Re: [PATCH v2] net: allwinner: emac: Fix double spinlock in emac_timeout

[Jernej Škrabec]

Dne ponedeljek, 12. september 2022 ob 08:33:31 CEST je qianfanguijin@xxxxxxx 
napisal(a):
> From: qianfan Zhao <qianfanguijin@xxxxxxx>
> 
> emac_timeout() callback acquires lock and so does emac_init_device(),
> which called inside lock protected region. This hangs the system and
> produces following warning:
> 
> WARNING: CPU: 2 PID: 0 at net/sched/sch_generic.c:478
> dev_watchdog+0x2e4/0x2e8
> NETDEV WATCHDOG: FE0 (sun4i-emac): transmit queue 0 timed out
> Modules linked in:
> CPU: 2 PID: 0 Comm: swapper/2 Tainted: G W 5.15.0-00047-g0848e4aeb313
> Hardware name: Wisdom T3 based CCT Family
> [<c010f740>] (unwind_backtrace) from [<c010b744>] (show_stack+0x10/0x14)
> [<c010b744>] (show_stack) from [<c0a4d978>] (dump_stack_lvl+0x40/0x4c)
> [<c0a4d978>] (dump_stack_lvl) from [<c0120734>] (__warn+0x104/0x108)
> [<c0120734>] (__warn) from [<c01207b0>] (warn_slowpath_fmt+0x78/0xbc)
> [<c01207b0>] (warn_slowpath_fmt) from [<c0898a54>]
> (dev_watchdog+0x2e4/0x2e8)
> [<c0898a54>] (dev_watchdog) from [<c019a888>] (call_timer_fn+0x3c/0x178)
> [<c019a888>] (call_timer_fn) from [<c019bddc>]
> (run_timer_softirq+0x540/0x624)
> [<c019bddc>] (run_timer_softirq) from [<c0101298>]
> (__do_softirq+0x130/0x3bc)
> [<c0101298>] (__do_softirq) from [<c0127ea0>] (irq_exit+0xbc/0x100)
> [<c0127ea0>] (irq_exit) from [<c017e514>] (handle_domain_irq+0x60/0x78)
> [<c017e514>] (handle_domain_irq) from [<c05a6130>]
> (gic_handle_irq+0x7c/0x90)
> [<c05a6130>] (gic_handle_irq) from [<c0100afc>] (__irq_svc+0x5c/0x78)
> Exception stack(0xc14f3f70 to 0xc14f3fb8)
> 3f60:                                     0003475c 00000000 00000001
> c01188a0
> 3f80: c107b200 c0f06b4c c0f06b90 00000004 c1079ff8 c0c62774 00000000
> 00000000
> 3fa0: c10b2198 c14f3fc0 c0107fcc c0107fd0 60030013 ffffffff
> [<c0100afc>] (__irq_svc) from [<c0107fd0>] (arch_cpu_idle+0x38/0x3c)
> [<c0107fd0>] (arch_cpu_idle) from [<c0a57f38>]
> (default_idle_call+0x3c/0xcc)
> [<c0a57f38>] (default_idle_call) from [<c0157258>] (do_idle+0xdc/0x13c)
> [<c0157258>] (do_idle) from [<c01575a4>] (cpu_startup_entry+0x18/0x1c)
> [<c01575a4>] (cpu_startup_entry) from [<401015d0>] (0x401015d0)
> ---[ end trace a70942a1265338f1 ]---
> rcu: INFO: rcu_sched detected stalls on CPUs/tasks:
> rcu: \x092-...0: (1 GPs behind) idle=75d/0/0x1 softirq=8288/8289 fqs=931
> \x09(detected by 0, t=2102 jiffies, g=13485, q=1635)
> Sending NMI from CPU 0 to CPUs 2:
> spi_master spi2: spi2.1: timeout transferring 4 bytes@100000Hz for
> 110(100)ms
> spidev spi2.1: SPI transfer failed: -110
> 
> Fixes: 492205050d77b (net: Add EMAC ethernet driver found on Allwinner A10
> SoC's) Signed-off-by: qianfan Zhao <qianfanguijin@xxxxxxx>

Reviewed-by: Jernej Skrabec <jernej.skrabec@xxxxxxxxx>

Best regards,
Jernej