Message ID | 1400112251.12666.14.camel@joe-AO725 |
---|---|
State | Changes Requested, archived |
Delegated to: | David Miller |
Headers | show |
On Wed, May 14, 2014 at 5:04 PM, Joe Perches <joe@perches.com> wrote: > Newer versions of gcc produce better code > so convert some macros to static inlines. > > $ gcc --version > gcc (Ubuntu 4.8.2-19ubuntu1) 4.8.2 > Copyright (C) 2013 Free Software Foundation, Inc. > This is free software; see the source for copying conditions. There is NO > warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. > > (x86/defconfig) > > $ size drivers/net/ethernet/broadcom/tg3.o.* > text data bss dec hex filename > 134282 963 0 135245 2104d drivers/net/ethernet/broadcom/tg3.o.new > 134613 963 0 135576 21198 drivers/net/ethernet/broadcom/tg3.o.old > > Signed-off-by: Joe Perches <joe@perches.com> > --- > On Wed, 2014-05-14 at 16:37 -0700, Alexei Starovoitov wrote: >> On Wed, May 14, 2014 at 3:03 PM, Joe Perches <joe@perches.com> wrote: >> > On Wed, 2014-05-14 at 14:52 -0700, Alexei Starovoitov wrote: >> >> I cannot imagine the case where macro would be faster than static inline >> >> unless it wasn't inlined. >> > >> > For an example, look at commit 4153577a8d >> > ("tg3: Use different macros for pci_chip_rev_id accesses") >> > >> > Converting these macros to static inline produces >> > larger/slower code. (at least with gcc 4.7.3) >> > >> > +#define tg3_chip_rev_id(tp) \ >> > + ((tp)->pci_chip_rev_id) >> > +#define tg3_asic_rev(tp) \ >> > + ((tp)->pci_chip_rev_id >> 12) >> > +#define tg3_chip_rev(tp) \ >> > + ((tp)->pci_chip_rev_id >> 8) >> > >> >> hmm. interesting. >> Using gcc 4.7.2 object file size is larger with 'static inline' >> 2893016 vs 2868112 >> but that's due to larger debug info. >> .text is actually smaller 000207c4 vs 00020824 >> and these three calls were inlined (even without __always_inline), >> so I suspect it's better optimized.. >> though better optimized can very well mean slower. > > Compiler optimizers change with every version too. Nice. Would interesting to hear whether performance stayed the same or not. > drivers/net/ethernet/broadcom/tg3.h | 20 ++++++++++++++------ > 1 file changed, 14 insertions(+), 6 deletions(-) > > diff --git a/drivers/net/ethernet/broadcom/tg3.h b/drivers/net/ethernet/broadcom/tg3.h > index 461acca..3d0cf6b 100644 > --- a/drivers/net/ethernet/broadcom/tg3.h > +++ b/drivers/net/ethernet/broadcom/tg3.h > @@ -3416,11 +3416,19 @@ struct tg3 { > * Using statement expression macros to check tp with > * typecheck(struct tg3 *, tp) also creates larger objects. > */ nit: the comment above needs updating. > -#define tg3_chip_rev_id(tp) \ > - ((tp)->pci_chip_rev_id) > -#define tg3_asic_rev(tp) \ > - ((tp)->pci_chip_rev_id >> 12) > -#define tg3_chip_rev(tp) \ > - ((tp)->pci_chip_rev_id >> 8) > +static inline u32 tg3_chip_rev_id(const struct tg3 *tp) > +{ > + return tp->pci_chip_rev_id; > +} > + > +static inline u32 tg3_asic_rev(const struct tg3 *tp) > +{ > + return tp->pci_chip_rev_id >> 12; > +} > + > +static inline u32 tg3_chip_rev(const struct tg3 *tp) > +{ > + return tp->pci_chip_rev_id >> 8; > +} > > #endif /* !(_T3_H) */ > > -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/drivers/net/ethernet/broadcom/tg3.h b/drivers/net/ethernet/broadcom/tg3.h index 461acca..3d0cf6b 100644 --- a/drivers/net/ethernet/broadcom/tg3.h +++ b/drivers/net/ethernet/broadcom/tg3.h @@ -3416,11 +3416,19 @@ struct tg3 { * Using statement expression macros to check tp with * typecheck(struct tg3 *, tp) also creates larger objects. */ -#define tg3_chip_rev_id(tp) \ - ((tp)->pci_chip_rev_id) -#define tg3_asic_rev(tp) \ - ((tp)->pci_chip_rev_id >> 12) -#define tg3_chip_rev(tp) \ - ((tp)->pci_chip_rev_id >> 8) +static inline u32 tg3_chip_rev_id(const struct tg3 *tp) +{ + return tp->pci_chip_rev_id; +} + +static inline u32 tg3_asic_rev(const struct tg3 *tp) +{ + return tp->pci_chip_rev_id >> 12; +} + +static inline u32 tg3_chip_rev(const struct tg3 *tp) +{ + return tp->pci_chip_rev_id >> 8; +} #endif /* !(_T3_H) */