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: <20230508162126.39146-1-kuniyu@amazon.com>
Date: Mon, 8 May 2023 09:21:26 -0700
From: Kuniyuki Iwashima <kuniyu@...zon.com>
To: <leon@...nel.org>
CC: <netdev@...r.kernel.org>, <patryk.sondej@...il.com>
Subject: Re: [PATCH net] inet_diag: fix inet_diag_msg_attrs_fill() for net_cls cgroup

From: Leon Romanovsky <leon@...nel.org>
Date: Mon, 8 May 2023 09:17:49 +0300
> On Mon, May 08, 2023 at 05:32:33AM +0200, Patryk Sondej wrote:
> > This commit fixes inet_diag_msg_attrs_fill() function in the ipv4/inet_diag.c file.
> > The problem was that the function was using CONFIG_SOCK_CGROUP_DATA to check for the net_cls cgroup.
> > However, the net_cls cgroup is defined by CONFIG_CGROUP_NET_CLASSID instead.
> > 
> > Therefore, this commit updates the #ifdef statement to CONFIG_CGROUP_NET_CLASSID,
> > and uses the sock_cgroup_classid() function to retrieve the classid from the socket cgroup.
> > 
> > This change ensures that the function correctly retrieves the classid for the net_cls cgroup
> > and fixes any issues related to the use of the function in this context.
> > 
> 
> Please add Fixes line here.
> 
> > Signed-off-by: Patryk Sondej <patryk.sondej@...il.com>
> > ---
> >  net/ipv4/inet_diag.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> > diff --git a/net/ipv4/inet_diag.c b/net/ipv4/inet_diag.c
> > index b812eb36f0e3..7017f88911a6 100644
> > --- a/net/ipv4/inet_diag.c
> > +++ b/net/ipv4/inet_diag.c
> > @@ -157,7 +157,7 @@ int inet_diag_msg_attrs_fill(struct sock *sk, struct sk_buff *skb,
> >  	    ext & (1 << (INET_DIAG_TCLASS - 1))) {
> >  		u32 classid = 0;
> >  
> > -#ifdef CONFIG_SOCK_CGROUP_DATA
> > +#ifdef CONFIG_CGROUP_NET_CLASSID
> 
> This ifdef should be deleted as sock_cgroup_classid() already has right ifdef.

sock_cgroup_classid() is defined under #ifdef CONFIG_SOCK_CGROUP_DATA,
so removing this guard will cause an error at compile time if we
disable CONFIG_SOCK_CGROUP_DATA.

I think we can keep the old #ifdef as is for consistensy with
another CONFIG_SOCK_CGROUP_DATA use just below for sock_cgroup_ptr()
in inet_diag_msg_attrs_fill().

If we enable CONFIG_SOCK_CGROUP_DATA without CONFIG_CGROUP_NET_CLASSID,
I guess gcc will optimize it and remove the zero assignment.


> 
>   809 static inline u32 sock_cgroup_classid(const struct sock_cgroup_data *skcd)
>   810 {
>   811 #ifdef CONFIG_CGROUP_NET_CLASSID
>   812         return READ_ONCE(skcd->classid);
>   813 #else
>   814         return 0;
>   815 #endif
>   816 }
>   817
> 
> 
> >  		classid = sock_cgroup_classid(&sk->sk_cgrp_data);
> >  #endif
> >  		/* Fallback to socket priority if class id isn't set.
> > -- 
> > 2.37.1 (Apple Git-137.1)

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ