linux-kernel - Re: [PATCH][next] scsi: Replace one-element arrays with flexible-array members

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]

Date:   Thu, 2 Dec 2021 11:23:26 -0800
From:   Kees Cook <keescook@...omium.org>
To:     Steffen Maier <maier@...ux.ibm.com>
Cc:     "Gustavo A. R. Silva" <gustavoars@...nel.org>,
        "James E.J. Bottomley" <jejb@...ux.ibm.com>,
        "Martin K. Petersen" <martin.petersen@...cle.com>,
        Hannes Reinecke <hare@...e.de>, linux-scsi@...r.kernel.org,
        linux-kernel@...r.kernel.org, linux-hardening@...r.kernel.org,
        Javed Hasan <jhasan@...vell.com>,
        Himanshu Madhani <himanshu.madhani@...cle.com>,
        Neerav Parikh <neerav.parikh@...el.com>,
        Ross Brattain <ross.b.brattain@...el.com>,
        Robert Love <robert.w.love@...el.com>
Subject: Re: [PATCH][next] scsi: Replace one-element arrays with
 flexible-array members

On Fri, Nov 05, 2021 at 12:24:13PM +0100, Steffen Maier wrote:
> On 11/5/21 10:11, Gustavo A. R. Silva wrote:
> > Use flexible-array members in struct fc_fdmi_attr_entry and
> > fs_fdmi_attrs instead of one-element arrays, and refactor the
> > code accordingly.
> > 
> > Also, turn the one-element array _port_ in struct fc_fdmi_rpl
> > into a simple object of type struct fc_fdmi_port_name, as it
> > seems there is no more than just one port expected:
> > 
> > $ git grep -nw numport drivers/scsi/
> > drivers/scsi/csiostor/csio_lnode.c:447: reg_pl->numport = htonl(1);
> > drivers/scsi/libfc/fc_encode.h:232:             put_unaligned_be32(1, &ct->payload.rhba.port.numport);
> > 
> > Also, this helps with the ongoing efforts to globally enable
> > -Warray-bounds and get us closer to being able to tighten the
> > FORTIFY_SOURCE routines on memcpy().
> > 
> > https://github.com/KSPP/linux/issues/79
> > Signed-off-by: Gustavo A. R. Silva <gustavoars@...nel.org>
> > ---
> >   drivers/scsi/csiostor/csio_lnode.c | 2 +-
> >   drivers/scsi/libfc/fc_encode.h     | 4 ++--
> >   include/scsi/fc/fc_ms.h            | 6 +++---
> >   3 files changed, 6 insertions(+), 6 deletions(-)
> > 
> > diff --git a/drivers/scsi/csiostor/csio_lnode.c b/drivers/scsi/csiostor/csio_lnode.c
> > index d5ac93897023..cf9dd79ee488 100644
> > --- a/drivers/scsi/csiostor/csio_lnode.c
> > +++ b/drivers/scsi/csiostor/csio_lnode.c
> > @@ -445,7 +445,7 @@ csio_ln_fdmi_dprt_cbfn(struct csio_hw *hw, struct csio_ioreq *fdmi_req)
> >   	/* Register one port per hba */
> >   	reg_pl = (struct fc_fdmi_rpl *)pld;
> >   	reg_pl->numport = htonl(1);
> > -	memcpy(&reg_pl->port[0].portname, csio_ln_wwpn(ln), 8);
> > +	memcpy(&reg_pl->port.portname, csio_ln_wwpn(ln), 8);
> >   	pld += sizeof(*reg_pl);
> > 
> >   	/* Start appending HBA attributes hba */
> > diff --git a/drivers/scsi/libfc/fc_encode.h b/drivers/scsi/libfc/fc_encode.h
> > index 74ae7fd15d8d..5806f99e4061 100644
> > --- a/drivers/scsi/libfc/fc_encode.h
> > +++ b/drivers/scsi/libfc/fc_encode.h
> > @@ -232,7 +232,7 @@ static inline int fc_ct_ms_fill(struct fc_lport *lport,
> >   		put_unaligned_be32(1, &ct->payload.rhba.port.numport);
> >   		/* Port Name */
> >   		put_unaligned_be64(lport->wwpn,
> > -				   &ct->payload.rhba.port.port[0].portname);
> > +				   &ct->payload.rhba.port.port.portname);
> > 
> >   		/* HBA Attributes */
> >   		put_unaligned_be32(numattrs,
> 
> > diff --git a/include/scsi/fc/fc_ms.h b/include/scsi/fc/fc_ms.h
> > index 00191695233a..44fbe84fa664 100644
> > --- a/include/scsi/fc/fc_ms.h
> > +++ b/include/scsi/fc/fc_ms.h
> 
> > @@ -174,7 +174,7 @@ struct fs_fdmi_attrs {
> 
> /*
>  * Registered Port List
> 
> >    */
> >   struct fc_fdmi_rpl {
> >   	__be32				numport;
> > -	struct fc_fdmi_port_name	port[1];
> > +	struct fc_fdmi_port_name	port;
> >   } __attribute__((__packed__));
> 
> While I'm not affected by the change, it feels to me as if these are
> protocol definitions originating in a T11 Fibre Channel standard FC-GS. It's
> a port *list*. Can you "modify" the standard here?
> 
> The fact, that currently existing code users only ever seem to use one
> single port in the list, would be an independent thing to me.

There are three changes made here, and I suspect it might make sense to
split them up.

In a quick look, I see "struct fc_fdmi_attr_entry" has a sizeof() call
against it, so it's not clear if it's safe to switch it to a flexible
array without other changes.

The change to struct fs_fdmi_attrs looks okay, since it appears to be
used only in casts, but it might make sense to use diffoscope on the
changed .o files to validate nothing weird has happened.

For struct fc_dmi_rpl, as long as "numport" is always set/validated to
1, I think this change is fine.

-- 
Kees Cook