linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH] driver core: make device_set_deferred_probe_reason a no-op when !CONFIG_DEBUG_FS
@ 2021-04-19 10:42 Rasmus Villemoes
  2021-04-19 11:03 ` Greg Kroah-Hartman
  2021-05-13 18:41 ` Greg Kroah-Hartman
  0 siblings, 2 replies; 3+ messages in thread
From: Rasmus Villemoes @ 2021-04-19 10:42 UTC (permalink / raw)
  To: Greg Kroah-Hartman, Rafael J. Wysocki
  Cc: Andrzej Hajda, Rasmus Villemoes, linux-kernel

When debugfs is not enabled, the deferred_probe_reason string is never
read. So there's no reason to spend time and memory on recording it.

There's still a bunch of redundant kfree(NULL) calls and NULL
assignments, but this gives most of the benefit (avoiding two
vsnprintf() and a kmalloc()) for the minimal amount of ifdeffery.

Signed-off-by: Rasmus Villemoes <linux@rasmusvillemoes.dk>
---
 drivers/base/dd.c | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/drivers/base/dd.c b/drivers/base/dd.c
index 37a5e5f8b221..6a197336c6a4 100644
--- a/drivers/base/dd.c
+++ b/drivers/base/dd.c
@@ -216,9 +216,13 @@ void device_unblock_probing(void)
  * device_set_deferred_probe_reason() - Set defer probe reason message for device
  * @dev: the pointer to the struct device
  * @vaf: the pointer to va_format structure with message
+ *
+ * The ->deferred_probe_reason string is only ever read when debugfs
+ * is enabled, so this is a no-op for !CONFIG_DEBUG_FS.
  */
 void device_set_deferred_probe_reason(const struct device *dev, struct va_format *vaf)
 {
+#ifdef CONFIG_DEBUG_FS
 	const char *drv = dev_driver_string(dev);
 
 	mutex_lock(&deferred_probe_mutex);
@@ -227,6 +231,7 @@ void device_set_deferred_probe_reason(const struct device *dev, struct va_format
 	dev->p->deferred_probe_reason = kasprintf(GFP_KERNEL, "%s: %pV", drv, vaf);
 
 	mutex_unlock(&deferred_probe_mutex);
+#endif
 }
 
 /*
-- 
2.29.2


^ permalink raw reply related	[flat|nested] 3+ messages in thread

* Re: [PATCH] driver core: make device_set_deferred_probe_reason a no-op when !CONFIG_DEBUG_FS
  2021-04-19 10:42 [PATCH] driver core: make device_set_deferred_probe_reason a no-op when !CONFIG_DEBUG_FS Rasmus Villemoes
@ 2021-04-19 11:03 ` Greg Kroah-Hartman
  2021-05-13 18:41 ` Greg Kroah-Hartman
  1 sibling, 0 replies; 3+ messages in thread
From: Greg Kroah-Hartman @ 2021-04-19 11:03 UTC (permalink / raw)
  To: Rasmus Villemoes; +Cc: Rafael J. Wysocki, Andrzej Hajda, linux-kernel

On Mon, Apr 19, 2021 at 12:42:56PM +0200, Rasmus Villemoes wrote:
> When debugfs is not enabled, the deferred_probe_reason string is never
> read. So there's no reason to spend time and memory on recording it.

How much time and memory is it?

> There's still a bunch of redundant kfree(NULL) calls and NULL
> assignments, but this gives most of the benefit (avoiding two
> vsnprintf() and a kmalloc()) for the minimal amount of ifdeffery.
> 
> Signed-off-by: Rasmus Villemoes <linux@rasmusvillemoes.dk>
> ---
>  drivers/base/dd.c | 5 +++++
>  1 file changed, 5 insertions(+)
> 
> diff --git a/drivers/base/dd.c b/drivers/base/dd.c
> index 37a5e5f8b221..6a197336c6a4 100644
> --- a/drivers/base/dd.c
> +++ b/drivers/base/dd.c
> @@ -216,9 +216,13 @@ void device_unblock_probing(void)
>   * device_set_deferred_probe_reason() - Set defer probe reason message for device
>   * @dev: the pointer to the struct device
>   * @vaf: the pointer to va_format structure with message
> + *
> + * The ->deferred_probe_reason string is only ever read when debugfs
> + * is enabled, so this is a no-op for !CONFIG_DEBUG_FS.
>   */
>  void device_set_deferred_probe_reason(const struct device *dev, struct va_format *vaf)
>  {
> +#ifdef CONFIG_DEBUG_FS
>  	const char *drv = dev_driver_string(dev);
>  
>  	mutex_lock(&deferred_probe_mutex);
> @@ -227,6 +231,7 @@ void device_set_deferred_probe_reason(const struct device *dev, struct va_format
>  	dev->p->deferred_probe_reason = kasprintf(GFP_KERNEL, "%s: %pV", drv, vaf);
>  
>  	mutex_unlock(&deferred_probe_mutex);
> +#endif

I really don't like #ifdef in .c files.  Any way to put this in the .h
file instead like should be done properly?

thanks,

greg k-h

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [PATCH] driver core: make device_set_deferred_probe_reason a no-op when !CONFIG_DEBUG_FS
  2021-04-19 10:42 [PATCH] driver core: make device_set_deferred_probe_reason a no-op when !CONFIG_DEBUG_FS Rasmus Villemoes
  2021-04-19 11:03 ` Greg Kroah-Hartman
@ 2021-05-13 18:41 ` Greg Kroah-Hartman
  1 sibling, 0 replies; 3+ messages in thread
From: Greg Kroah-Hartman @ 2021-05-13 18:41 UTC (permalink / raw)
  To: Rasmus Villemoes; +Cc: Rafael J. Wysocki, Andrzej Hajda, linux-kernel

On Mon, Apr 19, 2021 at 12:42:56PM +0200, Rasmus Villemoes wrote:
> When debugfs is not enabled, the deferred_probe_reason string is never
> read. So there's no reason to spend time and memory on recording it.
> 
> There's still a bunch of redundant kfree(NULL) calls and NULL
> assignments, but this gives most of the benefit (avoiding two
> vsnprintf() and a kmalloc()) for the minimal amount of ifdeffery.
> 
> Signed-off-by: Rasmus Villemoes <linux@rasmusvillemoes.dk>
> ---
>  drivers/base/dd.c | 5 +++++
>  1 file changed, 5 insertions(+)
> 
> diff --git a/drivers/base/dd.c b/drivers/base/dd.c
> index 37a5e5f8b221..6a197336c6a4 100644
> --- a/drivers/base/dd.c
> +++ b/drivers/base/dd.c
> @@ -216,9 +216,13 @@ void device_unblock_probing(void)
>   * device_set_deferred_probe_reason() - Set defer probe reason message for device
>   * @dev: the pointer to the struct device
>   * @vaf: the pointer to va_format structure with message
> + *
> + * The ->deferred_probe_reason string is only ever read when debugfs
> + * is enabled, so this is a no-op for !CONFIG_DEBUG_FS.
>   */
>  void device_set_deferred_probe_reason(const struct device *dev, struct va_format *vaf)
>  {
> +#ifdef CONFIG_DEBUG_FS
>  	const char *drv = dev_driver_string(dev);
>  
>  	mutex_lock(&deferred_probe_mutex);
> @@ -227,6 +231,7 @@ void device_set_deferred_probe_reason(const struct device *dev, struct va_format
>  	dev->p->deferred_probe_reason = kasprintf(GFP_KERNEL, "%s: %pV", drv, vaf);
>  
>  	mutex_unlock(&deferred_probe_mutex);
> +#endif

Can you move the #ifdef to a .h file and do this properly instead of
cutting up the function like this?

thanks,

greg k-h

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2021-05-13 18:42 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-04-19 10:42 [PATCH] driver core: make device_set_deferred_probe_reason a no-op when !CONFIG_DEBUG_FS Rasmus Villemoes
2021-04-19 11:03 ` Greg Kroah-Hartman
2021-05-13 18:41 ` Greg Kroah-Hartman

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).