From: Lorenzo Bianconi lorenzo@kernel.org
[ Upstream commit d74c4b5667425c35d74906795a08e02e29df5b46 ]
Do not schedule hw full reset if the device is not fully initialized (e.g if the channel has not been configured yet). This patch fixes the kernel crash reported below
[ 44.440266] mt7921e 0000:01:00.0: chip reset failed [ 44.527575] Unable to handle kernel paging request at virtual address ffffffc02f3e0000 [ 44.535771] Mem abort info: [ 44.538646] ESR = 0x96000006 [ 44.541792] EC = 0x25: DABT (current EL), IL = 32 bits [ 44.547268] SET = 0, FnV = 0 [ 44.550413] EA = 0, S1PTW = 0 [ 44.553648] Data abort info: [ 44.556613] ISV = 0, ISS = 0x00000006 [ 44.560563] CM = 0, WnR = 0 [ 44.563619] swapper pgtable: 4k pages, 39-bit VAs, pgdp=0000000000955000 [ 44.570530] [ffffffc02f3e0000] pgd=100000003ffff003, p4d=100000003ffff003, pud=100000003ffff003, pmd=0000000000000000 [ 44.581489] Internal error: Oops: 96000006 [#1] SMP [ 44.606406] CPU: 0 PID: 0 Comm: swapper/0 Tainted: G W 5.13.0-rc1-espressobin-12875-g6dc7f82ebc26 #33 [ 44.617264] Hardware name: Globalscale Marvell ESPRESSOBin Board (DT) [ 44.623905] pstate: 600000c5 (nZCv daIF -PAN -UAO -TCO BTYPE=--) [ 44.630100] pc : __queue_work+0x1f0/0x500 [ 44.634249] lr : __queue_work+0x1e8/0x500 [ 44.638384] sp : ffffffc010003d70 [ 44.641798] x29: ffffffc010003d70 x28: 0000000000000000 x27: ffffff8003989200 [ 44.649166] x26: ffffffc010c08510 x25: 0000000000000002 x24: ffffffc010ad90b0 [ 44.656533] x23: ffffffc010c08508 x22: 0000000000000012 x21: 0000000000000000 [ 44.663899] x20: ffffff8006385238 x19: ffffffc02f3e0000 x18: 00000000000003c9 [ 44.671266] x17: 0000000000000000 x16: 0000000000000000 x15: 000009b1a8a3bf90 [ 44.678632] x14: 0098968000000000 x13: 0000000000000000 x12: 0000000000000325 [ 44.685998] x11: ffffff803fda1928 x10: 0000000000000001 x9 : ffffffc010003e98 [ 44.693365] x8 : 0000000000000032 x7 : fff8000000000000 x6 : 0000000000000035 [ 44.700732] x5 : 0000000000000000 x4 : 0000000000000000 x3 : ffffffc010adf700 [ 44.708098] x2 : ffffff8006385238 x1 : 000000007fffffff x0 : 0000000000000000 [ 44.715465] Call trace: [ 44.717982] __queue_work+0x1f0/0x500 [ 44.721760] delayed_work_timer_fn+0x18/0x20 [ 44.726167] call_timer_fn+0x2c/0x178 [ 44.729947] run_timer_softirq+0x488/0x5c8 [ 44.734172] _stext+0x11c/0x378 [ 44.737411] irq_exit+0x100/0x108 [ 44.740830] __handle_domain_irq+0x60/0xb0 [ 44.745059] gic_handle_irq+0x70/0x2b4 [ 44.748929] el1_irq+0xb8/0x13c [ 44.752167] arch_cpu_idle+0x14/0x30 [ 44.755858] default_idle_call+0x38/0x168 [ 44.759994] do_idle+0x1fc/0x210 [ 44.763325] cpu_startup_entry+0x20/0x58 [ 44.767372] rest_init+0xb8/0xc8 [ 44.770703] arch_call_rest_init+0xc/0x14 [ 44.774841] start_kernel+0x408/0x424 [ 44.778623] Code: aa1403e0 97fff54f aa0003f5 b5fff500 (f9400275) [ 44.784907] ---[ end trace be73c3142d8c36a9 ]--- [ 44.789668] Kernel panic - not syncing: Oops: Fatal exception in interrupt
Fixes: 0c1ce9884607 ("mt76: mt7921: add wifi reset support") Signed-off-by: Lorenzo Bianconi lorenzo@kernel.org Signed-off-by: Felix Fietkau nbd@nbd.name Signed-off-by: Sasha Levin sashal@kernel.org --- drivers/net/wireless/mediatek/mt76/mt7921/mac.c | 3 +++ 1 file changed, 3 insertions(+)
diff --git a/drivers/net/wireless/mediatek/mt76/mt7921/mac.c b/drivers/net/wireless/mediatek/mt76/mt7921/mac.c index 4eac7c3206f9..5e00dd589331 100644 --- a/drivers/net/wireless/mediatek/mt76/mt7921/mac.c +++ b/drivers/net/wireless/mediatek/mt76/mt7921/mac.c @@ -1308,6 +1308,9 @@ void mt7921_reset(struct mt76_dev *mdev) { struct mt7921_dev *dev = container_of(mdev, struct mt7921_dev, mt76);
+ if (!test_bit(MT76_STATE_RUNNING, &dev->mphy.state)) + return; + if (!dev->hw_full_reset) queue_work(dev->mt76.wq, &dev->reset_work); }