CFLAGS=-ggdb2 is fine CFLAGS=-ggdb3 fails Tried both on gcc-5.4.0 and 6.3.0. My guess is -ggdb3 injects something into preprocessor output that makes guile-snarf shell script incorrectly strip source when it generates .x files.
Created attachment 462418 [details] build.log.gz
$ x86_64-pc-linux-gnu-gcc -E alist.c -DHAVE_CONFIG_H -DBUILDING_LIBGUILE=1 -I.. -I.. -I../lib -I../lib -I/usr/lib64/libffi-3.2.1/include -I/dev/shm/portage/dev-scheme/guile-2.0.13-r1/work/guile-2.0.13 -O1 -ggdb3 > alist.ggdb3 $ x86_64-pc-linux-gnu-gcc -E alist.c -DHAVE_CONFIG_H -DBUILDING_LIBGUILE=1 -I.. -I.. -I../lib -I../lib -I/usr/lib64/libffi-3.2.1/include -I/dev/shm/portage/dev-scheme/guile-2.0.13-r1/work/guile-2.0.13 -O1 -ggdb2 > alist.ggdb2 $ diff -U1 alist.ggdb2 alist.ggdb3 That shows a lot of #defines left (i guess for debugging purposes): --- alist.ggdb2 2017-02-04 10:16:59.302577748 +0000 +++ alist.ggdb3 2017-02-04 10:16:54.598566885 +0000 @@ -3,4 +3,267 @@ # 1 "<built-in>" +#define __STDC__ 1 +#define __STDC_VERSION__ 201112L +#define __STDC_UTF_16__ 1 +#define __STDC_UTF_32__ 1 +#define __STDC_HOSTED__ 1 ... As a result post -snarf files change output by containing type declarations: $ ../libguile/guile-snarf -o alist.x.ggdb3 alist.c -DHAVE_CONFIG_H -DBUILDING_LIBGUILE=1 -I.. -I.. -I../lib -I../lib -I/usr/lib64/libffi-3.2.1/include -I/dev/shm/portage/dev-scheme/guile-2.0.13-r1/work/guile-2.0.13 -O1 -ggdb3 $ ../libguile/guile-snarf -o alist.x.ggdb2 alist.c -DHAVE_CONFIG_H -DBUILDING_LIBGUILE=1 -I.. -I.. -I../lib -I../lib -I/usr/lib64/libffi-3.2.1/include -I/dev/shm/portage/dev-scheme/guile-2.0.13-r1/work/guile-2.0.13 -O1 -ggdb2 --- alist.x.ggdb2<----->2017-02-04 10:37:13.286006316 +0000 +++ alist.x.ggdb3<----->2017-02-04 10:36:58.929014223 +0000 @@ -1,3 +1,3 @@ -/* cpp arguments: alist.c -DHAVE_CONFIG_H -DBUILDING_LIBGUILE=1 -I.. -I.. -I../lib -I../lib -I/usr/lib64/libffi-3.2.1/include -I/dev/shm/portage/dev-scheme/guile-2.0.13-r1/ -scm_acons__name = scm_string_to_symbol (scm_acons__name_string); +/* cpp arguments: alist.c -DHAVE_CONFIG_H -DBUILDING_LIBGUILE=1 -I.. -I.. -I../lib -I../lib -I/usr/lib64/libffi-3.2.1/include -I/dev/shm/portage/dev-scheme/guile-2.0.13-r1/ +enum scm_t_foreign_type { scm_acons__raw_objtable[2] = scm_acons__subr_foreign; scm_acons__raw_objtable[3] = scm_acons__name; (((((SCM *)((scm_t_cell *) (((scm_t_bits) (0? (*(volatile SCM *)0=((sc
One of workarounds would be to downgrade debugging level: diff --git a/dev-scheme/guile/guile-2.0.13-r1.ebuild b/dev-scheme/guile/guile-2.0.13-r1.ebuild index ab7f9c651b..7c70ffa47e 100644 --- a/dev-scheme/guile/guile-2.0.13-r1.ebuild +++ b/dev-scheme/guile/guile-2.0.13-r1.ebuild @@ -44,2 +44,5 @@ src_configure() { + # see bug #608190 + replace-flags -ggdb3 -ggdb2 + econf \
Pushed as: https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=aa3c28baa056846ebd14dfbb48058b91018188fd
It is not resolved you changed only: diff --git a/dev-scheme/guile/guile-2.0.14.ebuild b/dev-scheme/guile/guile-2.0.14.ebuild index ab7f9c6..87f2621 100644 --- a/dev-scheme/guile/guile-2.0.14.ebuild +++ b/dev-scheme/guile/guile-2.0.14.ebuild @@ -42,6 +42,9 @@ src_configure() { # see bug #178499 filter-flags -ftree-vectorize + # see bug #608190 + replace-flags -ggdb[3-9] -ggdb2 + econf \ --disable-error-on-warning \ --disable-rpath \
Reverted as: https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=c67a520c832e9021889d428669e4002df03e702d The plan is to get upstream fix first.
Filed http://debbugs.gnu.org/cgi/bugreport.cgi?bug=25803 upstream.
Upstream applied patch as: http://git.savannah.gnu.org/cgit/guile.git/commit/?id=f775ab3654357fcaad294b95efb0b1c16de1eda8
commit 76de3d3b5ee141070edc5af750d6cd9f7281aec6 Author: Amy Liffey <amynka@gentoo.org> Date: Sat Apr 22 16:36:14 2017 +0200 dev-scheme/guile: add upstream patch bug #608190 Good job! Thanks
