summaryrefslogtreecommitdiff
path: root/gdb/testsuite/gdb.base/premature-dummy-frame-removal.exp
blob: e21ff2efab90e568e27d5a9f988420a71259734a (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
# Copyright 2021-2023 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 3 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, see <http://www.gnu.org/licenses/>.

# Make an inferior call to a function which uses longjmp.  However,
# the backtrace for the function that is called is broken at the point
# where the longjmp is handled.  This test is checking to see if the
# inferior call still completes successfully.
#
# This test forces a broken backtrace using Python, but in real life a
# broken backtrace can easily occur when calling through code for
# which there is no debug information if the prologue unwinder fails,
# which can often happen if the code has been optimized.
#
# The problem was that, due to the broken backtrace, GDB failed to
# find the inferior call's dummy frame.  GDB then concluded that the
# inferior had longjmp'd backward past the dummy frame and so garbage
# collected the dummy frame, this causes the breakpoint within the
# dummy frame to be deleted.
#
# When the inferior continued, and eventually returned to the dummy
# frame, it would try to execute instruction from the dummy frame
# (which for most, or even all, targets, is on the stack), and then
# experience undefined behaviuor, often a SIGSEGV.

standard_testfile .c

if { [prepare_for_testing "failed to prepare" $testfile $srcfile] } {
    return -1
}

if {![runto_main]} {
    return 0
}

# Skip this test if Python scripting is not enabled.
if { [skip_python_tests] } { continue }

set pyfile [gdb_remote_download host ${srcdir}/${subdir}/${testfile}.py]
gdb_test_no_output "source ${pyfile}" "load python file"

gdb_test "p some_func ()" " = 0"

# When frame debugging is turned on, this test has (previously)
# revealed some crashes due to the Python frame unwinder trying to
# read registers.
#
# Enable frame debug and rerun the test.  We don't bother checking the
# output of calling 'p some_func ()' as the output will be full of
# debug, to format of which isn't fixed.  All we care about is that
# GDB is still running afterwards.
#
# All of the debug output makes this really slow when testing with the
# special read1 version of expect, hence the timeout factor.
with_read1_timeout_factor 10 {
    gdb_test_no_output "set debug frame on"
    gdb_test "p some_func ()" ".*" \
	"repeat p some_func () with frame debug on"
    gdb_test_no_output "set debug frame off"
}
gdb_test "p 1 + 2 + 3" " = 6"