netdev - Re: [PATCH net-next] cxgb4: fix memory leak in cxgb4_init_ethtool

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <o4o32xf7oejvzyd3cb7sr4whvganh2uds3rvkxzcaqyhllaaum@iovzdahpu3ha>
Date: Fri, 11 Apr 2025 21:52:29 +0530
From: Abdun Nihaal <abdun.nihaal@...il.com>
To: Simon Horman <horms@...nel.org>
Cc: Markus Elfring <Markus.Elfring@....de>, netdev@...r.kernel.org, 
	LKML <linux-kernel@...r.kernel.org>, Andrew Lunn <andrew+netdev@...n.ch>, 
	"David S. Miller" <davem@...emloft.net>, Eric Dumazet <edumazet@...gle.com>, 
	Jakub Kicinski <kuba@...nel.org>, Paolo Abeni <pabeni@...hat.com>, 
	Potnuri Bharat Teja <bharat@...lsio.com>, Rahul Lakkireddy <rahul.lakkireddy@...lsio.com>, 
	Vishal Kulkarni <vishal@...lsio.com>
Subject: Re: [PATCH net-next] cxgb4: fix memory leak in
 cxgb4_init_ethtool_filters() error path

On Fri, Apr 11, 2025 at 03:57:34PM +0100, Simon Horman wrote:
> On Wed, Apr 09, 2025 at 05:47:46PM +0200, Markus Elfring wrote:
> > …
> > > +++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_ethtool.c
> > > @@ -2270,6 +2270,7 @@ int cxgb4_init_ethtool_filters(struct adapter *adap)
> > >  		eth_filter->port[i].bmap = bitmap_zalloc(nentries, GFP_KERNEL);
> > >  		if (!eth_filter->port[i].bmap) {
> > >  			ret = -ENOMEM;
> > > +			kvfree(eth_filter->port[i].loc_array);
> > >  			goto free_eth_finfo;
> > >  		}
> > >  	}
> > 
> > How do you think about to move the shown error code assignment behind the mentioned label
> > (so that another bit of duplicate source code could be avoided)?
> 
> Hi Markus,
> 
> If you mean something like the following. Then I agree that it
> is both in keeping with the existing error handling in this function
> and addresses the problem at hand.
> 
> diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_ethtool.c b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_ethtool.c
> index 7f3f5afa864f..df26d3388c00 100644
> --- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_ethtool.c
> +++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_ethtool.c
> @@ -2270,13 +2270,15 @@ int cxgb4_init_ethtool_filters(struct adapter *adap)
>                 eth_filter->port[i].bmap = bitmap_zalloc(nentries, GFP_KERNEL);
>                 if (!eth_filter->port[i].bmap) {
>                         ret = -ENOMEM;
> -                       goto free_eth_finfo;
> +                       goto free_eth_finfo_loc_array;
>                 }
>         }
> 
>         adap->ethtool_filters = eth_filter;
>         return 0;
> 
> +free_eth_finfo_loc_array:
> +       kvfree(eth_filter->port[i].loc_array);
>  free_eth_finfo:
>         while (i-- > 0) {
>                 bitmap_free(eth_filter->port[i].bmap);
> 

I think what Markus meant, was to move the ret = -ENOMEM from both the
allocations in the loop, to after the free_eth_finfo label because it is
-ENOMEM on both goto jumps.

But personally I would prefer having the ret code right after the call 
that is failing. Also I would avoid creating new goto labels unless
necessary, because it is easier to see the kvfree in context inside the
loop, than to put it in a separate label.

I just tried to make the most minimal code change to fix the memory leak.

Regards,
Nihaal