5f18a52d9fe8 x24: 0000000000000010

[  190.538193] x23: 0000000000000000 x22: 0000000000000000

[  190.538194] x21: 0000000000000510 x20: 0000000000000000

[  190.538196] x19: ffffd9cfa08d6080 x18: 00000000fffffffe

[  190.538197] x17: 0000000000000000 x16: 0000000000000000

[  190.538199] x15: 0000000000000000 x14: 0000000000000000

[  190.538200] x13: 0000000000000000 x12: 000000000000028d

[  190.538202] x11: 0000000000000040 x10: ffff5f18a52fe340

[  190.538203] x9 : ffffd9cfaf4a28f0 x8 : 0000000000000000

[  190.538205] x7 : ffffd9cfaf401b88 x6 : ffffd9cfaf401b60

[  190.538206] x5 : 0000000000000000 x4 : 0000000000000000

[  190.538208] x3 : 0000000000000000 x2 : ffff000008125000

[  190.538209] x1 : 0000000000000000 x0 : 0000000000000000

[  190.538211] Call trace:

[  190.538213]  i2c_dw_isr+0x2e4/0x614 [i2c_designware_core]

[  190.538218]  __handle_irq_event_percpu+0x68/0x230

[  190.538219]  handle_irq_event+0x6c/0x130

[  190.538222]  handle_fasteoi_irq+0xc0/0x220

[  190.538223]  __handle_domain_irq+0x80/0xe0

[  190.538226]  gic_handle_irq+0x84/0x188

[  190.538227]  el1_irq+0xb0/0x140

[  190.538229]  arch_cpu_idle+0x38/0x1c0

[  190.538232]  do_idle+0x238/0x2a4

[  190.538234]  cpu_startup_entry+0x2c/0x50

[  190.538237]  rest_init+0xbc/0xc8

[  190.538240]  start_kernel+0x4a8/0x4dc

[  190.538242] Code: 937c7c80 b940727a 8b0002c1 f9403e77 (78606ac0)

[  190.538248] SMP: stopping secondary CPUs

[  190.538400] Starting crashdump kernel...

[  190.538406] Bye!

---

   drivers/i2c/busses/i2c-designware-master.c | 7 ++++++-

   1 file changed, 6 insertions(+), 1 deletion(-)
drivers/i2c/busses/i2c-designware-master.c +0 -0

index ca1035e010c7..849e8a3e85ed 100644

--- a/drivers/i2c/busses/i2c-designware-master.c

+++ b/drivers/i2c/busses/i2c-designware-master.c

drivers/i2c/busses/i2c-designware-master.c +1 -1
@@ -429,12 +429,17 @@ i2c_dw_xfer_msg(struct dw_i2c_dev *dev)
 
 
 
    	struct i2c_msg *msgs = dev->msgs;
 
 
 
    	u32 intr_mask;
 
 
 
    	int tx_limit, rx_limit;
 
 
 
-	u32 addr = msgs[dev->msg_write_idx].addr;
 
 
 
+	u32 addr;

   u32 buf_len = dev->tx_buf_len;

   u8 *buf = dev->tx_buf;

   bool need_restart = false;

   unsigned int flr;

drivers/i2c/busses/i2c-designware-master.c

Quoted 5 lines
drivers/i2c/busses/i2c-designware-master.c +2 -0
+	if (WARN_ON(!msgs))
 
 
 
+		return;

Instead of treating the symptom, I suggest figuring out why this happens

in the first place. If the controller interrupt is enabled and triggered

Quoted 5 lines

it is expected that there is some transfer going on and dev->msgs !=

NULL.

What is the kernel version this happens? Does it include commit

301c8f5c32c8 ("i2c: designware: Fix handling of real but unexpected

device interrupts")?