[PATCH v3 1/3] Disable SFrame support by default

Indu Bhagat indu.bhagat@oracle.com
Wed Jul 23 00:14:11 GMT 2025


On 7/22/25 8:55 AM, Adhemerval Zanella wrote:
> diff --git a/debug/Makefile b/debug/Makefile
> index 919bf91c42..6fcb30429e 100644
> --- a/debug/Makefile
> +++ b/debug/Makefile
> @@ -285,14 +285,17 @@ LDFLAGS-tst-backtrace4 = -rdynamic
>   LDFLAGS-tst-backtrace5 = -rdynamic
>   LDFLAGS-tst-backtrace6 = -rdynamic
>   
> +$(objpfx)tst-backtrace1: $(shared-thread-library)
> +
>   # When SFrame is enabled, make sure the dwarf unwinder is also exercised.
>   ifeq ($(enable-gsframe),yes)
>   dw_unwind_pair := \
> -tst-backtrace7:tst-backtrace2 \
> -tst-backtrace8:tst-backtrace3 \
> -tst-backtrace9:tst-backtrace4 \
> -tst-backtrace10:tst-backtrace5 \
> -tst-backtrace11:tst-backtrace6
> +tst-backtrace1-nosframe:tst-backtrace1 \
> +tst-backtrace2-nosframe:tst-backtrace2 \
> +tst-backtrace3-nosframe:tst-backtrace3 \
> +tst-backtrace4-nosframe:tst-backtrace4 \
> +tst-backtrace5-nosframe:tst-backtrace5 \
> +tst-backtrace6-nosframe:tst-backtrace6
>   
>   first_column = $(foreach pair,$(dw_unwind_pair),$(word 1,$(subst :, ,$(pair))))
>   tests-dw-unwind = $(patsubst %,$(objpfx)%.out,$(first_column))
> @@ -301,7 +304,7 @@ endif
>   
>   define make-strip-rule
>   $(objpfx)$(word 1,$(subst :, ,$(1))): $(objpfx)$(word 2,$(subst :, ,$(1)))
> -	strip --remove-section=.sframe $$< -o $$@
> +	$(STRIP) --remove-section=.sframe $$< -o $$@
>   endef
>   

Hi Adhemerval,

When I use this patchset with Binutils 2.45, use --enable-sframe and run 
the make check for debug/ dir, I get:

remove-section=.sframe 
/scratch/ibhagat/GLIBC/objdir-sframe/debug/tst-backtrace1 -o 
/scratch/ibhagat/GLIBC/objdir-sframe/debug/tst-backtrace1-nosframe
/bin/sh: line 1: remove-section=.sframe: command not found
make: [Makefile:310: 
/scratch/ibhagat/GLIBC/objdir-sframe/debug/tst-backtrace1-nosframe] 
Error 127 (ignored)

Looks like $(STRIP) is not set.  I tried to look around in the configure 
scripts to see what needs fixing, but I havent found it yet...

Thanks



More information about the Libc-alpha mailing list