Re: [PATCH 029/190] Revert "EDAC: Fix reference count leaks"

[Borislav Petkov]

On Wed, Apr 21, 2021 at 02:58:24PM +0200, Greg Kroah-Hartman wrote:
> This reverts commit 17ed808ad243192fb923e4e653c1338d3ba06207.
> 
> Commits from @umn.edu addresses have been found to be submitted in "bad
> faith" to try to test the kernel community's ability to review "known
> malicious" changes.  The result of these submissions can be found in a
> paper published at the 42nd IEEE Symposium on Security and Privacy
> entitled, "Open Source Insecurity: Stealthily Introducing
> Vulnerabilities via Hypocrite Commits" written by Qiushi Wu (University
> of Minnesota) and Kangjie Lu (University of Minnesota).
> 
> Because of this, all submissions from this group must be reverted from
> the kernel tree and will need to be re-reviewed again to determine if
> they actually are a valid fix.  Until that work is complete, remove this
> change to ensure that no problems are being introduced into the
> codebase.
> 
> Cc: Qiushi Wu <wu000273@xxxxxxx>
> Cc: Borislav Petkov <bp@xxxxxxx>
> Cc: https
> Cc: https
> Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
> ---
>  drivers/edac/edac_device_sysfs.c | 1 -
>  drivers/edac/edac_pci_sysfs.c    | 2 +-
>  2 files changed, 1 insertion(+), 2 deletions(-)
> 
> diff --git a/drivers/edac/edac_device_sysfs.c b/drivers/edac/edac_device_sysfs.c
> index 5e7593753799..0e7ea3591b78 100644
> --- a/drivers/edac/edac_device_sysfs.c
> +++ b/drivers/edac/edac_device_sysfs.c
> @@ -275,7 +275,6 @@ int edac_device_register_sysfs_main_kobj(struct edac_device_ctl_info *edac_dev)
>  
>  	/* Error exit stack */
>  err_kobj_reg:
> -	kobject_put(&edac_dev->kobj);
>  	module_put(edac_dev->owner);
>  
>  err_out:
> diff --git a/drivers/edac/edac_pci_sysfs.c b/drivers/edac/edac_pci_sysfs.c
> index 53042af7262e..72c9eb9fdffb 100644
> --- a/drivers/edac/edac_pci_sysfs.c
> +++ b/drivers/edac/edac_pci_sysfs.c
> @@ -386,7 +386,7 @@ static int edac_pci_main_kobj_setup(void)
>  
>  	/* Error unwind statck */
>  kobject_init_and_add_fail:
> -	kobject_put(edac_pci_top_main_kobj);
> +	kfree(edac_pci_top_main_kobj);
>  
>  kzalloc_fail:
>  	module_put(THIS_MODULE);
> -- 

That one is actually correct. 

AFAIR, that was a reaction to kobject_init_and_add()'s comment saying
users should do kobject_put() when that function fails.

So the added kobject_put() would make sure the ->release method is
called which will kfree the kobject and drop the module reference. And
both ->release methods don't do anything else so there are no side
effects from it like with other cases earlier today.

So yeah, that one could stay in.

Can't say I didn't spend some time staring at this today again, thanks
to this "experiment". :-\

Thx.

-- 
Regards/Gruss,
    Boris.

SUSE Software Solutions Germany GmbH, GF: Felix Imendörffer, HRB 36809, AG Nürnberg