public inbox for linux-arm-msm@vger.kernel.org
 help / color / mirror / Atom feed
* [PATCH v4 1/3, RESEND] media: camss: vfe: Use trace_printk for debugging only
@ 2020-08-20  9:14 Nicolas Boichat
  2020-08-20 14:21 ` Steven Rostedt
  0 siblings, 1 reply; 2+ messages in thread
From: Nicolas Boichat @ 2020-08-20  9:14 UTC (permalink / raw)
  To: Mauro Carvalho Chehab, Steven Rostedt, Greg Kroah-Hartman
  Cc: Nicolas Boichat, Andy Gross, Bjorn Andersson, Todor Tomov,
	linux-arm-msm, linux-kernel, linux-media

trace_printk should not be used in production code. Since
tracing interrupts is presumably latency sensitive, pr_dbg is
not appropriate, so guard the call with a preprocessor symbol
that can be defined for debugging purpose.

Signed-off-by: Nicolas Boichat <drinkcat@chromium.org>
---

(resending this patch as part of the whole series, since we need a new
patch 3/3 now).

 drivers/media/platform/qcom/camss/camss-vfe-4-1.c | 2 ++
 drivers/media/platform/qcom/camss/camss-vfe-4-7.c | 2 ++
 2 files changed, 4 insertions(+)

diff --git a/drivers/media/platform/qcom/camss/camss-vfe-4-1.c b/drivers/media/platform/qcom/camss/camss-vfe-4-1.c
index 174a36be6f5d866..0c57171fae4f9e9 100644
--- a/drivers/media/platform/qcom/camss/camss-vfe-4-1.c
+++ b/drivers/media/platform/qcom/camss/camss-vfe-4-1.c
@@ -936,8 +936,10 @@ static irqreturn_t vfe_isr(int irq, void *dev)
 
 	vfe->ops->isr_read(vfe, &value0, &value1);
 
+#ifdef CAMSS_VFE_TRACE_IRQ
 	trace_printk("VFE: status0 = 0x%08x, status1 = 0x%08x\n",
 		     value0, value1);
+#endif
 
 	if (value0 & VFE_0_IRQ_STATUS_0_RESET_ACK)
 		vfe->isr_ops.reset_ack(vfe);
diff --git a/drivers/media/platform/qcom/camss/camss-vfe-4-7.c b/drivers/media/platform/qcom/camss/camss-vfe-4-7.c
index 0dca8bf9281e774..307675925e5c779 100644
--- a/drivers/media/platform/qcom/camss/camss-vfe-4-7.c
+++ b/drivers/media/platform/qcom/camss/camss-vfe-4-7.c
@@ -1058,8 +1058,10 @@ static irqreturn_t vfe_isr(int irq, void *dev)
 
 	vfe->ops->isr_read(vfe, &value0, &value1);
 
+#ifdef CAMSS_VFE_TRACE_IRQ
 	trace_printk("VFE: status0 = 0x%08x, status1 = 0x%08x\n",
 		     value0, value1);
+#endif
 
 	if (value0 & VFE_0_IRQ_STATUS_0_RESET_ACK)
 		vfe->isr_ops.reset_ack(vfe);
-- 
2.28.0.220.ged08abb693-goog


^ permalink raw reply related	[flat|nested] 2+ messages in thread

* Re: [PATCH v4 1/3, RESEND] media: camss: vfe: Use trace_printk for debugging only
  2020-08-20  9:14 [PATCH v4 1/3, RESEND] media: camss: vfe: Use trace_printk for debugging only Nicolas Boichat
@ 2020-08-20 14:21 ` Steven Rostedt
  0 siblings, 0 replies; 2+ messages in thread
From: Steven Rostedt @ 2020-08-20 14:21 UTC (permalink / raw)
  To: Nicolas Boichat
  Cc: Mauro Carvalho Chehab, Greg Kroah-Hartman, Andy Gross,
	Bjorn Andersson, Todor Tomov, linux-arm-msm, linux-kernel,
	linux-media

On Thu, 20 Aug 2020 17:14:10 +0800
Nicolas Boichat <drinkcat@chromium.org> wrote:

> trace_printk should not be used in production code. Since
> tracing interrupts is presumably latency sensitive, pr_dbg is
> not appropriate, so guard the call with a preprocessor symbol
> that can be defined for debugging purpose.
> 
> Signed-off-by: Nicolas Boichat <drinkcat@chromium.org>
> ---
> 
> (resending this patch as part of the whole series, since we need a new
> patch 3/3 now).
> 
>  drivers/media/platform/qcom/camss/camss-vfe-4-1.c | 2 ++
>  drivers/media/platform/qcom/camss/camss-vfe-4-7.c | 2 ++
>  2 files changed, 4 insertions(+)
> 
> diff --git a/drivers/media/platform/qcom/camss/camss-vfe-4-1.c b/drivers/media/platform/qcom/camss/camss-vfe-4-1.c
> index 174a36be6f5d866..0c57171fae4f9e9 100644
> --- a/drivers/media/platform/qcom/camss/camss-vfe-4-1.c
> +++ b/drivers/media/platform/qcom/camss/camss-vfe-4-1.c
> @@ -936,8 +936,10 @@ static irqreturn_t vfe_isr(int irq, void *dev)
>  
>  	vfe->ops->isr_read(vfe, &value0, &value1);
>  
> +#ifdef CAMSS_VFE_TRACE_IRQ
>  	trace_printk("VFE: status0 = 0x%08x, status1 = 0x%08x\n",
>  		     value0, value1);

Why are these not trace events?

The reason I have that ugly banner is to keep people from doing EXACTLY THIS!

trace_printk() is really easy to add, but it also is not configurable.
When a trace_printk() is in the code, it is always enabled (well, you
can turn trace_printk off, but that's an all or nothing. That is, by
turning trace_printk off, you turn off *all* trace_printks).

Instead, people should add trace events. This here is a perfect place
to have a trace event. You don't need to add #ifdef around trace events
because when not enabled, they are simply a nop. When enabled, the nop
is turned into a jump to the tracing code. It should not affect
performance. And as a trace event, you get a bunch of features with it
(filtering, histograms, etc).

-- Steve


> +#endif
>  
>  	if (value0 & VFE_0_IRQ_STATUS_0_RESET_ACK)
>  		vfe->isr_ops.reset_ack(vfe);
> diff --git a/drivers/media/platform/qcom/camss/camss-vfe-4-7.c b/drivers/media/platform/qcom/camss/camss-vfe-4-7.c
> index 0dca8bf9281e774..307675925e5c779 100644
> --- a/drivers/media/platform/qcom/camss/camss-vfe-4-7.c
> +++ b/drivers/media/platform/qcom/camss/camss-vfe-4-7.c
> @@ -1058,8 +1058,10 @@ static irqreturn_t vfe_isr(int irq, void *dev)
>  
>  	vfe->ops->isr_read(vfe, &value0, &value1);
>  
> +#ifdef CAMSS_VFE_TRACE_IRQ
>  	trace_printk("VFE: status0 = 0x%08x, status1 = 0x%08x\n",
>  		     value0, value1);
> +#endif
>  
>  	if (value0 & VFE_0_IRQ_STATUS_0_RESET_ACK)
>  		vfe->isr_ops.reset_ack(vfe);


^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2020-08-20 14:21 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2020-08-20  9:14 [PATCH v4 1/3, RESEND] media: camss: vfe: Use trace_printk for debugging only Nicolas Boichat
2020-08-20 14:21 ` Steven Rostedt

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox