tty warning

tty warning

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 1514 bytes --]

Hi Alan,

A sparc allmodconfig build gives these warnings:

drivers/char/n_tty.c: In function 'n_tty_poll':
drivers/char/n_tty.c:1592: warning: array subscript is above array bounds
drivers/char/n_tty.c:1601: warning: array subscript is above array bounds
drivers/char/n_tty.c: In function 'n_tty_read':
drivers/char/n_tty.c:1306: warning: array subscript is above array bounds
drivers/char/tty_ioctl.c: In function 'set_termios':
drivers/char/tty_ioctl.c:574: warning: array subscript is above array bounds
drivers/char/tty_ioctl.c:578: warning: array subscript is above array bounds
drivers/char/tty_ioctl.c: In function 'tty_mode_ioctl':
drivers/char/tty_ioctl.c:729: warning: array subscript is above array bounds
drivers/char/tty_ioctl.c:963: warning: array subscript is above array bounds
drivers/char/tty_ioctl.c:969: warning: array subscript is above array bounds
drivers/char/tty_ioctl.c:618: warning: array subscript is above array bounds
drivers/char/tty_ioctl.c:1007: warning: array subscript is above array bounds
drivers/char/tty_ioctl.c:1015: warning: array subscript is above array bounds

These are things like:

                termios->c_cc[VTIME] = 0;

termios is a (struct ktermios *) which has "cc_t c_cc[NCCS];".  NCCS is
17 and VTIME is also 17.  I assume the intention is to actually fill in
_x_cc[0], but why is it done this way and not more directly?

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 197 bytes --]

Re: tty warning

[David Miller]

From: Stephen Rothwell <sfr@canb.auug.org.au>
Date: Thu, 6 Nov 2008 15:24:21 +1100

> These are things like:
> 
>                 termios->c_cc[VTIME] = 0;
> 
> termios is a (struct ktermios *) which has "cc_t c_cc[NCCS];".  NCCS is
> 17 and VTIME is also 17.  I assume the intention is to actually fill in
> _x_cc[0], but why is it done this way and not more directly?

It's just some historical stuff.

I've tried to rectify it, but it is non-trivial to fix and
all it does is pacify a warning rather than fix a real problem
so... ;-)

This has been brought up perhaps 5 or 6 times in the past half
year :)

Re: tty warning

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 1029 bytes --]

Hi Dave,

On Wed, 05 Nov 2008 20:25:57 -0800 (PST) David Miller <davem@davemloft.net> wrote:
>
> From: Stephen Rothwell <sfr@canb.auug.org.au>
> Date: Thu, 6 Nov 2008 15:24:21 +1100
> 
> > These are things like:
> > 
> >                 termios->c_cc[VTIME] = 0;
> > 
> > termios is a (struct ktermios *) which has "cc_t c_cc[NCCS];".  NCCS is
> > 17 and VTIME is also 17.  I assume the intention is to actually fill in
> > _x_cc[0], but why is it done this way and not more directly?
> 
> It's just some historical stuff.
> 
> I've tried to rectify it, but it is non-trivial to fix and
> all it does is pacify a warning rather than fix a real problem
> so... ;-)
> 
> This has been brought up perhaps 5 or 6 times in the past half
> year :)

Yeah, I guess that is when the compilers started complaining.

Oh, well :-(

[BTW, the number of warnings is itself becoming a real problem ... :-( ]
-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 197 bytes --]

Re: tty warning

[Alan Cox]

On Thu, 6 Nov 2008 15:24:21 +1100
Stephen Rothwell <sfr@canb.auug.org.au> wrote:

> Hi Alan,
> 
> A sparc allmodconfig build gives these warnings:

Blame Dave, although its a bit unfair as the Sparc32 does some weirdness
that the compiler didn't use to be smart enough to complain about and is
actually valid for sparc32 at least.

Alan

Re: tty warning

[Robert Reif]

Stephen Rothwell wrote:
> Hi Alan,
>
> A sparc allmodconfig build gives these warnings:
>
> drivers/char/n_tty.c: In function 'n_tty_poll':
> drivers/char/n_tty.c:1592: warning: array subscript is above array bounds
> drivers/char/n_tty.c:1601: warning: array subscript is above array bounds
> drivers/char/n_tty.c: In function 'n_tty_read':
> drivers/char/n_tty.c:1306: warning: array subscript is above array bounds
> drivers/char/tty_ioctl.c: In function 'set_termios':
> drivers/char/tty_ioctl.c:574: warning: array subscript is above array bounds
> drivers/char/tty_ioctl.c:578: warning: array subscript is above array bounds
> drivers/char/tty_ioctl.c: In function 'tty_mode_ioctl':
> drivers/char/tty_ioctl.c:729: warning: array subscript is above array bounds
> drivers/char/tty_ioctl.c:963: warning: array subscript is above array bounds
> drivers/char/tty_ioctl.c:969: warning: array subscript is above array bounds
> drivers/char/tty_ioctl.c:618: warning: array subscript is above array bounds
> drivers/char/tty_ioctl.c:1007: warning: array subscript is above array bounds
> drivers/char/tty_ioctl.c:1015: warning: array subscript is above array bounds
>
> These are things like:
>
>                 termios->c_cc[VTIME] = 0;
>
> termios is a (struct ktermios *) which has "cc_t c_cc[NCCS];".  NCCS is
> 17 and VTIME is also 17.  I assume the intention is to actually fill in
> _x_cc[0], but why is it done this way and not more directly?
>
>   
There is a patch here that fixes it:  
http://marc.info/?l=linux-sparc&m=122048348932653&w=2

Re: tty warning

[David Miller]

From: Robert Reif <reif@earthlink.net>
Date: Thu, 06 Nov 2008 06:52:03 -0500

> There is a patch here that fixes it:  http://marc.info/?l=linux-sparc&m=122048348932653&w=2

Ok, I've applied this, let's see how it goes ;-)