Re: [PATCH] drivers: of: Export of_reserved_mem_device_{init,release}

[George G. Davis]

On Fri, Jan 09, 2015 at 01:33:10PM -0500, George G. Davis wrote:
> On Fri, Jan 09, 2015 at 11:53:01AM -0600, Rob Herring wrote:
> > On Fri, Jan 9, 2015 at 8:29 AM, George G. Davis <ggdavisiv@xxxxxxxxx> wrote:
> > > Export of_reserved_mem_device_{init,release} so that modules
> > > can initialize and release their assigned per-device cma_area.
> > 
> > I believe the original intent was for the core bus code to call these
> > functions. While the commit adding them says "automated assignment"
> > that part did not go in.
> 
> I suspected that there may have been a bit of missing magic.  The only
> use of these functions that I was able to find was in an earlier out of
> tree patch series where these functions were called from the s5p-mfc
> driver:
> 
> http://lists.linaro.org/pipermail/linaro-mm-sig/2014-August/004140.html
> 
> I agree that "automated assignment" via core bus code would be a better
> solution but I don't know where to add the calls.

How about something like the following for automated assignment?:

diff --git a/drivers/base/platform.c b/drivers/base/platform.c
index 9421fed..231b427 100644
--- a/drivers/base/platform.c
+++ b/drivers/base/platform.c
@@ -14,6 +14,7 @@
 #include <linux/platform_device.h>
 #include <linux/of_device.h>
 #include <linux/of_irq.h>
+#include <linux/of_reserved_mem.h>
 #include <linux/module.h>
 #include <linux/init.h>
 #include <linux/dma-mapping.h>
@@ -509,9 +510,12 @@ static int platform_drv_probe(struct device *_dev)
 
 	ret = dev_pm_domain_attach(_dev, true);
 	if (ret != -EPROBE_DEFER) {
+		of_reserved_mem_device_init(_dev);
 		ret = drv->probe(dev);
-		if (ret)
+		if (ret) {
+			of_reserved_mem_device_release(_dev);
 			dev_pm_domain_detach(_dev, true);
+		}
 	}
 
 	if (drv->prevent_deferred_probe && ret == -EPROBE_DEFER) {
@@ -534,6 +538,7 @@ static int platform_drv_remove(struct device *_dev)
 	int ret;
 
 	ret = drv->remove(dev);
+	of_reserved_mem_device_release(_dev);
 	dev_pm_domain_detach(_dev, true);
 
 	return ret;


--
Regards,
George
> 
> > As it stands, there are no in tree users of
> > these functions.
> > 
> > Marek, comments?
> 
> Thanks!
> 
> --
> Regards,
> George
> 
> > 
> > Rob
> > 
> > >
> > > Signed-off-by: George G. Davis <george_davis@xxxxxxxxxx>
> > > ---
> > >  drivers/of/of_reserved_mem.c | 2 ++
> > >  1 file changed, 2 insertions(+)
> > >
> > > diff --git a/drivers/of/of_reserved_mem.c b/drivers/of/of_reserved_mem.c
> > > index dc566b3..5eeb5b2 100644
> > > --- a/drivers/of/of_reserved_mem.c
> > > +++ b/drivers/of/of_reserved_mem.c
> > > @@ -265,6 +265,7 @@ int of_reserved_mem_device_init(struct device *dev)
> > >
> > >         return ret;
> > >  }
> > > +EXPORT_SYMBOL(of_reserved_mem_device_init);
> > >
> > >  /**
> > >   * of_reserved_mem_device_release() - release reserved memory device structures
> > > @@ -289,3 +290,4 @@ void of_reserved_mem_device_release(struct device *dev)
> > >
> > >         rmem->ops->device_release(rmem, dev);
> > >  }
> > > +EXPORT_SYMBOL(of_reserved_mem_device_release);
> > > --
> > > 1.9.3
> > >
--
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/