devicetree.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH] of: dma_configure: Free old DMA map range
@ 2022-02-04  8:45 Mårten Lindahl
  2022-02-11 14:48 ` Rob Herring
  0 siblings, 1 reply; 3+ messages in thread
From: Mårten Lindahl @ 2022-02-04  8:45 UTC (permalink / raw)
  To: Rob Herring, Frank Rowand; +Cc: kernel, devicetree, Mårten Lindahl

When unbinding/binding a driver with DMA mapped memory, the DMA map is
not freed when the driver is reloaded. This leads to a memory leak when
the DMA map is overwritten when reprobing the driver.

This can be reproduced with a platform driver having a dma-range:

dummy {
	...
	#address-cells = <0x2>;
	#size-cells = <0x2>;
	ranges;
	dma-ranges = <...>;
	...
};

and then unbinding/binding it:

~# echo soc:dummy >/sys/bus/platform/drivers/<driver>/unbind

DMA map object 0xffffff800b0ae540 still being held by &pdev->dev

~# echo soc:dummy >/sys/bus/platform/drivers/<driver>/bind
~# echo scan > /sys/kernel/debug/kmemleak
~# cat /sys/kernel/debug/kmemleak
unreferenced object 0xffffff800b0ae540 (size 64):
  comm "sh", pid 833, jiffies 4295174550 (age 2535.352s)
  hex dump (first 32 bytes):
    00 00 00 80 00 00 00 00 00 00 00 00 00 00 00 00  ................
    00 00 00 80 00 00 00 00 00 00 00 80 00 00 00 00  ................
  backtrace:
    [<ffffffefd1694708>] create_object.isra.0+0x108/0x344
    [<ffffffefd1d1a850>] kmemleak_alloc+0x8c/0xd0
    [<ffffffefd167e2d0>] __kmalloc+0x440/0x6f0
    [<ffffffefd1a960a4>] of_dma_get_range+0x124/0x220
    [<ffffffefd1a8ce90>] of_dma_configure_id+0x40/0x2d0
    [<ffffffefd198b68c>] platform_dma_configure+0x5c/0xa4
    [<ffffffefd198846c>] really_probe+0x8c/0x514
    [<ffffffefd1988990>] __driver_probe_device+0x9c/0x19c
    [<ffffffefd1988cd8>] device_driver_attach+0x54/0xbc
    [<ffffffefd1986634>] bind_store+0xc4/0x120
    [<ffffffefd19856e0>] drv_attr_store+0x30/0x44
    [<ffffffefd173c9b0>] sysfs_kf_write+0x50/0x60
    [<ffffffefd173c1c4>] kernfs_fop_write_iter+0x124/0x1b4
    [<ffffffefd16a013c>] new_sync_write+0xdc/0x160
    [<ffffffefd16a256c>] vfs_write+0x23c/0x2a0
    [<ffffffefd16a2758>] ksys_write+0x64/0xec

Prevent overwriting the dma_range_map by freeing it before saving the
new map.

Signed-off-by: Mårten Lindahl <marten.lindahl@axis.com>
---
 drivers/of/device.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/drivers/of/device.c b/drivers/of/device.c
index 874f031442dc..a67703d09bfd 100644
--- a/drivers/of/device.c
+++ b/drivers/of/device.c
@@ -156,6 +156,12 @@ int of_dma_configure_id(struct device *dev, struct device_node *np,
 			kfree(map);
 			return -EINVAL;
 		}
+
+		/*
+		 * Since we are about to set a new range map we should make sure we
+		 * do not overwrite any existing one without having freed it first.
+		 */
+		kfree(dev->dma_range_map);
 	}
 
 	/*
-- 
2.30.2


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

* Re: [PATCH] of: dma_configure: Free old DMA map range
  2022-02-04  8:45 [PATCH] of: dma_configure: Free old DMA map range Mårten Lindahl
@ 2022-02-11 14:48 ` Rob Herring
  2022-02-11 21:37   ` Marten Lindahl
  0 siblings, 1 reply; 3+ messages in thread
From: Rob Herring @ 2022-02-11 14:48 UTC (permalink / raw)
  To: Mårten Lindahl; +Cc: Frank Rowand, kernel, devicetree

On Fri, Feb 04, 2022 at 09:45:56AM +0100, Mårten Lindahl wrote:
> When unbinding/binding a driver with DMA mapped memory, the DMA map is
> not freed when the driver is reloaded. This leads to a memory leak when
> the DMA map is overwritten when reprobing the driver.
> 
> This can be reproduced with a platform driver having a dma-range:
> 
> dummy {
> 	...
> 	#address-cells = <0x2>;
> 	#size-cells = <0x2>;
> 	ranges;
> 	dma-ranges = <...>;
> 	...
> };
> 
> and then unbinding/binding it:
> 
> ~# echo soc:dummy >/sys/bus/platform/drivers/<driver>/unbind
> 
> DMA map object 0xffffff800b0ae540 still being held by &pdev->dev
> 
> ~# echo soc:dummy >/sys/bus/platform/drivers/<driver>/bind
> ~# echo scan > /sys/kernel/debug/kmemleak
> ~# cat /sys/kernel/debug/kmemleak
> unreferenced object 0xffffff800b0ae540 (size 64):
>   comm "sh", pid 833, jiffies 4295174550 (age 2535.352s)
>   hex dump (first 32 bytes):
>     00 00 00 80 00 00 00 00 00 00 00 00 00 00 00 00  ................
>     00 00 00 80 00 00 00 00 00 00 00 80 00 00 00 00  ................
>   backtrace:
>     [<ffffffefd1694708>] create_object.isra.0+0x108/0x344
>     [<ffffffefd1d1a850>] kmemleak_alloc+0x8c/0xd0
>     [<ffffffefd167e2d0>] __kmalloc+0x440/0x6f0
>     [<ffffffefd1a960a4>] of_dma_get_range+0x124/0x220
>     [<ffffffefd1a8ce90>] of_dma_configure_id+0x40/0x2d0
>     [<ffffffefd198b68c>] platform_dma_configure+0x5c/0xa4
>     [<ffffffefd198846c>] really_probe+0x8c/0x514
>     [<ffffffefd1988990>] __driver_probe_device+0x9c/0x19c
>     [<ffffffefd1988cd8>] device_driver_attach+0x54/0xbc
>     [<ffffffefd1986634>] bind_store+0xc4/0x120
>     [<ffffffefd19856e0>] drv_attr_store+0x30/0x44
>     [<ffffffefd173c9b0>] sysfs_kf_write+0x50/0x60
>     [<ffffffefd173c1c4>] kernfs_fop_write_iter+0x124/0x1b4
>     [<ffffffefd16a013c>] new_sync_write+0xdc/0x160
>     [<ffffffefd16a256c>] vfs_write+0x23c/0x2a0
>     [<ffffffefd16a2758>] ksys_write+0x64/0xec
> 
> Prevent overwriting the dma_range_map by freeing it before saving the
> new map.
> 
> Signed-off-by: Mårten Lindahl <marten.lindahl@axis.com>
> ---
>  drivers/of/device.c | 6 ++++++
>  1 file changed, 6 insertions(+)
> 
> diff --git a/drivers/of/device.c b/drivers/of/device.c
> index 874f031442dc..a67703d09bfd 100644
> --- a/drivers/of/device.c
> +++ b/drivers/of/device.c
> @@ -156,6 +156,12 @@ int of_dma_configure_id(struct device *dev, struct device_node *np,
>  			kfree(map);
>  			return -EINVAL;
>  		}
> +
> +		/*
> +		 * Since we are about to set a new range map we should make sure we
> +		 * do not overwrite any existing one without having freed it first.
> +		 */
> +		kfree(dev->dma_range_map);


The map can't change, can't we just reuse the existing map. Or just bail 
early? I suppose the driver could have modified the dma masks. Or the 
IOMMU config could have changed.

Rob

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

* Re: [PATCH] of: dma_configure: Free old DMA map range
  2022-02-11 14:48 ` Rob Herring
@ 2022-02-11 21:37   ` Marten Lindahl
  0 siblings, 0 replies; 3+ messages in thread
From: Marten Lindahl @ 2022-02-11 21:37 UTC (permalink / raw)
  To: Rob Herring
  Cc: Mårten Lindahl, Frank Rowand, kernel,
	devicetree@vger.kernel.org

On Fri, Feb 11, 2022 at 03:48:06PM +0100, Rob Herring wrote:
> On Fri, Feb 04, 2022 at 09:45:56AM +0100, Mårten Lindahl wrote:
> > When unbinding/binding a driver with DMA mapped memory, the DMA map is
> > not freed when the driver is reloaded. This leads to a memory leak when
> > the DMA map is overwritten when reprobing the driver.
> > 
> > This can be reproduced with a platform driver having a dma-range:
> > 
> > dummy {
> > 	...
> > 	#address-cells = <0x2>;
> > 	#size-cells = <0x2>;
> > 	ranges;
> > 	dma-ranges = <...>;
> > 	...
> > };
> > 
> > and then unbinding/binding it:
> > 
> > ~# echo soc:dummy >/sys/bus/platform/drivers/<driver>/unbind
> > 
> > DMA map object 0xffffff800b0ae540 still being held by &pdev->dev
> > 
> > ~# echo soc:dummy >/sys/bus/platform/drivers/<driver>/bind
> > ~# echo scan > /sys/kernel/debug/kmemleak
> > ~# cat /sys/kernel/debug/kmemleak
> > unreferenced object 0xffffff800b0ae540 (size 64):
> >   comm "sh", pid 833, jiffies 4295174550 (age 2535.352s)
> >   hex dump (first 32 bytes):
> >     00 00 00 80 00 00 00 00 00 00 00 00 00 00 00 00  ................
> >     00 00 00 80 00 00 00 00 00 00 00 80 00 00 00 00  ................
> >   backtrace:
> >     [<ffffffefd1694708>] create_object.isra.0+0x108/0x344
> >     [<ffffffefd1d1a850>] kmemleak_alloc+0x8c/0xd0
> >     [<ffffffefd167e2d0>] __kmalloc+0x440/0x6f0
> >     [<ffffffefd1a960a4>] of_dma_get_range+0x124/0x220
> >     [<ffffffefd1a8ce90>] of_dma_configure_id+0x40/0x2d0
> >     [<ffffffefd198b68c>] platform_dma_configure+0x5c/0xa4
> >     [<ffffffefd198846c>] really_probe+0x8c/0x514
> >     [<ffffffefd1988990>] __driver_probe_device+0x9c/0x19c
> >     [<ffffffefd1988cd8>] device_driver_attach+0x54/0xbc
> >     [<ffffffefd1986634>] bind_store+0xc4/0x120
> >     [<ffffffefd19856e0>] drv_attr_store+0x30/0x44
> >     [<ffffffefd173c9b0>] sysfs_kf_write+0x50/0x60
> >     [<ffffffefd173c1c4>] kernfs_fop_write_iter+0x124/0x1b4
> >     [<ffffffefd16a013c>] new_sync_write+0xdc/0x160
> >     [<ffffffefd16a256c>] vfs_write+0x23c/0x2a0
> >     [<ffffffefd16a2758>] ksys_write+0x64/0xec
> > 
> > Prevent overwriting the dma_range_map by freeing it before saving the
> > new map.
> > 
> > Signed-off-by: Mårten Lindahl <marten.lindahl@axis.com>
> > ---
> >  drivers/of/device.c | 6 ++++++
> >  1 file changed, 6 insertions(+)
> > 
> > diff --git a/drivers/of/device.c b/drivers/of/device.c
> > index 874f031442dc..a67703d09bfd 100644
> > --- a/drivers/of/device.c
> > +++ b/drivers/of/device.c
> > @@ -156,6 +156,12 @@ int of_dma_configure_id(struct device *dev, struct device_node *np,
> >  			kfree(map);
> >  			return -EINVAL;
> >  		}
> > +
> > +		/*
> > +		 * Since we are about to set a new range map we should make sure we
> > +		 * do not overwrite any existing one without having freed it first.
> > +		 */
> > +		kfree(dev->dma_range_map);
>
Hi Rob!
Thanks for looking at this.
> 
> The map can't change, can't we just reuse the existing map. Or just bail 
> early? I suppose the driver could have modified the dma masks. Or the 
> IOMMU config could have changed.

Yes, we can reuse the existing map. I will make a check for that instead.

Kind regards
Mårten
> 
> Rob

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

end of thread, other threads:[~2022-02-11 21:38 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2022-02-04  8:45 [PATCH] of: dma_configure: Free old DMA map range Mårten Lindahl
2022-02-11 14:48 ` Rob Herring
2022-02-11 21:37   ` Marten Lindahl

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).