rpms/gdb/F-12 gdb-testsuite-unknown-output.patch, NONE, 1.1 gdb-watchpoint-cond-gone.patch, NONE, 1.1 gdb-6.3-ia64-info-frame-fix-20050725.patch, 1.2, 1.3 gdb-6.5-bz216711-clone-is-outermost.patch, 1.5, 1.6 gdb-7.0-upstream.patch, 1.5, 1.6 gdb.spec, 1.406, 1.407 gdb-6.3-watchpoint-cond-gone-test.patch, 1.2, NONE
Jan Kratochvil
jkratoch at fedoraproject.org
Fri Jan 1 11:06:09 UTC 2010
Author: jkratoch
Update of /cvs/pkgs/rpms/gdb/F-12
In directory cvs1.fedora.phx.redhat.com:/tmp/cvs-serv18725
Modified Files:
gdb-6.3-ia64-info-frame-fix-20050725.patch
gdb-6.5-bz216711-clone-is-outermost.patch
gdb-7.0-upstream.patch gdb.spec
Added Files:
gdb-testsuite-unknown-output.patch
gdb-watchpoint-cond-gone.patch
Removed Files:
gdb-6.3-watchpoint-cond-gone-test.patch
Log Message:
* Fri Jan 1 2009 Jan Kratochvil <jan.kratochvil at redhat.com> - 7.0-15.fc12
- Fix error on a sw watchpoint active at function epilogue (hit on s390x).
- testsuite: Fix false MI "unknown output after running" regression.
- testsuite: Update ia64-sigtramp.exp for recent GDB.
- Implement bt-clone-stop.exp fix also for ia64.
- testsuite: Upstream condbreak.exp results stability fix (Daniel Jacobowitz).
gdb-testsuite-unknown-output.patch:
mi-support.exp | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
--- NEW FILE gdb-testsuite-unknown-output.patch ---
http://sourceware.org/ml/gdb-patches/2009-12/msg00473.html
Subject: [patch] testsuite: false MI "unknown output after running" regression
Hi,
the attached patch will turn the results this way:
-KFAIL: gdb.mi/mi-until.exp: until after while loop (unknown output after running) (PRMS: gdb/2104)
+KFAIL: gdb.mi/mi-until.exp: until after while loop (stopped at wrong place) (PRMS: gdb/2104)
-XFAIL: gdb.mi/mi-watch.exp: sw: watchpoint trigger (unknown output after running)
+XFAIL: gdb.mi/mi-watch.exp: sw: watchpoint trigger (stopped at wrong place)
-KFAIL: gdb.mi/mi2-until.exp: until after while loop (unknown output after running) (PRMS: gdb/2104)
+KFAIL: gdb.mi/mi2-until.exp: until after while loop (stopped at wrong place) (PRMS: gdb/2104)
-XFAIL: gdb.mi/mi2-watch.exp: sw: watchpoint trigger (unknown output after running)
+XFAIL: gdb.mi/mi2-watch.exp: sw: watchpoint trigger (stopped at wrong place)
This is a fix of KFAIL-kind regression due to:
commit 26d086d741fb1bb0eee9d50e0bafa7c5e388023f
Re: [RFA] fix *stopped for CLI commands
http://sourceware.org/ml/gdb-patches/2009-02/msg00278.html
http://sourceware.org/ml/gdb-cvs/2009-02/msg00084.html
part:
gdb/testsuite/
2009-02-14 Vladimir Prus <vladimir at codesourcery.com>
* lib/mi-support.exp (mi_expect_stop): Adjust the order of fields.
(mi_expect_interrupt): Likewise.
* gdb.mi/mi-cli.exp: Check that "step" results in proper *stopped
response.
re-wdiff-ed for your convenience as:
verbose -log "mi_expect_stop: expecting: \\*stopped,${r}${a}${bn}[-thread-id=\"$decimal\",stopped-threads=$any,-]frame=\{addr=\"$hex\",func=\"$func\",args=$args,file=\"$any$file\",fullname=\"${fullname_syntax}$file\",line=\"$line\"\}$after_stopped{+,thread-id=\"$decimal\",stopped-threads=$any+}\r\n($thread_selected_re)?$prompt_re"
-re "\\*stopped,${r}${a}${bn}[-thread-id=\"$decimal\",stopped-threads=$any,-]frame=\{addr=\"$hex\",func=\"$func\",args=$args,file=\"$any$file\",fullname=\"${fullname_syntax}$file\",line=\"($line)\"\}$after_stopped{+,thread-id=\"$decimal\",stopped-threads=$any+}\r\n($thread_selected_re)?$prompt_re" {
-re "\\*stopped,${r}${a}${bn}[-thread-id=\"$decimal\",stopped-threads=$any,-]frame=\{addr=\"$hex\",func=\"$any\",args=\[\\\[\{\]$any\[\\\]\}\],file=\"$any\",fullname=\"${fullname_syntax}$any\",line=\"\[0-9\]*\"\}[-$any-]{+thread-id=\"$decimal\",stopped-threads=$any+}\r\n$prompt_re" {
FYI there is: set any "\[^\n\]*"
"stopped at wrong place" is the right kind of KFAIL/XFAIL (tested
gcc-4.1.2-46.el5_4.1.x86_64 and gcc-4.4.2-20.fc12.x86_64) as discussed
in PR gdb/2104.
In the last case there is missing comma (,) after the former patch above.
$after_stopped was IMO missing there even before. It is in use only by
gdb.mi/mi-simplerun.exp and gdb.mi/mi2-simplerun.exp. Also fixed below.
No regressions on {x86_64,x86_64-m32,i686}-fedora12-linux-gnu.
Thanks,
Jan
gdb/testsuite/
2009-12-31 Jan Kratochvil <jan.kratochvil at redhat.com>
* lib/mi-support.exp (mi_expect_stop <stopped at wrong place>): Insert
missing $after_stopped and comma (,) expectation.
--- a/gdb/testsuite/lib/mi-support.exp
+++ b/gdb/testsuite/lib/mi-support.exp
@@ -1038,7 +1038,7 @@ proc mi_expect_stop { reason func args file line extra test } {
pass "$test"
return $expect_out(2,string)
}
- -re "\\*stopped,${r}${a}${bn}frame=\{addr=\"$hex\",func=\"$any\",args=\[\\\[\{\]$any\[\\\]\}\],file=\"$any\",fullname=\"${fullname_syntax}$any\",line=\"\[0-9\]*\"\}thread-id=\"$decimal\",stopped-threads=$any\r\n$prompt_re" {
+ -re "\\*stopped,${r}${a}${bn}frame=\{addr=\"$hex\",func=\"$any\",args=\[\\\[\{\]$any\[\\\]\}\],file=\"$any\",fullname=\"${fullname_syntax}$any\",line=\"\[0-9\]*\"\}$after_stopped,thread-id=\"$decimal\",stopped-threads=$any\r\n$prompt_re" {
verbose -log "got $expect_out(buffer)"
fail "$test (stopped at wrong place)"
return -1
gdb-watchpoint-cond-gone.patch:
b/gdb/testsuite/gdb.base/watchpoint-cond-gone-stripped.c | 23 ++++++
b/gdb/testsuite/gdb.base/watchpoint-cond-gone.c | 37 ++++++++++
b/gdb/testsuite/gdb.base/watchpoint-cond-gone.exp | 51 +++++++++++++++
gdb/breakpoint.c | 22 +++---
4 files changed, 122 insertions(+), 11 deletions(-)
--- NEW FILE gdb-watchpoint-cond-gone.patch ---
http://sourceware.org/ml/gdb-patches/2009-12/msg00394.html
Subject: [patch] Fix watchpoint-at-epilogue 7.0 regression (s390x)
Hi,
attached testcase regressed 6.8->7.0 on some s390x systems.
At the moment PC is at epilogue (after frame pointer got destroyed and PC is
still in the calleed function) we must not try any unwinding as it can try to
unwind a garbage.
Correct way would be to implement epilogue unwinder (such as
amd64_epilogue_frame_unwind or i386_epilogue_frame_unwind ones) but it would
not be so easy on s390x as it would need full prologue-analyzer: After SP gets
restored there can be various instructions before the last jump - GDB needs to
interpret the jump to find the register value where it is going to jump to.
00000048279aa09c <round_and_return>:
[...]
48279aa1b8: eb 6f f0 d0 00 04 lmg %r6,%r15,208(%r15)
48279aa1be: 50 32 10 00 st %r3,0(%r2,%r1) # SP (%r15) is gone hone
48279aa1c2: b3 74 00 00 lzer %r0
48279aa1c6: 07 f4 br %r4 # %r4 is the target
[...]
Just an easy fix is to be more careful and do not try any unwinding if we find
PC is in the epilogue (previous instruction modifies SP=%r15).
It is even a performance optimization and I see no regression risk there.
The specific s390x error is:
#0 throw_error (error=MEMORY_ERROR, fmt=0x8042676a "Cannot access memory at address %s") at exceptions.c:415
#1 in memory_error (status=5, memaddr=0) at corefile.c:220
#2 in read_memory (memaddr=0, myaddr=0x3ffffc7f7f0 "", len=8) at corefile.c:238
#3 in read_memory_unsigned_integer (memaddr=0, len=8, byte_order=BFD_ENDIAN_BIG) at corefile.c:321
#4 in s390_backchain_frame_unwind_cache (this_frame=0x80608958, info=0x80608a08) at s390-tdep.c:1525
#5 in s390_frame_unwind_cache (this_frame=0x80608958, this_prologue_cache=0x80608970) at s390-tdep.c:1572
#6 in s390_frame_this_id (this_frame=0x80608958, this_prologue_cache=0x80608970, this_id=0x806089b8) at s390-tdep.c:1583
#7 in get_frame_id (fi=0x80608958) at frame.c:335
#8 in frame_find_by_id (id={stack_addr = 4398044824952, code_addr = 2147484952, special_addr = 0, stack_addr_p = 1, code_addr_p = 1, special_addr_p = 0, inline_depth = 0}) at frame.c:587
#9 in watchpoint_check (p=0x8095bbc0) at breakpoint.c:3203
#10 in catch_errors (func=0x801262c0 <watchpoint_check>, func_args=0x8095bbc0, errstring=0x80a26410 "Error evaluating expression for watchpoint 3\n", mask=6) at exceptions.c:510
#11 in bpstat_check_watchpoint (bs=0x8095bbc0) at breakpoint.c:3404
#12 in bpstat_stop_status (aspace=0x8061fcb0, bp_addr=2147485040, ptid={pid = 17372, lwp = 17372, tid = 0}) at breakpoint.c:3594
#13 in handle_inferior_event (ecs=0x3ffffc80520) at infrun.c:3588
I did not analyze why 6.8 did not error out, the GDB code is similar there.
No regressions on:
{x86_64,x86_64-m32,i686}-fedora12-linux-gnu (CVS HEAD GDB)
s390x-rhel48-linux-gnu (CVS HEAD GDB)
s390-rhel48-linux-gnu (CVS HEAD GDB)
s390x-rhel54-linux-gnu (Fedora 12 GDB)
s390-rhel54-linux-gnu (Fedora 12 GDB)
Thanks,
Jan
gdb/
2009-12-25 Jan Kratochvil <jan.kratochvil at redhat.com>
* breakpoint.c (watchpoint_check): Check the call
gdbarch_in_function_epilogue_p before calling frame_find_by_id.
Extend the comment.
gdb/testsuite/
2009-12-25 Jan Kratochvil <jan.kratochvil at redhat.com>
* gdb.base/watchpoint-cond-gone.exp, gdb.base/watchpoint-cond-gone.c,
gdb.base/watchpoint-cond-gone-stripped.c: New.
--- a/gdb/breakpoint.c
+++ b/gdb/breakpoint.c
@@ -3200,6 +3200,17 @@ watchpoint_check (void *p)
struct gdbarch *frame_arch = get_frame_arch (frame);
CORE_ADDR frame_pc = get_frame_pc (frame);
+ /* in_function_epilogue_p() returns a non-zero value if we're still
+ in the function but the stack frame has already been invalidated.
+ Since we can't rely on the values of local variables after the
+ stack has been destroyed, we are treating the watchpoint in that
+ state as `not changed' without further checking. Don't mark
+ watchpoints as changed if the current frame is in an epilogue -
+ even if they are in some other frame, our view of the stack
+ is likely to be wrong and frame_find_by_id could error out. */
+ if (gdbarch_in_function_epilogue_p (frame_arch, frame_pc))
+ return WP_VALUE_NOT_CHANGED;
+
fr = frame_find_by_id (b->watchpoint_frame);
within_current_scope = (fr != NULL);
@@ -3216,17 +3227,6 @@ watchpoint_check (void *p)
within_current_scope = 0;
}
- /* in_function_epilogue_p() returns a non-zero value if we're still
- in the function but the stack frame has already been invalidated.
- Since we can't rely on the values of local variables after the
- stack has been destroyed, we are treating the watchpoint in that
- state as `not changed' without further checking. Don't mark
- watchpoints as changed if the current frame is in an epilogue -
- even if they are in some other frame, our view of the stack
- is likely to be wrong. */
- if (gdbarch_in_function_epilogue_p (frame_arch, frame_pc))
- return WP_VALUE_NOT_CHANGED;
-
if (within_current_scope)
/* If we end up stopping, the current frame will get selected
in normal_stop. So this call to select_frame won't affect
--- /dev/null
+++ b/gdb/testsuite/gdb.base/watchpoint-cond-gone-stripped.c
@@ -0,0 +1,23 @@
+/* This testcase is part of GDB, the GNU debugger.
+
+ Copyright 2009 Free Software Foundation, Inc.
+
+ This program is free software; you can redistribute it and/or modify
+ it under the terms of the GNU General Public License as published by
+ the Free Software Foundation; either version 2 of the License, or
+ (at your option) any later version.
+
+ This program is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ GNU General Public License for more details.
+
+ You should have received a copy of the GNU General Public License
+ along with this program; if not, write to the Free Software
+ Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */
+
+void
+jumper (void (*jumpto) (void))
+{
+ (*jumpto) ();
+}
--- /dev/null
+++ b/gdb/testsuite/gdb.base/watchpoint-cond-gone.c
@@ -0,0 +1,37 @@
+/* This testcase is part of GDB, the GNU debugger.
+
+ Copyright 2009 Free Software Foundation, Inc.
+
+ This program is free software; you can redistribute it and/or modify
+ it under the terms of the GNU General Public License as published by
+ the Free Software Foundation; either version 2 of the License, or
+ (at your option) any later version.
+
+ This program is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ GNU General Public License for more details.
+
+ You should have received a copy of the GNU General Public License
+ along with this program; if not, write to the Free Software
+ Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */
+
+extern void jumper (void (*jumpto) (void));
+
+static void
+func (void)
+{
+ volatile int c;
+
+ c = 5;
+ c = 10; /* watchpoint-here */
+ c = 20;
+}
+
+int
+main (void)
+{
+ jumper (func);
+
+ return 0;
+}
--- /dev/null
+++ b/gdb/testsuite/gdb.base/watchpoint-cond-gone.exp
@@ -0,0 +1,51 @@
+# Copyright 2009 Free Software Foundation, Inc.
+
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 2 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write to the Free Software
+# Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+
+set testfile "watchpoint-cond-gone"
+set srcfile ${testfile}.c
+set srcfilestripped ${testfile}-stripped.c
+set objfilestripped ${objdir}/${subdir}/${testfile}-stripped.o
+set binfile ${objdir}/${subdir}/${testfile}
+
+# We need to generate a function without DWARF to crash older GDB.
+# Stepping into a dynamic function trampoline or stepping out of MAIN may work
+# but it is not a reliable FAIL case.
+
+if { [gdb_compile "${srcdir}/${subdir}/${srcfilestripped}" "${objfilestripped}" object {}] != ""
+ || [gdb_compile "${srcdir}/${subdir}/${srcfile} ${objfilestripped}" "${binfile}" executable {debug}] != "" } {
+ gdb_suppress_entire_file "Testcase compile failed, so all tests in this file will automatically fail."
+}
+
+clean_restart ${testfile}
+
+# Problem does not occur otherwise.
+gdb_test "set can-use-hw-watchpoints 0"
+
+if ![runto_main] {
+ return -1
+}
+
+gdb_breakpoint [gdb_get_line_number "watchpoint-here"]
+gdb_continue_to_breakpoint "Place to set the watchpoint"
+
+# The condition `c == 30' is the subject being tested.
+gdb_test "watch c if c == 30" "" "Place the watchpoint"
+
+# We may stay either in the function itself or only at the first instruction of
+# its caller depending on the epilogue unwinder (or valid epilogue CFI) presence.
+gdb_test "finish" \
+ "Watchpoint .* deleted because the program has left the block in.*which its expression is valid..*in (jumper|func).*" \
+ "Catch the no longer valid watchpoint"
gdb-6.3-ia64-info-frame-fix-20050725.patch:
ia64-sigtramp.c | 23 +++++++++++++++++
ia64-sigtramp.exp | 73 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
2 files changed, 96 insertions(+)
Index: gdb-6.3-ia64-info-frame-fix-20050725.patch
===================================================================
RCS file: /cvs/pkgs/rpms/gdb/F-12/gdb-6.3-ia64-info-frame-fix-20050725.patch,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -p -r1.2 -r1.3
--- gdb-6.3-ia64-info-frame-fix-20050725.patch 3 Mar 2008 16:13:47 -0000 1.2
+++ gdb-6.3-ia64-info-frame-fix-20050725.patch 1 Jan 2010 11:06:08 -0000 1.3
@@ -41,7 +41,7 @@ testsuite:
+
--- gdb-6.3/gdb/testsuite/gdb.arch/ia64-sigtramp.exp.fix 2005-07-25 16:42:50.000000000 -0400
+++ gdb-6.3/gdb/testsuite/gdb.arch/ia64-sigtramp.exp 2005-07-25 16:42:01.000000000 -0400
-@@ -0,0 +1,66 @@
+@@ -0,0 +1,73 @@
+# Copyright 2005 Free Software Foundation, Inc.
+
+# This program is free software; you can redistribute it and/or modify
@@ -90,7 +90,10 @@ testsuite:
+}
+
+gdb_exit
++set match_max_old [match_max]
++match_max -d 1000000
+gdb_start
++match_max -d $match_max_old
+gdb_reinitialize_dir $srcdir/$subdir
+gdb_load ${binfile}
+
@@ -106,5 +109,9 @@ testsuite:
+gdb_test "continue" "Breakpoint.*x.*" "continue to x"
+
+gdb_test "f 1" ".*signal handler called.*" "frame 1"
-+gdb_test "info frame" "Stack level 1.*p63 at .*" "info sigtramp frame"
-+
++
++# gdb-7.0+ no longer prints the pseudo registers as they are computed.
++# frame_info says: /* For moment, only display registers that were saved on the
++# stack. */
++gdb_test "set debug frame 1"
++gdb_test "info frame" "Stack level 1, .*frame_unwind_register_value \\(frame=1,regnum=750\\(p63\\),\[^\r\n\]*\r\n\[^\r\n\]*-> computed bytes=.*" "info sigtramp frame"
gdb-6.5-bz216711-clone-is-outermost.patch:
gdb-6.8.50.20090802/gdb/amd64-linux-tdep.c | 76 +++++
gdb-6.8.50.20090802/gdb/amd64-tdep.c | 5
gdb-6.8.50.20090802/gdb/testsuite/gdb.threads/bt-clone-stop.c | 39 ++
gdb-6.8.50.20090802/gdb/testsuite/gdb.threads/bt-clone-stop.exp | 61 ++++
gdb-6.8.50.20090811/gdb/i386-tdep.c | 3
gdb-6.8.50.20090811/gdb/i386-tdep.h | 3
gdb/ia64-tdep.c | 132 ++++++++++
7 files changed, 319 insertions(+)
Index: gdb-6.5-bz216711-clone-is-outermost.patch
===================================================================
RCS file: /cvs/pkgs/rpms/gdb/F-12/gdb-6.5-bz216711-clone-is-outermost.patch,v
retrieving revision 1.5
retrieving revision 1.6
diff -u -p -r1.5 -r1.6
--- gdb-6.5-bz216711-clone-is-outermost.patch 11 Aug 2009 14:42:46 -0000 1.5
+++ gdb-6.5-bz216711-clone-is-outermost.patch 1 Jan 2010 11:06:08 -0000 1.6
@@ -162,6 +162,154 @@ Index: gdb-6.8.50.20090802/gdb/amd64-tde
};
/* Floating-point registers. */
+--- a/gdb/ia64-tdep.c
++++ b/gdb/ia64-tdep.c
+@@ -2122,6 +2122,137 @@ static const struct frame_unwind ia64_frame_unwind =
+ default_frame_sniffer
+ };
+
++/* Detect the outermost frame; during unwind of
++ #6 0x2000000000347100 in __clone2 () from /lib/libc.so.6.1
++ avoid the additional bogus frame
++ #7 0x0000000000000000 in ?? () */
++
++static char linux_clone2_code[] =
++{
++/* libc/sysdeps/unix/sysv/linux/ia64/clone2.S */
++ 0x09, 0x00, 0x20, 0x12, 0x90, 0x11, 0x00, 0x40,
++ 0x28, 0x20, 0x23, 0x00, 0x00, 0x00, 0x04, 0x00,
++/* st4 [r9]=r8 */
++/* st4 [r10]=r8 */
++/* ;; */
++/* #endif */
++ 0x02, 0x50, 0x21, 0x40, 0x18, 0x14, 0x90, 0x02,
++ 0x90, 0x00, 0x42, 0x00, 0x00, 0x00, 0x04, 0x00,
++/* 1: ld8 out1=[in0],8 |* Retrieve code pointer. *| */
++/* mov out0=in4 |* Pass proper argument to fn *| */
++/* ;; */
++ 0x11, 0x08, 0x00, 0x40, 0x18, 0x10, 0x60, 0x50,
++ 0x05, 0x80, 0x03, 0x00, 0x68, 0x00, 0x80, 0x12,
++/* ld8 gp=[in0] |* Load function gp. *| */
++/* mov b6=out1 */
++/* br.call.dptk.many rp=b6 |* Call fn(arg) in the child *| */
++/* ;; */
++ 0x10, 0x48, 0x01, 0x10, 0x00, 0x21, 0x10, 0x00,
++ 0xa0, 0x00, 0x42, 0x00, 0x98, 0xdf, 0xf7, 0x5b,
++/* mov out0=r8 |* Argument to _exit *| */
++/* mov gp=loc0 */
++/* .globl HIDDEN_JUMPTARGET(_exit) */
++/* br.call.dpnt.many rp=HIDDEN_JUMPTARGET(_exit) */
++/* |* call _exit with result from fn. *| */
++ 0x11, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00,
++ 0x00, 0x02, 0x00, 0x80, 0x00, 0x00, 0x84, 0x00
++/* ret |* Not reached. *| */
++};
++
++#define LINUX_CLONE_PRE_SLOTS 3 /* Number of slots before PC. */
++#define LINUX_CLONE_LEN (sizeof linux_clone2_code)
++
++static int
++ia64_linux_clone2_running (struct frame_info *this_frame)
++{
++ CORE_ADDR pc = get_frame_pc (this_frame);
++ char buf[LINUX_CLONE_LEN];
++ struct minimal_symbol *minsym;
++ long long instr;
++
++ if (!safe_frame_unwind_memory (this_frame, pc - LINUX_CLONE_PRE_SLOTS * 16,
++ buf, LINUX_CLONE_LEN))
++ return 0;
++
++ if (memcmp (buf, linux_clone2_code, LINUX_CLONE_PRE_SLOTS * 16) != 0)
++ return 0;
++
++ /* Adjust the expected "_exit" address. */
++ minsym = lookup_minimal_symbol_text ("_exit", NULL);
++ if (minsym == NULL)
++ return 0;
++
++ instr = slotN_contents (&buf[LINUX_CLONE_PRE_SLOTS * 16], 2);
++ instr &= ~(((1L << 20) - 1) << 13);
++ /* Address is relative to the jump instruction slot, not the next one. */
++ instr |= (((SYMBOL_VALUE_ADDRESS (minsym) - (pc & ~0xfL)) >> 4)
++ & ((1L << 20) - 1)) << 13;
++ replace_slotN_contents (&buf[LINUX_CLONE_PRE_SLOTS * 16], instr, 2);
++
++ if (memcmp (&buf[LINUX_CLONE_PRE_SLOTS * 16],
++ &linux_clone2_code[LINUX_CLONE_PRE_SLOTS * 16],
++ LINUX_CLONE_LEN - (LINUX_CLONE_PRE_SLOTS * 16)) != 0)
++ return 0;
++
++ return 1;
++}
++
++static int
++ia64_outermost_frame (struct frame_info *this_frame)
++{
++ CORE_ADDR pc = get_frame_pc (this_frame);
++ char *name;
++
++ find_pc_partial_function (pc, &name, NULL, NULL);
++
++ /* If we have NAME, we can optimize the search.
++ `clone' NAME still needs to have the code checked as its name may be
++ present in the user code.
++ `__clone' NAME should not be present in the user code but in the initial
++ parts of the `__clone' implementation the unwind still makes sense.
++ More detailed unwinding decision would be too much sensitive to possible
++ subtle changes in specific glibc revisions. */
++ if (name == NULL || strcmp (name, "clone2") == 0
++ || strcmp ("__clone2", name) == 0)
++ return (ia64_linux_clone2_running (this_frame) != 0);
++
++ return 0;
++}
++
++static void
++ia64_clone2_frame_this_id (struct frame_info *this_frame, void **this_cache,
++ struct frame_id *this_id)
++{
++ /* Leave the default outermost frame at *THIS_ID. */
++}
++
++static struct value *
++ia64_clone2_frame_prev_register (struct frame_info *this_frame,
++ void **this_cache, int regnum)
++{
++ return frame_unwind_got_register (this_frame, regnum, regnum);
++}
++
++static int
++ia64_clone2_frame_sniffer (const struct frame_unwind *self,
++ struct frame_info *this_frame,
++ void **this_prologue_cache)
++{
++ if (ia64_outermost_frame (this_frame))
++ return 1;
++
++ return 0;
++}
++
++static const struct frame_unwind ia64_clone2_frame_unwind =
++{
++ NORMAL_FRAME,
++ &ia64_clone2_frame_this_id,
++ &ia64_clone2_frame_prev_register,
++ NULL,
++ &ia64_clone2_frame_sniffer
++};
++
+ /* Signal trampolines. */
+
+ static void
+@@ -3824,6 +3955,7 @@ ia64_gdbarch_init (struct gdbarch_info info, struct gdbarch_list *arches)
+ set_gdbarch_dummy_id (gdbarch, ia64_dummy_id);
+
+ set_gdbarch_unwind_pc (gdbarch, ia64_unwind_pc);
++ frame_unwind_append_unwinder (gdbarch, &ia64_clone2_frame_unwind);
+ #ifdef HAVE_LIBUNWIND_IA64_H
+ frame_unwind_append_unwinder (gdbarch,
+ &ia64_libunwind_sigtramp_frame_unwind);
Index: gdb-6.8.50.20090802/gdb/testsuite/gdb.threads/bt-clone-stop.c
===================================================================
--- /dev/null 1970-01-01 00:00:00.000000000 +0000
gdb-7.0-upstream.patch:
gdb/COPYING | 916 ++++++++++++++++++++---------
gdb/ChangeLog | 75 ++
gdb/amd64-tdep.c | 2
gdb/amd64fbsd-nat.c | 3
gdb/arm-tdep.c | 212 ++++++
gdb/completer.c | 4
gdb/dbxread.c | 9
gdb/dcache.c | 139 +++-
gdb/doc/ChangeLog | 4
gdb/doc/gpl.texi | 958 ++++++++++++++++++++-----------
gdb/dwarf2-frame.c | 8
gdb/dwarf2read.c | 3
gdb/elfread.c | 3
gdb/objfiles.c | 16
gdb/score-tdep.c | 52 -
gdb/score-tdep.h | 64 --
gdb/testsuite/ChangeLog | 21
gdb/testsuite/gdb.base/bigcore.exp | 2
gdb/testsuite/gdb.base/foll-fork.c | 3
gdb/testsuite/gdb.base/foll-fork.exp | 3
gdb/testsuite/gdb.base/structs.c | 6
gdb/testsuite/gdb.base/structs.exp | 62 +-
src/gdb/doc/gdb.texinfo | 13
src/gdb/testsuite/gdb.base/condbreak.exp | 9
src/gdb/testsuite/gdb.base/default.exp | 7
src/gdb/testsuite/gdb.cp/expand-sals.cc | 2
src/gdb/testsuite/gdb.cp/expand-sals.exp | 6
src/gdb/top.c | 7
28 files changed, 1797 insertions(+), 812 deletions(-)
Index: gdb-7.0-upstream.patch
===================================================================
RCS file: /cvs/pkgs/rpms/gdb/F-12/gdb-7.0-upstream.patch,v
retrieving revision 1.5
retrieving revision 1.6
diff -u -p -r1.5 -r1.6
--- gdb-7.0-upstream.patch 24 Dec 2009 22:39:35 -0000 1.5
+++ gdb-7.0-upstream.patch 1 Jan 2010 11:06:08 -0000 1.6
@@ -3221,3 +3221,65 @@ http://sourceware.org/ml/gdb-cvs/2009-12
-# Verify GDB really could not catch any other breakpoint location.
-
-gdb_continue_to_breakpoint "uncaught return" ".*exit-line.*"
+
+
+
+http://sourceware.org/ml/gdb-cvs/2009-11/msg00213.html
+
+[ cut ]
+
+--- src/gdb/testsuite/gdb.base/condbreak.exp 2009/01/03 05:58:03 1.13
++++ src/gdb/testsuite/gdb.base/condbreak.exp 2009/11/25 20:43:29 1.14
+@@ -207,10 +207,10 @@
+ setup_xfail hppa2.0w-*-* 11512CLLbs
+ send_gdb "continue\n"
+ gdb_expect {
+- -re "Continuing\\..*Breakpoint \[0-9\]+, marker2 \\(a=43\\) at .*$srcfile1:($bp_location8|$bp_location9).*($bp_location8|$bp_location9)\[\t \]+.*" {
++ -re "Continuing\\..*Breakpoint \[0-9\]+, marker2 \\(a=43\\) at .*$srcfile1:($bp_location8|$bp_location9).*($bp_location8|$bp_location9)\[\t \]+.*$gdb_prompt $" {
+ pass "run until breakpoint at marker2"
+ }
+- -re "Continuing\\..*Breakpoint \[0-9\]+, $hex in marker2 \\(a=43\\) at .*$srcfile1:($bp_location8|$bp_location9).*($bp_location8|$bp_location9)\[\t \]+.*" {
++ -re "Continuing\\..*Breakpoint \[0-9\]+, $hex in marker2 \\(a=43\\) at .*$srcfile1:($bp_location8|$bp_location9).*($bp_location8|$bp_location9)\[\t \]+.*$gdb_prompt $" {
+ xfail "run until breakpoint at marker2"
+ }
+ -re "$gdb_prompt $" {
+
+
+
+Fix for gfortran-4.1:
+ -PASS: gdb.mi/mi-var-child-f.exp: mi runto MAIN__
+ +FAIL: gdb.mi/mi-var-child-f.exp: mi runto MAIN__ (unknown output after running)
+
+RFC: Move language-changed message to verbose
+http://sourceware.org/ml/gdb-patches/2009-11/msg00031.html
+http://sourceware.org/ml/gdb-cvs/2009-11/msg00034.html
+
+### src/gdb/ChangeLog 2009/11/05 19:53:03 1.11030
+### src/gdb/ChangeLog 2009/11/05 20:43:52 1.11031
+## -1,3 +1,8 @@
++2009-11-05 Daniel Jacobowitz <dan at codesourcery.com>
++
++ * top.c (execute_command): Select a frame before checking the current
++ language. Only output a message if verbose.
++
+ 2009-11-05 Tom Tromey <tromey at redhat.com>
+
+ * symtab.h (SYMBOL_SET_LINKAGE_NAME): Update comment.
+--- src/gdb/top.c 2009/10/19 09:51:42 1.172
++++ src/gdb/top.c 2009/11/05 20:43:52 1.173
+@@ -457,10 +457,13 @@
+
+ }
+
+- /* Tell the user if the language has changed (except first time). */
++ /* Tell the user if the language has changed (except first time).
++ First make sure that a new frame has been selected, in case this
++ command or the hooks changed the program state. */
++ deprecated_safe_get_selected_frame ();
+ if (current_language != expected_language)
+ {
+- if (language_mode == language_mode_auto)
++ if (language_mode == language_mode_auto && info_verbose)
+ {
+ language_info (1); /* Print what changed. */
+ }
Index: gdb.spec
===================================================================
RCS file: /cvs/pkgs/rpms/gdb/F-12/gdb.spec,v
retrieving revision 1.406
retrieving revision 1.407
diff -u -p -r1.406 -r1.407
--- gdb.spec 24 Dec 2009 22:39:35 -0000 1.406
+++ gdb.spec 1 Jan 2010 11:06:08 -0000 1.407
@@ -36,7 +36,7 @@ Version: 7.0
# The release always contains a leading reserved number, start it at 1.
# `upstream' is not a part of `name' to stay fully rpm dependencies compatible for the testing.
-Release: 14%{?_with_upstream:.upstream}%{dist}
+Release: 15%{?_with_upstream:.upstream}%{dist}
License: GPLv3+
Group: Development/Debuggers
@@ -318,8 +318,8 @@ Patch309: gdb-6.3-mapping-zero-inode-tes
# Test a crash on `focus cmd', `focus prev' commands.
Patch311: gdb-6.3-focus-cmd-prev-test.patch
-# Test crash on a sw watchpoint condition getting out of the scope.
-Patch314: gdb-6.3-watchpoint-cond-gone-test.patch
+# Fix error on a sw watchpoint active at function epilogue (hit on s390x).
+Patch314: gdb-watchpoint-cond-gone.patch
# Test various forms of threads tracking across exec() (BZ 442765).
Patch315: gdb-6.8-bz442765-threaded-exec-test.patch
@@ -423,6 +423,9 @@ Patch396: gdb-ppc-hw-watchpoint-twice.pa
# Fix regression by python on ia64 due to stale current frame.
Patch397: gdb-follow-child-stale-parent.patch
+# testsuite: Fix false MI "unknown output after running" regression.
+Patch398: gdb-testsuite-unknown-output.patch
+
BuildRequires: ncurses-devel texinfo gettext flex bison expat-devel
Requires: readline
BuildRequires: readline-devel
@@ -667,6 +670,7 @@ rm -f gdb/jv-exp.c gdb/m2-exp.c gdb/objc
%patch395 -p1
%patch396 -p1
%patch397 -p1
+%patch398 -p1
find -name "*.orig" | xargs rm -f
! find -name "*.rej" # Should not happen.
@@ -978,6 +982,13 @@ fi
%endif
%changelog
+* Fri Jan 1 2009 Jan Kratochvil <jan.kratochvil at redhat.com> - 7.0-15.fc12
+- Fix error on a sw watchpoint active at function epilogue (hit on s390x).
+- testsuite: Fix false MI "unknown output after running" regression.
+- testsuite: Update ia64-sigtramp.exp for recent GDB.
+- Implement bt-clone-stop.exp fix also for ia64.
+- testsuite: Upstream condbreak.exp results stability fix (Daniel Jacobowitz).
+
* Thu Dec 24 2009 Jan Kratochvil <jan.kratochvil at redhat.com> - 7.0-14.fc12
- testsuite: Fix constructortest.exp and expand-sals.exp for gcc-4.4.2-20.fc12.
--- gdb-6.3-watchpoint-cond-gone-test.patch DELETED ---
More information about the fedora-extras-commits
mailing list