| Message ID | 1453243871-22360-1-git-send-email-arnout@mind.be |
|---|---|
| State | Accepted |
| Commit | 86e97855724f77817749cf6a036a352ad9b99cca |
| Headers | show |
>>>>> "Arnout" == Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be> writes: > libnspr currently passes --enable-thumb2 if the CPU has thumb > instructions. This option will pass -mthumb to the compiler. However, > if an external multilib toolchain is used that has a thumb-specific > variant (e.g. Sourcery), it will try to use that one. But we only copy > a single variant to the sysroot, so the build will fail with: > .../arm-none-linux-gnueabi/bin/ld: cannot find crti.o: No such file or directory > ... > collect2: error: ld returned 1 exit status > ../../config/rules.mk:303: recipe for target 'libnspr4.so' failed > We can in fact just remove the thumb2 handling. With current libnspr, > the thumb and thumb2 options just add -marm and -mthumb. But we already > pass that in our toolchain wrapper so it's completely redundant. > Note that when nothing is passed, the configure script still tries to > autodetect whether thumb2 is available (but doesn't do it correctly, > see the error above), but in the end it doesn't use the result for > anything. In other words, even if it detects that thumb2 is available, > it will _not_ pass -mthumb to the compiler. > Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be> > --- > Build-tested with various combinations of thumb, thumb2, floating point, > and external toolchains. I didn't test with an internal toolchain but > what could go wrong there? :-) > Also tested with the original build failure which led to the > introduction of the --enable-thumb2 handling: > http://autobuild.buildroot.org/results/d7323831372050e425a34f5104a46d8cbd6be214 > Changes v2: added missing build failure output to commit message > Changes v3: Remove thumb2 handling completely Committed, thanks.
diff --git a/package/libnspr/libnspr.mk b/package/libnspr/libnspr.mk index 8e58986..9f49eef 100644 --- a/package/libnspr/libnspr.mk +++ b/package/libnspr/libnspr.mk @@ -49,12 +49,4 @@ LIBNSPR_INSTALL_TARGET_OPTS = DESTDIR=$(TARGET_DIR) LIBRARY= install LIBNSPR_INSTALL_STAGING_OPTS = DESTDIR=$(STAGING_DIR) LIBRARY= install endif -ifeq ($(BR2_arm),y) -ifeq ($(BR2_ARM_CPU_HAS_THUMB2),y) -LIBNSPR_CONF_OPTS += --enable-thumb2 -else -LIBNSPR_CONF_OPTS += --disable-thumb2 -endif -endif - $(eval $(autotools-package))
libnspr currently passes --enable-thumb2 if the CPU has thumb instructions. This option will pass -mthumb to the compiler. However, if an external multilib toolchain is used that has a thumb-specific variant (e.g. Sourcery), it will try to use that one. But we only copy a single variant to the sysroot, so the build will fail with: .../arm-none-linux-gnueabi/bin/ld: cannot find crti.o: No such file or directory ... collect2: error: ld returned 1 exit status ../../config/rules.mk:303: recipe for target 'libnspr4.so' failed We can in fact just remove the thumb2 handling. With current libnspr, the thumb and thumb2 options just add -marm and -mthumb. But we already pass that in our toolchain wrapper so it's completely redundant. Note that when nothing is passed, the configure script still tries to autodetect whether thumb2 is available (but doesn't do it correctly, see the error above), but in the end it doesn't use the result for anything. In other words, even if it detects that thumb2 is available, it will _not_ pass -mthumb to the compiler. Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be> --- Build-tested with various combinations of thumb, thumb2, floating point, and external toolchains. I didn't test with an internal toolchain but what could go wrong there? :-) Also tested with the original build failure which led to the introduction of the --enable-thumb2 handling: http://autobuild.buildroot.org/results/d7323831372050e425a34f5104a46d8cbd6be214 Changes v2: added missing build failure output to commit message Changes v3: Remove thumb2 handling completely --- package/libnspr/libnspr.mk | 8 -------- 1 file changed, 8 deletions(-)