Bug#274367: glibc: [amd64] New GLIBC pass to create 32bit libc6-i386 and libc6-dev-i386 packages
I have some small comments on your patch.
On 06-Feb-09 19:52, Aurelien Jarno wrote:
> +# build 32-bit (i386) alternative library
> +GLIBC_PASSES += i386
> +DEB_ARCH_REGULAR_PACKAGES += libc6-i386 libc6-dev-i386
> +libc6-i386_shlib_dep = libc6-i386 (>= $(shlib_dep_ver))
> +i386_configure_target = i486-linux
> +# __i386__ is defined here because Makeconfig uses -undef and the
> +# /usr/include/asm wrappers need that symbol.
> +i386_CC = $(CC) -m32 -D__i386__
> +i386_add-ons = nptl $(add-ons)
> +i386_extra_cflags = -O3 -fomit-frame-pointer -D__USE_STRING_INLINES
> +i386_extra_cflags += -g0 -fno-unit-at-a-time
> +i386_extra_config_options = $(extra_config_options) --with-tls --with-__thread \
> + --includedir=/usr/include/i386-linux --prefix=/emul/ia32-linux/usr
I think that 'i386_extra_cflags' should just be set to '-O3 -g1'. The
other flags were used to circumvent some compilation problems that
were solved with gcc-4.0.
Also 'i386_extra_config_options' does not need to have '--with-tls'
and '--with-__thread' specified explicitly because the new glibc
version 2.3.5 uses these options by default.
The '--includedir=/usr/include/i386-linux' is also no longer necessary
because this is handled automagically by the wrappers in
I also do not have the '-D__i386__' in my own (/usr/lib32 based)
amd64 glibc patch and it compiles and works fine. So I am not sure
if that definition is still necessary.
Thanks again for working on this. I really hope that we will get
the new libc6(-dev)-i386 packages for etch.