[PATCH] platform/x86: dell-wmi-sysman: fix kobject leak on populate failure

[PATCH] platform/x86: dell-wmi-sysman: fix kobject leak on populate failure

[Daniel Hodges]

When populate_enum_data(), populate_int_data(), populate_str_data(),
or populate_po_data() fails after a successful kobject_init_and_add(),
the code jumps to err_attr_init without calling kobject_put() on
attr_name_kobj, leaking the kobject and its associated memory.

Add the missing kobject_put() call before the goto to properly release
the kobject on error.

Fixes: e8a60aa7404b ("platform/x86: Introduce support for Systems Management Driver over WMI for Dell Systems")
Cc: stable@vger.kernel.org
Signed-off-by: Daniel Hodges <git@danielhodges.dev>
---
 drivers/platform/x86/dell/dell-wmi-sysman/sysman.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/platform/x86/dell/dell-wmi-sysman/sysman.c b/drivers/platform/x86/dell/dell-wmi-sysman/sysman.c
index f5402b714657..d9f6d24c84d6 100644
--- a/drivers/platform/x86/dell/dell-wmi-sysman/sysman.c
+++ b/drivers/platform/x86/dell/dell-wmi-sysman/sysman.c
@@ -497,6 +497,7 @@ static int init_bios_attributes(int attr_type, const char *guid)
 		if (retval) {
 			pr_debug("failed to populate %s\n",
 				elements[ATTR_NAME].string.pointer);
+			kobject_put(attr_name_kobj);
 			goto err_attr_init;
 		}
 
-- 
2.52.0

Re: [PATCH] platform/x86: dell-wmi-sysman: fix kobject leak on populate failure

[Greg KH]

On Fri, Feb 06, 2026 at 06:16:42PM -0500, Daniel Hodges wrote:
> When populate_enum_data(), populate_int_data(), populate_str_data(),
> or populate_po_data() fails after a successful kobject_init_and_add(),
> the code jumps to err_attr_init without calling kobject_put() on
> attr_name_kobj, leaking the kobject and its associated memory.
> 
> Add the missing kobject_put() call before the goto to properly release
> the kobject on error.
> 
> Fixes: e8a60aa7404b ("platform/x86: Introduce support for Systems Management Driver over WMI for Dell Systems")
> Cc: stable@vger.kernel.org
> Signed-off-by: Daniel Hodges <git@danielhodges.dev>
> ---
>  drivers/platform/x86/dell/dell-wmi-sysman/sysman.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/drivers/platform/x86/dell/dell-wmi-sysman/sysman.c b/drivers/platform/x86/dell/dell-wmi-sysman/sysman.c
> index f5402b714657..d9f6d24c84d6 100644
> --- a/drivers/platform/x86/dell/dell-wmi-sysman/sysman.c
> +++ b/drivers/platform/x86/dell/dell-wmi-sysman/sysman.c
> @@ -497,6 +497,7 @@ static int init_bios_attributes(int attr_type, const char *guid)
>  		if (retval) {
>  			pr_debug("failed to populate %s\n",
>  				elements[ATTR_NAME].string.pointer);
> +			kobject_put(attr_name_kobj);
>  			goto err_attr_init;
>  		}
>  

The "larger" problem with this driver is its use of raw kobjects.  No
driver should be doing that, it is hiding all of this information from
userspace tools by doing so, and there's loads of race conditions
happening with the creation of these files.  It should be fixed by just
using normal device attributes and not attempting to custom create
kobjects by hand like this (as it is very easy to get things wrong, as
this patch shows.)

thanks,

greg k-h

Re: [PATCH] platform/x86: dell-wmi-sysman: fix kobject leak on populate failure

[Daniel Hodges]

On Sat, Feb 07, 2026 at 08:48:34AM +0100, Greg KH wrote:
> On Fri, Feb 06, 2026 at 06:16:42PM -0500, Daniel Hodges wrote:
> > When populate_enum_data(), populate_int_data(), populate_str_data(),
> > or populate_po_data() fails after a successful kobject_init_and_add(),
> > the code jumps to err_attr_init without calling kobject_put() on
> > attr_name_kobj, leaking the kobject and its associated memory.
> > 
> > Add the missing kobject_put() call before the goto to properly release
> > the kobject on error.
> > 
> > Fixes: e8a60aa7404b ("platform/x86: Introduce support for Systems Management Driver over WMI for Dell Systems")
> > Cc: stable@vger.kernel.org
> > Signed-off-by: Daniel Hodges <git@danielhodges.dev>
> > ---
> >  drivers/platform/x86/dell/dell-wmi-sysman/sysman.c | 1 +
> >  1 file changed, 1 insertion(+)
> > 
> > diff --git a/drivers/platform/x86/dell/dell-wmi-sysman/sysman.c b/drivers/platform/x86/dell/dell-wmi-sysman/sysman.c
> > index f5402b714657..d9f6d24c84d6 100644
> > --- a/drivers/platform/x86/dell/dell-wmi-sysman/sysman.c
> > +++ b/drivers/platform/x86/dell/dell-wmi-sysman/sysman.c
> > @@ -497,6 +497,7 @@ static int init_bios_attributes(int attr_type, const char *guid)
> >  		if (retval) {
> >  			pr_debug("failed to populate %s\n",
> >  				elements[ATTR_NAME].string.pointer);
> > +			kobject_put(attr_name_kobj);
> >  			goto err_attr_init;
> >  		}
> >  
> 
> The "larger" problem with this driver is its use of raw kobjects.  No
> driver should be doing that, it is hiding all of this information from
> userspace tools by doing so, and there's loads of race conditions
> happening with the creation of these files.  It should be fixed by just
> using normal device attributes and not attempting to custom create
> kobjects by hand like this (as it is very easy to get things wrong, as
> this patch shows.)
> 
> thanks,
> 
> greg k-h

Yeah, that makes sense. I think your point of finding "bad patterns" in
usage would also be really helpful for code analysis tools. If there's
some canonical docs on various antipatterns it would be pretty helpful.

-Daniel

Re: [PATCH] platform/x86: dell-wmi-sysman: fix kobject leak on populate failure

[Greg KH]

On Sat, Feb 07, 2026 at 11:52:11AM -0500, Daniel Hodges wrote:
> On Sat, Feb 07, 2026 at 08:48:34AM +0100, Greg KH wrote:
> > On Fri, Feb 06, 2026 at 06:16:42PM -0500, Daniel Hodges wrote:
> > > When populate_enum_data(), populate_int_data(), populate_str_data(),
> > > or populate_po_data() fails after a successful kobject_init_and_add(),
> > > the code jumps to err_attr_init without calling kobject_put() on
> > > attr_name_kobj, leaking the kobject and its associated memory.
> > > 
> > > Add the missing kobject_put() call before the goto to properly release
> > > the kobject on error.
> > > 
> > > Fixes: e8a60aa7404b ("platform/x86: Introduce support for Systems Management Driver over WMI for Dell Systems")
> > > Cc: stable@vger.kernel.org
> > > Signed-off-by: Daniel Hodges <git@danielhodges.dev>
> > > ---
> > >  drivers/platform/x86/dell/dell-wmi-sysman/sysman.c | 1 +
> > >  1 file changed, 1 insertion(+)
> > > 
> > > diff --git a/drivers/platform/x86/dell/dell-wmi-sysman/sysman.c b/drivers/platform/x86/dell/dell-wmi-sysman/sysman.c
> > > index f5402b714657..d9f6d24c84d6 100644
> > > --- a/drivers/platform/x86/dell/dell-wmi-sysman/sysman.c
> > > +++ b/drivers/platform/x86/dell/dell-wmi-sysman/sysman.c
> > > @@ -497,6 +497,7 @@ static int init_bios_attributes(int attr_type, const char *guid)
> > >  		if (retval) {
> > >  			pr_debug("failed to populate %s\n",
> > >  				elements[ATTR_NAME].string.pointer);
> > > +			kobject_put(attr_name_kobj);
> > >  			goto err_attr_init;
> > >  		}
> > >  
> > 
> > The "larger" problem with this driver is its use of raw kobjects.  No
> > driver should be doing that, it is hiding all of this information from
> > userspace tools by doing so, and there's loads of race conditions
> > happening with the creation of these files.  It should be fixed by just
> > using normal device attributes and not attempting to custom create
> > kobjects by hand like this (as it is very easy to get things wrong, as
> > this patch shows.)
> > 
> > thanks,
> > 
> > greg k-h
> 
> Yeah, that makes sense. I think your point of finding "bad patterns" in
> usage would also be really helpful for code analysis tools. If there's
> some canonical docs on various antipatterns it would be pretty helpful.

Huge hint, any driver that calls any sysfs_*() function or touches a
kobject at all, is probably doing things wrong.  Start there and see
what you find :)

thanks,

greg k-h