driver core: fix driver_set_override() issue with empty strings

[PATCH] driver core: fix driver_set_override() issue with empty strings

Posted by Greg Kroah-Hartman 3 years, 7 months ago

Python likes to send an empty string for some sysfs files, including the
driver_override field.  When commit 23d99baf9d72 ("PCI: Use
driver_set_override() instead of open-coding") moved the PCI core to use
the driver core function instead of hand-rolling their own handler, this
showed up as a regression from some userspace tools, like DPDK.

Fix this up by actually looking at the length of the string first
instead of trusting that userspace got it correct.

Reported-by: Stephen Hemminger <stephen@networkplumber.org>
Fixes: 23d99baf9d72 ("PCI: Use driver_set_override() instead of open-coding")
Cc: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Cc: Bjorn Helgaas <bhelgaas@google.com>
Cc: "Rafael J. Wysocki" <rafael@kernel.org>
Cc: Andy Shevchenko <andy.shevchenko@gmail.com>
Cc: stable <stable@kernel.org>
Tested-by: Huisong Li <lihuisong@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
 drivers/base/driver.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/drivers/base/driver.c b/drivers/base/driver.c
index 15a75afe6b84..676b6275d5b5 100644
--- a/drivers/base/driver.c
+++ b/drivers/base/driver.c
@@ -63,6 +63,12 @@ int driver_set_override(struct device *dev, const char **override,
 	if (len >= (PAGE_SIZE - 1))
 		return -EINVAL;
 
+	/*
+	 * Compute the real length of the string in case userspace sends us a
+	 * bunch of \0 characters like python likes to do.
+	 */
+	len = strlen(s);
+
 	if (!len) {
 		/* Empty string passed - clear override */
 		device_lock(dev);
-- 
2.37.3

Re: [PATCH] driver core: fix driver_set_override() issue with empty strings

Posted by Krzysztof Kozlowski 3 years, 7 months ago

On 01/09/2022 19:37, Greg Kroah-Hartman wrote:
> Python likes to send an empty string for some sysfs files, including the
> driver_override field.  When commit 23d99baf9d72 ("PCI: Use
> driver_set_override() instead of open-coding") moved the PCI core to use
> the driver core function instead of hand-rolling their own handler, this
> showed up as a regression from some userspace tools, like DPDK.
> 
> Fix this up by actually looking at the length of the string first
> instead of trusting that userspace got it correct.
> 
> Reported-by: Stephen Hemminger <stephen@networkplumber.org>
> Fixes: 23d99baf9d72 ("PCI: Use driver_set_override() instead of open-coding")
> Cc: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
> Cc: Bjorn Helgaas <bhelgaas@google.com>
> Cc: "Rafael J. Wysocki" <rafael@kernel.org>
> Cc: Andy Shevchenko <andy.shevchenko@gmail.com>
> Cc: stable <stable@kernel.org>
> Tested-by: Huisong Li <lihuisong@huawei.com>


Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>


Best regards,
Krzysztof

Re: [PATCH] driver core: fix driver_set_override() issue with empty strings

Posted by Stephen Hemminger 3 years, 7 months ago

On Thu,  1 Sep 2022 18:37:34 +0200
Greg Kroah-Hartman <gregkh@linuxfoundation.org> wrote:

> Python likes to send an empty string for some sysfs files, including the
> driver_override field.  When commit 23d99baf9d72 ("PCI: Use
> driver_set_override() instead of open-coding") moved the PCI core to use
> the driver core function instead of hand-rolling their own handler, this
> showed up as a regression from some userspace tools, like DPDK.
> 
> Fix this up by actually looking at the length of the string first
> instead of trusting that userspace got it correct.
> 
> Reported-by: Stephen Hemminger <stephen@networkplumber.org>
> Fixes: 23d99baf9d72 ("PCI: Use driver_set_override() instead of open-coding")
> Cc: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
> Cc: Bjorn Helgaas <bhelgaas@google.com>
> Cc: "Rafael J. Wysocki" <rafael@kernel.org>
> Cc: Andy Shevchenko <andy.shevchenko@gmail.com>
> Cc: stable <stable@kernel.org>
> Tested-by: Huisong Li <lihuisong@huawei.com>
> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
> ---
>  drivers/base/driver.c | 6 ++++++
>  1 file changed, 6 insertions(+)
> 
> diff --git a/drivers/base/driver.c b/drivers/base/driver.c
> index 15a75afe6b84..676b6275d5b5 100644
> --- a/drivers/base/driver.c
> +++ b/drivers/base/driver.c
> @@ -63,6 +63,12 @@ int driver_set_override(struct device *dev, const char **override,
>  	if (len >= (PAGE_SIZE - 1))
>  		return -EINVAL;
>  
> +	/*
> +	 * Compute the real length of the string in case userspace sends us a
> +	 * bunch of \0 characters like python likes to do.
> +	 */
> +	len = strlen(s);
> +
>  	if (!len) {
>  		/* Empty string passed - clear override */
>  		device_lock(dev);

Thanks for addressing this.

Reviewed-by: Stephen Hemminger <stephen@networkplumber.org>