Re: [PATCH v2 4/4] ax.25: Remove the now superfluous sentinel elements from ctl_table array

Joel Granados <[email protected]> Mon, 8 Apr 2024 09:20:43 +0200

Newsgroups	org.kernel.vger.dccp,dev.linux.lists.bridge,dev.linux.lists.mptcp,org.kernel.vger.linux-hams,org.kernel.vger.linux-kernel,org.kernel.vger.linux-nfs,org.kernel.vger.linux-rdma,org.kernel.vger.linux-s390,org.kernel.vger.linux-sctp,org.kernel.vger.linux-wpan,org.kernel.vger.linux-x25,org.kernel.vger.lvs-devel,org.kernel.vger.netdev,org.kernel.vger.netfilter-devel
Message-ID	<[email protected]>

On Fri, Apr 05, 2024 at 03:26:58PM -0700, Kuniyuki Iwashima wrote:
> From: Joel Granados <[email protected]>
> Date: Fri, 5 Apr 2024 09:15:31 +0200
> > On Thu, Mar 28, 2024 at 12:49:34PM -0700, Kuniyuki Iwashima wrote:
> > > From: Joel Granados via B4 Relay <[email protected]>
> > > Date: Thu, 28 Mar 2024 16:40:05 +0100
> > > > This commit comes at the tail end of a greater effort to remove the
> > > > empty elements at the end of the ctl_table arrays (sentinels) which will
> > > > reduce the overall build time size of the kernel and run time memory
> > > > bloat by ~64 bytes per sentinel (further information Link :
> > > > https://lore.kernel.org/all/ZO5Yx5JFogGi%[email protected]/)
> > > > 
> > > > When we remove the sentinel from ax25_param_table a buffer overflow
> > > > shows its ugly head. The sentinel's data element used to be changed when
> > > > CONFIG_AX25_DAMA_SLAVE was not defined.
> > > 
> > > I think it's better to define the relation explicitly between the
> > > enum and sysctl table by BUILD_BUG_ON() in ax25_register_dev_sysctl()
> > > 
> > >   BUILD_BUG_ON(AX25_MAX_VALUES != ARRAY_SIZE(ax25_param_table));
> > > 
> > > and guard AX25_VALUES_DS_TIMEOUT with #ifdef CONFIG_AX25_DAMA_SLAVE
> > > as done for other enum.
> > 
> > When I remove AX25_VALUES_DS_TIMEOUT from the un-guarded build it
> > complains in net/ax25/ax25_ds_timer.c (ax25_ds_set_timer). Here is the
> > report https://lore.kernel.org/oe-kbuild-all/[email protected]/.
> > 
> > How best to address this? Should we just guard the whole function and do
> > nothing when not set? like this:
> 
> It seems fine to me.
> 
> ax25_ds_timeout() checks !ax25_dev->dama.slave_timeout, but it's
> initialised by kzalloc() during dev setup, so it will be a noop.
thx. I'll solve it like this then

> 
> 
> > 
> > ```
> > void ax25_ds_set_timer(ax25_dev *ax25_dev)
> > {
> > #ifdef COFNIG_AX25_DAMA_SLAVE
> >         if (ax25_dev == NULL)        ···/* paranoia */
> >                 return;
> > 
> >         ax25_dev->dama.slave_timeout =
> >                 msecs_to_jiffies(ax25_dev->values[AX25_VALUES_DS_TIMEOUT]) / 10;
> >         mod_timer(&ax25_dev->dama.slave_timer, jiffies + HZ);
> > #else
> >         return;
> > #endif
> > }
> > 
> > ```
> > 
> > I'm not too familiar with this, so pointing me to the "correct" way to
> > handle this would be helpfull.
> 
> Also, you will need to guard another use of AX25_VALUES_DS_TIMEOUT in
> ax25_dev_device_up().
Yes. I had noticed this already. This was a trivial one though, so I did
not ask about it.

Thx.

Best

-- 

Joel Granados

signature.asc (application/pgp-signature, 659 B)