[PATCH v2 1/3] spi: fix resource leak for drivers without .remove callback

[PATCH v2 1/3] spi: fix resource leak for drivers without .remove callback

[Uwe Kleine-König]

Consider an spi driver with a .probe but without a .remove callback (e.g.
rtc-ds1347). The function spi_drv_probe() is called to bind a device and
so dev_pm_domain_attach() is called. As there is no remove callback
spi_drv_remove() isn't called at unbind time however and so calling
dev_pm_domain_detach() is missed and the pm domain keeps active.

To fix this always use both spi_drv_probe() and spi_drv_remove() and
make them handle the respective callback not being set. This has the
side effect that for a (hypothetical) driver that has neither .probe nor
remove the clk and pm domain setup is done.

Fixes: 33cf00e57082 ("spi: attach/detach SPI device to the ACPI power domain")
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
---
Changes since (implicit) v1:

 - make use of spi_drv_probe and spi_drv_remove unconditionally.

 drivers/spi/spi.c | 19 ++++++++++---------
 1 file changed, 10 insertions(+), 9 deletions(-)

diff --git a/drivers/spi/spi.c b/drivers/spi/spi.c
index 0cab239d8e7f..5becf6c2c409 100644
--- a/drivers/spi/spi.c
+++ b/drivers/spi/spi.c
@@ -405,9 +405,11 @@ static int spi_drv_probe(struct device *dev)
 	if (ret)
 		return ret;
 
-	ret = sdrv->probe(spi);
-	if (ret)
-		dev_pm_domain_detach(dev, true);
+	if (sdrv->probe) {
+		ret = sdrv->probe(spi);
+		if (ret)
+			dev_pm_domain_detach(dev, true);
+	}
 
 	return ret;
 }
@@ -415,9 +417,10 @@ static int spi_drv_probe(struct device *dev)
 static int spi_drv_remove(struct device *dev)
 {
 	const struct spi_driver		*sdrv = to_spi_driver(dev->driver);
-	int ret;
+	int ret = 0;
 
-	ret = sdrv->remove(to_spi_device(dev));
+	if (sdrv->remove)
+		ret = sdrv->remove(to_spi_device(dev));
 	dev_pm_domain_detach(dev, true);
 
 	return ret;
@@ -442,10 +445,8 @@ int __spi_register_driver(struct module *owner, struct spi_driver *sdrv)
 {
 	sdrv->driver.owner = owner;
 	sdrv->driver.bus = &spi_bus_type;
-	if (sdrv->probe)
-		sdrv->driver.probe = spi_drv_probe;
-	if (sdrv->remove)
-		sdrv->driver.remove = spi_drv_remove;
+	sdrv->driver.probe = spi_drv_probe;
+	sdrv->driver.remove = spi_drv_remove;
 	if (sdrv->shutdown)
 		sdrv->driver.shutdown = spi_drv_shutdown;
 	return driver_register(&sdrv->driver);

base-commit: 09162bc32c880a791c6c0668ce0745cf7958f576
-- 
2.28.0

[PATCH v2 2/3] spi: Use bus_type functions for probe, remove and shutdown

[Uwe Kleine-König]

The eventual goal is to get rid of the callbacks in struct
device_driver. Other than not using driver callbacks there should be no
side effect of this patch.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
---
 drivers/spi/spi.c | 33 ++++++++++++++++-----------------
 1 file changed, 16 insertions(+), 17 deletions(-)

diff --git a/drivers/spi/spi.c b/drivers/spi/spi.c
index 5becf6c2c409..e8c0a000ee19 100644
--- a/drivers/spi/spi.c
+++ b/drivers/spi/spi.c
@@ -374,16 +374,7 @@ static int spi_uevent(struct device *dev, struct kobj_uevent_env *env)
 	return add_uevent_var(env, "MODALIAS=%s%s", SPI_MODULE_PREFIX, spi->modalias);
 }
 
-struct bus_type spi_bus_type = {
-	.name		= "spi",
-	.dev_groups	= spi_dev_groups,
-	.match		= spi_match_device,
-	.uevent		= spi_uevent,
-};
-EXPORT_SYMBOL_GPL(spi_bus_type);
-
-
-static int spi_drv_probe(struct device *dev)
+static int spi_probe(struct device *dev)
 {
 	const struct spi_driver		*sdrv = to_spi_driver(dev->driver);
 	struct spi_device		*spi = to_spi_device(dev);
@@ -414,7 +405,7 @@ static int spi_drv_probe(struct device *dev)
 	return ret;
 }
 
-static int spi_drv_remove(struct device *dev)
+static int spi_remove(struct device *dev)
 {
 	const struct spi_driver		*sdrv = to_spi_driver(dev->driver);
 	int ret = 0;
@@ -426,13 +417,25 @@ static int spi_drv_remove(struct device *dev)
 	return ret;
 }
 
-static void spi_drv_shutdown(struct device *dev)
+static void spi_shutdown(struct device *dev)
 {
 	const struct spi_driver		*sdrv = to_spi_driver(dev->driver);
 
-	sdrv->shutdown(to_spi_device(dev));
+	if (sdrv->shutdown)
+		sdrv->shutdown(to_spi_device(dev));
 }
 
+struct bus_type spi_bus_type = {
+	.name		= "spi",
+	.dev_groups	= spi_dev_groups,
+	.match		= spi_match_device,
+	.uevent		= spi_uevent,
+	.probe		= spi_probe,
+	.remove		= spi_remove,
+	.shutdown	= spi_shutdown,
+};
+EXPORT_SYMBOL_GPL(spi_bus_type);
+
 /**
  * __spi_register_driver - register a SPI driver
  * @owner: owner module of the driver to register
@@ -445,10 +448,6 @@ int __spi_register_driver(struct module *owner, struct spi_driver *sdrv)
 {
 	sdrv->driver.owner = owner;
 	sdrv->driver.bus = &spi_bus_type;
-	sdrv->driver.probe = spi_drv_probe;
-	sdrv->driver.remove = spi_drv_remove;
-	if (sdrv->shutdown)
-		sdrv->driver.shutdown = spi_drv_shutdown;
 	return driver_register(&sdrv->driver);
 }
 EXPORT_SYMBOL_GPL(__spi_register_driver);
-- 
2.28.0

Re: [PATCH v2 2/3] spi: Use bus_type functions for probe, remove and shutdown

[Marek Szyprowski]

Hi Uwe,

On 19.11.2020 17:16, Uwe Kleine-König wrote:
> The eventual goal is to get rid of the callbacks in struct
> device_driver. Other than not using driver callbacks there should be no
> side effect of this patch.
>
> Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>

This patch landed recently in linux-next as commit 9db34ee64ce4 ("spi: 
Use bus_type functions for probe, remove and shutdown").

It causes a regression on some of my test boards:

Unable to handle kernel NULL pointer dereference at virtual address 
0000000000000018
Mem abort info:
   ESR = 0x96000004
   EC = 0x25: DABT (current EL), IL = 32 bits
   SET = 0, FnV = 0
   EA = 0, S1PTW = 0
Data abort info:
   ISV = 0, ISS = 0x00000004
   CM = 0, WnR = 0
user pgtable: 4k pages, 48-bit VAs, pgdp=00000000318ed000
[0000000000000018] pgd=0000000000000000, p4d=0000000000000000
Internal error: Oops: 96000004 [#1] PREEMPT SMP
Modules linked in: cpufreq_powersave cpufreq_conservative brcmfmac 
brcmutil cfg80211 crct10dif_ce s3fwrn5_i2c s3fwrn5 nci nfc s5p_mfc 
s5p_jpeg hci_uart btqca btbc
buf2_dma_contig videobuf2_memops videobuf2_v4l2 bluetooth 
videobuf2_common videodev panfrost gpu_sched ecdh_generic mc ecc rfkill 
ip_tables x_tables ipv6
CPU: 0 PID: 1 Comm: systemd-shutdow Not tainted 
5.10.0-rc5-next-20201124+ #9771
Hardware name: Samsung TM2E board (DT)
pstate: 60000005 (nZCv daif -PAN -UAO -TCO BTYPE=--)
pc : spi_shutdown+0x10/0x38
lr : device_shutdown+0x10c/0x350
sp : ffff80001311bc70
...
Call trace:
  spi_shutdown+0x10/0x38
  kernel_restart_prepare+0x34/0x40
  kernel_restart+0x14/0x88
  __do_sys_reboot+0x148/0x248
  __arm64_sys_reboot+0x1c/0x28
  el0_svc_common.constprop.3+0x74/0x198
  do_el0_svc+0x20/0x98
  el0_sync_handler+0x140/0x1a8
  el0_sync+0x140/0x180
Code: f9403402 d1008041 f100005f 9a9f1021 (f9400c21)
---[ end trace 266c07205a2d632e ]---
Kernel panic - not syncing: Oops: Fatal exception
Kernel Offset: disabled
CPU features: 0x0240022,65006087
Memory Limit: none
---[ end Kernel panic - not syncing: Oops: Fatal exception ]---

> ---
>   drivers/spi/spi.c | 33 ++++++++++++++++-----------------
>   1 file changed, 16 insertions(+), 17 deletions(-)
>
> diff --git a/drivers/spi/spi.c b/drivers/spi/spi.c
> index 5becf6c2c409..e8c0a000ee19 100644
> --- a/drivers/spi/spi.c
> +++ b/drivers/spi/spi.c
> @@ -374,16 +374,7 @@ static int spi_uevent(struct device *dev, struct kobj_uevent_env *env)
>   	return add_uevent_var(env, "MODALIAS=%s%s", SPI_MODULE_PREFIX, spi->modalias);
>   }
>   
> -struct bus_type spi_bus_type = {
> -	.name		= "spi",
> -	.dev_groups	= spi_dev_groups,
> -	.match		= spi_match_device,
> -	.uevent		= spi_uevent,
> -};
> -EXPORT_SYMBOL_GPL(spi_bus_type);
> -
> -
> -static int spi_drv_probe(struct device *dev)
> +static int spi_probe(struct device *dev)
>   {
>   	const struct spi_driver		*sdrv = to_spi_driver(dev->driver);
>   	struct spi_device		*spi = to_spi_device(dev);
> @@ -414,7 +405,7 @@ static int spi_drv_probe(struct device *dev)
>   	return ret;
>   }
>   
> -static int spi_drv_remove(struct device *dev)
> +static int spi_remove(struct device *dev)
>   {
>   	const struct spi_driver		*sdrv = to_spi_driver(dev->driver);
>   	int ret = 0;
> @@ -426,13 +417,25 @@ static int spi_drv_remove(struct device *dev)
>   	return ret;
>   }
>   
> -static void spi_drv_shutdown(struct device *dev)
> +static void spi_shutdown(struct device *dev)
>   {
>   	const struct spi_driver		*sdrv = to_spi_driver(dev->driver);
>   
> -	sdrv->shutdown(to_spi_device(dev));
> +	if (sdrv->shutdown)
> +		sdrv->shutdown(to_spi_device(dev));
>   }

In the above function dev->driver might be NULL, so its use in 
to_spi_driver() and sdrv->shutdown leads to NULL pointer dereference. I 
didn't check the details, but a simple check for NULL dev->driver and 
return is enough to fix this issue. I can send such fix if you want.

> +struct bus_type spi_bus_type = {
> +	.name		= "spi",
> +	.dev_groups	= spi_dev_groups,
> +	.match		= spi_match_device,
> +	.uevent		= spi_uevent,
> +	.probe		= spi_probe,
> +	.remove		= spi_remove,
> +	.shutdown	= spi_shutdown,
> +};
> +EXPORT_SYMBOL_GPL(spi_bus_type);
> +
>   /**
>    * __spi_register_driver - register a SPI driver
>    * @owner: owner module of the driver to register
> @@ -445,10 +448,6 @@ int __spi_register_driver(struct module *owner, struct spi_driver *sdrv)
>   {
>   	sdrv->driver.owner = owner;
>   	sdrv->driver.bus = &spi_bus_type;
> -	sdrv->driver.probe = spi_drv_probe;
> -	sdrv->driver.remove = spi_drv_remove;
> -	if (sdrv->shutdown)
> -		sdrv->driver.shutdown = spi_drv_shutdown;
>   	return driver_register(&sdrv->driver);
>   }
>   EXPORT_SYMBOL_GPL(__spi_register_driver);

Best regards
-- 
Marek Szyprowski, PhD
Samsung R&D Institute Poland

Re: [PATCH v2 2/3] spi: Use bus_type functions for probe, remove and shutdown

[Uwe Kleine-König]

[-- Attachment #1: Type: text/plain, Size: 4676 bytes --]

Hello Marek,

On Tue, Nov 24, 2020 at 01:03:25PM +0100, Marek Szyprowski wrote:
> On 19.11.2020 17:16, Uwe Kleine-König wrote:
> > The eventual goal is to get rid of the callbacks in struct
> > device_driver. Other than not using driver callbacks there should be no
> > side effect of this patch.
> >
> > Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
> 
> This patch landed recently in linux-next as commit 9db34ee64ce4 ("spi: 
> Use bus_type functions for probe, remove and shutdown").
> 
> It causes a regression on some of my test boards:
> 
> Unable to handle kernel NULL pointer dereference at virtual address 
> 0000000000000018
> Mem abort info:
>    ESR = 0x96000004
>    EC = 0x25: DABT (current EL), IL = 32 bits
>    SET = 0, FnV = 0
>    EA = 0, S1PTW = 0
> Data abort info:
>    ISV = 0, ISS = 0x00000004
>    CM = 0, WnR = 0
> user pgtable: 4k pages, 48-bit VAs, pgdp=00000000318ed000
> [0000000000000018] pgd=0000000000000000, p4d=0000000000000000
> Internal error: Oops: 96000004 [#1] PREEMPT SMP
> Modules linked in: cpufreq_powersave cpufreq_conservative brcmfmac 
> brcmutil cfg80211 crct10dif_ce s3fwrn5_i2c s3fwrn5 nci nfc s5p_mfc 
> s5p_jpeg hci_uart btqca btbc
> buf2_dma_contig videobuf2_memops videobuf2_v4l2 bluetooth 
> videobuf2_common videodev panfrost gpu_sched ecdh_generic mc ecc rfkill 
> ip_tables x_tables ipv6
> CPU: 0 PID: 1 Comm: systemd-shutdow Not tainted 
> 5.10.0-rc5-next-20201124+ #9771
> Hardware name: Samsung TM2E board (DT)
> pstate: 60000005 (nZCv daif -PAN -UAO -TCO BTYPE=--)
> pc : spi_shutdown+0x10/0x38
> lr : device_shutdown+0x10c/0x350
> sp : ffff80001311bc70
> ...
> Call trace:
>   spi_shutdown+0x10/0x38
>   kernel_restart_prepare+0x34/0x40
>   kernel_restart+0x14/0x88
>   __do_sys_reboot+0x148/0x248
>   __arm64_sys_reboot+0x1c/0x28
>   el0_svc_common.constprop.3+0x74/0x198
>   do_el0_svc+0x20/0x98
>   el0_sync_handler+0x140/0x1a8
>   el0_sync+0x140/0x180
> Code: f9403402 d1008041 f100005f 9a9f1021 (f9400c21)
> ---[ end trace 266c07205a2d632e ]---
> Kernel panic - not syncing: Oops: Fatal exception
> Kernel Offset: disabled
> CPU features: 0x0240022,65006087
> Memory Limit: none
> ---[ end Kernel panic - not syncing: Oops: Fatal exception ]---
> 
> > ---
> >   drivers/spi/spi.c | 33 ++++++++++++++++-----------------
> >   1 file changed, 16 insertions(+), 17 deletions(-)
> >
> > diff --git a/drivers/spi/spi.c b/drivers/spi/spi.c
> > index 5becf6c2c409..e8c0a000ee19 100644
> > --- a/drivers/spi/spi.c
> > +++ b/drivers/spi/spi.c
> > @@ -374,16 +374,7 @@ static int spi_uevent(struct device *dev, struct kobj_uevent_env *env)
> >   	return add_uevent_var(env, "MODALIAS=%s%s", SPI_MODULE_PREFIX, spi->modalias);
> >   }
> >   
> > -struct bus_type spi_bus_type = {
> > -	.name		= "spi",
> > -	.dev_groups	= spi_dev_groups,
> > -	.match		= spi_match_device,
> > -	.uevent		= spi_uevent,
> > -};
> > -EXPORT_SYMBOL_GPL(spi_bus_type);
> > -
> > -
> > -static int spi_drv_probe(struct device *dev)
> > +static int spi_probe(struct device *dev)
> >   {
> >   	const struct spi_driver		*sdrv = to_spi_driver(dev->driver);
> >   	struct spi_device		*spi = to_spi_device(dev);
> > @@ -414,7 +405,7 @@ static int spi_drv_probe(struct device *dev)
> >   	return ret;
> >   }
> >   
> > -static int spi_drv_remove(struct device *dev)
> > +static int spi_remove(struct device *dev)
> >   {
> >   	const struct spi_driver		*sdrv = to_spi_driver(dev->driver);
> >   	int ret = 0;
> > @@ -426,13 +417,25 @@ static int spi_drv_remove(struct device *dev)
> >   	return ret;
> >   }
> >   
> > -static void spi_drv_shutdown(struct device *dev)
> > +static void spi_shutdown(struct device *dev)
> >   {
> >   	const struct spi_driver		*sdrv = to_spi_driver(dev->driver);
> >   
> > -	sdrv->shutdown(to_spi_device(dev));
> > +	if (sdrv->shutdown)
> > +		sdrv->shutdown(to_spi_device(dev));
> >   }
> 
> In the above function dev->driver might be NULL, so its use in 
> to_spi_driver() and sdrv->shutdown leads to NULL pointer dereference. I 
> didn't check the details, but a simple check for NULL dev->driver and 
> return is enough to fix this issue. I can send such fix if you want.

Ah, I see. shutdown is called for unbound devices, too. Assuming that
Mark prefers a fix on top instead of an updated patch: Yes, please send
a fix. Otherwise I can do this, too, as I introduced the problem.

Best regards and thanks,
Uwe

-- 
Pengutronix e.K.                           | Uwe Kleine-König            |
Industrial Linux Solutions                 | https://www.pengutronix.de/ |

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: [PATCH v2 2/3] spi: Use bus_type functions for probe, remove and shutdown

[Mark Brown]

[-- Attachment #1: Type: text/plain, Size: 904 bytes --]

On Tue, Nov 24, 2020 at 02:01:07PM +0100, Uwe Kleine-König wrote:
> On Tue, Nov 24, 2020 at 01:03:25PM +0100, Marek Szyprowski wrote:

> > > +	if (sdrv->shutdown)
> > > +		sdrv->shutdown(to_spi_device(dev));
> > >   }

> > In the above function dev->driver might be NULL, so its use in 
> > to_spi_driver() and sdrv->shutdown leads to NULL pointer dereference. I 
> > didn't check the details, but a simple check for NULL dev->driver and 
> > return is enough to fix this issue. I can send such fix if you want.

> Ah, I see. shutdown is called for unbound devices, too. Assuming that
> Mark prefers a fix on top instead of an updated patch: Yes, please send
> a fix. Otherwise I can do this, too, as I introduced the problem.

Yes, please send an incremental fix (in general in a situation like this
I'd just send a fix as part of the original report, it's quicker if the
fix is OK).

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

[PATCH v2 3/3] spi: Warn when a driver's remove callback returns an error

[Uwe Kleine-König]

The driver core ignores the return value of struct device_driver::remove
(because in general there is nothing that can be done about that). So
add a warning when an spi driver returns an error.

This simplifies the quest to make struct device_driver::remove return void.
A consequent change would be to make struct spi_driver::remove return void,
but I'm keeping this quest for later (or someone else).

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
---
 drivers/spi/spi.c | 13 ++++++++++---
 1 file changed, 10 insertions(+), 3 deletions(-)

diff --git a/drivers/spi/spi.c b/drivers/spi/spi.c
index e8c0a000ee19..6b7c19bf7715 100644
--- a/drivers/spi/spi.c
+++ b/drivers/spi/spi.c
@@ -408,13 +408,20 @@ static int spi_probe(struct device *dev)
 static int spi_remove(struct device *dev)
 {
 	const struct spi_driver		*sdrv = to_spi_driver(dev->driver);
-	int ret = 0;
 
-	if (sdrv->remove)
+	if (sdrv->remove) {
+		int ret;
+
 		ret = sdrv->remove(to_spi_device(dev));
+		if (ret)
+			dev_warn(dev,
+				 "Failed to unbind driver (%pe), ignoring\n",
+				 ERR_PTR(ret));
+	}
+
 	dev_pm_domain_detach(dev, true);
 
-	return ret;
+	return 0;
 }
 
 static void spi_shutdown(struct device *dev)
-- 
2.28.0