[gdb/testsuite] Rewrite catch-follow-exec.exp
Commit Message
On 10/24/18 1:37 AM, Simon Marchi wrote:
> On 2018-10-23 6:38 p.m., Tom de Vries wrote:
>> On 10/23/18 11:05 PM, Tom de Vries wrote:
>>> On 10/23/18 11:04 PM, Simon Marchi wrote:
>>>> On 2018-10-15 3:54 p.m., Tom de Vries wrote:
>>>>>> Just wondering. Would it make life easier if we fixed PR 23368, which
>>>>>> is the reason we have to do the test in an unnatural way?
>>>>>
>>>>> Yes.
>>>>
>>>> Hi Tom,
>>>>
>>>> PR 23368 should be fixed now. Do you plan on updating catch-follow-exec.exp
>>>> to be written in a more standard way?
>>>
>>> Sure, will do.
>>
>> How does this look?
>
> Hi Tom,
>
> Thanks for looking into this so quickly.
And thanks for the quick review.
> I have some superficial suggestions that
> can help shorten the test a bit and make it more readable (some of them can be personal
> preference though...).
>
> When the test name is omitted, it defaults to the command. So instead of
>
> gdb_test "catch exec" \
> {Catchpoint [0-9][0-9]* \(exec\)} \
> "catch exec"
>
> You can write
>
> gdb_test "catch exec" {Catchpoint [0-9][0-9]* \(exec\)}
>
> and the test name will be "catch exec".
Done.
> Instead of [0-9][0-9]*, I am
> pretty sure you can use [0-9]+,
Done.
> or $decimal, which is provided by DejaGnu
> (/usr/share/dejagnu/runtest.exp):
>
> 101: set decimal "\[0-9\]+"
>
> Except in the {} string, $decimal won't work, because it won't get
> substituted.
Indeed. I prefer the {} quoting over "" quoting if that means less
escaping, so I went with {} here.
>
> For this:
>
> gdb_test "set follow-exec-mode new" \
> "" \
> "set follow-exec-mode new"
>
> You can use
>
> gdb_test_no_output "set follow-exec-mode new"
>
Done.
> (again, omitting the test name makes it default to the command)
>
> I'd suggest replacing
>
> gdb_test_multiple "info prog" "info prog" {
> -i "$gdb_spawn_id" eof {
> fail "info prog"
> }
> -i "$gdb_spawn_id" "No selected thread\." {
> pass "info prog"
> }
> }
>
> with the simpler
>
> gdb_test "info prog" "No selected thread."
>
> If GDB crashes as it did before your fix, the test will be unresolved, which is
> treated the same as a FAIL.
Done.
> While at it, could you update the comment at the top of the file, which currently
> says:
>
> # Check whether finish respects the print pretty user setting when printing the
> # function result.
>
Done.
Also, I realized that by using runto_main at the start, I could replace
gdb_run_cmd/gdb_expect with a regular gdb_test continue.
Committed as attached.
Thanks,
- Tom
[gdb/testsuite] Rewrite catch-follow-exec.exp using gdb_test
The testcase catch-follow-exec.exp is written use gdb -batch in order to avoid
a GDB SIGTTOU. After the commit of "Avoid GDB SIGTTOU on catch exec + set
follow-exec-mode new (PR 23368)", that no longer is necessary.
Rewrite the test using regular gdb_test commands.
Tested with x86_64-linux.
2018-10-24 Tom de Vries <tdevries@suse.de>
* gdb.base/catch-follow-exec.exp: Rewrite using gdb_test.
---
gdb/testsuite/gdb.base/catch-follow-exec.exp | 63 ++++++----------------------
1 file changed, 13 insertions(+), 50 deletions(-)
@@ -13,72 +13,35 @@
# You should have received a copy of the GNU General Public License
# along with this program. If not, see <http://www.gnu.org/licenses/>.
-# Check whether finish respects the print pretty user setting when printing the
-# function result.
+# Test whether info prog crashes gdb at a catch point in follow-exec-mode new.
standard_testfile
-if { [target_info gdb_protocol] != "" } {
- # Even though the feature under features being tested are supported by
- # gdbserver, the way this test is written doesn't make it easy with a
- # remote target.
- unsupported "not native"
- return
-}
-
if { ![remote_file target exists /bin/ls] } {
unsupported "no ls"
return
}
-if { [build_executable "failed to prepare" $testfile $srcfile debug] == -1 } {
- return -1
+if { [prepare_for_testing "failed to prepare" $testfile $srcfile debug] } {
+ return
}
proc catch_follow_exec { } {
- global binfile
- global gdb_spawn_id
-
- set test "catch-follow-exec"
-
- append FLAGS " \"$binfile\""
- append FLAGS " -batch"
- append FLAGS " -ex \"catch exec\""
- append FLAGS " -ex \"set follow-exec-mode new\""
- append FLAGS " -ex \"run\""
- append FLAGS " -ex \"info prog\""
-
- gdb_exit
- if {[gdb_spawn_with_cmdline_opts "$FLAGS"] != 0} {
- fail "spawn"
- return
+ if { ![runto_main] } {
+ untested "could not run to main"
+ return -1
}
- gdb_test_multiple "" "run til exit" {
- "runtime error:" {
- # Error in case of --enable-ubsan
- fail "no runtime error"
- }
- eof {
- set result [wait -i $gdb_spawn_id]
- verbose $result
+ gdb_test "catch exec" \
+ {Catchpoint [0-9]+ \(exec\)}
- gdb_assert { [lindex $result 2] == 0 }
+ gdb_test_no_output "set follow-exec-mode new"
- # We suspect this will be zero instead of one after fixing PR23368
- # - "gdb goes to into background when hitting exec catchpoint with
- # follow-exec-mode new"
- gdb_assert { [lindex $result 3] != 0 }
+ gdb_test "continue" \
+ ".*hit Catchpoint.*"
- # Error in case of --disable-ubsan, we get
- # "CHILDKILLED SIGSEGV {segmentation violation}" as extra
- # argument(s).
- gdb_assert { [llength $result] == 4 }
- }
-
- remote_close host
- clear_gdb_spawn_id
- }
+ gdb_test "info prog" \
+ "No selected thread."
}
catch_follow_exec