diff mbox series

[1/2] mailbox: Add power_get/power_put API to mbox_chan_ops

Message ID 20240614040133.24967-2-jason-jh.lin@mediatek.com (mailing list archive)
State New
Headers show
Series Fix sleeping function called from invalid context | expand

Commit Message

Jason-JH Lin (林睿祥) June 14, 2024, 4:01 a.m. UTC 
To avoid pm_runtime APIs calling sleep in all the atomic context APIs
in mbox_chan_ops, we add power_get/power_put API to let the controllers
implement them with pm_runtime APIs outside the other atomic context APIs
in mbox_chan_ops.

Signed-off-by: Jason-JH.Lin <jason-jh.lin@mediatek.com>
---
 drivers/mailbox/mailbox.c          | 55 ++++++++++++++++++++++++++++++
 include/linux/mailbox_controller.h | 11 ++++++
 2 files changed, 66 insertions(+)

Comments

Markus Elfring June 14, 2024, 11:35 a.m. UTC | #1 
> To avoid pm_runtime APIs calling sleep in all the atomic context APIs

           PM-runtime?


> in mbox_chan_ops, we add power_get/power_put API to let the controllers
> implement them with pm_runtime APIs outside the other atomic context APIs
> in mbox_chan_ops.

* How do you think about to omit the word “we” from such a change description?
  https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/process/submitting-patches.rst?h=v6.10-rc3#n94

* Would you like to improve any indications for adjustments of data structures?


> Signed-off-by: Jason-JH.Lin <jason-jh.lin@mediatek.com>

Does a dot really belong to the personal name
(according to the Developer's Certificate of Origin)?
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/process/submitting-patches.rst?h=v6.10-rc3#n438

Regards,
Markus
kernel test robot June 14, 2024, 4:23 p.m. UTC | #2 
Hi Jason-JH.Lin,

kernel test robot noticed the following build errors:

[auto build test ERROR on linus/master]
[also build test ERROR on v6.10-rc3 next-20240613]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]

url:    https://github.com/intel-lab-lkp/linux/commits/Jason-JH-Lin/mailbox-Add-power_get-power_put-API-to-mbox_chan_ops/20240614-120412
base:   linus/master
patch link:    https://lore.kernel.org/r/20240614040133.24967-2-jason-jh.lin%40mediatek.com
patch subject: [PATCH 1/2] mailbox: Add power_get/power_put API to mbox_chan_ops
config: riscv-defconfig (https://download.01.org/0day-ci/archive/20240615/202406150022.LXLvvkeI-lkp@intel.com/config)
compiler: clang version 19.0.0git (https://github.com/llvm/llvm-project 78ee473784e5ef6f0b19ce4cb111fb6e4d23c6b2)
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20240615/202406150022.LXLvvkeI-lkp@intel.com/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202406150022.LXLvvkeI-lkp@intel.com/

All errors (new ones prefixed by >>):

   In file included from drivers/mailbox/mailbox.c:9:
   In file included from include/linux/interrupt.h:21:
   In file included from arch/riscv/include/asm/sections.h:9:
   In file included from include/linux/mm.h:2253:
   include/linux/vmstat.h:514:36: warning: arithmetic between different enumeration types ('enum node_stat_item' and 'enum lru_list') [-Wenum-enum-conversion]
     514 |         return node_stat_name(NR_LRU_BASE + lru) + 3; // skip "nr_"
         |                               ~~~~~~~~~~~ ^ ~~~
>> drivers/mailbox/mailbox.c:382:11: error: incompatible pointer to integer conversion returning 'void *' from a function with result type 'int' [-Wint-conversion]
     382 |                         return ERR_PTR(ret);
         |                                ^~~~~~~~~~~~
   1 warning and 1 error generated.


vim +382 drivers/mailbox/mailbox.c

   355	
   356	static int __mbox_bind_client(struct mbox_chan *chan, struct mbox_client *cl)
   357	{
   358		struct device *dev = cl->dev;
   359		unsigned long flags;
   360		int ret;
   361	
   362		if (chan->cl || !try_module_get(chan->mbox->dev->driver->owner)) {
   363			dev_dbg(dev, "%s: mailbox not free\n", __func__);
   364			return -EBUSY;
   365		}
   366	
   367		spin_lock_irqsave(&chan->lock, flags);
   368		chan->msg_free = 0;
   369		chan->msg_count = 0;
   370		chan->active_req = NULL;
   371		chan->cl = cl;
   372		init_completion(&chan->tx_complete);
   373	
   374		if (chan->txdone_method	== TXDONE_BY_POLL && cl->knows_txdone)
   375			chan->txdone_method = TXDONE_BY_ACK;
   376	
   377		spin_unlock_irqrestore(&chan->lock, flags);
   378	
   379		if (chan->mbox->ops->power_get) {
   380			ret = chan->mbox->ops->power_get(chan);
   381			if (ret < 0)
 > 382				return ERR_PTR(ret);
   383		}
   384	
   385		if (chan->mbox->ops->startup) {
   386			ret = chan->mbox->ops->startup(chan);
   387	
   388			if (ret) {
   389				dev_err(dev, "Unable to startup the chan (%d)\n", ret);
   390				mbox_free_channel(chan);
   391				return ret;
   392			}
   393		}
   394	
   395		return 0;
   396	}
   397
kernel test robot June 14, 2024, 4:55 p.m. UTC | #3 
Hi Jason-JH.Lin,

kernel test robot noticed the following build warnings:

[auto build test WARNING on linus/master]
[also build test WARNING on fujitsu-integration/mailbox-for-next v6.10-rc3 next-20240613]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]

url:    https://github.com/intel-lab-lkp/linux/commits/Jason-JH-Lin/mailbox-Add-power_get-power_put-API-to-mbox_chan_ops/20240614-120412
base:   linus/master
patch link:    https://lore.kernel.org/r/20240614040133.24967-2-jason-jh.lin%40mediatek.com
patch subject: [PATCH 1/2] mailbox: Add power_get/power_put API to mbox_chan_ops
config: arc-randconfig-002-20240614 (https://download.01.org/0day-ci/archive/20240615/202406150041.QTui5Xgq-lkp@intel.com/config)
compiler: arceb-elf-gcc (GCC) 13.2.0
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20240615/202406150041.QTui5Xgq-lkp@intel.com/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202406150041.QTui5Xgq-lkp@intel.com/

All warnings (new ones prefixed by >>):

   drivers/mailbox/mailbox.c: In function '__mbox_bind_client':
>> drivers/mailbox/mailbox.c:382:32: warning: returning 'void *' from a function with return type 'int' makes integer from pointer without a cast [-Wint-conversion]
     382 |                         return ERR_PTR(ret);
         |                                ^~~~~~~~~~~~

Kconfig warnings: (for reference only)
   WARNING: unmet direct dependencies detected for REGMAP_SPI
   Depends on [n]: SPI [=n]
   Selected by [y]:
   - AD9739A [=y] && IIO [=y] && (SPI [=n] || COMPILE_TEST [=y])


vim +382 drivers/mailbox/mailbox.c

   355	
   356	static int __mbox_bind_client(struct mbox_chan *chan, struct mbox_client *cl)
   357	{
   358		struct device *dev = cl->dev;
   359		unsigned long flags;
   360		int ret;
   361	
   362		if (chan->cl || !try_module_get(chan->mbox->dev->driver->owner)) {
   363			dev_dbg(dev, "%s: mailbox not free\n", __func__);
   364			return -EBUSY;
   365		}
   366	
   367		spin_lock_irqsave(&chan->lock, flags);
   368		chan->msg_free = 0;
   369		chan->msg_count = 0;
   370		chan->active_req = NULL;
   371		chan->cl = cl;
   372		init_completion(&chan->tx_complete);
   373	
   374		if (chan->txdone_method	== TXDONE_BY_POLL && cl->knows_txdone)
   375			chan->txdone_method = TXDONE_BY_ACK;
   376	
   377		spin_unlock_irqrestore(&chan->lock, flags);
   378	
   379		if (chan->mbox->ops->power_get) {
   380			ret = chan->mbox->ops->power_get(chan);
   381			if (ret < 0)
 > 382				return ERR_PTR(ret);
   383		}
   384	
   385		if (chan->mbox->ops->startup) {
   386			ret = chan->mbox->ops->startup(chan);
   387	
   388			if (ret) {
   389				dev_err(dev, "Unable to startup the chan (%d)\n", ret);
   390				mbox_free_channel(chan);
   391				return ret;
   392			}
   393		}
   394	
   395		return 0;
   396	}
   397
kernel test robot June 14, 2024, 6:51 p.m. UTC | #4 
Hi Jason-JH.Lin,

kernel test robot noticed the following build warnings:

[auto build test WARNING on linus/master]
[also build test WARNING on fujitsu-integration/mailbox-for-next v6.10-rc3 next-20240613]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]

url:    https://github.com/intel-lab-lkp/linux/commits/Jason-JH-Lin/mailbox-Add-power_get-power_put-API-to-mbox_chan_ops/20240614-120412
base:   linus/master
patch link:    https://lore.kernel.org/r/20240614040133.24967-2-jason-jh.lin%40mediatek.com
patch subject: [PATCH 1/2] mailbox: Add power_get/power_put API to mbox_chan_ops
config: i386-randconfig-062-20240614 (https://download.01.org/0day-ci/archive/20240615/202406150214.pVxsbllC-lkp@intel.com/config)
compiler: gcc-13 (Ubuntu 13.2.0-4ubuntu3) 13.2.0
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20240615/202406150214.pVxsbllC-lkp@intel.com/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202406150214.pVxsbllC-lkp@intel.com/

sparse warnings: (new ones prefixed by >>)
>> drivers/mailbox/mailbox.c:382:39: sparse: sparse: incorrect type in return expression (different base types) @@     expected int @@     got void * @@
   drivers/mailbox/mailbox.c:382:39: sparse:     expected int
   drivers/mailbox/mailbox.c:382:39: sparse:     got void *
   drivers/mailbox/mailbox.c: note: in included file (through include/linux/smp.h, include/linux/alloc_tag.h, include/linux/percpu.h, ...):
   include/linux/list.h:83:21: sparse: sparse: self-comparison always evaluates to true

vim +382 drivers/mailbox/mailbox.c

   355	
   356	static int __mbox_bind_client(struct mbox_chan *chan, struct mbox_client *cl)
   357	{
   358		struct device *dev = cl->dev;
   359		unsigned long flags;
   360		int ret;
   361	
   362		if (chan->cl || !try_module_get(chan->mbox->dev->driver->owner)) {
   363			dev_dbg(dev, "%s: mailbox not free\n", __func__);
   364			return -EBUSY;
   365		}
   366	
   367		spin_lock_irqsave(&chan->lock, flags);
   368		chan->msg_free = 0;
   369		chan->msg_count = 0;
   370		chan->active_req = NULL;
   371		chan->cl = cl;
   372		init_completion(&chan->tx_complete);
   373	
   374		if (chan->txdone_method	== TXDONE_BY_POLL && cl->knows_txdone)
   375			chan->txdone_method = TXDONE_BY_ACK;
   376	
   377		spin_unlock_irqrestore(&chan->lock, flags);
   378	
   379		if (chan->mbox->ops->power_get) {
   380			ret = chan->mbox->ops->power_get(chan);
   381			if (ret < 0)
 > 382				return ERR_PTR(ret);
   383		}
   384	
   385		if (chan->mbox->ops->startup) {
   386			ret = chan->mbox->ops->startup(chan);
   387	
   388			if (ret) {
   389				dev_err(dev, "Unable to startup the chan (%d)\n", ret);
   390				mbox_free_channel(chan);
   391				return ret;
   392			}
   393		}
   394	
   395		return 0;
   396	}
   397
Dan Carpenter June 17, 2024, 12:39 p.m. UTC | #5 
Hi Jason-JH.Lin,

kernel test robot noticed the following build warnings:

https://git-scm.com/docs/git-format-patch#_base_tree_information]

url:    https://github.com/intel-lab-lkp/linux/commits/Jason-JH-Lin/mailbox-Add-power_get-power_put-API-to-mbox_chan_ops/20240614-120412
base:   linus/master
patch link:    https://lore.kernel.org/r/20240614040133.24967-2-jason-jh.lin%40mediatek.com
patch subject: [PATCH 1/2] mailbox: Add power_get/power_put API to mbox_chan_ops
config: x86_64-randconfig-161-20240615 (https://download.01.org/0day-ci/archive/20240615/202406151520.TDnOFFbV-lkp@intel.com/config)
compiler: gcc-13 (Ubuntu 13.2.0-4ubuntu3) 13.2.0

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Reported-by: Dan Carpenter <dan.carpenter@linaro.org>
| Closes: https://lore.kernel.org/r/202406151520.TDnOFFbV-lkp@intel.com/

smatch warnings:
drivers/mailbox/mailbox.c:485 mbox_request_channel() error: 'chan' dereferencing possible ERR_PTR()

vim +/chan +485 drivers/mailbox/mailbox.c

2b6d83e2b8b7de Jassi Brar      2014-06-12  444  struct mbox_chan *mbox_request_channel(struct mbox_client *cl, int index)
2b6d83e2b8b7de Jassi Brar      2014-06-12  445  {
2b6d83e2b8b7de Jassi Brar      2014-06-12  446  	struct device *dev = cl->dev;
2b6d83e2b8b7de Jassi Brar      2014-06-12  447  	struct mbox_controller *mbox;
2b6d83e2b8b7de Jassi Brar      2014-06-12  448  	struct of_phandle_args spec;
2b6d83e2b8b7de Jassi Brar      2014-06-12  449  	struct mbox_chan *chan;
2b6d83e2b8b7de Jassi Brar      2014-06-12  450  	int ret;
2b6d83e2b8b7de Jassi Brar      2014-06-12  451  
2b6d83e2b8b7de Jassi Brar      2014-06-12  452  	if (!dev || !dev->of_node) {
2b6d83e2b8b7de Jassi Brar      2014-06-12  453  		pr_debug("%s: No owner device node\n", __func__);
2b6d83e2b8b7de Jassi Brar      2014-06-12  454  		return ERR_PTR(-ENODEV);
2b6d83e2b8b7de Jassi Brar      2014-06-12  455  	}
2b6d83e2b8b7de Jassi Brar      2014-06-12  456  
2b6d83e2b8b7de Jassi Brar      2014-06-12  457  	mutex_lock(&con_mutex);
2b6d83e2b8b7de Jassi Brar      2014-06-12  458  
2b6d83e2b8b7de Jassi Brar      2014-06-12  459  	if (of_parse_phandle_with_args(dev->of_node, "mboxes",
2b6d83e2b8b7de Jassi Brar      2014-06-12  460  				       "#mbox-cells", index, &spec)) {
b6d83e2b8b7de Jassi Brar      2014-06-12  461  		dev_dbg(dev, "%s: can't parse \"mboxes\" property\n", __func__);
2b6d83e2b8b7de Jassi Brar      2014-06-12  462  		mutex_unlock(&con_mutex);
2b6d83e2b8b7de Jassi Brar      2014-06-12  463  		return ERR_PTR(-ENODEV);
2b6d83e2b8b7de Jassi Brar      2014-06-12  464  	}
2b6d83e2b8b7de Jassi Brar      2014-06-12  465  
2d805fc1c6b2ec Benson Leung    2015-05-04  466  	chan = ERR_PTR(-EPROBE_DEFER);
2b6d83e2b8b7de Jassi Brar      2014-06-12  467  	list_for_each_entry(mbox, &mbox_cons, node)
2b6d83e2b8b7de Jassi Brar      2014-06-12  468  		if (mbox->dev->of_node == spec.np) {
2b6d83e2b8b7de Jassi Brar      2014-06-12  469  			chan = mbox->of_xlate(mbox, &spec);
8ed82e23875e60 Mikko Perttunen 2018-11-28  470  			if (!IS_ERR(chan))
2b6d83e2b8b7de Jassi Brar      2014-06-12  471  				break;
2b6d83e2b8b7de Jassi Brar      2014-06-12  472  		}
2b6d83e2b8b7de Jassi Brar      2014-06-12  473  
2b6d83e2b8b7de Jassi Brar      2014-06-12  474  	of_node_put(spec.np);
2b6d83e2b8b7de Jassi Brar      2014-06-12  475  
2d805fc1c6b2ec Benson Leung    2015-05-04  476  	if (IS_ERR(chan)) {
2d805fc1c6b2ec Benson Leung    2015-05-04  477  		mutex_unlock(&con_mutex);
2d805fc1c6b2ec Benson Leung    2015-05-04  478  		return chan;
2d805fc1c6b2ec Benson Leung    2015-05-04  479  	}
2d805fc1c6b2ec Benson Leung    2015-05-04  480  
85a953806557db Elliot Berman   2023-04-10  481  	ret = __mbox_bind_client(chan, cl);
85a953806557db Elliot Berman   2023-04-10  482  	if (ret)
2b6d83e2b8b7de Jassi Brar      2014-06-12  483  		chan = ERR_PTR(ret);

change is set to error pointer

2b6d83e2b8b7de Jassi Brar      2014-06-12  484  
9c57f48cd805b1 Jason-JH.Lin    2024-06-14 @485  	if (chan->mbox->ops->power_put)
                                                            ^^^^^^
Dereferenced on the next line

9c57f48cd805b1 Jason-JH.Lin    2024-06-14  486  		chan->mbox->ops->power_put(chan);
9c57f48cd805b1 Jason-JH.Lin    2024-06-14  487  
2b6d83e2b8b7de Jassi Brar      2014-06-12  488  	mutex_unlock(&con_mutex);
9c57f48cd805b1 Jason-JH.Lin    2024-06-14  489  
2b6d83e2b8b7de Jassi Brar      2014-06-12  490  	return chan;
2b6d83e2b8b7de Jassi Brar      2014-06-12  491  }
Jason-JH Lin (林睿祥) June 18, 2024, 6:26 a.m. UTC | #6 
On Fri, 2024-06-14 at 13:35 +0200, Markus Elfring wrote:
>  	 
> External email : Please do not click links or open attachments until
> you have verified the sender or the content.
>  > To avoid pm_runtime APIs calling sleep in all the atomic context
> APIs
> 
>            PM-runtime?
> 
OK, I'll change this.

> 
> > in mbox_chan_ops, we add power_get/power_put API to let the
> controllers
> > implement them with pm_runtime APIs outside the other atomic
> context APIs
> > in mbox_chan_ops.
> 
> * How do you think about to omit the word “we” from such a change
> description?
>   
> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/process/submitting-patches.rst?h=v6.10-rc3#n94
> 

OK, I'll remove `we` and change the sentence below:
Add power_get/power_put API into mbox_chan_ops to let controllers
implement them with PM-runtime APIs outside the other atomic context
APIs in mbox_chan_ops to avoid calling sleep in the atomic context.


> * Would you like to improve any indications for adjustments of data
> structures?
> 

You mean the description of power_get and power_put in struct
mbox_chan_ops? Is this better?
@power_get: Called when the controller needs to increase a reference
count to keep power on for the device of mailbox controller. It's
optional to use PM-runtime API or more complex operations to implement
this function.

@power_put: Called when the controller needs to decrease a reference
count to release power for the device of mailbox controller. It's
optional to use PM-runtime API to more complex operations to implement
this function.

> 
> > Signed-off-by: Jason-JH.Lin <jason-jh.lin@mediatek.com>
> 
> Does a dot really belong to the personal name
> (according to the Developer's Certificate of Origin)?
> 
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/process/submitting-patches.rst?h=v6.10-rc3#n438
> 
A dot is not in my name, but I've been using it for a long time.
It's binding to my environment everywhere and it's not easy to change
in a short time. But I'll try to change it before sending upstream
patch. Thanks for your advice.

Regards
Jason-JH Lin


> Regards,
> Markus
diff mbox series

Patch

diff --git a/drivers/mailbox/mailbox.c b/drivers/mailbox/mailbox.c
index ebff3baf3045..bafcc7b0c0b8 100644
--- a/drivers/mailbox/mailbox.c
+++ b/drivers/mailbox/mailbox.c
@@ -58,6 +58,12 @@  static void msg_submit(struct mbox_chan *chan)
 	void *data;
 	int err = -EBUSY;
 
+	if (chan->mbox->ops->power_get) {
+		err = chan->mbox->ops->power_get(chan);
+		if (err < 0)
+			return;
+	}
+
 	spin_lock_irqsave(&chan->lock, flags);
 
 	if (!chan->msg_count || chan->active_req)
@@ -89,12 +95,16 @@  static void msg_submit(struct mbox_chan *chan)
 		hrtimer_start(&chan->mbox->poll_hrt, 0, HRTIMER_MODE_REL);
 		spin_unlock_irqrestore(&chan->mbox->poll_hrt_lock, flags);
 	}
+
+	if (chan->mbox->ops->power_put)
+		chan->mbox->ops->power_put(chan);
 }
 
 static void tx_tick(struct mbox_chan *chan, int r)
 {
 	unsigned long flags;
 	void *mssg;
+	int ret;
 
 	spin_lock_irqsave(&chan->lock, flags);
 	mssg = chan->active_req;
@@ -107,10 +117,19 @@  static void tx_tick(struct mbox_chan *chan, int r)
 	if (!mssg)
 		return;
 
+	if (chan->mbox->ops->power_get) {
+		ret = chan->mbox->ops->power_get(chan);
+		if (ret < 0)
+			return;
+	}
+
 	/* Notify the client */
 	if (chan->cl->tx_done)
 		chan->cl->tx_done(chan->cl, mssg, r);
 
+	if (chan->mbox->ops->power_put)
+		chan->mbox->ops->power_put(chan);
+
 	if (r != -ETIME && chan->cl->tx_block)
 		complete(&chan->tx_complete);
 }
@@ -223,9 +242,16 @@  EXPORT_SYMBOL_GPL(mbox_client_txdone);
  */
 bool mbox_client_peek_data(struct mbox_chan *chan)
 {
+	if (chan->mbox->ops->power_get)
+		if (chan->mbox->ops->power_get(chan) < 0)
+			return false;
+
 	if (chan->mbox->ops->peek_data)
 		return chan->mbox->ops->peek_data(chan);
 
+	if (chan->mbox->ops->power_put)
+		chan->mbox->ops->power_put(chan);
+
 	return false;
 }
 EXPORT_SYMBOL_GPL(mbox_client_peek_data);
@@ -310,10 +336,19 @@  int mbox_flush(struct mbox_chan *chan, unsigned long timeout)
 	if (!chan->mbox->ops->flush)
 		return -ENOTSUPP;
 
+	if (chan->mbox->ops->power_get) {
+		ret = chan->mbox->ops->power_get(chan);
+		if (ret < 0)
+			return ret;
+	}
+
 	ret = chan->mbox->ops->flush(chan, timeout);
 	if (ret < 0)
 		tx_tick(chan, ret);
 
+	if (chan->mbox->ops->power_put)
+		chan->mbox->ops->power_put(chan);
+
 	return ret;
 }
 EXPORT_SYMBOL_GPL(mbox_flush);
@@ -341,6 +376,12 @@  static int __mbox_bind_client(struct mbox_chan *chan, struct mbox_client *cl)
 
 	spin_unlock_irqrestore(&chan->lock, flags);
 
+	if (chan->mbox->ops->power_get) {
+		ret = chan->mbox->ops->power_get(chan);
+		if (ret < 0)
+			return ERR_PTR(ret);
+	}
+
 	if (chan->mbox->ops->startup) {
 		ret = chan->mbox->ops->startup(chan);
 
@@ -441,7 +482,11 @@  struct mbox_chan *mbox_request_channel(struct mbox_client *cl, int index)
 	if (ret)
 		chan = ERR_PTR(ret);
 
+	if (chan->mbox->ops->power_put)
+		chan->mbox->ops->power_put(chan);
+
 	mutex_unlock(&con_mutex);
+
 	return chan;
 }
 EXPORT_SYMBOL_GPL(mbox_request_channel);
@@ -485,13 +530,23 @@  EXPORT_SYMBOL_GPL(mbox_request_channel_byname);
 void mbox_free_channel(struct mbox_chan *chan)
 {
 	unsigned long flags;
+	int ret;
 
 	if (!chan || !chan->cl)
 		return;
 
+	if (chan->mbox->ops->power_get) {
+		ret = chan->mbox->ops->power_get(chan);
+		if (ret < 0)
+			return;
+	}
+
 	if (chan->mbox->ops->shutdown)
 		chan->mbox->ops->shutdown(chan);
 
+	if (chan->mbox->ops->power_put)
+		chan->mbox->ops->power_put(chan);
+
 	/* The queued TX requests are simply aborted, no callbacks are made */
 	spin_lock_irqsave(&chan->lock, flags);
 	chan->cl = NULL;
diff --git a/include/linux/mailbox_controller.h b/include/linux/mailbox_controller.h
index 6fee33cb52f5..e8f26e7dabfd 100644
--- a/include/linux/mailbox_controller.h
+++ b/include/linux/mailbox_controller.h
@@ -42,6 +42,15 @@  struct mbox_chan;
  *		  Used only if txdone_poll:=true && txdone_irq:=false
  * @peek_data: Atomic check for any received data. Return true if controller
  *		  has some data to push to the client. False otherwise.
+ * @power_get:	Called when the controller need to get the reference to keep
+ *		the power on for the device of mailbox controller. It is
+ *		optional to implement this function with pm_runtime APIs or
+ *		more complicated operation.
+ *		Return 0 is success, otherwise are fail.
+ * @power_put:	Called when the controller need to put the reference to release
+ *		the power for the device of mailbox controller. It is optional
+ *		to implement this function with pm_runtime APIs or more
+ *		complicated operation.
  */
 struct mbox_chan_ops {
 	int (*send_data)(struct mbox_chan *chan, void *data);
@@ -50,6 +59,8 @@  struct mbox_chan_ops {
 	void (*shutdown)(struct mbox_chan *chan);
 	bool (*last_tx_done)(struct mbox_chan *chan);
 	bool (*peek_data)(struct mbox_chan *chan);
+	int (*power_get)(struct mbox_chan *chan);
+	void (*power_put)(struct mbox_chan *chan);
 };
 
 /**