Skip to contentSkip to conversation

m68k: fix compiler warning by properly inlining flat_set_persistent()

RFC/PATCH Ezequiel Garcia 3 participants 6 messages Apr 23, 2012 – Apr 24, 2012 m68k architecture

lore
Share & tools
lore
#0 lore raw

---
To define or to inline that is the question:
The current definition of flat_set_persistent produces a compiler warning; arch/sh/ does it in a different way defining it to a macro that uses persistent var. IMHO, an inline is easier to read.

---
 arch/m68k/include/asm/flat.h |    7 ++++++-
 1 files changed, 6 insertions(+), 1 deletions(-)
arch/m68k/include/asm/flat.h +6 -1
index a0e2907..f9454b8 100644
@@ -11,6 +11,11 @@
 #define	flat_get_addr_from_rp(rp, relval, flags, p)	get_unaligned(rp)
 #define	flat_put_addr_at_rp(rp, val, relval)	put_unaligned(val,rp)
 #define	flat_get_relocate_addr(rel)		(rel)
-#define	flat_set_persistent(relval, p)		0
+
+static inline int flat_set_persistent(unsigned long relval,
+				      unsigned long *persistent)
+{
+	return 0;
+}
 
 #endif /* __M68KNOMMU_FLAT_H__ */
Signature · 1 line
1.7.3.4
#1 lore raw
Quoted 6 lines · Ezequiel Garcia

---
To define or to inline that is the question:
The current definition of flat_set_persistent produces a compiler warning; arch/sh/ does it in a different way defining it to a macro that uses persistent var. IMHO, an inline is easier to read.

What's the compiler warning?
It seems several other nommu arches use the same definition for flat_set_persistent()?

Quoted 22 lines

---
 arch/m68k/include/asm/flat.h |    7 ++++++-

 1 files changed, 6 insertions(+), 1 deletions(-)
arch/m68k/include/asm/flat.h +6 -2
index a0e2907..f9454b8 100644
@@ -11,6 +11,11 @@
  #define        flat_get_addr_from_rp(rp, relval, flags, p)     get_unaligned(rp)
  #define        flat_put_addr_at_rp(rp, val, relval)    put_unaligned(val,rp)
  #define        flat_get_relocate_addr(rel)             (rel)
-#define        flat_set_persistent(relval, p)          0
+
+static inline int flat_set_persistent(unsigned long relval,
+                                     unsigned long *persistent)
+{
+       return 0;
+}
 
  #endif /* __M68KNOMMU_FLAT_H__ */
--
Quoted 1 line

1.7.3.4

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that. -- Linus Torvalds

#2 lore raw

Hi Geert,

Quoted 8 lines · Geert Uytterhoeven
Quoted 6 lines · Ezequiel Garcia

---
To define or to inline that is the question:
The current definition of flat_set_persistent produces a compiler warning; arch/sh/ does it in a different way defining it to a macro that uses persistent var. IMHO, an inline is easier to read.

Quoted 4 lines

What's the compiler warning?
It seems several other nommu arches use the same definition for flat_set_persistent()?

Here's the warning:
fs/binfmt_flat.c: In function ‘load_flat_file’:
fs/binfmt_flat.c:752: warning: unused variable ‘persistent’

Yes, every arch except sh is using the same definition. My first thought was to extend the arch/sh definition:

  #define flat_set_persistent(relval, p)          ({ (void)p; 0; })

but in a conversation in the janitors list I was told that inlining to a return-only function is a common pattern.
Plus, if you compare fs/built-in.o there is no extra code generated:

$ size fs-built-in-flat-inline
   text	   data	    bss	    dec	    hex	filename
 233332	   1908	   1640	 236880	  39d50	built-in-flat-inline
$ size fs/built-in.o
   text	   data	    bss	    dec	    hex	filename
 233332	   1908	   1640	 236880	  39d50	fs/built-in.o

FWIW, this is my compiler, I built it using gentoo's crossdev tool.
gcc version 4.4.5 (Gentoo 4.4.5 p1.3, pie-0.4.5)

So, what's your opinion?
Thanks,
Ezequiel.

#3 lore raw

Hi Ezequiel,

Quoted 4 lines · Ezequiel García
Quoted 2 lines · Geert Uytterhoeven
Quoted 1 line · Ezequiel Garcia

Normally this signed-off-by goes after your patch description (Documentation/SubmittingPatches section 12).

Quoted 6 lines
Quoted 6 lines
Quoted 5 lines

---
To define or to inline that is the question:
The current definition of flat_set_persistent produces a compiler warning; arch/sh/ does it in a different way defining it to a macro that uses persistent var. IMHO, an inline is easier to read.

Quoted 8 lines
Quoted 4 lines

What's the compiler warning?
It seems several other nommu arches use the same definition for flat_set_persistent()?

Here's the warning:
fs/binfmt_flat.c: In function æload_flat_fileÆ:
fs/binfmt_flat.c:752: warning: unused variable æpersistentÆ

I like to see the actual compiler messages in the git message description as well.

Quoted 16 lines

Yes, every arch except sh is using the same definition. My first thought was to extend the arch/sh definition:

   #define flat_set_persistent(relval, p)          ({ (void)p; 0; })

but in a conversation in the janitors list I was told that inlining to a return-only function is a common pattern.
Plus, if you compare fs/built-in.o there is no extra code generated:

$ size fs-built-in-flat-inline
    text	   data	    bss	    dec	    hex	filename
  233332	   1908	   1640	 236880	  39d50	built-in-flat-inline
$ size fs/built-in.o
    text	   data	    bss	    dec	    hex	filename
  233332	   1908	   1640	 236880	  39d50	fs/built-in.o
Quoted 4 lines

FWIW, this is my compiler, I built it using gentoo's crossdev tool.
gcc version 4.4.5 (Gentoo 4.4.5 p1.3, pie-0.4.5)

So, what's your opinion?

I see the same warning on my gcc-4.5.1 based toolchain as well. No surprise really given none of the m68k macros use the persistent arg.

If you want to move that signed-off-by and put the warning text in the message description I will carry this in the m68knommu git tree.

Thanks
Greg

------------------------------------------------------------------------

Greg Ungerer  --  Principal Engineer        EMAIL:     [email protected]
SnapGear Group, McAfee                      PHONE:       +61 7 3435 2888
8 Gardner Close                             FAX:         +61 7 3217 5323
Milton, QLD, 4064, Australia                WEB: http://www.SnapGear.com
#4 lore raw

Hi Greg,

2012/4/23 Greg Ungerer <[email protected]>:

Quoted 3 lines

Normally this signed-off-by goes after your patch description (Documentation/SubmittingPatches section 12).

I know. That text wasn't intended to be part of the commit message, it was to trigger the discussion. I'll resend the patch with a proper commit message, including the warning.

Same warning will appear on every other arch, since the only one using flat_set_persistent() is arch/blackfin...
do you think it would be nice to submit that also (to the proper lists, of course)?

Thanks for the review,
Ezequiel.

#5 lore raw

Hi Ezequiel,

Quoted 8 lines · Ezequiel García

2012/4/23 Greg Ungerer<[email protected]>:

Quoted 3 lines

Normally this signed-off-by goes after your patch description (Documentation/SubmittingPatches section 12).

I know. That text wasn't intended to be part of the commit message, it was to trigger the discussion. I'll resend the patch with a proper commit message, including the warning.

Ok, sorry, I thought that was supposed to be the commit message.

Quoted 4 lines

Same warning will appear on every other arch, since the only one using flat_set_persistent() is arch/blackfin...
do you think it would be nice to submit that also (to the proper lists, of course)?

Yes, I think that is a good idea.

Regards
Greg

------------------------------------------------------------------------

Greg Ungerer  --  Principal Engineer        EMAIL:     [email protected]
SnapGear Group, McAfee                      PHONE:       +61 7 3435 2888
8 Gardner Close                             FAX:         +61 7 3217 5323
Milton, QLD, 4064, Australia                WEB: http://www.SnapGear.com