| Message ID | 1342184620-31448-1-git-send-email-hkronsto@frequentis.com |
|---|---|
| State | Accepted |
| Headers | show |
On Friday 13 July 2012 09:03:40 Horst Kronstorfer wrote: > Add '-undef' to DTS_CPPFLAGS to avoid unwanted expansion of dts content > that matches system-specific or gcc-specific predefined macros. > > Example: A number of PowerPC related *.dts files in the kernel define a > property named 'linux,network-index' which (w/o '-undef') is expanded to > '1,network-index' by the preprocessor because of '#define linux 1.' i think you should use -ansi instead. that's what we use in other places for the same reason. -mike
On 07/19/2012 05:22 AM, Mike Frysinger wrote: > On Friday 13 July 2012 09:03:40 Horst Kronstorfer wrote: >> Add '-undef' to DTS_CPPFLAGS to avoid unwanted expansion of dts content >> that matches system-specific or gcc-specific predefined macros. >> >> Example: A number of PowerPC related *.dts files in the kernel define a >> property named 'linux,network-index' which (w/o '-undef') is expanded to >> '1,network-index' by the preprocessor because of '#define linux 1.' > > i think you should use -ansi instead. that's what we use in other places for > the same reason. > -mike > this would increase the probability of a name clash. -h > > > _______________________________________________ > U-Boot mailing list > U-Boot@lists.denx.de > http://lists.denx.de/mailman/listinfo/u-boot >
On Tuesday 24 July 2012 06:11:04 Horst Kronstorfer wrote: > On 07/19/2012 05:22 AM, Mike Frysinger wrote: > > On Friday 13 July 2012 09:03:40 Horst Kronstorfer wrote: > >> Add '-undef' to DTS_CPPFLAGS to avoid unwanted expansion of dts content > >> that matches system-specific or gcc-specific predefined macros. > >> > >> Example: A number of PowerPC related *.dts files in the kernel define a > >> property named 'linux,network-index' which (w/o '-undef') is expanded to > >> '1,network-index' by the preprocessor because of '#define linux 1.' > > > > i think you should use -ansi instead. that's what we use in other places > > for the same reason. > > this would increase the probability of a name clash. no idea what you're talking about. have you actually looked at the output of `gcc -E -dD -ansi` ? -mike
On 07/24/2012 05:28 PM, Mike Frysinger wrote: > On Tuesday 24 July 2012 06:11:04 Horst Kronstorfer wrote: >> On 07/19/2012 05:22 AM, Mike Frysinger wrote: >>> On Friday 13 July 2012 09:03:40 Horst Kronstorfer wrote: >>>> Add '-undef' to DTS_CPPFLAGS to avoid unwanted expansion of dts content >>>> that matches system-specific or gcc-specific predefined macros. >>>> >>>> Example: A number of PowerPC related *.dts files in the kernel define a >>>> property named 'linux,network-index' which (w/o '-undef') is expanded to >>>> '1,network-index' by the preprocessor because of '#define linux 1.' >>> >>> i think you should use -ansi instead. that's what we use in other places >>> for the same reason. >> >> this would increase the probability of a name clash. > > no idea what you're talking about. have you actually looked at the output of > `gcc -E -dD -ansi` ? $ gcc -E -dM -ansi - </dev/null | wc -l 229 $ gcc -E -dM -undef - </dev/null | wc -l 2 -h > -mike >
On Tuesday 24 July 2012 15:38:55 Horst Kronstorfer wrote: > On 07/24/2012 05:28 PM, Mike Frysinger wrote: > > On Tuesday 24 July 2012 06:11:04 Horst Kronstorfer wrote: > >> On 07/19/2012 05:22 AM, Mike Frysinger wrote: > >>> On Friday 13 July 2012 09:03:40 Horst Kronstorfer wrote: > >>>> Add '-undef' to DTS_CPPFLAGS to avoid unwanted expansion of dts > >>>> content that matches system-specific or gcc-specific predefined > >>>> macros. > >>>> > >>>> Example: A number of PowerPC related *.dts files in the kernel define > >>>> a property named 'linux,network-index' which (w/o '-undef') is > >>>> expanded to '1,network-index' by the preprocessor because of '#define > >>>> linux 1.' > >>> > >>> i think you should use -ansi instead. that's what we use in other > >>> places for the same reason. > >> > >> this would increase the probability of a name clash. > > > > no idea what you're talking about. have you actually looked at the > > output of `gcc -E -dD -ansi` ? > > $ gcc -E -dM -ansi - </dev/null | wc -l > 229 > $ gcc -E -dM -undef - </dev/null | wc -l > 2 and ? did you *look* at the output ? they're all of the form __foo__. do any device trees really use __foo__ names ? i don't think so. -mike
On 07/25/2012 06:06 PM, Mike Frysinger wrote: > On Tuesday 24 July 2012 15:38:55 Horst Kronstorfer wrote: >> On 07/24/2012 05:28 PM, Mike Frysinger wrote: >>> On Tuesday 24 July 2012 06:11:04 Horst Kronstorfer wrote: >>>> On 07/19/2012 05:22 AM, Mike Frysinger wrote: >>>>> On Friday 13 July 2012 09:03:40 Horst Kronstorfer wrote: >>>>>> Add '-undef' to DTS_CPPFLAGS to avoid unwanted expansion of dts >>>>>> content that matches system-specific or gcc-specific predefined >>>>>> macros. >>>>>> >>>>>> Example: A number of PowerPC related *.dts files in the kernel define >>>>>> a property named 'linux,network-index' which (w/o '-undef') is >>>>>> expanded to '1,network-index' by the preprocessor because of '#define >>>>>> linux 1.' >>>>> >>>>> i think you should use -ansi instead. that's what we use in other >>>>> places for the same reason. >>>> >>>> this would increase the probability of a name clash. >>> >>> no idea what you're talking about. have you actually looked at the >>> output of `gcc -E -dD -ansi` ? >> >> $ gcc -E -dM -ansi - </dev/null | wc -l >> 229 >> $ gcc -E -dM -undef - </dev/null | wc -l >> 2 > > and ? did you *look* at the output ? they're all of the form __foo__. i already considered that. > do any device trees really use __foo__ names ? i don't think so. is there any drawback using '-undef' in this particular case besides "we use -ansi in other places for the same reason"? -h > -mike >
On Thursday 26 July 2012 03:06:06 Horst Kronstorfer wrote: > On 07/25/2012 06:06 PM, Mike Frysinger wrote: > > On Tuesday 24 July 2012 15:38:55 Horst Kronstorfer wrote: > >> On 07/24/2012 05:28 PM, Mike Frysinger wrote: > >>> On Tuesday 24 July 2012 06:11:04 Horst Kronstorfer wrote: > >>>> On 07/19/2012 05:22 AM, Mike Frysinger wrote: > >>>>> On Friday 13 July 2012 09:03:40 Horst Kronstorfer wrote: > >>>>>> Add '-undef' to DTS_CPPFLAGS to avoid unwanted expansion of dts > >>>>>> content that matches system-specific or gcc-specific predefined > >>>>>> macros. > >>>>>> > >>>>>> Example: A number of PowerPC related *.dts files in the kernel > >>>>>> define a property named 'linux,network-index' which (w/o '-undef') > >>>>>> is expanded to '1,network-index' by the preprocessor because of > >>>>>> '#define linux 1.' > >>>>> > >>>>> i think you should use -ansi instead. that's what we use in other > >>>>> places for the same reason. > >>>> > >>>> this would increase the probability of a name clash. > >>> > >>> no idea what you're talking about. have you actually looked at the > >>> output of `gcc -E -dD -ansi` ? > >> > >> $ gcc -E -dM -ansi - </dev/null | wc -l > >> 229 > >> $ gcc -E -dM -undef - </dev/null | wc -l > >> 2 > > > > and ? did you *look* at the output ? they're all of the form __foo__. > > i already considered that. > > > do any device trees really use __foo__ names ? i don't think so. > > is there any drawback using '-undef' in this particular case besides > "we use -ansi in other places for the same reason"? we use -ansi because it's more portable and has known behavior in the build system -mike
Dear Horst Kronstorfer, In message <1342184620-31448-1-git-send-email-hkronsto@frequentis.com> you wrote: > Add '-undef' to DTS_CPPFLAGS to avoid unwanted expansion of dts content > that matches system-specific or gcc-specific predefined macros. > > Example: A number of PowerPC related *.dts files in the kernel define a > property named 'linux,network-index' which (w/o '-undef') is expanded to > '1,network-index' by the preprocessor because of '#define linux 1.' > > With '-undef' in place only the standard predefined macros are visible: > > $ gcc -dM -E -undef - </dev/null > #define __STDC_HOSTED__ 1 > #define __STDC__ 1 > > Signed-off-by: Horst Kronstorfer <hkronsto@frequentis.com> > --- > dts/Makefile | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) Applied, thanks. Best regards, Wolfgang Denk
diff --git a/dts/Makefile b/dts/Makefile index 402dfe1..3c6bc18 100644 --- a/dts/Makefile +++ b/dts/Makefile @@ -36,7 +36,7 @@ $(error Your architecture does not have device tree support enabled. \ Please define CONFIG_ARCH_DEVICE_TREE)) # We preprocess the device tree file provide a useful define -DTS_CPPFLAGS := -DARCH_CPU_DTS=\"$(SRCTREE)/arch/$(ARCH)/dts/$(CONFIG_ARCH_DEVICE_TREE).dtsi\" +DTS_CPPFLAGS := -undef -DARCH_CPU_DTS=\"$(SRCTREE)/arch/$(ARCH)/dts/$(CONFIG_ARCH_DEVICE_TREE).dtsi\" all: $(obj).depend $(LIB)
Add '-undef' to DTS_CPPFLAGS to avoid unwanted expansion of dts content that matches system-specific or gcc-specific predefined macros. Example: A number of PowerPC related *.dts files in the kernel define a property named 'linux,network-index' which (w/o '-undef') is expanded to '1,network-index' by the preprocessor because of '#define linux 1.' With '-undef' in place only the standard predefined macros are visible: $ gcc -dM -E -undef - </dev/null #define __STDC_HOSTED__ 1 #define __STDC__ 1 Signed-off-by: Horst Kronstorfer <hkronsto@frequentis.com> --- dts/Makefile | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)