linux-kernel - Re: [PATCH v2] scsi: NCR5380: Mark expected switch fall-through

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:   Fri, 1 Mar 2019 12:16:57 +1100 (AEDT)
From:   Finn Thain <fthain@...egraphics.com.au>
To:     "Gustavo A. R. Silva" <gustavo@...eddedor.com>
cc:     Michael Schmitz <schmitzmic@...il.com>,
        "James E.J. Bottomley" <jejb@...ux.ibm.com>,
        "Martin K. Petersen" <martin.petersen@...cle.com>,
        linux-scsi@...r.kernel.org, linux-kernel@...r.kernel.org,
        Kees Cook <keescook@...omium.org>
Subject: Re: [PATCH v2] scsi: NCR5380: Mark expected switch fall-through

On Thu, 28 Feb 2019, Gustavo A. R. Silva wrote:

> In preparation to enabling -Wimplicit-fallthrough, mark switch
> cases where we are expecting to fall through.
> 

This switch case is already marked. So I think the patch description 
should state that this patch is actually a workaround for a gcc deficiency 
which prevents it from locating the marker.

> This patch fixes the following warning:
> 
> In file included from drivers/scsi/dmx3191d.c:48:
> drivers/scsi/NCR5380.c: In function ?NCR5380_information_transfer?:
> drivers/scsi/NCR5380.c:1933:9: warning: this statement may fall through [-Wimplicit-fallthrough=]
>       if (!hostdata->connected)
>          ^
> drivers/scsi/NCR5380.c:1937:5: note: here
>      default:
>      ^~~~~~~
> 
> Warning level 3 was used: -Wimplicit-fallthrough=3
> 
> Notice that, in this particular case, the code comment is modified
> in accordance with what GCC is expecting to find.
> 
> This patch is part of the ongoing efforts to enable
> -Wimplicit-fallthrough.
> 
> Signed-off-by: Gustavo A. R. Silva <gustavo@...eddedor.com>
> ---
> Changes in v2:
>  - Update commit log.
>  - Move code comment after the default label and
>    retain reason for fall-through in comment as
>    requested by Michael Schmitz.
> 
>  drivers/scsi/NCR5380.c | 9 ++++-----
>  1 file changed, 4 insertions(+), 5 deletions(-)
> 
> diff --git a/drivers/scsi/NCR5380.c b/drivers/scsi/NCR5380.c
> index 01c23d27f290..985d1c053578 100644
> --- a/drivers/scsi/NCR5380.c
> +++ b/drivers/scsi/NCR5380.c
> @@ -1933,13 +1933,12 @@ static void NCR5380_information_transfer(struct Scsi_Host *instance)
>  					if (!hostdata->connected)
>  						return;
>  
> -					/* Fall through to reject message */
> -
> +					/* Fall through - to reject message */

This new hyphen is wrong and harms readability for humans.

I did confirm that gcc can be appeased by the use of a hyphen but not by 
correct grammar such as "Fall through to reject message" or "Fall through. 
Reject message."

> +				default:
>  					/*
> -					 * If we get something weird that we aren't expecting,
> -					 * reject it.
> +					 * If we get something weird that we
> +					 * aren't expecting, reject it.

This reformatting isn't relevant to this patch. The comments can be 
improved however (see below).

>  					 */
> -				default:

Moving the 'default' keyword closer to the 'fall through' comment makes 
sense to me -- I could understand if gcc had simple, unambiguous rules for 
annotations.

Do compilers and static analysers agree as to what a correctly annotated 
switch label should look like? If not, we would have to try to mangle code 
and comments in such a way that might satisfy all of the failings in all 
of the tools.

>  					if (tmp == EXTENDED_MESSAGE)
>  						scmd_printk(KERN_INFO, cmd,
>  						            "rejecting unknown extended message code %02x, length %d\n",
> 

Here's an alternative patch, which has the virtue that a simple heuristic 
will work. This patch does not require that other static analysis tools 
will follow gcc's weird rules about hyphens. (I assume they don't but I 
didn't check.)

diff --git a/drivers/scsi/NCR5380.c b/drivers/scsi/NCR5380.c
index 7fed9bb72784..fe0535affc14 100644
--- a/drivers/scsi/NCR5380.c
+++ b/drivers/scsi/NCR5380.c
@@ -1932,13 +1932,13 @@ static void NCR5380_information_transfer(struct Scsi_Host *instance)
 					if (!hostdata->connected)
 						return;
 
-					/* Fall through to reject message */
-
+					/* Reject message */
+					/* Fall through */
+				default:
 					/*
 					 * If we get something weird that we aren't expecting,
-					 * reject it.
+					 * log it.
 					 */
-				default:
 					if (tmp == EXTENDED_MESSAGE)
 						scmd_printk(KERN_INFO, cmd,
 						            "rejecting unknown extended message code %02x, length %d\n",

--