Re: [PATCH v2] thermal: Intel SoC DTS: Don't do thermal zone update inside spin_lock

[Zhang Rui]

On Fri, 2014-11-28 at 10:11 +0800, Maurice Petallo wrote:
> The driver calls spin_lock_irqsave during DTS interrupt. The interrupt
> handle then calls thermal_zone_device_update which implicitly calls
> a sleep function and produce the following bug:
> 
> BUG: sleeping function called from invalid context at kernel/locking/mutex.c:97
> in_atomic(): 1, irqs_disabled(): 1, pid: 920, name: irq/86-soc_dts
> CPU: 0 PID: 920 Comm: irq/86-soc_dts Tainted: G            E  3.17.0-rc2+ #1
> Hardware name: Intel Corp. VALLEYVIEW B3 PLATFORM/NOTEBOOK, BIOS BYTICRB1.86C.0092.R31.1408290850 08/29/2014
>  00000000 00000000 c25dbe74 c1818cfd f3cc488c c25dbe9c c1059305 c1b4063b
>  00000001 00000001 00000398 f3cc488c f6817644 f6817644 f3ecc6c0 c25dbea8
>  c18208f2 f6817400 c25dbebc c159b0bb c25dbedc f6817400 f32a2300 c25dbee8
> Call Trace:
>  [<c1818cfd>] dump_stack+0x48/0x60
>  [<c1059305>] __might_sleep+0xec/0xf4
>  [<c18208f2>] mutex_lock+0x1c/0x34
>  [<c159b0bb>] thermal_zone_get_temp+0x34/0x59
>  [<c159bde5>] thermal_zone_device_update+0x2d/0xcb
>  [<f85da16a>] ? iosf_mbi_write+0x6c/0x74 [iosf_mbi]
>  [<f7c7445d>] soc_irq_thread_fn+0x10c/0x163 [intel_soc_dts_thermal]
>  [<c107b72b>] irq_thread_fn+0x18/0x2a
>  [<c107bedb>] irq_thread+0x81/0x11f
>  [<c107b713>] ? irq_finalize_oneshot+0x7c/0x7c
>  [<c107bf79>] ? irq_thread+0x11f/0x11f
>  [<c107be5a>] ? wake_threads_waitq+0x31/0x31
>  [<c1054217>] kthread+0x87/0x8c
>  [<c1821e41>] ret_from_kernel_thread+0x21/0x30
>  [<c1054190>] ? __kthread_parkme+0x55/0x55
> 
> Signed-off-by: Maurice Petallo <mauricex.r.petallo@xxxxxxxxx>
> Acked-by: Srinivas Pandruvada <srinivas.pandruvada@xxxxxxxxxxxxxxx>
> Acked-by: Eduardo Valentin <edubezval@xxxxxxxxx>
> CC: Kweh, Hock Leong <hock.leong.kweh@xxxxxxxxx>

applied.

thanks,
rui
> ---
> Changelog v2:
> * Change subject to emphasize that this patch affects only Intel SoC DTS
>   (intel_soc_dts_thermal.c)
> 
>  drivers/thermal/intel_soc_dts_thermal.c |   12 +++++++-----
>  1 file changed, 7 insertions(+), 5 deletions(-)
> 
> diff --git a/drivers/thermal/intel_soc_dts_thermal.c b/drivers/thermal/intel_soc_dts_thermal.c
> index a6a0a18..5580f5b 100644
> --- a/drivers/thermal/intel_soc_dts_thermal.c
> +++ b/drivers/thermal/intel_soc_dts_thermal.c
> @@ -360,6 +360,9 @@ static void proc_thermal_interrupt(void)
>  	u32 sticky_out;
>  	int status;
>  	u32 ptmc_out;
> +	unsigned long flags;
> +
> +	spin_lock_irqsave(&intr_notify_lock, flags);
>  
>  	/* Clear APIC interrupt */
>  	status = iosf_mbi_read(BT_MBI_UNIT_PMC, BT_MBI_BUNIT_READ,
> @@ -378,21 +381,20 @@ static void proc_thermal_interrupt(void)
>  		/* reset sticky bit */
>  		status = iosf_mbi_write(BT_MBI_UNIT_PMC, BT_MBI_BUNIT_WRITE,
>  					SOC_DTS_OFFSET_PTTSS, sticky_out);
> +		spin_unlock_irqrestore(&intr_notify_lock, flags);
> +
>  		for (i = 0; i < SOC_MAX_DTS_SENSORS; ++i) {
>  			pr_debug("TZD update for zone %d\n", i);
>  			thermal_zone_device_update(soc_dts[i]->tzone);
>  		}
> -	}
> +	} else
> +		spin_unlock_irqrestore(&intr_notify_lock, flags);
>  
>  }
>  
>  static irqreturn_t soc_irq_thread_fn(int irq, void *dev_data)
>  {
> -	unsigned long flags;
> -
> -	spin_lock_irqsave(&intr_notify_lock, flags);
>  	proc_thermal_interrupt();
> -	spin_unlock_irqrestore(&intr_notify_lock, flags);
>  	pr_debug("proc_thermal_interrupt\n");
>  
>  	return IRQ_HANDLED;


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/