[PATCH] fix warning in 8250.c

[PATCH] fix warning in 8250.c

[Denis Vlasenko]

[-- Attachment #1: Type: text/plain, Size: 410 bytes --]

  CC      drivers/serial/8250.o
/.1/usr/srcdevel/kernel/linux-2.6.15-rc7.src/drivers/serial/8250.c:1085: warning: 'transmit_chars' declared inline after being called
/.1/usr/srcdevel/kernel/linux-2.6.15-rc7.src/drivers/serial/8250.c:1085: warning: previous declaration of 'transmit_chars' was here

Since this function is not small, inlining effect is way below noise floor.
Let's just remove _INLINE_.
--
vda

[-- Attachment #2: linux-2.6.15-rc7.inline.patch --]
[-- Type: text/x-diff, Size: 809 bytes --]

  CC      drivers/serial/8250.o
/.1/usr/srcdevel/kernel/linux-2.6.15-rc7.src/drivers/serial/8250.c:1085: warning: 'transmit_chars' declared inline after being called
/.1/usr/srcdevel/kernel/linux-2.6.15-rc7.src/drivers/serial/8250.c:1085: warning: previous declaration of 'transmit_chars' was here


diff -urpN linux-2.6.15-rc7.src/drivers/serial/8250.c linux-2.6.15-rc7.fix/drivers/serial/8250.c
--- linux-2.6.15-rc7.src/drivers/serial/8250.c	Fri Dec 30 14:18:03 2005
+++ linux-2.6.15-rc7.fix/drivers/serial/8250.c	Sun Jan  1 16:56:17 2006
@@ -1217,7 +1217,7 @@ receive_chars(struct uart_8250_port *up,
 	*status = lsr;
 }
 
-static _INLINE_ void transmit_chars(struct uart_8250_port *up)
+static void transmit_chars(struct uart_8250_port *up)
 {
 	struct circ_buf *xmit = &up->port.info->xmit;
 	int count;

Re: [PATCH] fix warning in 8250.c

[Russell King]

On Tue, Jan 03, 2006 at 10:12:48AM +0200, Denis Vlasenko wrote:
>   CC      drivers/serial/8250.o
> /.1/usr/srcdevel/kernel/linux-2.6.15-rc7.src/drivers/serial/8250.c:1085: warning: 'transmit_chars' declared inline after being called
> /.1/usr/srcdevel/kernel/linux-2.6.15-rc7.src/drivers/serial/8250.c:1085: warning: previous declaration of 'transmit_chars' was here
> 
> Since this function is not small, inlining effect is way below noise floor.
> Let's just remove _INLINE_.

I think we want to remove _INLINE_ from both receive_chars and
transmit_chars.  Both functions aren't small, so...

-- 
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:  2.6 Serial core

Re: [PATCH] fix warning in 8250.c

[Christoph Hellwig]

On Wed, Jan 04, 2006 at 06:14:25PM +0000, Russell King wrote:
> On Tue, Jan 03, 2006 at 10:12:48AM +0200, Denis Vlasenko wrote:
> >   CC      drivers/serial/8250.o
> > /.1/usr/srcdevel/kernel/linux-2.6.15-rc7.src/drivers/serial/8250.c:1085: warning: 'transmit_chars' declared inline after being called
> > /.1/usr/srcdevel/kernel/linux-2.6.15-rc7.src/drivers/serial/8250.c:1085: warning: previous declaration of 'transmit_chars' was here
> > 
> > Since this function is not small, inlining effect is way below noise floor.
> > Let's just remove _INLINE_.
> 
> I think we want to remove _INLINE_ from both receive_chars and
> transmit_chars.  Both functions aren't small, so...

While we're at it can we please kill _INLINE_?  Those functions that should
be inlined can become inline, but this macro just obsfucates the serial code.

Re: [PATCH] fix warning in 8250.c

[Russell King]

On Wed, Jan 04, 2006 at 06:18:01PM +0000, Christoph Hellwig wrote:
> On Wed, Jan 04, 2006 at 06:14:25PM +0000, Russell King wrote:
> > On Tue, Jan 03, 2006 at 10:12:48AM +0200, Denis Vlasenko wrote:
> > >   CC      drivers/serial/8250.o
> > > /.1/usr/srcdevel/kernel/linux-2.6.15-rc7.src/drivers/serial/8250.c:1085: warning: 'transmit_chars' declared inline after being called
> > > /.1/usr/srcdevel/kernel/linux-2.6.15-rc7.src/drivers/serial/8250.c:1085: warning: previous declaration of 'transmit_chars' was here
> > > 
> > > Since this function is not small, inlining effect is way below noise floor.
> > > Let's just remove _INLINE_.
> > 
> > I think we want to remove _INLINE_ from both receive_chars and
> > transmit_chars.  Both functions aren't small, so...
> 
> While we're at it can we please kill _INLINE_?  Those functions that should
> be inlined can become inline, but this macro just obsfucates the serial code.

No idea - I don't know about x86 nuances and why they wanted:

#if defined(__i386__) && (defined(CONFIG_M386) || defined(CONFIG_M486))
#define _INLINE_ inline
#else
#define _INLINE_
#endif

Maybe someone in the x86 world needs to comment?  Does the above even
mean that we'll ever inline anything marked _INLINE_ ?

-- 
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:  2.6 Serial core

Re: [PATCH] fix warning in 8250.c

[Andrew Morton]

Russell King <rmk+lkml@arm.linux.org.uk> wrote:
>
> > While we're at it can we please kill _INLINE_?  Those functions that should
> > be inlined can become inline, but this macro just obsfucates the serial code.
> 
> No idea - I don't know about x86 nuances and why they wanted:
> 
> #if defined(__i386__) && (defined(CONFIG_M386) || defined(CONFIG_M486))
> #define _INLINE_ inline
> #else
> #define _INLINE_
> #endif
> 
> Maybe someone in the x86 world needs to comment?  Does the above even
> mean that we'll ever inline anything marked _INLINE_ ?

I suspect the reasoning behind this is lost in the mists of pre-bk time. 
I'd be inclined to just nuke it and see what happens.  Nothing, I expect.