This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
[Bug bootstrap/12974] [3.4/3.5 Regression] gcc-3.4/3.5-20031105 bootstrap failure [MinGW]
- From: "dannysmith at gcc dot gnu dot org" <gcc-bugzilla at gcc dot gnu dot org>
- To: gcc-bugs at gcc dot gnu dot org
- Date: 19 Feb 2004 01:50:57 -0000
- Subject: [Bug bootstrap/12974] [3.4/3.5 Regression] gcc-3.4/3.5-20031105 bootstrap failure [MinGW]
- References: <20031109014258.12974.stl@caltech.edu>
- Reply-to: gcc-bugzilla at gcc dot gnu dot org
------- Additional Comments From dannysmith at gcc dot gnu dot org 2004-02-19 01:50 -------
The relevant part of the gcc makefile is this:
in 3.4
INCLUDES = -I. -I$(@D) -I$(srcdir) -I$(srcdir)/$(@D) \
-I$(srcdir)/../include
in 3.3
INCLUDES = -I. -I$(@D) -I$(srcdir) -I$(srcdir)/$(@D) \
-I$(srcdir)/config -I$(srcdir)/../include
How does msys version of make expand @D automatic variable,
when the target filename is of the form libgcc/./_muldi3.o
-I$(@D) -I$(srcdir)
may cause problems if it is interpreted as -I -I$(srcdir).
(which is what your output seems to suggest)
The search will be for a nonexistent directory named -I$(srcdir)
rather than two directories: -Ilibgcc -I$(srcdir)
In 3.3 the include of -I$(srcdir)/config would have allowed it to find
config/i386/ regardless of handling of @D.
Can you add -v to TARGET_LIBGCC2_CFLAGS to see what directories are being
searched.
Danny
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12974