From: Lancelot SIX <lancelot.six@amd.com>
To: "Alexandra Hájková" <ahajkova@khirnov.net>
Cc: <gdb-patches@sourceware.org>, <ahajkova@redhat.com>,
Jan Kratochvil <jan.kratochvil@redhat.com>
Subject: Re: [PATCH v6] Add a test for the gcore script
Date: Mon, 3 Jun 2024 21:16:52 +0100 [thread overview]
Message-ID: <20240603201541.g5ly35xlknqamnvq@khazad-dum> (raw)
In-Reply-To: <20240602155100.533195-1-ahajkova@khirnov.net>
On Sun, Jun 02, 2024 at 05:50:59PM +0200, Alexandra Hájková wrote:
> From: Alexandra Hájková <ahajkova@redhat.com>
>
> It also tests the gcore script being run without its accessible
> terminal.
>
> This test was written by Jan Kratochvil a long time ago. I modernized
> the test making it use various procs from lib/gdb.exp, reorganizing it
> and added some comments.
>
> Modify the gcore script to make it possible to pass the --data-directory to
> it. This prevents a lot of these warnings:
>
> Python Exception <class 'AttributeError'>: module 'gdb' has no attribute
> '_handle_missing_debuginfo'
>
> Tested by using make check-all-boards.
>
> Co-Authored-By: Jan Kratochvil <jan.kratochvil@redhat.com>
Hi Alexandra,
If I run this testcase with `sysctl kernel.yama.ptrace_scope=1`, I have:
Running ..../gdb/testsuite/gdb.base/gcorebg.exp ...
FAIL: gdb.base/gcorebg.exp: detached = standard: Spawned gcore finished
FAIL: gdb.base/gcorebg.exp: detached = standard: Core file generated by gcore
FAIL: gdb.base/gcorebg.exp: detached = detached: Spawned gcore finished
FAIL: gdb.base/gcorebg.exp: detached = detached: Core file generated by gcore
The log is:
PASS: gdb.base/gcorebg.exp: detached = standard: Spawned gcore
ptrace: Operation not permitted.
You can't do that without a process to debug.
The program is not being run.
gcore: failed to create ..../gdb/testsuite/outputs/gdb.base/gcorebg/gcorebg.core.275084
FAIL: gdb.base/gcorebg.exp: detached = standard: Spawned gcore finished
FAIL: gdb.base/gcorebg.exp: detached = standard: Core file generated by gcore
I think you can avoid this by using
require can_spawn_for_attach
I do have minor remarks inlined in the patch as well.
> ---
> v6:
> - fixed typos
> - used Jan Kratochvil's old invalid email for the record
> - improved the "usage" message
> - fixed has_gcore_script proc
>
> gdb/doc/gdb.texinfo | 2 +-
> gdb/gcore.in | 12 +++-
> gdb/testsuite/gdb.base/gcorebg.c | 86 +++++++++++++++++++++++++++++
> gdb/testsuite/gdb.base/gcorebg.exp | 88 ++++++++++++++++++++++++++++++
> gdb/testsuite/lib/gdb.exp | 18 ++++++
> 5 files changed, 202 insertions(+), 4 deletions(-)
> create mode 100644 gdb/testsuite/gdb.base/gcorebg.c
> create mode 100644 gdb/testsuite/gdb.base/gcorebg.exp
>
> diff --git a/gdb/doc/gdb.texinfo b/gdb/doc/gdb.texinfo
> index 750f368f980..67edfdaf800 100644
> --- a/gdb/doc/gdb.texinfo
> +++ b/gdb/doc/gdb.texinfo
> @@ -50984,7 +50984,7 @@ Richard M. Stallman and Roland H. Pesch, July 1991.
>
> @format
> @c man begin SYNOPSIS gcore
> -gcore [-a] [-o @var{prefix}] @var{pid1} [@var{pid2}...@var{pidN}]
> +gcore [-a] [-o @var{prefix}] [-d @var{path}] @var{pid1} [@var{pid2}...@var{pidN}]
> @c man end
> @end format
>
> diff --git a/gdb/gcore.in b/gdb/gcore.in
> index 982c854eb70..0c40e5a54cd 100644
> --- a/gdb/gcore.in
> +++ b/gdb/gcore.in
> @@ -27,7 +27,9 @@ prefix=core
> # to ensure gdb dumps all mappings (OS dependent).
> dump_all_cmds=()
>
> -while getopts :ao: opt; do
> +data_directory_opt=()
> +
> +while getopts :ao:d: opt; do
> case "$opt" in
> a)
> case "$OSTYPE" in
> @@ -40,8 +42,11 @@ while getopts :ao: opt; do
> o)
> prefix=$OPTARG
> ;;
> + d)
> + data_directory_opt=("--data-directory" "$OPTARG")
> + ;;
> *)
> - echo "usage: @GCORE_TRANSFORM_NAME@ [-a] [-o prefix] pid1 [pid2...pidN]"
> + echo "usage: @GCORE_TRANSFORM_NAME@ [-a] [-o prefix] [-d data-directory] pid1 [pid2...pidN]"
> exit 2
> ;;
> esac
> @@ -51,7 +56,7 @@ shift $((OPTIND-1))
>
> if [ "$#" -eq "0" ]
> then
> - echo "usage: @GCORE_TRANSFORM_NAME@ [-a] [-o prefix] pid1 [pid2...pidN]"
> + echo "usage: @GCORE_TRANSFORM_NAME@ [-a] [-o prefix] [-d data-directory] pid1 [pid2...pidN]"
> exit 2
> fi
>
> @@ -98,6 +103,7 @@ do
> # `</dev/null' to avoid touching interactive terminal if it is
> # available but not accessible as GDB would get stopped on SIGTTIN.
> "$binary_path/@GDB_TRANSFORM_NAME@" </dev/null \
> + "${data_directory_opt[@]}" \
> --nx --batch --readnever -iex 'set debuginfod enabled off' \
> -ex "set pagination off" -ex "set height 0" -ex "set width 0" \
> "${dump_all_cmds[@]}" \
> diff --git a/gdb/testsuite/gdb.base/gcorebg.c b/gdb/testsuite/gdb.base/gcorebg.c
> new file mode 100644
> index 00000000000..1e9ddbd8649
> --- /dev/null
> +++ b/gdb/testsuite/gdb.base/gcorebg.c
> @@ -0,0 +1,86 @@
> +/* Copyright 2007-2024 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/>. */
> +
> +#include <stdio.h>
> +#include <sys/types.h>
> +#include <unistd.h>
> +#include <stdlib.h>
> +#include <signal.h>
> +#include <string.h>
> +#include <assert.h>
> +
> +/* Expects 4 arguments:
> +
> + 1. Either 'standard' or 'detached', where 'standard' tests
> + a general gcore script spawn with its controlling terminal available
> + and 'detached' tests gcore script spawn without its controlling
> + terminal available.
> + 2. Path to the gcore script.
> + 3. Path to the data-directory to pass to the gcore script.
> + 4. The core file output name. */
> +
> +int
> +main (int argc, char **argv)
> +{
> + pid_t pid = 0;
> + pid_t ppid;
> + char buf[1024*2 + 500];
> + int gotint, res;
> + int fd[2];
> +
> + assert (argc == 5);
> +
> + if (pipe(fd) == -1)
> + {
> + perror ("pipe err\n");
> + exit (1);
> + }
> + pid = fork ();
> +
> + switch (pid)
> + {
> + case 0:
> + close (fd[0]);
> + if (strcmp (argv[1], "detached") == 0)
> + setpgrp ();
> + ppid = getppid ();
> + gotint = snprintf (buf, sizeof (buf), "%s -d %s -o %s %d",
> + argv[2], argv[3], argv[4], (int) ppid);
> + assert (gotint < sizeof (buf));
> + res = system (buf);
> + assert (res != -1);
> + close(fd[1]);
> + break;
> +
> + case -1:
> + close (fd[0]);
> + close (fd[1]);
> + perror ("fork err\n");
> + exit (1);
> + break;
> +
> + default:
> + close (fd[1]);
> + /* Wait here until the child is done with gcore-ing us. When the
> + child has finished it will close its end of the pipe and this read
> + call will return. */
> + res = read (fd[0], buf, 1);
I guess in theory read could return -1 and errno be EINTR. I was
wondering if there is a reason to prefer creating a pipe rather than
using waitpid to wait for the child finish?
> + assert (res == 0);
> + close (fd[0]);
> + break;
> + }
> +
> + return 0;
> +}
> diff --git a/gdb/testsuite/gdb.base/gcorebg.exp b/gdb/testsuite/gdb.base/gcorebg.exp
> new file mode 100644
> index 00000000000..5bcb547b9e1
> --- /dev/null
> +++ b/gdb/testsuite/gdb.base/gcorebg.exp
> @@ -0,0 +1,88 @@
> +# Copyright 2007-2024 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/>.
> +#
> +# This is a test for the gcore script (not the gcore command from
> +# inside GDB). It also tests the gcore script being run without its
> +# accessible terminal.
> +
> +standard_testfile
> +require {!is_remote host}
> +require {!is_remote target}
> +require has_gcore_script
As mentioned above, I think you also need:
require can_spawn_for_attach
> +
> +set corefile [standard_output_file ${testfile}.core]
> +
> +if {[build_executable "failed to build" $testfile ${srcfile}] == -1 } {
> + return -1
> +}
> +
> +# Cleanup.
> +
> +proc core_clean {} {
> + global corefile
> +
> + foreach file [glob -nocomplain [join [list $corefile *] ""]] {
> + verbose "Delete file $file" 1
> + remote_file target delete $file
> + }
> +}
> +core_clean
> +
> +# Generate the core file.
> +proc test_body { detached } {
> + global binfile
> + global GCORE
> + global corefile
> + global GDB_DATA_DIRECTORY
> +
> + with_test_prefix "detached = $detached" {
See below, I gues you could use foreach_with_prefix outside of this
function which would be more "usual".
> + # We can't use gdb_test_multiple here because GDB is not started.
> + set res [remote_spawn target "$binfile $detached $GCORE $GDB_DATA_DIRECTORY $corefile"]
> + if { $res < 0 || $res == "" } {
> + fail "Spawning gcore"
> + return 1
> + }
> + pass "Spawned gcore"
> +
> + set saw_corefile_created false
> + set testname "Spawned gcore finished"
> + remote_expect target 20 {
> + timeout {
> + fail "$testname (timeout)"
> + remote_exec target "kill -9 -[exp_pid -i $res]"
> + return
> + }
> + -re "Saved corefile \[^\r\n\]+\r\n" {
> + set saw_corefile_created true
> + exp_continue
> + }
> + eof {
> + gdb_assert { $saw_corefile_created } $testname
> + }
> + }
> +
> + gdb_assert {1 == [llength [glob -nocomplain [join [list $corefile *] ""]]]} "Core file generated by gcore"
> + core_clean
> + }
> +}
> +
> +# First a general gcore script spawn with its controlling terminal available.
> +
> +test_body standard
> +
> +# And now gcore script spawn without its controlling terminal available.
> +# It is spawned through `gcorebg.c' using setpgrp ().
> +
> +test_body detached
Could be
foreach_with_prefix detached { detached standard } {
test_body $detached
}
> diff --git a/gdb/testsuite/lib/gdb.exp b/gdb/testsuite/lib/gdb.exp
> index cdc3721a1cd..25a9cec5c34 100644
> --- a/gdb/testsuite/lib/gdb.exp
> +++ b/gdb/testsuite/lib/gdb.exp
> @@ -158,6 +158,24 @@ load_lib gdb-utils.exp
> load_lib memory.exp
> load_lib check-test-names.exp
>
> +# The path to the GCORE script to test.
> +global GCORE
> +if {![info exists GCORE]} {
> + set GCORE [findfile $base_dir/../../gdb/gcore]
> +} else {
> + set GCORE ""
> +}
> +verbose "using GCORE = $GCORE" 2
> +
> +proc has_gcore_script {} {
> + global GCORE
> + if {$GCORE == ""} {
> + return 0
> + } else {
> + return 1
> + }
> +}
> +
> # The path to the GDB binary to test.
> global GDB
>
>
> --
> 2.45.0
>
Best,
Lancelot.
prev parent reply other threads:[~2024-06-03 20:17 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-06-02 15:50 Alexandra Hájková
2024-06-03 20:16 ` Lancelot SIX [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20240603201541.g5ly35xlknqamnvq@khazad-dum \
--to=lancelot.six@amd.com \
--cc=ahajkova@khirnov.net \
--cc=ahajkova@redhat.com \
--cc=gdb-patches@sourceware.org \
--cc=jan.kratochvil@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox