diff mbox series

mt76: fix BUG: spinlock bad magic in mt76_dma_init

Message ID 54dc821bf69fe8e7ae017f221069a4f41ae37893.1590260648.git.sean.wang@mediatek.com (mailing list archive)
State Superseded
Delegated to: Felix Fietkau
Headers show
Series mt76: fix BUG: spinlock bad magic in mt76_dma_init | expand

Commit Message

Sean Wang May 24, 2020, 10:14 a.m. UTC 
From: Sean Wang <sean.wang@mediatek.com>

dma init should be done when dma have been allocated

[    2.361127] BUG: spinlock bad magic on CPU#0, modprobe/456
[    2.361583]  lock: 0xffffa1287525b3b8, .magic: 00000000, .owner: <none>/-1, .owner_cpu: 0
[    2.362250] CPU: 0 PID: 456 Comm: modprobe Not tainted 4.14.177 #5
[    2.362751] Hardware name: HP Meep/Meep, BIOS Google_Meep.11297.75.0 06/17/2019
[    2.363343] Call Trace:
[    2.363552]  dump_stack+0x97/0xdb
[    2.363826]  ? spin_bug+0xa6/0xb3
[    2.364096]  do_raw_spin_lock+0x6a/0x9a
[    2.364417]  mt76_dma_rx_fill+0x44/0x1de [mt76]
[    2.364787]  ? mt76_dma_kick_queue+0x18/0x18 [mt76]
[    2.365184]  mt76_dma_init+0x53/0x85 [mt76]
[    2.365532]  mt7615_dma_init+0x3d7/0x546 [mt7615e]
[    2.365928]  mt7615_register_device+0xe6/0x1a0 [mt7615e]
[    2.366364]  mt7615_mmio_probe+0x14b/0x171 [mt7615e]
[    2.366771]  mt7615_pci_probe+0x118/0x13b [mt7615e]
[    2.367169]  pci_device_probe+0xaf/0x13d
[    2.367491]  driver_probe_device+0x284/0x2ca
[    2.367840]  __driver_attach+0x7a/0x9e
[    2.368146]  ? driver_attach+0x1f/0x1f
[    2.368451]  bus_for_each_dev+0xa0/0xdb
[    2.368765]  bus_add_driver+0x132/0x204
[    2.369078]  driver_register+0x8e/0xcd
[    2.369384]  do_one_initcall+0x160/0x257
[    2.369706]  ? 0xffffffffc0240000
[    2.369980]  do_init_module+0x60/0x1bb
[    2.370286]  load_module+0x18c2/0x1a2b
[    2.370596]  ? kernel_read_file+0x141/0x1b9
[    2.370937]  ? kernel_read_file_from_fd+0x46/0x71
[    2.371320]  SyS_finit_module+0xcc/0xf0
[    2.371636]  do_syscall_64+0x6b/0xf7
[    2.371930]  entry_SYSCALL_64_after_hwframe+0x3d/0xa2
[    2.372344] RIP: 0033:0x7da218ae4199
[    2.372637] RSP: 002b:00007fffd0608398 EFLAGS: 00000246 ORIG_RAX: 0000000000000139
[    2.373252] RAX: ffffffffffffffda RBX: 00005a705449df90 RCX: 00007da218ae4199
[    2.373833] RDX: 0000000000000000 RSI: 00005a7052e73bd8 RDI: 0000000000000006
[    2.374411] RBP: 00007fffd06083e0 R08: 0000000000000000 R09: 00005a705449d540
[    2.374989] R10: 0000000000000006 R11: 0000000000000246 R12: 0000000000000000
[    2.375569] R13: 00005a705449def0 R14: 00005a7052e73bd8 R15: 0000000000000000

Fixes: d3377b78cec6 ("mt76: add HE phy modes and hardware queue")
Signed-off-by: Sean Wang <sean.wang@mediatek.com>
---
 drivers/net/wireless/mediatek/mt76/dma.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Comments

Felix Fietkau May 24, 2020, 1:08 p.m. UTC | #1 
On 2020-05-24 12:14, sean.wang@mediatek.com wrote:
> From: Sean Wang <sean.wang@mediatek.com>
> 
> dma init should be done when dma have been allocated
This fix seems incomplete to me. I've sent a replacement patch,
please take a look.

Thanks,

- Felix
diff mbox series

Patch

diff --git a/drivers/net/wireless/mediatek/mt76/dma.c b/drivers/net/wireless/mediatek/mt76/dma.c
index 75e659774e07..eca39799f050 100644
--- a/drivers/net/wireless/mediatek/mt76/dma.c
+++ b/drivers/net/wireless/mediatek/mt76/dma.c
@@ -576,7 +576,7 @@  mt76_dma_init(struct mt76_dev *dev)
 
 	init_dummy_netdev(&dev->napi_dev);
 
-	for (i = 0; i < ARRAY_SIZE(dev->q_rx); i++) {
+	for (i = 0; i < ARRAY_SIZE(dev->q_rx) && dev->q_rx[i].desc; i++) {
 		netif_napi_add(&dev->napi_dev, &dev->napi[i], mt76_dma_rx_poll,
 			       64);
 		mt76_dma_rx_fill(dev, &dev->q_rx[i]);