Re: [PATCH] ip2: fix compile warnings

[Jim Nelson]

Randy.Dunlap wrote:
> james4765@xxxxxxxxxxx wrote:
>
> > This fixes the following compile errors in the ip2 and ip2main drivers:
> >
> >   CC      drivers/char/ip2main.o
> > drivers/char/ip2main.c:470: warning: initialization from incompatible 
> > pointer type
>
>
>
> > diff -urN --exclude='*~' 
> > linux-2.6.10-rc3-mm1-original/drivers/char/ip2main.c 
> > linux-2.6.10-rc3-mm1/drivers/char/ip2main.c
> > --- linux-2.6.10-rc3-mm1-original/drivers/char/ip2main.c    2004-12-03 
> > 16:55:03.000000000 -0500
> > +++ linux-2.6.10-rc3-mm1/drivers/char/ip2main.c    2004-12-17 
> > 16:24:24.094730049 -0500
> > @@ -467,7 +466,7 @@
> >  static struct tty_operations ip2_ops = {
> >      .open            = ip2_open,
> >      .close           = ip2_close,
> > -    .write           = ip2_write,
> > +    .write           = (void *) ip2_write,
> >      .put_char        = ip2_putchar,
> >      .flush_chars     = ip2_flush_chars,
> >      .write_room      = ip2_write_room,
>
>
> The write() prototype in tty_operations is:
>     int  (*write)(struct tty_struct * tty,
>               const unsigned char *buf, int count);
>
> Somehow the cast does eliminate the compiler warning (and give
> a false sense of correctness).
>
> However, ip2main.c::ip2_write() should be modified like so:
>
> static int
> ip2_write( PTTY tty, const unsigned char *pData, int count)
>
> and drop the cast and fix the ip2_write comment (drop old arg 2),
> and fix the ip2_write() prototype.
> But then you (someone) will have to decide how to handle the
> dropped <user> parameter when calling i2Output()...
> I don't know the answer to that.
> I just changed <user> to 0 to get a clean build of ip2main.o,
> but ip2/i2lib.c still needs some work.

Sorry for the constant n00b questions, but:

Is there anything outside the kernel that could call tty_operations.write?

drivers/input/serio/serport.c uses: (as an example)

static int serport_serio_write(struct serio *serio, unsigned char data)
{
	struct serport *serport = serio->port_data;
	return -(serport->tty->driver->write(serport->tty, &data, 1) != 1);
}

I'm guessing that something does copy_from_user() before tty_operations.write is 
called, but I don't know quite what that is.

Can anyone to point me in the direction of where the user/kernel interface for tty 
devices is?

Given the way this is set up -

int  (*write)(struct tty_struct * tty, const unsigned char *buf, int count);

vs.

ip2_write( PTTY tty, int user, const unsigned char *pData, int count)

I don't even know if the driver would work - I think you'd have serious problems 
as it tries to dereference a pointer that is half-integer.

Am I reading this wrong?

Jim
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/