Date: Sun, 26 Aug 2012 20:32:09 +0200 (CEST) From: Tijl Coosemans <tijl@coosemans.org> To: FreeBSD-gnats-submit@FreeBSD.org Subject: ports/171086: [patch] devel/gdb: various fixes Message-ID: <201208261832.q7QIW87Z010021@kalimero.tijl.coosemans.org> Resent-Message-ID: <201208261840.q7QIe3Y9047196@freefall.freebsd.org>
next in thread | raw e-mail | index | archive | help
>Number: 171086
>Category: ports
>Synopsis: [patch] devel/gdb: various fixes
>Confidential: no
>Severity: non-critical
>Priority: low
>Responsible: freebsd-ports-bugs
>State: open
>Quarter:
>Keywords:
>Date-Required:
>Class: sw-bug
>Submitter-Id: current-users
>Arrival-Date: Sun Aug 26 18:40:03 UTC 2012
>Closed-Date:
>Last-Modified:
>Originator: Tijl Coosemans
>Release: FreeBSD 10.0-CURRENT i386
>Organization:
>Environment:
>Description:
The attached patch fixes some issues with devel/gdb:
1) Some files are extracted with 0666 permissions which allows anyone
to modify the source while building. The fix is to add --no-same-owner
--no-same-permissions to EXTRACT_AFTER_ARGS like bsd.port.mk does.
2) Threading support is currently broken. When you run gdb it prints an
error that says it failed to load libthread_db.so because it cannot
resolve ps_pread. This function is defined in fbsd-threads.c and gdb
has to be linked with -Wl,-E to export it. This flag used to be added by
patch-gdb-Makefile.in, but that has been removed in the update to gdb 7.5.
I've added it back to extra-patch-gdb-Makefile.in.
3) Threading support on FreeBSD/i386 10-CURRENT is currently broken
because two configure tests that test existence of r_fs and r_gs fields
in struct reg fail. This causes gdb to overwrite %fs and %gs registers
with zero which makes threaded programs crash. The tests are missing
sys/types.h as a prerequisite header. (The test for the existence of
struct reg itself already uses sys/types.h, so I simply copied that.)
>How-To-Repeat:
>Fix:
--- gdb.patch begins here ---
diff -urN /usr/ports/devel/gdb/Makefile ./Makefile
--- /usr/ports/devel/gdb/Makefile 2012-08-24 04:54:50.000000000 +0200
+++ ./Makefile 2012-08-26 20:16:49.000000000 +0200
@@ -7,6 +7,7 @@
PORTNAME= gdb
PORTVERSION= 7.5
+PORTREVISION= 1
CATEGORIES= devel
MASTER_SITES= GNU
@@ -30,7 +31,8 @@
CFLAGS:= ${CFLAGS:C/ +$//} # blanks at EOL creep in sometimes
CFLAGS+= -DRL_NO_COMPAT
EXCLUDE= dejagnu expect readline sim texinfo intl
-EXTRACT_AFTER_ARGS=| ${TAR} -xf - ${EXCLUDE:S/^/--exclude /}
+EXTRACT_AFTER_ARGS= | ${TAR} -xf - ${EXCLUDE:S/^/--exclude /} \
+ --no-same-owner --no-same-permissions
VER= ${PORTVERSION:S/.//g}
PLIST_SUB= VER=${VER}
MAN1= gdb${VER}.1
diff -urN /usr/ports/devel/gdb/files/extrapatch-gdb-Makefile.in ./files/extrapatch-gdb-Makefile.in
--- /usr/ports/devel/gdb/files/extrapatch-gdb-Makefile.in 2012-08-24 04:54:50.000000000 +0200
+++ ./files/extrapatch-gdb-Makefile.in 2012-08-26 18:40:22.000000000 +0200
@@ -18,6 +18,15 @@
mdebugread.h m88k-tdep.h stabsread.h hppa-linux-offsets.h linux-fork.h \
ser-unix.h inf-ptrace.h terminal.h ui-out.h frame-base.h \
f-lang.h dwarf2loc.h value.h sparc-tdep.h defs.h target-descriptions.h \
+@@ -1164,7 +1164,7 @@
+ # Removing the old gdb first works better if it is running, at least on SunOS.
+ gdb$(EXEEXT): gdb.o $(LIBGDB_OBS) $(ADD_DEPS) $(CDEPS) $(TDEPLIBS)
+ rm -f gdb$(EXEEXT)
+- $(CC_LD) $(INTERNAL_LDFLAGS) $(WIN32LDAPP) \
++ $(CC_LD) $(INTERNAL_LDFLAGS) $(WIN32LDAPP) -Wl,-E \
+ -o gdb$(EXEEXT) gdb.o $(LIBGDB_OBS) \
+ $(TDEPLIBS) $(TUI_LIBRARY) $(CLIBS) $(LOADLIBES)
+
@@ -1429,7 +1429,7 @@
armnbsd-nat.c armbsd-tdep.c armnbsd-tdep.c armobsd-tdep.c \
avr-tdep.c \
diff -urN /usr/ports/devel/gdb/files/patch-gdb-configure ./files/patch-gdb-configure
--- /usr/ports/devel/gdb/files/patch-gdb-configure 1970-01-01 01:00:00.000000000 +0100
+++ ./files/patch-gdb-configure 2012-08-24 14:06:02.000000000 +0200
@@ -0,0 +1,22 @@
+--- gdb/configure.orig 2012-08-24 14:03:52.000000000 +0200
++++ gdb/configure 2012-08-24 14:05:06.000000000 +0200
+@@ -10590,7 +10590,8 @@
+
+ # See if <machine/reg.h> supports the %fs and %gs i386 segment registers.
+ # Older i386 BSD's don't have the r_fs and r_gs members of `struct reg'.
+-ac_fn_c_check_member "$LINENO" "struct reg" "r_fs" "ac_cv_member_struct_reg_r_fs" "#include <machine/reg.h>
++ac_fn_c_check_member "$LINENO" "struct reg" "r_fs" "ac_cv_member_struct_reg_r_fs" "#include <sys/types.h>
++#include <machine/reg.h>
+ "
+ if test "x$ac_cv_member_struct_reg_r_fs" = x""yes; then :
+
+@@ -10600,7 +10601,8 @@
+
+
+ fi
+-ac_fn_c_check_member "$LINENO" "struct reg" "r_gs" "ac_cv_member_struct_reg_r_gs" "#include <machine/reg.h>
++ac_fn_c_check_member "$LINENO" "struct reg" "r_gs" "ac_cv_member_struct_reg_r_gs" "#include <sys/types.h>
++#include <machine/reg.h>
+ "
+ if test "x$ac_cv_member_struct_reg_r_gs" = x""yes; then :
+
--- gdb.patch ends here ---
>Release-Note:
>Audit-Trail:
>Unformatted:
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?201208261832.q7QIW87Z010021>