* [PATCH v3 0/3] Make linux checkpoints work with multiple inferiors
@ 2024-04-14 19:44 Kevin Buettner
2024-04-14 19:44 ` [PATCH v3 1/3] " Kevin Buettner
` (3 more replies)
0 siblings, 4 replies; 8+ messages in thread
From: Kevin Buettner @ 2024-04-14 19:44 UTC (permalink / raw)
To: gdb-patches; +Cc: Pedro Alves, Kevin Buettner
This series fixes some problems with the current checkpoint code. The
first patch makes the checkpoint code inferior aware, fixing a number
of bugs. The second and third patches are largely cosmetic - they
make changes to checkpoint related output.
The v2 series incorporates Pedro's suggestions regarding the
numbering of checkpoint ids. See the first patch for details. The
tests have been revised to account for these changes and new tests
have been added as well.
This v3 series splits out a cosmetic change from the first patch.
It capitalizes the output of a successful checkpoint command.
This was prompted by the Linaro regression tester, which, due to the
capitalization change, found two regressions in
gdb.base/kill-during-detach.exp. (I thought I was running this test
during my own testing, but found that it was not being run due to a
typo in the TESTS= portion of the "make check" line.)
Kevin Buettner (3):
Make linux checkpoints work with multiple inferiors
Capitalize output of successful checkpoint command
Make thread_db_target::pid_to_str checkpoint-aware
gdb/linux-fork.c | 519 +++++++++----
gdb/linux-fork.h | 15 +-
gdb/linux-nat.c | 20 +-
gdb/linux-thread-db.c | 4 +-
gdb/testsuite/gdb.base/checkpoint.exp | 26 +-
gdb/testsuite/gdb.base/kill-during-detach.exp | 2 +-
gdb/testsuite/gdb.multi/checkpoint-multi.exp | 687 ++++++++++++++++++
7 files changed, 1084 insertions(+), 189 deletions(-)
create mode 100644 gdb/testsuite/gdb.multi/checkpoint-multi.exp
--
2.44.0
^ permalink raw reply [flat|nested] 8+ messages in thread
* [PATCH v3 1/3] Make linux checkpoints work with multiple inferiors
2024-04-14 19:44 [PATCH v3 0/3] Make linux checkpoints work with multiple inferiors Kevin Buettner
@ 2024-04-14 19:44 ` Kevin Buettner
2024-05-03 15:54 ` Pedro Alves
2024-04-14 19:44 ` [PATCH v3 2/3] Capitalize output of successful checkpoint command Kevin Buettner
` (2 subsequent siblings)
3 siblings, 1 reply; 8+ messages in thread
From: Kevin Buettner @ 2024-04-14 19:44 UTC (permalink / raw)
To: gdb-patches; +Cc: Pedro Alves, Kevin Buettner
The current linux checkpoint code, most of which may be found in
linux-fork.c, is quite broken when attempting to use more than
one inferior. Running GDB will show internal errors when starting
two inferiors, placing a checkpoint in one, then switching to
the other and doing one of the following commands, "restart",
"detach", "kill", or continue (to program exit). Test cases
for two of those scenarios may be found in this bug:
https://sourceware.org/bugzilla/show_bug.cgi?id=31065
I've tested for each of the scenarios and many more in the new
test case, gdb.multi/checkpoint-multi.exp.
I started off with the goal of fixing just those problems, and was
mostly successful with a much smaller patch, but doing "info
checkpoints" with more than one inferior didn't work correctly due to
some of the inferiors being in the wrong program space. That led me
to making the linux-fork code fully inferior-aware.
Prior to this commit, the list of forks was being maintained in a
global named named 'fork_list'. I turned this into a per-inferior
data structure. There was also global named 'highest_fork_num' which
is also now part of the per-inferior struct. A registry key named
'checkpoint_inferior_data_key' along with function
'get_checkpoint_inferior_data' is used to access the per-inferior
data. This new function, get_checkpoint_inferior_data, is only
called by the new functions 'fork_list' and 'highest_fork_num', each
of which is passed a pointer to the inferior. Most occurrences
referring to the (previously) global 'fork_list' have been replaced
by 'fork_list (inf)'. In some functions, where the 'fork_list' is
referenced multiple times, a local named 'fork_list' is declared and
initialized instead, like this:
auto &fork_list = ::fork_list (inf);
The constructor for 'struct fork_info' has gained an additional
parameter. In addition to passing the pid of the new fork, we now
also pass the fork identifier, fork_num, to the constructor. This
integer is shown to the user in the "info checkpoints" command and
is provided by the user, perhaps in conjunction with the inferior
number, in commands which manipulate checkpoints, e.g. 'restart' and
'delete checkpoint'.
When checkpoints are used in only one inferior, this commit will
present information to the user and will accept checkpoint identifiers
to commands in much the same way as the code did before this commit.
Per Pedro Alves's recommendations, the "info checkpoints" command
has been changed somewhat. "info checkpoints" used to display
"(main process)" for the first process in the checkpoint list.
This is no longer done because it does not provide useful information.
It also used to display "<running>", presumably when in non-stop
mode. I've changed this to print a state flag 'R'. Additionally,
a state flag 'A' is used to indicate the active checkpoint for
each inferior. For the active inferior a '*' is also printed
preceding the checkpoint identifier. Here's what things look(ed)
like before and after for just one inferior:
Before:
(gdb) info checkpoints
* 0 Thread 0x7ffff7cd3740 (LWP 84201) (main process) at 0x40114a, file hello.c, line 28
1 process 84205 at 0x401199, file hello.c, line 51
2 process 84206 at 0x4011a3, file hello.c, line 53
After:
(gdb) info checkpoints
* 0 A process 84159 at 0x40114a, file hello.c, line 28
1 process 84162 at 0x401199, file hello.c, line 51
2 process 84163 at 0x4011a3, file hello.c, line 53
(The Thread versus process distinction is handled by another
patch - the "After" example assumes that patch is applied too.)
When there are multiple inferiors, the "info checkpoints" output looks
like this:
(gdb) info checkpoints
1.0 process 84159 at 0x40114a, file hello.c, line 28
* 1.1 A process 84162 at 0x401199, file hello.c, line 51
1.2 process 84163 at 0x4011a3, file hello.c, line 53
2.0 A process 84223 at 0x40116f, file hangout.c, line 35
2.1 process 84224 at 0x40115c, file hangout.c, line 31
2.2 process 84225 at 0x401166, file hangout.c, line 33
3.0 A process 84226 at 0x401218, file goodbye.c, line 46
3.1 process 84229 at 0x40124e, file goodbye.c, line 61
A new function named 'parse_checkpoint_id' has been added. As its
name suggests, it's responsible for parsing a string representing a
checkpoint identifier. These identifiers may be either a decimal
number representing the checkpoint number in the current inferior or
two decimal numbers separated by '.', in which case the first is the
inferior number and the second is the checkpoint number in that
inferior. It is called by delete_checkpoint_command,
detach_checkpoint_command, info_checkpoints_command, and
restart_command. Calls to 'parse_checkpoint_id' replace calls to
'parse_and_eval_long', plus error checking and error reporting code
near the calls to 'parse_and_eval_long'. As such, error checking and
reporting has been consolidated into a single function and the
messages output are more uniform, though this has necessitated changes
to the existing test case gdb.base/checkpoint.exp.
The functions 'find_fork_ptid' and 'find_fork_pid' used to return a
pointer to a fork_info struct. They now return a pair consisting of
the pointer to a fork_info struct in addition to a pointer to the
inferior containing that checkpoint.
'find_fork_id' returns a a pointer to a fork_info struct just as
it did before, but it's now gained a new parameter, 'inf', which
is the inferior in which to look.
info_checkpoints_command used to simply iterate over the list of
forks (checkpoints), printing each one out. It now needs to iterate
over all inferiors and, for those which have checkpoints, it needs
to iterate over the list of checkpoints in that inferior. As noted
earlier, the format of the output has been changed so that checkpoint
identifers incorporating an inferior number may be printed.
linux_fork_context, called by restart_command, now contains code to
switch inferiors when the fork being restarted is in a different
inferior from the current one. The scoped_switch_fork_info class now
also contains code for switching inferiors in both the constructor and
destructor.
gdb/linux-nat.c has a few changes. All but one of them are related
to passing the inferior to one of the linux-fork functions. But
one of the tests in linux_nat_target::detach has also changed in
a non-obvious way. In attempting to determine whether to call
linux_fork_detach(), that code used to do:
if (pid == inferior_ptid.pid () && forks_exist_p ())
It's been simplified to:
if (forks_exist_p (inf))
I had added the 'pid == inferior_ptid.pid ()' condition in late 2023
while working on a detach bug. It was kind of a hack to prevent
calling linux_fork_detach() when in a different inferior. That's no
longer needed since the call to forks_exist_p does this directly -
i.e. it is now inferior-aware.
Finally, the header file 'linux-fork.h' has been updated to reflect
the fact that add_fork, linux_fork_killall, linux_fork_mourn_inferior,
linux_fork_detach, and forks_exist_p all now require that a pointer to
an inferior be passed to these functions. Additionally (as mentioned
earlier), find_fork_pid now returns std::pair<fork_info *, inferior *>
instead 'of fork_info *'.
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=31065
---
gdb/linux-fork.c | 519 +++++++++-----
gdb/linux-fork.h | 15 +-
gdb/linux-nat.c | 20 +-
gdb/testsuite/gdb.base/checkpoint.exp | 24 +-
gdb/testsuite/gdb.multi/checkpoint-multi.exp | 687 +++++++++++++++++++
5 files changed, 1079 insertions(+), 186 deletions(-)
create mode 100644 gdb/testsuite/gdb.multi/checkpoint-multi.exp
diff --git a/gdb/linux-fork.c b/gdb/linux-fork.c
index bb8119a36d0..987c925bdbc 100644
--- a/gdb/linux-fork.c
+++ b/gdb/linux-fork.c
@@ -28,6 +28,7 @@
#include "linux-nat.h"
#include "gdbthread.h"
#include "source.h"
+#include "progspace-and-thread.h"
#include "nat/gdb_ptrace.h"
#include "gdbsupport/gdb_wait.h"
@@ -38,10 +39,11 @@
#include <list>
/* Fork list data structure: */
+
struct fork_info
{
- explicit fork_info (pid_t pid)
- : ptid (pid, pid)
+ explicit fork_info (pid_t pid, int fork_num)
+ : ptid (pid, pid), num (fork_num)
{
}
@@ -69,7 +71,7 @@ struct fork_info
ptid_t parent_ptid = null_ptid;
/* Convenient handle (GDB fork id). */
- int num = 0;
+ int num;
/* Convenient for info fork, saves having to actually switch
contexts. */
@@ -83,115 +85,253 @@ struct fork_info
int maxfd = 0;
};
-static std::list<fork_info> fork_list;
-static int highest_fork_num;
+/* Per-inferior checkpoint data. */
+
+struct checkpoint_inferior_data
+{
+ /* List of forks (checkpoints) in particular inferior. Once a
+ checkpoint has been created, fork_list will contain at least two
+ items, the first in the list will be the original (or, if not
+ original, then the oldest) fork. */
+ std::list<fork_info> fork_list;
+
+ /* Most recently assigned fork number; when 0, no checkpoints have
+ been created yet. */
+ int highest_fork_num = 0;
+};
+
+/* Per-inferior data key. */
+
+static const registry<inferior>::key<checkpoint_inferior_data>
+ checkpoint_inferior_data_key;
+
+/* Fetch per-inferior checkpoint data. It always returns a valid pointer
+ to a checkpoint_inferior_info struct. */
+
+static struct checkpoint_inferior_data *
+get_checkpoint_inferior_data (struct inferior *inf)
+{
+ struct checkpoint_inferior_data *data;
+
+ data = checkpoint_inferior_data_key.get (inf);
+ if (data == nullptr)
+ data = checkpoint_inferior_data_key.emplace (inf);
+
+ return data;
+}
+
+/* Return a reference to the per-inferior fork list. */
+
+static std::list<fork_info> &
+fork_list (inferior *inf)
+{
+ return get_checkpoint_inferior_data (inf)->fork_list;
+}
+
+/* Return a reference to 'highest_fork_num' for the given inferior INF. */
+
+static int &
+highest_fork_num (inferior *inf)
+{
+ return get_checkpoint_inferior_data (inf)->highest_fork_num;
+}
+
+/* Predicate which returns true when more than one inferior contains
+ a checkpoint, false otherwise. */
+
+static bool
+forks_exist_in_multiple_inferiors_p ()
+{
+ int count = 0;
+ for (inferior *inf : all_inferiors ())
+ {
+ if (forks_exist_p (inf))
+ count++;
+ if (count > 1)
+ return true;
+ }
+ return false;
+}
/* Fork list methods: */
-int
-forks_exist_p (void)
+/* Predicate which returns true if checkpoint(s) exist in the inferior
+ INF, false otherwise. */
+
+bool
+forks_exist_p (inferior *inf)
{
- return !fork_list.empty ();
+ return !fork_list (inf).empty ();
}
-/* Return the last fork in the list. */
+/* Return the last fork in the list for inferior INF. */
static struct fork_info *
-find_last_fork (void)
+find_last_fork (inferior *inf)
{
+ auto &fork_list = ::fork_list (inf);
+
if (fork_list.empty ())
return NULL;
return &fork_list.back ();
}
-/* Return true iff there's one fork in the list. */
+/* Return true iff there's one fork in the list for inferior INF. */
static bool
-one_fork_p ()
+one_fork_p (inferior *inf)
{
- return fork_list.size () == 1;
+ return fork_list (inf).size () == 1;
}
/* Add a new fork to the internal fork list. */
void
-add_fork (pid_t pid)
+add_fork (pid_t pid, inferior *inf)
{
- fork_list.emplace_back (pid);
-
- if (one_fork_p ())
- highest_fork_num = 0;
-
- fork_info *fp = &fork_list.back ();
- fp->num = ++highest_fork_num;
+ fork_list (inf).emplace_back (pid, ++highest_fork_num (inf));
}
+/* Delete a fork for PTID in inferior INF. When the last fork is
+ deleted, HIGHEST_FORK_NUM for the given inferior is reset to 0.
+ The fork list may also be made to be empty when only one fork
+ remains. */
+
static void
-delete_fork (ptid_t ptid)
+delete_fork (ptid_t ptid, inferior *inf)
{
linux_target->low_forget_process (ptid.pid ());
+ auto &fork_list = ::fork_list (inf);
for (auto it = fork_list.begin (); it != fork_list.end (); ++it)
if (it->ptid == ptid)
{
fork_list.erase (it);
+ if (fork_list.empty ())
+ {
+ /* Reset highest_fork_num. */
+ highest_fork_num (inf) = 0;
+ }
+
/* Special case: if there is now only one process in the list,
and if it is (hopefully!) the current inferior_ptid, then
remove it, leaving the list empty -- we're now down to the
default case of debugging a single process. */
- if (one_fork_p () && fork_list.front ().ptid == inferior_ptid)
+ if (one_fork_p (inf) && fork_list.front ().ptid == inferior_ptid)
{
/* Last fork -- delete from list and handle as solo
process (should be a safe recursion). */
- delete_fork (inferior_ptid);
+ delete_fork (inferior_ptid, inf);
}
return;
}
}
-/* Find a fork_info by matching PTID. */
-static struct fork_info *
+/* Find a fork_info and inferior by matching PTID. */
+
+static std::pair<fork_info *, inferior *>
find_fork_ptid (ptid_t ptid)
{
- for (fork_info &fi : fork_list)
- if (fi.ptid == ptid)
- return &fi;
+ for (inferior *inf : all_inferiors ())
+ {
+ for (fork_info &fi : fork_list (inf))
+ if (fi.ptid == ptid)
+ return { &fi, inf };
+ }
- return NULL;
+ return { nullptr, nullptr };
}
-/* Find a fork_info by matching ID. */
-static struct fork_info *
-find_fork_id (int num)
+/* Find a fork_info by matching NUM in inferior INF. */
+
+static fork_info *
+find_fork_id (inferior *inf, int num)
{
- for (fork_info &fi : fork_list)
+ for (fork_info &fi : fork_list (inf))
if (fi.num == num)
return &fi;
- return NULL;
+ return nullptr;
}
-/* Find a fork_info by matching pid. */
-extern struct fork_info *
+/* Find a fork_info and inferior by matching pid. */
+
+extern std::pair<fork_info *, inferior *>
find_fork_pid (pid_t pid)
{
- for (fork_info &fi : fork_list)
- if (pid == fi.ptid.pid ())
- return &fi;
+ for (inferior *inf : all_inferiors ())
+ {
+ for (fork_info &fi : fork_list (inf))
+ if (pid == fi.ptid.pid ())
+ return { &fi, inf };
+ }
- return NULL;
+ return { nullptr, nullptr };
}
-static ptid_t
-fork_id_to_ptid (int num)
+/* Parse a command argument representing a checkpoint id. This
+ can take one of two forms:
+
+ Num
+
+ -or-
+
+ Inf.Num
+
+ where Num is a non-negative decimal integer and Inf, if present, is
+ a positive decimal integer.
+
+ Return a pair with a pointer to the fork_info struct and pointer
+ to the inferior. This function will throw an error if there's
+ a problem with the parsing or if either the inferior or checkpoint
+ id does not exist. */
+
+static std::pair<fork_info *, inferior *>
+parse_checkpoint_id (const char *ckptstr)
{
- struct fork_info *fork = find_fork_id (num);
- if (fork)
- return fork->ptid;
+ const char *number = ckptstr;
+ const char *p1;
+ struct inferior *inf;
+
+ const char *dot = strchr (number, '.');
+
+ if (dot != nullptr)
+ {
+ /* Parse number to the left of the dot. */
+ int inf_num;
+
+ p1 = number;
+ inf_num = get_number_trailer (&p1, '.');
+ if (inf_num <= 0)
+ error (_("Inferior number must be a positive integer"));
+
+ inf = find_inferior_id (inf_num);
+ if (inf == NULL)
+ error (_("No inferior number '%d'"), inf_num);
+
+ p1 = dot + 1;
+ }
else
- return ptid_t (-1);
+ {
+ inf = current_inferior ();
+ p1 = number;
+ }
+
+ int fork_num = get_number_trailer (&p1, 0);
+ if (fork_num < 0)
+ error (_("Checkpoint number must be a non-negative integer"));
+
+ if (!forks_exist_p (inf))
+ error (_("Inferior %d has no checkpoints"), inf->num);
+
+ auto fork_ptr = find_fork_id (inf, fork_num);
+ if (fork_ptr == nullptr)
+ error (_("Invalid checkpoint number %d for inferior %d"),
+ fork_num, inf->num);
+
+ return { fork_ptr, inf };
}
/* Fork list <-> gdb interface. */
@@ -296,7 +436,7 @@ fork_save_infrun_state (struct fork_info *fp)
/* Kill 'em all, let God sort 'em out... */
void
-linux_fork_killall (void)
+linux_fork_killall (inferior *inf)
{
/* Walk list and kill every pid. No need to treat the
current inferior_ptid as special (we do not return a
@@ -304,6 +444,7 @@ linux_fork_killall (void)
or a parent, so may get a SIGCHLD from a previously
killed child. Wait them all out. */
+ auto &fork_list = ::fork_list (inf);
for (fork_info &fi : fork_list)
{
pid_t pid = fi.ptid.pid ();
@@ -322,6 +463,7 @@ linux_fork_killall (void)
/* Clear list, prepare to start fresh. */
fork_list.clear ();
+ highest_fork_num (inf) = 0;
}
/* The current inferior_ptid has exited, but there are other viable
@@ -329,7 +471,7 @@ linux_fork_killall (void)
first available. */
void
-linux_fork_mourn_inferior (void)
+linux_fork_mourn_inferior (inferior *inf)
{
struct fork_info *last;
int status;
@@ -341,23 +483,23 @@ linux_fork_mourn_inferior (void)
waitpid (inferior_ptid.pid (), &status, 0);
/* OK, presumably inferior_ptid is the one who has exited.
- We need to delete that one from the fork_list, and switch
+ We need to delete that one from the fork list, and switch
to the next available fork. */
- delete_fork (inferior_ptid);
+ delete_fork (inferior_ptid, inf);
/* There should still be a fork - if there's only one left,
delete_fork won't remove it, because we haven't updated
inferior_ptid yet. */
- gdb_assert (!fork_list.empty ());
+ gdb_assert (!fork_list (inf).empty ());
- last = find_last_fork ();
+ last = find_last_fork (inf);
fork_load_infrun_state (last);
gdb_printf (_("[Switching to %s]\n"),
target_pid_to_str (inferior_ptid).c_str ());
/* If there's only one fork, switch back to non-fork mode. */
- if (one_fork_p ())
- delete_fork (inferior_ptid);
+ if (one_fork_p (inf))
+ delete_fork (inferior_ptid, inf);
}
/* The current inferior_ptid is being detached, but there are other
@@ -365,13 +507,13 @@ linux_fork_mourn_inferior (void)
the first available. */
void
-linux_fork_detach (int from_tty, lwp_info *lp)
+linux_fork_detach (int from_tty, lwp_info *lp, inferior *inf)
{
gdb_assert (lp != nullptr);
gdb_assert (lp->ptid == inferior_ptid);
/* OK, inferior_ptid is the one we are detaching from. We need to
- delete it from the fork_list, and switch to the next available
+ delete it from the fork list, and switch to the next available
fork. But before doing the detach, do make sure that the lwp
hasn't exited or been terminated first. */
@@ -384,11 +526,12 @@ linux_fork_detach (int from_tty, lwp_info *lp)
target_pid_to_str (inferior_ptid).c_str ());
}
- delete_fork (inferior_ptid);
+ delete_fork (inferior_ptid, inf);
/* There should still be a fork - if there's only one left,
delete_fork won't remove it, because we haven't updated
inferior_ptid yet. */
+ auto &fork_list = ::fork_list (inf);
gdb_assert (!fork_list.empty ());
fork_load_infrun_state (&fork_list.front ());
@@ -398,8 +541,8 @@ linux_fork_detach (int from_tty, lwp_info *lp)
target_pid_to_str (inferior_ptid).c_str ());
/* If there's only one fork, switch back to non-fork mode. */
- if (one_fork_p ())
- delete_fork (inferior_ptid);
+ if (one_fork_p (inf))
+ delete_fork (inferior_ptid, inf);
}
/* Temporarily switch to the infrun state stored on the fork_info
@@ -412,19 +555,26 @@ class scoped_switch_fork_info
/* Switch to the infrun state held on the fork_info identified by
PPTID. If PPTID is the current inferior then no switch is done. */
explicit scoped_switch_fork_info (ptid_t pptid)
- : m_oldfp (nullptr)
+ : m_oldfp (nullptr), m_oldinf (nullptr)
{
if (pptid != inferior_ptid)
{
- struct fork_info *newfp = nullptr;
-
/* Switch to pptid. */
- m_oldfp = find_fork_ptid (inferior_ptid);
+ auto [oldfp, oldinf] = find_fork_ptid (inferior_ptid);
+ m_oldfp = oldfp;
gdb_assert (m_oldfp != nullptr);
- newfp = find_fork_ptid (pptid);
+ auto [newfp, newinf] = find_fork_ptid (pptid);
gdb_assert (newfp != nullptr);
fork_save_infrun_state (m_oldfp);
remove_breakpoints ();
+
+ if (oldinf != newinf)
+ {
+ thread_info *tp = any_thread_of_inferior (newinf);
+ switch_to_thread (tp);
+ m_oldinf = oldinf;
+ }
+
fork_load_infrun_state (newfp);
insert_breakpoints ();
}
@@ -434,12 +584,17 @@ class scoped_switch_fork_info
didn't need to switch states, then nothing is done here either. */
~scoped_switch_fork_info ()
{
- if (m_oldfp != nullptr)
+ if (m_oldinf != nullptr || m_oldfp != nullptr)
{
/* Switch back to inferior_ptid. */
try
{
remove_breakpoints ();
+ if (m_oldinf != nullptr)
+ {
+ thread_info *tp = any_thread_of_inferior (m_oldinf);
+ switch_to_thread (tp);
+ }
fork_load_infrun_state (m_oldfp);
insert_breakpoints ();
}
@@ -471,6 +626,7 @@ class scoped_switch_fork_info
we were already in the desired state, and nothing needs to be
restored. */
struct fork_info *m_oldfp;
+ inferior *m_oldinf;
};
static int
@@ -513,30 +669,35 @@ static void
delete_checkpoint_command (const char *args, int from_tty)
{
ptid_t ptid, pptid;
- struct fork_info *fi;
if (!args || !*args)
error (_("Requires argument (checkpoint id to delete)"));
- ptid = fork_id_to_ptid (parse_and_eval_long (args));
- if (ptid == minus_one_ptid)
- error (_("No such checkpoint id, %s"), args);
+ auto [fi, inf] = parse_checkpoint_id (args);
+ ptid = fi->ptid;
+ gdb_assert (fi != nullptr);
+ pptid = fi->parent_ptid;
- if (ptid == inferior_ptid)
- error (_("\
+ thread_info *inf_thr = any_thread_of_inferior (inf);
+ gdb_assert (inf_thr != nullptr);
+ if (ptid == inf_thr->ptid)
+ {
+ if (inf == current_inferior ())
+ error (_("\
Please switch to another checkpoint before deleting the current one"));
+ else
+ error (_("\
+Please switch to another checkpoint before deleting the current one\n\
+in inferior %d"), inf->num);
+ }
if (ptrace (PTRACE_KILL, ptid.pid (), 0, 0))
error (_("Unable to kill pid %s"), target_pid_to_str (ptid).c_str ());
- fi = find_fork_ptid (ptid);
- gdb_assert (fi);
- pptid = fi->parent_ptid;
-
if (from_tty)
gdb_printf (_("Killed %s\n"), target_pid_to_str (ptid).c_str ());
- delete_fork (ptid);
+ delete_fork (ptid, inf);
if (pptid == null_ptid)
{
@@ -554,7 +715,7 @@ Please switch to another checkpoint before deleting the current one"));
If fi->parent_ptid is a part of lwp and it is stopped, waitpid the
ptid. */
thread_info *parent = linux_target->find_thread (pptid);
- if ((parent == NULL && find_fork_ptid (pptid))
+ if ((parent == NULL && find_fork_ptid (pptid).first != nullptr)
|| (parent != NULL && parent->state == THREAD_STOPPED))
{
if (inferior_call_waitpid (pptid, ptid.pid ()))
@@ -571,9 +732,8 @@ detach_checkpoint_command (const char *args, int from_tty)
if (!args || !*args)
error (_("Requires argument (checkpoint id to detach)"));
- ptid = fork_id_to_ptid (parse_and_eval_long (args));
- if (ptid == minus_one_ptid)
- error (_("No such checkpoint id, %s"), args);
+ auto [fi, inf] = parse_checkpoint_id (args);
+ ptid = fi->ptid;
if (ptid == inferior_ptid)
error (_("\
@@ -585,7 +745,7 @@ Please switch to another checkpoint before detaching the current one"));
if (from_tty)
gdb_printf (_("Detached %s\n"), target_pid_to_str (ptid).c_str ());
- delete_fork (ptid);
+ delete_fork (ptid, current_inferior ());
}
/* Print information about currently known checkpoints. */
@@ -594,72 +754,96 @@ static void
info_checkpoints_command (const char *arg, int from_tty)
{
struct gdbarch *gdbarch = get_current_arch ();
- int requested = -1;
+ struct inferior *cur_inf = current_inferior ();
+ inferior *req_inf = nullptr;
+ fork_info *req_fi = nullptr;
bool printed = false;
if (arg && *arg)
- requested = (int) parse_and_eval_long (arg);
+ std::tie (req_fi, req_inf) = parse_checkpoint_id (arg);
- for (const fork_info &fi : fork_list)
+ /* Figure out whether to print the inferior number in the
+ checkpoint list. */
+ bool print_inf = (!forks_exist_p (current_inferior ())
+ || forks_exist_in_multiple_inferiors_p ());
+
+ for (inferior *inf : all_inferiors ())
{
- if (requested > 0 && fi.num != requested)
+ if (req_inf != nullptr && req_inf != inf)
continue;
- printed = true;
-
- bool is_current = fi.ptid == inferior_ptid;
- if (is_current)
- gdb_printf ("* ");
- else
- gdb_printf (" ");
- gdb_printf ("%d %s", fi.num, target_pid_to_str (fi.ptid).c_str ());
- if (fi.num == 0)
- gdb_printf (_(" (main process)"));
+ scoped_restore_current_pspace_and_thread restore_pspace_thread;
+ switch_to_program_space_and_thread (inf->pspace);
- if (is_current && inferior_thread ()->state == THREAD_RUNNING)
+ for (const fork_info &fi : fork_list (inf))
{
- gdb_printf (_(" <running>\n"));
- continue;
+ if (req_fi != nullptr && req_fi != &fi)
+ continue;
+ printed = true;
+
+ thread_info *t = any_thread_of_inferior (inf);
+ bool is_current = fi.ptid == t->ptid;
+ if (is_current && cur_inf == inf)
+ gdb_printf ("*");
+ else
+ gdb_printf (" ");
+
+ if (print_inf)
+ gdb_printf ("%3d.%-2d", inf->num, fi.num);
+ else
+ gdb_printf ("%3d ", fi.num);
+
+ /* Print out state associated with current checkpoint. */
+ if (is_current)
+ {
+ gdb_printf ("A");
+ if (t->state == THREAD_RUNNING)
+ gdb_printf ("R");
+ else
+ gdb_printf (" ");
+ gdb_printf (" ");
+ }
+ else
+ gdb_printf (" ");
+
+ gdb_printf ("%s", target_pid_to_str (fi.ptid).c_str ());
+
+ gdb_printf (_(" at "));
+ ULONGEST pc
+ = (is_current
+ ? regcache_read_pc (get_thread_regcache (t))
+ : fi.pc);
+ gdb_puts (paddress (gdbarch, pc));
+
+ symtab_and_line sal = find_pc_line (pc, 0);
+ if (sal.symtab)
+ gdb_printf (_(", file %s"),
+ symtab_to_filename_for_display (sal.symtab));
+ if (sal.line)
+ gdb_printf (_(", line %d"), sal.line);
+ if (!sal.symtab && !sal.line)
+ {
+ struct bound_minimal_symbol msym;
+
+ msym = lookup_minimal_symbol_by_pc (pc);
+ if (msym.minsym)
+ gdb_printf (", <%s>", msym.minsym->linkage_name ());
+ }
+
+ gdb_putc ('\n');
}
-
- gdb_printf (_(" at "));
- ULONGEST pc
- = (is_current
- ? regcache_read_pc (get_thread_regcache (inferior_thread ()))
- : fi.pc);
- gdb_puts (paddress (gdbarch, pc));
-
- symtab_and_line sal = find_pc_line (pc, 0);
- if (sal.symtab)
- gdb_printf (_(", file %s"),
- symtab_to_filename_for_display (sal.symtab));
- if (sal.line)
- gdb_printf (_(", line %d"), sal.line);
- if (!sal.symtab && !sal.line)
- {
- struct bound_minimal_symbol msym;
-
- msym = lookup_minimal_symbol_by_pc (pc);
- if (msym.minsym)
- gdb_printf (", <%s>", msym.minsym->linkage_name ());
- }
-
- gdb_putc ('\n');
}
if (!printed)
{
- if (requested > 0)
- gdb_printf (_("No checkpoint number %d.\n"), requested);
- else
- gdb_printf (_("No checkpoints.\n"));
+ gdb_printf (_("No checkpoints.\n"));
}
}
/* The PID of the process we're checkpointing. */
static int checkpointing_pid = 0;
-int
+bool
linux_fork_checkpointing_p (int pid)
{
return (checkpointing_pid == pid);
@@ -689,17 +873,16 @@ checkpoint_command (const char *args, int from_tty)
struct target_waitstatus last_target_waitstatus;
ptid_t last_target_ptid;
struct value *fork_fn = NULL, *ret;
- struct fork_info *fp;
pid_t retpid;
- if (!target_has_execution ())
+ if (!target_has_execution ())
error (_("The program is not being run."));
/* Ensure that the inferior is not multithreaded. */
update_thread_list ();
if (inf_has_multiple_threads ())
error (_("checkpoint: can't checkpoint multiple threads."));
-
+
/* Make the inferior fork, record its (and gdb's) state. */
if (lookup_minimal_symbol ("fork", NULL, NULL).minsym != NULL)
@@ -727,14 +910,22 @@ checkpoint_command (const char *args, int from_tty)
retpid = value_as_long (ret);
get_last_target_status (nullptr, &last_target_ptid, &last_target_waitstatus);
- fp = find_fork_pid (retpid);
+ auto [fp, inf] = find_fork_pid (retpid);
+
+ if (!fp)
+ error (_("Failed to find new fork"));
if (from_tty)
{
int parent_pid;
- gdb_printf (_("checkpoint %d: fork returned pid %ld.\n"),
- fp != NULL ? fp->num : -1, (long) retpid);
+ gdb_printf (_("checkpoint "));
+ if (forks_exist_in_multiple_inferiors_p ())
+ gdb_printf ("%d.%d: ", inf->num, fp->num);
+ else
+ gdb_printf ("%d: ", fp->num);
+ gdb_printf (_("fork returned pid %ld.\n"), (long) retpid);
+
if (info_verbose)
{
parent_pid = last_target_ptid.lwp ();
@@ -745,15 +936,12 @@ checkpoint_command (const char *args, int from_tty)
}
}
- if (!fp)
- error (_("Failed to find new fork"));
-
- if (one_fork_p ())
+ if (one_fork_p (inf))
{
/* Special case -- if this is the first fork in the list (the
- list was hitherto empty), then add inferior_ptid first, as a
- special zeroeth fork id. */
- fork_list.emplace_front (inferior_ptid.pid ());
+ list was hitherto empty), then add inferior_ptid as a special
+ zeroeth fork id. */
+ fork_list (inf).emplace_front (inferior_ptid.pid (), 0);
}
fork_save_infrun_state (fp);
@@ -761,40 +949,53 @@ checkpoint_command (const char *args, int from_tty)
}
static void
-linux_fork_context (struct fork_info *newfp, int from_tty)
+linux_fork_context (struct fork_info *newfp, int from_tty, inferior *newinf)
{
- /* Now we attempt to switch processes. */
- struct fork_info *oldfp;
+ bool inferior_changed = false;
+ /* Now we attempt to switch processes. */
gdb_assert (newfp != NULL);
- oldfp = find_fork_ptid (inferior_ptid);
- gdb_assert (oldfp != NULL);
+ if (newinf != current_inferior ())
+ {
+ thread_info *tp = any_thread_of_inferior (newinf);
+ switch_to_thread (tp);
+ inferior_changed = true;
+ }
- fork_save_infrun_state (oldfp);
- remove_breakpoints ();
- fork_load_infrun_state (newfp);
- insert_breakpoints ();
+ auto [oldfp, oldinf] = find_fork_ptid (inferior_ptid);
+ gdb_assert (oldfp != NULL);
- gdb_printf (_("Switching to %s\n"),
- target_pid_to_str (inferior_ptid).c_str ());
+ if (oldfp == newfp)
+ {
+ if (!inferior_changed)
+ error (_("Already using checkpoint %d"), oldfp->num);
+ }
+ else
+ {
+ fork_save_infrun_state (oldfp);
+ remove_breakpoints ();
+ fork_load_infrun_state (newfp);
+ insert_breakpoints ();
+ if (!inferior_changed)
+ gdb_printf (_("Switching to %s\n"),
+ target_pid_to_str (inferior_ptid).c_str ());
+ }
- print_stack_frame (get_selected_frame (NULL), 1, SRC_AND_LOC, 1);
+ notify_user_selected_context_changed
+ (inferior_changed ? (USER_SELECTED_INFERIOR | USER_SELECTED_FRAME)
+ : USER_SELECTED_FRAME);
}
/* Switch inferior process (checkpoint) context, by checkpoint id. */
static void
restart_command (const char *args, int from_tty)
{
- struct fork_info *fp;
-
if (!args || !*args)
error (_("Requires argument (checkpoint id to restart)"));
- if ((fp = find_fork_id (parse_and_eval_long (args))) == NULL)
- error (_("Not found: checkpoint id %s"), args);
-
- linux_fork_context (fp, from_tty);
+ auto [fp, inf] = parse_checkpoint_id (args);
+ linux_fork_context (fp, from_tty, inf);
}
void _initialize_linux_fork ();
diff --git a/gdb/linux-fork.h b/gdb/linux-fork.h
index c553aaf0740..0233a6c8cfc 100644
--- a/gdb/linux-fork.h
+++ b/gdb/linux-fork.h
@@ -22,12 +22,13 @@
struct fork_info;
struct lwp_info;
-extern void add_fork (pid_t);
-extern struct fork_info *find_fork_pid (pid_t);
-extern void linux_fork_killall (void);
-extern void linux_fork_mourn_inferior (void);
-extern void linux_fork_detach (int, lwp_info *);
-extern int forks_exist_p (void);
-extern int linux_fork_checkpointing_p (int);
+class inferior;
+extern void add_fork (pid_t, inferior *inf);
+extern std::pair<fork_info *, inferior *> find_fork_pid (pid_t);
+extern void linux_fork_killall (inferior *inf);
+extern void linux_fork_mourn_inferior (inferior *inf);
+extern void linux_fork_detach (int, lwp_info *, inferior *inf);
+extern bool forks_exist_p (inferior *inf);
+extern bool linux_fork_checkpointing_p (int);
#endif /* LINUX_FORK_H */
diff --git a/gdb/linux-nat.c b/gdb/linux-nat.c
index 2602e1f240d..e774691cb20 100644
--- a/gdb/linux-nat.c
+++ b/gdb/linux-nat.c
@@ -1552,13 +1552,13 @@ linux_nat_target::detach (inferior *inf, int from_tty)
gdb_assert (num_lwps (pid) == 1
|| (target_is_non_stop_p () && num_lwps (pid) == 0));
- if (pid == inferior_ptid.pid () && forks_exist_p ())
+ if (forks_exist_p (inf))
{
/* Multi-fork case. The current inferior_ptid is being detached
from, but there are other viable forks to debug. Detach from
the current fork, and context-switch to the first
available. */
- linux_fork_detach (from_tty, find_lwp_pid (ptid_t (pid)));
+ linux_fork_detach (from_tty, find_lwp_pid (ptid_t (pid)), inf);
}
else
{
@@ -2076,8 +2076,12 @@ linux_handle_extended_wait (struct lwp_info *lp, int status)
detach_breakpoints (ptid_t (new_pid, new_pid));
/* Retain child fork in ptrace (stopped) state. */
- if (!find_fork_pid (new_pid))
- add_fork (new_pid);
+ if (find_fork_pid (new_pid).first == nullptr)
+ {
+ struct inferior *inf = find_inferior_ptid (linux_target,
+ lp->ptid);
+ add_fork (new_pid, inf);
+ }
/* Report as spurious, so that infrun doesn't want to follow
this fork. We're actually doing an infcall in
@@ -3757,8 +3761,8 @@ linux_nat_target::kill ()
parent will be sleeping if this is a vfork. */
iterate_over_lwps (pid_ptid, kill_unfollowed_child_callback);
- if (forks_exist_p ())
- linux_fork_killall ();
+ if (forks_exist_p (current_inferior ()))
+ linux_fork_killall (current_inferior ());
else
{
/* Stop all threads before killing them, since ptrace requires
@@ -3789,14 +3793,14 @@ linux_nat_target::mourn_inferior ()
close_proc_mem_file (pid);
- if (! forks_exist_p ())
+ if (! forks_exist_p (current_inferior ()))
/* Normal case, no other forks available. */
inf_ptrace_target::mourn_inferior ();
else
/* Multi-fork case. The current inferior_ptid has exited, but
there are other viable forks to debug. Delete the exiting
one and context-switch to the first available. */
- linux_fork_mourn_inferior ();
+ linux_fork_mourn_inferior (current_inferior ());
/* Let the arch-specific native code know this process is gone. */
linux_target->low_forget_process (pid);
diff --git a/gdb/testsuite/gdb.base/checkpoint.exp b/gdb/testsuite/gdb.base/checkpoint.exp
index f87f528e167..b9a1e3161db 100644
--- a/gdb/testsuite/gdb.base/checkpoint.exp
+++ b/gdb/testsuite/gdb.base/checkpoint.exp
@@ -274,17 +274,17 @@ gdb_test "kill" "" "kill all one" \
# and confirm that all are gone
#
-gdb_test "restart 0" "Not found.*" "no more checkpoint 0"
-gdb_test "restart 1" "Not found.*" "no more checkpoint 1"
-gdb_test "restart 2" "Not found.*" "no more checkpoint 2"
-gdb_test "restart 3" "Not found.*" "no more checkpoint 3"
-gdb_test "restart 4" "Not found.*" "no more checkpoint 4"
-gdb_test "restart 5" "Not found.*" "no more checkpoint 5"
-gdb_test "restart 6" "Not found.*" "no more checkpoint 6"
-gdb_test "restart 7" "Not found.*" "no more checkpoint 7"
-gdb_test "restart 8" "Not found.*" "no more checkpoint 8"
-gdb_test "restart 9" "Not found.*" "no more checkpoint 9"
-gdb_test "restart 10" "Not found.*" "no more checkpoint 10"
+gdb_test "restart 0" "has no checkpoints" "no more checkpoint 0"
+gdb_test "restart 1" "has no checkpoints" "no more checkpoint 1"
+gdb_test "restart 2" "has no checkpoints" "no more checkpoint 2"
+gdb_test "restart 3" "has no checkpoints" "no more checkpoint 3"
+gdb_test "restart 4" "has no checkpoints" "no more checkpoint 4"
+gdb_test "restart 5" "has no checkpoints" "no more checkpoint 5"
+gdb_test "restart 6" "has no checkpoints" "no more checkpoint 6"
+gdb_test "restart 7" "has no checkpoints" "no more checkpoint 7"
+gdb_test "restart 8" "has no checkpoints" "no more checkpoint 8"
+gdb_test "restart 9" "has no checkpoints" "no more checkpoint 9"
+gdb_test "restart 10" "has no checkpoints" "no more checkpoint 10"
#
# Now let's try setting a large number of checkpoints (>600)
@@ -309,7 +309,7 @@ gdb_test "continue" "breakpoint 2.*" "break2 with many checkpoints"
set count 0
set msg "info checkpoints with at least 600 checkpoints"
gdb_test_multiple "info checkpoints" $msg {
- -re " $decimal process \[^\r\]*\r\n" {
+ -re " $decimal A? *process \[^\r\]*\r\n" {
incr count
exp_continue
}
diff --git a/gdb/testsuite/gdb.multi/checkpoint-multi.exp b/gdb/testsuite/gdb.multi/checkpoint-multi.exp
new file mode 100644
index 00000000000..929b09b6876
--- /dev/null
+++ b/gdb/testsuite/gdb.multi/checkpoint-multi.exp
@@ -0,0 +1,687 @@
+# Copyright 2009-2024 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 file tests various scenarios involving multiple inferiors
+# and the checkpoint command.
+
+# Checkpoint support works only on Linux.
+require {istarget "*-*-linux*"}
+
+# Checkpoint support is implemented for the (Linux) native target,
+# which is indicated when [target_info gdb_protocol] returns the empty
+# string.
+if { [target_info gdb_protocol] != "" } {
+ return
+}
+
+set testfile "checkpoint-multi"
+
+set exec1 "hello"
+set srcfile1 ${exec1}.c
+set binfile1 [standard_output_file ${exec1}]
+
+set exec2 "goodbye"
+set srcfile2 ${exec2}.c
+set binfile2 [standard_output_file ${exec2}]
+
+set exec3 "hangout"
+set srcfile3 ${exec3}.c
+set binfile3 [standard_output_file ${exec3}]
+
+if { [build_executable ${testfile}.exp ${exec1} "${srcfile1}" {debug}] == -1 } {
+ return -1
+}
+
+if { [build_executable ${testfile}.exp ${exec2} "${srcfile2}" {debug}] == -1 } {
+ return -1
+}
+
+if { [build_executable ${testfile}.exp ${exec3} "${srcfile3}" {debug}] == -1 } {
+ return -1
+}
+
+# Start two inferiors, place a checkpoint on inferior 2, but switch
+# back to inferior 1.
+proc start_2_inferiors_checkpoint_on_inf_2 {} {
+ clean_restart $::exec1
+
+ # Start inferior 1.
+ if {[gdb_start_cmd] < 0} {
+ fail "start first inferior"
+ } else {
+ gdb_test "" "main.*" "start first inferior"
+ }
+
+ # Add a new inferior and exec into it.
+ gdb_test "add-inferior -exec $::binfile2" \
+ "Added inferior 2.*" \
+ "add inferior 2 with -exec $::exec2"
+
+ # Check that we have multiple inferiors.
+ gdb_test "info inferiors" \
+ "Executable.*$::exec1.*$::exec2.*"
+
+ # Switch to inferior 2.
+ gdb_test "inferior 2" \
+ "Switching to inferior 2.*$::exec2.*"
+
+ # Start inferior 2:
+ if {[gdb_start_cmd] < 0} {
+ fail "start second inferior"
+ } else {
+ gdb_test "" "main.*" "start second inferior"
+ }
+
+ # Set a checkpoint in inferior 2
+ gdb_test "checkpoint" "\[Cc\]heckpoint 1: fork returned pid $::decimal.*"
+
+ # Step one line in inferior 2.
+ gdb_test "step" "glob = 46;"
+
+ # Switch back to inferior 1.
+ gdb_test "inferior 1" "Switching to inferior 1.*$::exec1.*"
+}
+
+# Start two inferiors, place a checkpoint on inferior 2, but switch
+# back to inferior 1. This is like the one above, except that it
+# swaps the executables loaded into inferior 1 and inferior 2. This
+# is important for being able to test "continue to exit". (Because...
+# hello.c has an infinite loop, but goodbye.c doesn't. In order to
+# test "continue to exit", we need to continue in an executable which
+# will actually exit.)
+
+proc start_2_inferiors_checkpoint_on_inf_2_alt {} {
+ clean_restart $::exec2
+
+ # Start inferior 1.
+ if {[gdb_start_cmd] < 0} {
+ fail "start first inferior"
+ } else {
+ gdb_test "" "main.*" "start first inferior"
+ }
+
+ # Add a new inferior and exec exec1 into it.
+ gdb_test "add-inferior -exec $::binfile1" \
+ "Added inferior 2.*" \
+ "add inferior 2 with -exec $::exec1"
+
+ # Check that we have two inferiors.
+ gdb_test "info inferiors" \
+ "Executable.*$::exec2.*$::exec1.*"
+
+ # Switch to inferior 2.
+ gdb_test "inferior 2" \
+ "Switching to inferior 2.*$::exec1.*"
+
+ # Start inferior 2:
+ if {[gdb_start_cmd] < 0} {
+ fail "start second inferior"
+ } else {
+ gdb_test "" "main.*" "start second inferior"
+ }
+
+ # Set a checkpoint in inferior 2
+ gdb_test "checkpoint" "\[Cc\]heckpoint 1: fork returned pid $::decimal.*"
+
+ # next one line in inferior 2.
+ gdb_test "next" "bar\\(\\).*"
+
+ # Switch back to inferior 1.
+ gdb_test "inferior 1" "Switching to inferior 1.*$::exec2.*"
+}
+
+with_test_prefix "check detach on non-checkpointed inferior" {
+ start_2_inferiors_checkpoint_on_inf_2
+ gdb_test "detach" "Detaching from program.*$::exec1.*Inferior 1.*detached.*"
+}
+
+with_test_prefix "check kill on non-checkpointed inferior" {
+ start_2_inferiors_checkpoint_on_inf_2
+ gdb_test "kill" "" "kill non-checkpointed inferior" \
+ "Kill the program being debugged.*y or n. $" "y"
+}
+
+with_test_prefix "check restart 0 on non-checkpointed inferior" {
+ start_2_inferiors_checkpoint_on_inf_2
+ gdb_test "restart 0" "Inferior 1 has no checkpoints"
+ gdb_test "restart 2.0" "Switching to inferior 2.*?goodbye.*?#0 +mailand .*?glob = 46;.*"
+}
+
+with_test_prefix "check restart 1 on non-checkpointed inferior" {
+ start_2_inferiors_checkpoint_on_inf_2
+ gdb_test "restart 1" "Inferior 1 has no checkpoints"
+ gdb_test "restart 2.1" "Switching to inferior 2.*?goodbye.*?#0 +main .*?mailand\\(\\);.*"
+}
+
+with_test_prefix "check continue to exit on non-checkpointed inferior" {
+ start_2_inferiors_checkpoint_on_inf_2_alt
+ gdb_test "continue" "Inferior 1.*? exited normally.*"
+}
+
+set proc_re "(?:process $::decimal|Thread $::hex \\(LWP $::decimal\\))"
+set main_proc "\\(main process\\)"
+set hello_c "hello\\.c"
+set goodbye_c "goodbye\\.c"
+set hangout_c "hangout\\.c"
+
+with_test_prefix "two inferiors with checkpoints" {
+ start_2_inferiors_checkpoint_on_inf_2
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 2.0 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2.1 $proc_re at $::hex, file.*?$goodbye_c.*?"] \
+ "info checkpoints 1"
+ gdb_test "checkpoint" "\[Cc\]heckpoint 1\\.1: fork returned pid $::decimal.*" \
+ "checkpoint in inferior 1"
+ gdb_test "info checkpoints" \
+ [multi_line \
+ "\\* 1\\.0 A $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.1 $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 2\\.0 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.1 $proc_re at $::hex, file.*?$goodbye_c.*?"] \
+ "info checkpoints 2"
+ gdb_test "restart 0" "Already using checkpoint 0"
+ gdb_test "restart 2.0" \
+ "\\\[Switching to inferior 2.*?mailand.*?glob = 46;.*"
+ gdb_test "next" "\}"
+
+ gdb_test "restart 1" "^Switching to $proc_re.*?#0 main \\(\\) at.*?$goodbye_c.*mailand\\(\\);"
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 1\\.0 A $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.1 $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 2\\.0 $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ "\\* 2\\.1 A $proc_re at $::hex, file.*?$goodbye_c.*?"] \
+ "info checkpoints 3"
+
+ # Doing "info_checkpoints" twice in a row might seem pointless,
+ # but during work on making the checkpoint code inferior aware,
+ # there was a point at which doing it twice in a row did not
+ # produce the same output.
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 1\\.0 A $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.1 $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 2\\.0 $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ "\\* 2\\.1 A $proc_re at $::hex, file.*?$goodbye_c.*?"] \
+ "info checkpoints 4"
+
+ # Switch back to checkpoint 0; again, there should be no
+ # "Switching to inferior" message.
+ gdb_test "restart 0" "^Switching to $proc_re.*?#0 mailand \\(\\) at.*?$goodbye_c.*\}" \
+ "restart 0 #2"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 1\\.0 A $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.1 $proc_re at $::hex, file.*?$hello_c.*?" \
+ "\\* 2\\.0 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.1 $proc_re at $::hex, file.*?$goodbye_c.*?"] \
+ "info checkpoints 5"
+
+ # Try switching to invalid checkpoints:
+ gdb_test "restart 3" "Invalid checkpoint number 3 for inferior 2"
+ gdb_test "restart 2" "Invalid checkpoint number 2 for inferior 2"
+ gdb_test "restart -1" "Checkpoint number must be a non-negative integer"
+ gdb_test "restart 2.3" "Invalid checkpoint number 3 for inferior 2"
+ gdb_test "restart 3.0" "No inferior number '3'"
+ gdb_test "restart 1.2" "Invalid checkpoint number 2 for inferior 1"
+ gdb_test "restart 1.3" "Invalid checkpoint number 3 for inferior 1"
+ gdb_test "restart 1.-1" "Checkpoint number must be a non-negative integer"
+ gdb_test "restart -1.0" "Inferior number must be a positive integer"
+
+ # Switch to checkpoint 3; this time, we should see a "Switching to
+ # inferior" message.
+ gdb_test "restart 1.1" \
+ "\\\[Switching to inferior 1.*?main.*?$hello_c.*?alarm \\(240\\);"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 1\\.0 $proc_re at $::hex, file.*?$hello_c.*?" \
+ "\\* 1\\.1 A $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 2\\.0 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.1 $proc_re at $::hex, file.*?$goodbye_c.*?"] \
+ "info checkpoints 6"
+
+ gdb_test "restart 2.1" \
+ "Switching to inferior 2.*?#0 main \\(\\) at.*?$goodbye_c.*mailand\\(\\);"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 1\\.0 $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.1 A $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 2\\.0 $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ "\\* 2\\.1 A $proc_re at $::hex, file.*?$goodbye_c.*?"] \
+ "info checkpoints 7"
+
+ gdb_test "checkpoint" "\[Cc\]heckpoint 2\\.2: fork returned pid $::decimal.*" \
+ "second checkpoint in inferior 2"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 1\\.0 $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.1 A $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 2\\.0 $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ "\\* 2\\.1 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.2 $proc_re at $::hex, file.*?$goodbye_c.*?"] \
+ "info checkpoints 8"
+
+ gdb_test "checkpoint" "\[Cc\]heckpoint 2.3: fork returned pid $::decimal.*" \
+ "third checkpoint in inferior 2"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 1\\.0 $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.1 A $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 2\\.0 $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ "\\* 2\\.1 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.2 $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.3 $proc_re at $::hex, file.*?$goodbye_c.*?"] \
+ "info checkpoints 9"
+
+ gdb_test "continue" \
+ "Inferior 2 \\(process $decimal\\) exited normally.*?Switching to $proc_re.*?" \
+ "continue to exit in checkpoint 2.1"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 1\\.0 $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.1 A $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 2\\.0 $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.2 $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ "\\* 2\\.3 A $proc_re at $::hex, file.*?$goodbye_c.*?"] \
+ "info checkpoints 10"
+
+ gdb_test "continue" \
+ "Inferior 2 \\(process $decimal\\) exited normally.*?Switching to process $decimal.*?" \
+ "continue to exit in checkpoint 2.3"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 1\\.0 $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.1 A $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 2\\.0 $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ "\\* 2\\.2 A $proc_re at $::hex, file.*?$goodbye_c.*?"] \
+ "info checkpoints 11"
+
+ gdb_test "continue" \
+ "Inferior 2 \\(process $decimal\\) exited normally.*?Switching to process $decimal.*?" \
+ "continue to exit in checkpoint 2.2"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 1\\.0 $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.1 A $proc_re at $::hex, file.*?$hello_c.*?"] \
+ "info checkpoints 12"
+
+ gdb_test "checkpoint" "\[Cc\]heckpoint 2.1: fork returned pid $::decimal.*" \
+ "fourth checkpoint in inferior 2"
+
+ gdb_test "checkpoint" "\[Cc\]heckpoint 2.2: fork returned pid $::decimal.*" \
+ "fifth checkpoint in inferior 2"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 1\\.0 $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.1 A $proc_re at $::hex, file.*?$hello_c.*?" \
+ "\\* 2\\.0 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.1 $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.2 $proc_re at $::hex, file.*?$goodbye_c.*?"] \
+ "info checkpoints 13"
+
+ gdb_test "delete checkpoint 2.0" \
+ "Please switch to another checkpoint before deleting the current one"
+
+ gdb_test "restart 2.1" \
+ "^Switching to process.*?#0 mailand \\(\\) at.*?$goodbye_c.*\}" \
+ "restart 2.1 #2"
+
+ gdb_test "delete checkpoint 2.0" \
+ "Killed process $::decimal" \
+ "delete checkpoint 2.0 #2"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 1\\.0 $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.1 A $proc_re at $::hex, file.*?$hello_c.*?" \
+ "\\* 2\\.1 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.2 $proc_re at $::hex, file.*?$goodbye_c.*?"] \
+ "info checkpoints 14"
+
+ gdb_test "delete checkpoint 2.2" \
+ "Killed process $::decimal"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 1\\.0 $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.1 A $proc_re at $::hex, file.*?$hello_c.*?"] \
+ "info checkpoints 15"
+
+ gdb_test "checkpoint" "\[Cc\]heckpoint 2.1: fork returned pid $::decimal.*" \
+ "sixth checkpoint in inferior 2"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 1\\.0 $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.1 A $proc_re at $::hex, file.*?$hello_c.*?" \
+ "\\* 2\\.0 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.1 $proc_re at $::hex, file.*?$goodbye_c.*?"] \
+ "info checkpoints 16"
+
+ gdb_test "inferior 1" "Switching to inferior 1.*?alarm \\(240\\);" \
+ "inferior 1 #2"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 1\\.0 $proc_re at $::hex, file.*?$hello_c.*?" \
+ "\\* 1\\.1 A $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 2\\.0 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.1 $proc_re at $::hex, file.*?$goodbye_c.*?"] \
+ "info checkpoints 17"
+
+ gdb_test "kill" "\\\[Inferior 1 \\(process $::decimal\\) killed\\\]" \
+ "kill inferior 1" \
+ "Kill the program being debugged.*y or n. $" "y"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 2\\.0 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.1 $proc_re at $::hex, file.*?$goodbye_c.*?"] \
+ "info checkpoints 18"
+
+ gdb_test "checkpoint" "The program is not being run\\." \
+ "checkpoint in non-running inferior"
+
+ gdb_test "start" "Starting program.*?hello.*?alarm \\(240\\);"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 2\\.0 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.1 $proc_re at $::hex, file.*?$goodbye_c.*?"] \
+ "info checkpoints 19"
+
+ gdb_test "checkpoint" "\[Cc\]heckpoint 1.1: fork returned pid $::decimal.*" \
+ "second checkpoint in inferior 1"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ "\\* 1\\.0 A $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.1 $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 2\\.0 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.1 $proc_re at $::hex, file.*?$goodbye_c.*?"] \
+ "info checkpoints 20"
+}
+
+with_test_prefix "three inferiors with checkpoints" {
+ start_2_inferiors_checkpoint_on_inf_2
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 2\\.0 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.1 $proc_re at $::hex, file.*?$goodbye_c.*?"] \
+ "info checkpoints 1"
+
+ # Add a third inferior and exec into it.
+ gdb_test "add-inferior -exec $::binfile3" \
+ "Added inferior 3.*" \
+ "add inferior 3 with -exec $::exec3"
+
+ # Check that we have three inferiors.
+ gdb_test "info inferiors" \
+ "Executable.*?\\* 1 .*?$::exec1.*? 2 .*?$::exec2.*? 3 .*?$::exec3.*?" \
+ "check for three inferiors"
+
+ # Switch to inferior 3.
+ gdb_test "inferior 3" \
+ "Switching to inferior 3.*$::exec3.*"
+
+ # Start inferior 2:
+ if {[gdb_start_cmd] < 0} {
+ fail "start third inferior"
+ } else {
+ gdb_test "" "main.*" "start third inferior"
+ }
+
+ gdb_test "checkpoint" "\[Cc\]heckpoint 3\\.1: fork returned pid $::decimal.*" \
+ "first checkpoint in inferior 3"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 2\\.0 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.1 $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ "\\* 3\\.0 A $proc_re at $::hex, file.*?$hangout_c.*?" \
+ " 3\\.1 $proc_re at $::hex, file.*?$hangout_c.*?"] \
+ "info checkpoints 2"
+
+ gdb_test "inferior 1" "Switching to inferior 1.*?alarm \\(240\\);" \
+ "inferior 1 #2"
+
+ gdb_test "checkpoint" "\[Cc\]heckpoint 1\\.1: fork returned pid $::decimal.*" \
+ "first checkpoint in inferior 1"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ "\\* 1\\.0 A $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.1 $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 2\\.0 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.1 $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 3\\.0 A $proc_re at $::hex, file.*?$hangout_c.*?" \
+ " 3\\.1 $proc_re at $::hex, file.*?$hangout_c.*?"] \
+ "info checkpoints 3"
+
+ gdb_test "restart 2.1" \
+ "Switching to inferior 2.*?#0 main \\(\\) at.*?$goodbye_c.*mailand\\(\\);" \
+ "restart 1"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 1\\.0 A $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.1 $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 2\\.0 $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ "\\* 2\\.1 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 3\\.0 A $proc_re at $::hex, file.*?$hangout_c.*?" \
+ " 3\\.1 $proc_re at $::hex, file.*?$hangout_c.*?"] \
+ "info checkpoints 4"
+
+ gdb_test "next" "foo\\(glob\\);"
+
+ gdb_test "checkpoint" "\[Cc\]heckpoint 2\\.2: fork returned pid $::decimal.*" \
+ "second checkpoint in inferior 2"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 1\\.0 A $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.1 $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 2\\.0 $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ "\\* 2\\.1 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.2 $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 3\\.0 A $proc_re at $::hex, file.*?$hangout_c.*?" \
+ " 3\\.1 $proc_re at $::hex, file.*?$hangout_c.*?"] \
+ "info checkpoints 5"
+
+ gdb_test "inferior 3" "Switching to inferior 3.*?alarm \\(30\\);" \
+ "inferior 3 #2"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 1\\.0 A $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.1 $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 2\\.0 $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.1 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.2 $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ "\\* 3\\.0 A $proc_re at $::hex, file.*?$hangout_c.*?" \
+ " 3\\.1 $proc_re at $::hex, file.*?$hangout_c.*?"] \
+ "info checkpoints 6"
+
+ gdb_test "kill" "\\\[Inferior 3 \\(process $::decimal\\) killed\\\]" \
+ "kill inferior 3" \
+ "Kill the program being debugged.*y or n. $" "y"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 1\\.0 A $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.1 $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 2\\.0 $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.1 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.2 $proc_re at $::hex, file.*?$goodbye_c.*?" ]\
+ "info checkpoints 7"
+
+ gdb_test "delete checkpoint 0" \
+ "Inferior 3 has no checkpoints"
+ gdb_test "delete checkpoint 2.0" \
+ "Killed process $::decimal"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 1\\.0 A $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.1 $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 2\\.1 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.2 $proc_re at $::hex, file.*?$goodbye_c.*?" ]\
+ "info checkpoints 8"
+
+ gdb_test "restart 2.2" \
+ "Switching to inferior 2.*?#0 main \\(\\) at.*?$goodbye_c.*foo\\(glob\\);"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 1\\.0 A $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.1 $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 2\\.1 $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ "\\* 2\\.2 A $proc_re at $::hex, file.*?$goodbye_c.*?" ]\
+ "info checkpoints 9"
+
+ gdb_test "inferior 3" "\\\[Switching to inferior 3 \\\[<null>\\\] \\(.*?$::exec3\\)\\\]" \
+ "inferior 3 #3"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 1\\.0 A $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.1 $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 2\\.1 $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.2 A $proc_re at $::hex, file.*?$goodbye_c.*?" ]\
+ "info checkpoints 10"
+
+ gdb_test "start" "Starting program.*?hangout.*?alarm \\(30\\);"
+
+ gdb_test "checkpoint" "\[Cc\]heckpoint 3\\.1: fork returned pid $::decimal.*" \
+ "second checkpoint in inferior 3"
+
+ gdb_test "checkpoint" "\[Cc\]heckpoint 3\\.2: fork returned pid $::decimal.*" \
+ "third checkpoint in inferior 3"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 1\\.0 A $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.1 $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 2\\.1 $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.2 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ "\\* 3\\.0 A $proc_re at $::hex, file.*?$hangout_c.*?" \
+ " 3\\.1 $proc_re at $::hex, file.*?$hangout_c.*?" \
+ " 3\\.2 $proc_re at $::hex, file.*?$hangout_c.*?"] \
+ "info checkpoints 11"
+
+ gdb_test "delete checkpoint 1" \
+ "Killed process $::decimal"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 1\\.0 A $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.1 $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 2\\.1 $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.2 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ "\\* 3\\.0 A $proc_re at $::hex, file.*?$hangout_c.*?" \
+ " 3\\.2 $proc_re at $::hex, file.*?$hangout_c.*?"] \
+ "info checkpoints 12"
+
+ # Switch to inferior 1, add another checkpoint - so that there are
+ # three of them in inferior 1 - then switch back to inferior 3 and
+ # delete active checkpoint in inferior 1. Then, switch to
+ # inferior 1 and attempt to add another checkpoint. During
+ # development, a "Cannot access memory at address ..." message was
+ # seen. This was a bug - there were several problems, but one
+ # of them was that the checkpoint in question was an "active"
+ # checkpoint. The fix was to disallow this case.
+ gdb_test "inferior 1" "Switching to inferior 1.*?alarm \\(240\\);" \
+ "inferior 1 #3"
+
+ gdb_test "checkpoint" "\[Cc\]heckpoint 1\\.2: fork returned pid $::decimal.*" \
+ "second checkpoint in inferior 1"
+
+ gdb_test "inferior 3" "Switching to inferior 3.*?alarm \\(30\\);" \
+ "inferior 3 #4"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 1\\.0 A $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.1 $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.2 $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 2\\.1 $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.2 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ "\\* 3\\.0 A $proc_re at $::hex, file.*?$hangout_c.*?" \
+ " 3\\.2 $proc_re at $::hex, file.*?$hangout_c.*?"] \
+ "info checkpoints 13"
+
+ # Check that deleting active checkpoints in other (non-current)
+ # inferiors is disallowed.
+ gdb_test "delete checkpoint 1.0" \
+ "Please switch to another checkpoint before deleting the current one\[ \r\n\]+in inferior 1"
+
+ # But deleting non-active checkpoints, even in other inferiors
+ # should work.
+ gdb_test "delete checkpoint 1.1" \
+ "Killed process $::decimal"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ " 1\\.0 A $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.2 $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 2\\.1 $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.2 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ "\\* 3\\.0 A $proc_re at $::hex, file.*?$hangout_c.*?" \
+ " 3\\.2 $proc_re at $::hex, file.*?$hangout_c.*?"] \
+ "info checkpoints 14"
+
+ gdb_test "inferior 1" "Switching to inferior 1.*?alarm \\(240\\);" \
+ "inferior 1 #4"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ "\\* 1\\.0 A $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.2 $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 2\\.1 $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.2 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 3\\.0 A $proc_re at $::hex, file.*?$hangout_c.*?" \
+ " 3\\.2 $proc_re at $::hex, file.*?$hangout_c.*?"] \
+ "info checkpoints 15"
+
+ gdb_test "checkpoint" "\[Cc\]heckpoint 1\\.3: fork returned pid $::decimal.*" \
+ "third checkpoint in inferior 1"
+
+ gdb_test "info checkpoints" \
+ [multi_line \
+ "\\* 1\\.0 A $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.2 $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 1\\.3 $proc_re at $::hex, file.*?$hello_c.*?" \
+ " 2\\.1 $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 2\\.2 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
+ " 3\\.0 A $proc_re at $::hex, file.*?$hangout_c.*?" \
+ " 3\\.2 $proc_re at $::hex, file.*?$hangout_c.*?"] \
+ "info checkpoints 16"
+
+ gdb_test "x/i \$pc" "=> $::hex <main.*"
+
+ gdb_test "remove-inferior 3" \
+ "warning: Can not remove active inferior 3\."
+}
--
2.44.0
^ permalink raw reply [flat|nested] 8+ messages in thread
* [PATCH v3 2/3] Capitalize output of successful checkpoint command
2024-04-14 19:44 [PATCH v3 0/3] Make linux checkpoints work with multiple inferiors Kevin Buettner
2024-04-14 19:44 ` [PATCH v3 1/3] " Kevin Buettner
@ 2024-04-14 19:44 ` Kevin Buettner
2024-05-03 15:54 ` Pedro Alves
2024-04-14 19:44 ` [PATCH v3 3/3] Make thread_db_target::pid_to_str checkpoint-aware Kevin Buettner
2024-05-02 1:49 ` [PATCH v3 0/3] Make linux checkpoints work with multiple inferiors Kevin Buettner
3 siblings, 1 reply; 8+ messages in thread
From: Kevin Buettner @ 2024-04-14 19:44 UTC (permalink / raw)
To: gdb-patches; +Cc: Pedro Alves, Kevin Buettner
This commit causes the output of a "checkpoint" command to be
changed from:
checkpoint N: fork returned pid DDDD
to:
Checkpoint N: fork returned pid DDDD
This change was made to bring the output of the "checkpoint" command in
line with that of other commands, e.g.:
(gdb) break main
Breakpoint 1 at ...
(gdb) catch exec
Catchpoint 2 (exec)
(gdb) add-inferior
[New inferior 2]
Added inferior 2
The tests gdb.base/checkpoint.exp and gdb.base/kill-during-detach.exp
have been updated to accept either form of output from the
"checkpoint" command. The test gdb.multi/checkpoint-multi.exp had
already been written with this change in mind.
---
gdb/linux-fork.c | 2 +-
gdb/testsuite/gdb.base/checkpoint.exp | 2 +-
gdb/testsuite/gdb.base/kill-during-detach.exp | 2 +-
3 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/gdb/linux-fork.c b/gdb/linux-fork.c
index 987c925bdbc..e686fdcb0f4 100644
--- a/gdb/linux-fork.c
+++ b/gdb/linux-fork.c
@@ -919,7 +919,7 @@ checkpoint_command (const char *args, int from_tty)
{
int parent_pid;
- gdb_printf (_("checkpoint "));
+ gdb_printf (_("Checkpoint "));
if (forks_exist_in_multiple_inferiors_p ())
gdb_printf ("%d.%d: ", inf->num, fp->num);
else
diff --git a/gdb/testsuite/gdb.base/checkpoint.exp b/gdb/testsuite/gdb.base/checkpoint.exp
index b9a1e3161db..9fcb91139de 100644
--- a/gdb/testsuite/gdb.base/checkpoint.exp
+++ b/gdb/testsuite/gdb.base/checkpoint.exp
@@ -345,7 +345,7 @@ with_test_prefix "delete checkpoint 0" {
clean_restart $binfile
runto_main
- gdb_test "checkpoint" "checkpoint 1: fork returned pid $decimal\\."
+ gdb_test "checkpoint" "\[Cc\]heckpoint 1: fork returned pid $decimal\\."
gdb_test "restart 1" "Switching to .*"
gdb_test "delete checkpoint 0" "Killed process $decimal"
gdb_test "info checkpoints" [string_to_regexp "No checkpoints."]
diff --git a/gdb/testsuite/gdb.base/kill-during-detach.exp b/gdb/testsuite/gdb.base/kill-during-detach.exp
index 68292cc3c51..d61ddedceaf 100644
--- a/gdb/testsuite/gdb.base/kill-during-detach.exp
+++ b/gdb/testsuite/gdb.base/kill-during-detach.exp
@@ -93,7 +93,7 @@ proc run_test { exit_p checkpoint_p } {
# Set the checkpoint.
gdb_test "checkpoint" \
- "checkpoint 1: fork returned pid $::decimal\\."
+ "\[Cc\]heckpoint 1: fork returned pid $::decimal\\."
}
# Must get the PID before we resume the inferior.
--
2.44.0
^ permalink raw reply [flat|nested] 8+ messages in thread
* [PATCH v3 3/3] Make thread_db_target::pid_to_str checkpoint-aware
2024-04-14 19:44 [PATCH v3 0/3] Make linux checkpoints work with multiple inferiors Kevin Buettner
2024-04-14 19:44 ` [PATCH v3 1/3] " Kevin Buettner
2024-04-14 19:44 ` [PATCH v3 2/3] Capitalize output of successful checkpoint command Kevin Buettner
@ 2024-04-14 19:44 ` Kevin Buettner
2024-05-03 15:55 ` Pedro Alves
2024-05-02 1:49 ` [PATCH v3 0/3] Make linux checkpoints work with multiple inferiors Kevin Buettner
3 siblings, 1 reply; 8+ messages in thread
From: Kevin Buettner @ 2024-04-14 19:44 UTC (permalink / raw)
To: gdb-patches; +Cc: Pedro Alves, Kevin Buettner
This commit prevents thread_db_target::pid_to_str from considering
a checkpoint as a thread. The reason for doing this is that pids
associated with checkpoints can never be a thread due to the fact
that checkpoints (which are implemented by forking a process) can
only work with single-threaded processes.
Without this commit, many of the "info checkpoints" commands
in gdb.multi/checkpoint-multi.exp will incorrectly show some
of the checkpoints as threads. E.g...
* 1.0 A Thread 0x7ffff7cd3740 (LWP 128534) at 0x401199, file hello.c, line 51
1.2 process 128546 at 0x401199, file hello.c, line 51
1.3 process 128547 at 0x401199, file hello.c, line 51
2.1 process 128538 at 0x401258, file goodbye.c, line 62
2.2 A Thread 0x7ffff7cd3740 (LWP 128542) at 0x401258, file goodbye.c, line 62
3.0 A Thread 0x7ffff7cd3740 (LWP 128543) at 0x40115c, file hangout.c, line 31
3.2 process 128545 at 0x40115c, file hangout.c, line 31
With this commit in place, the output looks like this instead:
* 1.0 A process 129961 at 0x401199, file hello.c, line 51
1.2 process 129974 at 0x401199, file hello.c, line 51
1.3 process 129975 at 0x401199, file hello.c, line 51
2.1 process 129965 at 0x401258, file goodbye.c, line 62
2.2 A process 129969 at 0x401258, file goodbye.c, line 62
3.0 A process 129970 at 0x40115c, file hangout.c, line 31
3.2 process 129972 at 0x40115c, file hangout.c, line 31
(For brevity, I've removed the directory elements in each of the paths
above.)
The testcase, gdb.multi/checkpoint-multi.exp, has been updated to
reflect the fact that only "process" should now appear in output
from "info checkpoints".
---
gdb/linux-thread-db.c | 4 +++-
gdb/testsuite/gdb.multi/checkpoint-multi.exp | 2 +-
2 files changed, 4 insertions(+), 2 deletions(-)
diff --git a/gdb/linux-thread-db.c b/gdb/linux-thread-db.c
index 65bf4a79fdf..488a7086acf 100644
--- a/gdb/linux-thread-db.c
+++ b/gdb/linux-thread-db.c
@@ -48,6 +48,7 @@
#include "gdbsupport/pathstuff.h"
#include "valprint.h"
#include "cli/cli-style.h"
+#include "linux-fork.h"
/* GNU/Linux libthread_db support.
@@ -1657,7 +1658,8 @@ thread_db_target::pid_to_str (ptid_t ptid)
{
thread_info *thread_info = current_inferior ()->find_thread (ptid);
- if (thread_info != NULL && thread_info->priv != NULL)
+ if (thread_info != NULL && thread_info->priv != NULL
+ && !forks_exist_p (current_inferior ()))
{
thread_db_thread_info *priv = get_thread_db_thread_info (thread_info);
diff --git a/gdb/testsuite/gdb.multi/checkpoint-multi.exp b/gdb/testsuite/gdb.multi/checkpoint-multi.exp
index 929b09b6876..491d070d902 100644
--- a/gdb/testsuite/gdb.multi/checkpoint-multi.exp
+++ b/gdb/testsuite/gdb.multi/checkpoint-multi.exp
@@ -170,7 +170,7 @@ with_test_prefix "check continue to exit on non-checkpointed inferior" {
gdb_test "continue" "Inferior 1.*? exited normally.*"
}
-set proc_re "(?:process $::decimal|Thread $::hex \\(LWP $::decimal\\))"
+set proc_re "(?:process $::decimal)"
set main_proc "\\(main process\\)"
set hello_c "hello\\.c"
set goodbye_c "goodbye\\.c"
--
2.44.0
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [PATCH v3 0/3] Make linux checkpoints work with multiple inferiors
2024-04-14 19:44 [PATCH v3 0/3] Make linux checkpoints work with multiple inferiors Kevin Buettner
` (2 preceding siblings ...)
2024-04-14 19:44 ` [PATCH v3 3/3] Make thread_db_target::pid_to_str checkpoint-aware Kevin Buettner
@ 2024-05-02 1:49 ` Kevin Buettner
3 siblings, 0 replies; 8+ messages in thread
From: Kevin Buettner @ 2024-05-02 1:49 UTC (permalink / raw)
To: gdb-patches; +Cc: Pedro Alves
Ping.
On Sun, 14 Apr 2024 12:44:41 -0700
Kevin Buettner <kevinb@redhat.com> wrote:
> This series fixes some problems with the current checkpoint code. The
> first patch makes the checkpoint code inferior aware, fixing a number
> of bugs. The second and third patches are largely cosmetic - they
> make changes to checkpoint related output.
>
> The v2 series incorporates Pedro's suggestions regarding the
> numbering of checkpoint ids. See the first patch for details. The
> tests have been revised to account for these changes and new tests
> have been added as well.
>
> This v3 series splits out a cosmetic change from the first patch.
> It capitalizes the output of a successful checkpoint command.
> This was prompted by the Linaro regression tester, which, due to the
> capitalization change, found two regressions in
> gdb.base/kill-during-detach.exp. (I thought I was running this test
> during my own testing, but found that it was not being run due to a
> typo in the TESTS= portion of the "make check" line.)
>
> Kevin Buettner (3):
> Make linux checkpoints work with multiple inferiors
> Capitalize output of successful checkpoint command
> Make thread_db_target::pid_to_str checkpoint-aware
>
> gdb/linux-fork.c | 519 +++++++++----
> gdb/linux-fork.h | 15 +-
> gdb/linux-nat.c | 20 +-
> gdb/linux-thread-db.c | 4 +-
> gdb/testsuite/gdb.base/checkpoint.exp | 26 +-
> gdb/testsuite/gdb.base/kill-during-detach.exp | 2 +-
> gdb/testsuite/gdb.multi/checkpoint-multi.exp | 687 ++++++++++++++++++
> 7 files changed, 1084 insertions(+), 189 deletions(-)
> create mode 100644 gdb/testsuite/gdb.multi/checkpoint-multi.exp
>
> --
> 2.44.0
>
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [PATCH v3 1/3] Make linux checkpoints work with multiple inferiors
2024-04-14 19:44 ` [PATCH v3 1/3] " Kevin Buettner
@ 2024-05-03 15:54 ` Pedro Alves
0 siblings, 0 replies; 8+ messages in thread
From: Pedro Alves @ 2024-05-03 15:54 UTC (permalink / raw)
To: Kevin Buettner, gdb-patches
Hi Kevin,
Thanks a lot for this. I think it looks so much better.
We should document the interface change in the manual. And probably multi-inferior
checkpoints in NEWS.
Comments below.
On 2024-04-14 20:44, Kevin Buettner wrote:
> The current linux checkpoint code, most of which may be found in
> linux-fork.c, is quite broken when attempting to use more than
> one inferior. Running GDB will show internal errors when starting
> two inferiors, placing a checkpoint in one, then switching to
> the other and doing one of the following commands, "restart",
> "detach", "kill", or continue (to program exit). Test cases
> for two of those scenarios may be found in this bug:
>
> https://sourceware.org/bugzilla/show_bug.cgi?id=31065
>
> I've tested for each of the scenarios and many more in the new
> test case, gdb.multi/checkpoint-multi.exp.
>
> I started off with the goal of fixing just those problems, and was
> mostly successful with a much smaller patch, but doing "info
> checkpoints" with more than one inferior didn't work correctly due to
> some of the inferiors being in the wrong program space. That led me
> to making the linux-fork code fully inferior-aware.
>
> Prior to this commit, the list of forks was being maintained in a
> global named named 'fork_list'. I turned this into a per-inferior
> data structure. There was also global named 'highest_fork_num' which
> is also now part of the per-inferior struct. A registry key named
> 'checkpoint_inferior_data_key' along with function
> 'get_checkpoint_inferior_data' is used to access the per-inferior
> data. This new function, get_checkpoint_inferior_data, is only
> called by the new functions 'fork_list' and 'highest_fork_num', each
> of which is passed a pointer to the inferior. Most occurrences
> referring to the (previously) global 'fork_list' have been replaced
> by 'fork_list (inf)'. In some functions, where the 'fork_list' is
> referenced multiple times, a local named 'fork_list' is declared and
> initialized instead, like this:
>
> auto &fork_list = ::fork_list (inf);
>
> The constructor for 'struct fork_info' has gained an additional
> parameter. In addition to passing the pid of the new fork, we now
> also pass the fork identifier, fork_num, to the constructor. This
> integer is shown to the user in the "info checkpoints" command and
> is provided by the user, perhaps in conjunction with the inferior
> number, in commands which manipulate checkpoints, e.g. 'restart' and
> 'delete checkpoint'.
>
> When checkpoints are used in only one inferior, this commit will
> present information to the user and will accept checkpoint identifiers
> to commands in much the same way as the code did before this commit.
> Per Pedro Alves's recommendations, the "info checkpoints" command
> has been changed somewhat. "info checkpoints" used to display
> "(main process)" for the first process in the checkpoint list.
> This is no longer done because it does not provide useful information.
> It also used to display "<running>", presumably when in non-stop
> mode. I've changed this to print a state flag 'R'. Additionally,
> a state flag 'A' is used to indicate the active checkpoint for
> each inferior. For the active inferior a '*' is also printed
> preceding the checkpoint identifier. Here's what things look(ed)
> like before and after for just one inferior:
>
> Before:
>
> (gdb) info checkpoints
> * 0 Thread 0x7ffff7cd3740 (LWP 84201) (main process) at 0x40114a, file hello.c, line 28
> 1 process 84205 at 0x401199, file hello.c, line 51
> 2 process 84206 at 0x4011a3, file hello.c, line 53
>
> After:
>
> (gdb) info checkpoints
> * 0 A process 84159 at 0x40114a, file hello.c, line 28
> 1 process 84162 at 0x401199, file hello.c, line 51
> 2 process 84163 at 0x4011a3, file hello.c, line 53
>
> (The Thread versus process distinction is handled by another
> patch - the "After" example assumes that patch is applied too.)
>
> When there are multiple inferiors, the "info checkpoints" output looks
> like this:
>
> (gdb) info checkpoints
> 1.0 process 84159 at 0x40114a, file hello.c, line 28
> * 1.1 A process 84162 at 0x401199, file hello.c, line 51
> 1.2 process 84163 at 0x4011a3, file hello.c, line 53
> 2.0 A process 84223 at 0x40116f, file hangout.c, line 35
> 2.1 process 84224 at 0x40115c, file hangout.c, line 31
> 2.2 process 84225 at 0x401166, file hangout.c, line 33
> 3.0 A process 84226 at 0x401218, file goodbye.c, line 46
> 3.1 process 84229 at 0x40124e, file goodbye.c, line 61
>
> A new function named 'parse_checkpoint_id' has been added. As its
> name suggests, it's responsible for parsing a string representing a
> checkpoint identifier. These identifiers may be either a decimal
> number representing the checkpoint number in the current inferior or
> two decimal numbers separated by '.', in which case the first is the
> inferior number and the second is the checkpoint number in that
> inferior. It is called by delete_checkpoint_command,
> detach_checkpoint_command, info_checkpoints_command, and
> restart_command. Calls to 'parse_checkpoint_id' replace calls to
> 'parse_and_eval_long', plus error checking and error reporting code
> near the calls to 'parse_and_eval_long'. As such, error checking and
> reporting has been consolidated into a single function and the
> messages output are more uniform, though this has necessitated changes
> to the existing test case gdb.base/checkpoint.exp.
>
> The functions 'find_fork_ptid' and 'find_fork_pid' used to return a
> pointer to a fork_info struct. They now return a pair consisting of
> the pointer to a fork_info struct in addition to a pointer to the
> inferior containing that checkpoint.
>
> 'find_fork_id' returns a a pointer to a fork_info struct just as
> it did before, but it's now gained a new parameter, 'inf', which
> is the inferior in which to look.
>
> info_checkpoints_command used to simply iterate over the list of
> forks (checkpoints), printing each one out. It now needs to iterate
> over all inferiors and, for those which have checkpoints, it needs
> to iterate over the list of checkpoints in that inferior. As noted
> earlier, the format of the output has been changed so that checkpoint
> identifers incorporating an inferior number may be printed.
>
> linux_fork_context, called by restart_command, now contains code to
> switch inferiors when the fork being restarted is in a different
> inferior from the current one. The scoped_switch_fork_info class now
> also contains code for switching inferiors in both the constructor and
> destructor.
>
> gdb/linux-nat.c has a few changes. All but one of them are related
> to passing the inferior to one of the linux-fork functions. But
> one of the tests in linux_nat_target::detach has also changed in
> a non-obvious way. In attempting to determine whether to call
> linux_fork_detach(), that code used to do:
>
> if (pid == inferior_ptid.pid () && forks_exist_p ())
>
> It's been simplified to:
>
> if (forks_exist_p (inf))
>
> I had added the 'pid == inferior_ptid.pid ()' condition in late 2023
> while working on a detach bug. It was kind of a hack to prevent
> calling linux_fork_detach() when in a different inferior. That's no
> longer needed since the call to forks_exist_p does this directly -
> i.e. it is now inferior-aware.
>
> Finally, the header file 'linux-fork.h' has been updated to reflect
> the fact that add_fork, linux_fork_killall, linux_fork_mourn_inferior,
> linux_fork_detach, and forks_exist_p all now require that a pointer to
> an inferior be passed to these functions. Additionally (as mentioned
> earlier), find_fork_pid now returns std::pair<fork_info *, inferior *>
> instead 'of fork_info *'.
>
> Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=31065
> ---
> gdb/linux-fork.c | 519 +++++++++-----
> gdb/linux-fork.h | 15 +-
> gdb/linux-nat.c | 20 +-
> gdb/testsuite/gdb.base/checkpoint.exp | 24 +-
> gdb/testsuite/gdb.multi/checkpoint-multi.exp | 687 +++++++++++++++++++
> 5 files changed, 1079 insertions(+), 186 deletions(-)
> create mode 100644 gdb/testsuite/gdb.multi/checkpoint-multi.exp
>
> diff --git a/gdb/linux-fork.c b/gdb/linux-fork.c
> index bb8119a36d0..987c925bdbc 100644
> --- a/gdb/linux-fork.c
> +++ b/gdb/linux-fork.c
> @@ -28,6 +28,7 @@
> #include "linux-nat.h"
> #include "gdbthread.h"
> #include "source.h"
> +#include "progspace-and-thread.h"
>
> #include "nat/gdb_ptrace.h"
> #include "gdbsupport/gdb_wait.h"
> @@ -38,10 +39,11 @@
> #include <list>
>
> /* Fork list data structure: */
> +
> struct fork_info
> {
> - explicit fork_info (pid_t pid)
> - : ptid (pid, pid)
> + explicit fork_info (pid_t pid, int fork_num)
> + : ptid (pid, pid), num (fork_num)
> {
> }
>
> @@ -69,7 +71,7 @@ struct fork_info
> ptid_t parent_ptid = null_ptid;
>
> /* Convenient handle (GDB fork id). */
> - int num = 0;
> + int num;
>
> /* Convenient for info fork, saves having to actually switch
> contexts. */
> @@ -83,115 +85,253 @@ struct fork_info
> int maxfd = 0;
> };
>
> -static std::list<fork_info> fork_list;
> -static int highest_fork_num;
> +/* Per-inferior checkpoint data. */
> +
> +struct checkpoint_inferior_data
> +{
> + /* List of forks (checkpoints) in particular inferior. Once a
> + checkpoint has been created, fork_list will contain at least two
> + items, the first in the list will be the original (or, if not
> + original, then the oldest) fork. */
> + std::list<fork_info> fork_list;
> +
> + /* Most recently assigned fork number; when 0, no checkpoints have
> + been created yet. */
> + int highest_fork_num = 0;
> +};
> +
> +/* Per-inferior data key. */
> +
> +static const registry<inferior>::key<checkpoint_inferior_data>
> + checkpoint_inferior_data_key;
> +
> +/* Fetch per-inferior checkpoint data. It always returns a valid pointer
> + to a checkpoint_inferior_info struct. */
> +
> +static struct checkpoint_inferior_data *
> +get_checkpoint_inferior_data (struct inferior *inf)
> +{
> + struct checkpoint_inferior_data *data;
> +
> + data = checkpoint_inferior_data_key.get (inf);
> + if (data == nullptr)
> + data = checkpoint_inferior_data_key.emplace (inf);
> +
> + return data;
> +}
> +
> +/* Return a reference to the per-inferior fork list. */
> +
> +static std::list<fork_info> &
> +fork_list (inferior *inf)
> +{
> + return get_checkpoint_inferior_data (inf)->fork_list;
> +}
> +
> +/* Return a reference to 'highest_fork_num' for the given inferior INF. */
> +
> +static int &
> +highest_fork_num (inferior *inf)
> +{
> + return get_checkpoint_inferior_data (inf)->highest_fork_num;
> +}
We've moved away from the style of code that returns an lval
into separate getter and setter elsewhere (mainly the symbols code).
I think the only write we do is to reset the counter, so we could have:
static int
get_highest_fork_num (inferior *inf)
{
return get_checkpoint_inferior_data (inf)->highest_fork_num;
}
static void
reset_highest_fork_num (inferior *inf)
{
get_checkpoint_inferior_data (inf)->highest_fork_num = 0;
}
IMO, that'd be better.
> +
> +/* Predicate which returns true when more than one inferior contains
> + a checkpoint, false otherwise. */
> +
> +static bool
> +forks_exist_in_multiple_inferiors_p ()
> +{
> + int count = 0;
> + for (inferior *inf : all_inferiors ())
> + {
This iterates over all inferiors, even ones debugging with targets
other than linux-nat, but it's harmless if forks_exist_p considers it.
See below.
> + if (forks_exist_p (inf))
> + count++;
> + if (count > 1)
> + return true;
> + }
> + return false;
> +}
>
> /* Fork list methods: */
>
> -int
> -forks_exist_p (void)
> +/* Predicate which returns true if checkpoint(s) exist in the inferior
> + INF, false otherwise. */
> +
> +bool
> +forks_exist_p (inferior *inf)
> {
> - return !fork_list.empty ();
> + return !fork_list (inf).empty ();
The call to fork_list always creates the checkpoint_inferior_data
if it wasn't created already, even for inferiors that aren't
debugging with linux-nat. You could add a check for
checkpoint_inferior_data_key.get (inf) != nullptr
to avoid this.
> }
>
> -/* Return the last fork in the list. */
> +/* Return the last fork in the list for inferior INF. */
>
> static struct fork_info *
> -find_last_fork (void)
> +find_last_fork (inferior *inf)
> {
> + auto &fork_list = ::fork_list (inf);
> +
> if (fork_list.empty ())
> return NULL;
>
> return &fork_list.back ();
> }
>
> -/* Return true iff there's one fork in the list. */
> +/* Return true iff there's one fork in the list for inferior INF. */
>
> static bool
> -one_fork_p ()
> +one_fork_p (inferior *inf)
> {
> - return fork_list.size () == 1;
> + return fork_list (inf).size () == 1;
> }
>
> /* Add a new fork to the internal fork list. */
>
> void
> -add_fork (pid_t pid)
> +add_fork (pid_t pid, inferior *inf)
> {
> - fork_list.emplace_back (pid);
> -
> - if (one_fork_p ())
> - highest_fork_num = 0;
> -
> - fork_info *fp = &fork_list.back ();
> - fp->num = ++highest_fork_num;
> + fork_list (inf).emplace_back (pid, ++highest_fork_num (inf));
> }
>
> +/* Delete a fork for PTID in inferior INF. When the last fork is
> + deleted, HIGHEST_FORK_NUM for the given inferior is reset to 0.
> + The fork list may also be made to be empty when only one fork
> + remains. */
> +
> static void
> -delete_fork (ptid_t ptid)
> +delete_fork (ptid_t ptid, inferior *inf)
> {
> linux_target->low_forget_process (ptid.pid ());
>
> + auto &fork_list = ::fork_list (inf);
> for (auto it = fork_list.begin (); it != fork_list.end (); ++it)
> if (it->ptid == ptid)
> {
> fork_list.erase (it);
>
> + if (fork_list.empty ())
> + {
> + /* Reset highest_fork_num. */
> + highest_fork_num (inf) = 0;
If we do instead:
reset_highest_fork_num (inf);
then we don't even need the comment. :-)
> + }
> +
> /* Special case: if there is now only one process in the list,
> and if it is (hopefully!) the current inferior_ptid, then
> remove it, leaving the list empty -- we're now down to the
> default case of debugging a single process. */
> - if (one_fork_p () && fork_list.front ().ptid == inferior_ptid)
> + if (one_fork_p (inf) && fork_list.front ().ptid == inferior_ptid)
> {
> /* Last fork -- delete from list and handle as solo
> process (should be a safe recursion). */
> - delete_fork (inferior_ptid);
> + delete_fork (inferior_ptid, inf);
> }
> return;
> }
> }
>
> -/* Find a fork_info by matching PTID. */
> -static struct fork_info *
> +/* Find a fork_info and inferior by matching PTID. */
> +
> +static std::pair<fork_info *, inferior *>
> find_fork_ptid (ptid_t ptid)
> {
> - for (fork_info &fi : fork_list)
> - if (fi.ptid == ptid)
> - return &fi;
> + for (inferior *inf : all_inferiors ())
> + {
> + for (fork_info &fi : fork_list (inf))
Similar remark about the fork_list call creating the per-inferior data
for inferiors that cannot have anything to do with checkpoints.
The minimal fix would be to pass "linux_target" as argument to
all_inferiors, so only inferiors of linux-target are iterated over.
Comments apply to all loops over inferiors, so I'll stop pointing at it.
> + if (fi.ptid == ptid)
> + return { &fi, inf };
> + }
>
> - return NULL;
> + return { nullptr, nullptr };
> }
>
> -/* Find a fork_info by matching ID. */
> -static struct fork_info *
> -find_fork_id (int num)
> +/* Find a fork_info by matching NUM in inferior INF. */
> +
> +static fork_info *
> +find_fork_id (inferior *inf, int num)
> {
> - for (fork_info &fi : fork_list)
> + for (fork_info &fi : fork_list (inf))
> if (fi.num == num)
> return &fi;
>
> - return NULL;
> + return nullptr;
> }
>
> -/* Find a fork_info by matching pid. */
> -extern struct fork_info *
> +/* Find a fork_info and inferior by matching pid. */
> +
> +extern std::pair<fork_info *, inferior *>
> find_fork_pid (pid_t pid)
> {
> - for (fork_info &fi : fork_list)
> - if (pid == fi.ptid.pid ())
> - return &fi;
> + for (inferior *inf : all_inferiors ())
> + {
> + for (fork_info &fi : fork_list (inf))
> + if (pid == fi.ptid.pid ())
> + return { &fi, inf };
> + }
>
> - return NULL;
> + return { nullptr, nullptr };
> }
>
> -static ptid_t
> -fork_id_to_ptid (int num)
> +/* Parse a command argument representing a checkpoint id. This
> + can take one of two forms:
> +
> + Num
> +
> + -or-
> +
> + Inf.Num
> +
> + where Num is a non-negative decimal integer and Inf, if present, is
> + a positive decimal integer.
> +
> + Return a pair with a pointer to the fork_info struct and pointer
> + to the inferior. This function will throw an error if there's
> + a problem with the parsing or if either the inferior or checkpoint
> + id does not exist. */
> +
> +static std::pair<fork_info *, inferior *>
> +parse_checkpoint_id (const char *ckptstr)
> {
> - struct fork_info *fork = find_fork_id (num);
> - if (fork)
> - return fork->ptid;
> + const char *number = ckptstr;
> + const char *p1;
> + struct inferior *inf;
> +
> + const char *dot = strchr (number, '.');
> +
> + if (dot != nullptr)
> + {
> + /* Parse number to the left of the dot. */
> + int inf_num;
> +
> + p1 = number;
> + inf_num = get_number_trailer (&p1, '.');
> + if (inf_num <= 0)
> + error (_("Inferior number must be a positive integer"));
> +
> + inf = find_inferior_id (inf_num);
> + if (inf == NULL)
> + error (_("No inferior number '%d'"), inf_num);
> +
> + p1 = dot + 1;
> + }
> else
> - return ptid_t (-1);
> + {
> + inf = current_inferior ();
> + p1 = number;
> + }
> +
> + int fork_num = get_number_trailer (&p1, 0);
> + if (fork_num < 0)
> + error (_("Checkpoint number must be a non-negative integer"));
> +
> + if (!forks_exist_p (inf))
> + error (_("Inferior %d has no checkpoints"), inf->num);
> +
> + auto fork_ptr = find_fork_id (inf, fork_num);
We tend to avoid auto (outside generic code) unless the expression already
uses the type on the rhs. But if you're going to use it, I'd suggest
at least writing:
auto *fork_ptr
so that it's less surpring to see the variable treated as pointer.
But really, writing:
fork_info *fork_ptr
isn't that much more.
> + if (fork_ptr == nullptr)
> + error (_("Invalid checkpoint number %d for inferior %d"),
> + fork_num, inf->num);
> +
> + return { fork_ptr, inf };
> }
>
> /* Fork list <-> gdb interface. */
> @@ -296,7 +436,7 @@ fork_save_infrun_state (struct fork_info *fp)
> /* Kill 'em all, let God sort 'em out... */
>
> void
> -linux_fork_killall (void)
> +linux_fork_killall (inferior *inf)
> {
> /* Walk list and kill every pid. No need to treat the
> current inferior_ptid as special (we do not return a
> @@ -304,6 +444,7 @@ linux_fork_killall (void)
> or a parent, so may get a SIGCHLD from a previously
> killed child. Wait them all out. */
>
> + auto &fork_list = ::fork_list (inf);
> for (fork_info &fi : fork_list)
> {
> pid_t pid = fi.ptid.pid ();
> @@ -322,6 +463,7 @@ linux_fork_killall (void)
>
> /* Clear list, prepare to start fresh. */
> fork_list.clear ();
> + highest_fork_num (inf) = 0;
> }
>
> /* The current inferior_ptid has exited, but there are other viable
> @@ -329,7 +471,7 @@ linux_fork_killall (void)
> first available. */
>
> void
> -linux_fork_mourn_inferior (void)
> +linux_fork_mourn_inferior (inferior *inf)
This is passing down an explicit inferior, but it assumes that that
inferior is the current, since it uses inferior_ptid anyhow.
I was going to point at the fact that the function's comment still
talks about "current inferior", and then I noticed that.
> {
> struct fork_info *last;
> int status;
> @@ -341,23 +483,23 @@ linux_fork_mourn_inferior (void)
> waitpid (inferior_ptid.pid (), &status, 0);
>
> /* OK, presumably inferior_ptid is the one who has exited.
> - We need to delete that one from the fork_list, and switch
> + We need to delete that one from the fork list, and switch
> to the next available fork. */
> - delete_fork (inferior_ptid);
> + delete_fork (inferior_ptid, inf);
>
> /* There should still be a fork - if there's only one left,
> delete_fork won't remove it, because we haven't updated
> inferior_ptid yet. */
> - gdb_assert (!fork_list.empty ());
> + gdb_assert (!fork_list (inf).empty ());
>
> - last = find_last_fork ();
> + last = find_last_fork (inf);
> fork_load_infrun_state (last);
> gdb_printf (_("[Switching to %s]\n"),
> target_pid_to_str (inferior_ptid).c_str ());
>
> /* If there's only one fork, switch back to non-fork mode. */
> - if (one_fork_p ())
> - delete_fork (inferior_ptid);
> + if (one_fork_p (inf))
> + delete_fork (inferior_ptid, inf);
> }
>
> /* The current inferior_ptid is being detached, but there are other
> @@ -365,13 +507,13 @@ linux_fork_mourn_inferior (void)
> the first available. */
>
> void
> -linux_fork_detach (int from_tty, lwp_info *lp)
> +linux_fork_detach (int from_tty, lwp_info *lp, inferior *inf)
> {
> gdb_assert (lp != nullptr);
> gdb_assert (lp->ptid == inferior_ptid);
>
> /* OK, inferior_ptid is the one we are detaching from. We need to
> - delete it from the fork_list, and switch to the next available
> + delete it from the fork list, and switch to the next available
> fork. But before doing the detach, do make sure that the lwp
> hasn't exited or been terminated first. */
>
> @@ -384,11 +526,12 @@ linux_fork_detach (int from_tty, lwp_info *lp)
> target_pid_to_str (inferior_ptid).c_str ());
> }
>
> - delete_fork (inferior_ptid);
> + delete_fork (inferior_ptid, inf);
>
> /* There should still be a fork - if there's only one left,
> delete_fork won't remove it, because we haven't updated
> inferior_ptid yet. */
> + auto &fork_list = ::fork_list (inf);
> gdb_assert (!fork_list.empty ());
>
> fork_load_infrun_state (&fork_list.front ());
> @@ -398,8 +541,8 @@ linux_fork_detach (int from_tty, lwp_info *lp)
> target_pid_to_str (inferior_ptid).c_str ());
>
> /* If there's only one fork, switch back to non-fork mode. */
> - if (one_fork_p ())
> - delete_fork (inferior_ptid);
> + if (one_fork_p (inf))
> + delete_fork (inferior_ptid, inf);
> }
>
> /* Temporarily switch to the infrun state stored on the fork_info
> @@ -412,19 +555,26 @@ class scoped_switch_fork_info
> /* Switch to the infrun state held on the fork_info identified by
> PPTID. If PPTID is the current inferior then no switch is done. */
> explicit scoped_switch_fork_info (ptid_t pptid)
> - : m_oldfp (nullptr)
> + : m_oldfp (nullptr), m_oldinf (nullptr)
> {
> if (pptid != inferior_ptid)
> {
> - struct fork_info *newfp = nullptr;
> -
> /* Switch to pptid. */
> - m_oldfp = find_fork_ptid (inferior_ptid);
> + auto [oldfp, oldinf] = find_fork_ptid (inferior_ptid);
> + m_oldfp = oldfp;
> gdb_assert (m_oldfp != nullptr);
> - newfp = find_fork_ptid (pptid);
> + auto [newfp, newinf] = find_fork_ptid (pptid);
> gdb_assert (newfp != nullptr);
> fork_save_infrun_state (m_oldfp);
> remove_breakpoints ();
> +
> + if (oldinf != newinf)
> + {
> + thread_info *tp = any_thread_of_inferior (newinf);
> + switch_to_thread (tp);
> + m_oldinf = oldinf;
> + }
> +
> fork_load_infrun_state (newfp);
> insert_breakpoints ();
> }
> @@ -434,12 +584,17 @@ class scoped_switch_fork_info
> didn't need to switch states, then nothing is done here either. */
> ~scoped_switch_fork_info ()
> {
> - if (m_oldfp != nullptr)
> + if (m_oldinf != nullptr || m_oldfp != nullptr)
> {
> /* Switch back to inferior_ptid. */
> try
> {
> remove_breakpoints ();
> + if (m_oldinf != nullptr)
> + {
> + thread_info *tp = any_thread_of_inferior (m_oldinf);
> + switch_to_thread (tp);
> + }
> fork_load_infrun_state (m_oldfp);
> insert_breakpoints ();
> }
> @@ -471,6 +626,7 @@ class scoped_switch_fork_info
> we were already in the desired state, and nothing needs to be
> restored. */
> struct fork_info *m_oldfp;
> + inferior *m_oldinf;
Comment?
> };
>
> static int
> @@ -513,30 +669,35 @@ static void
> delete_checkpoint_command (const char *args, int from_tty)
> {
> ptid_t ptid, pptid;
> - struct fork_info *fi;
>
> if (!args || !*args)
> error (_("Requires argument (checkpoint id to delete)"));
>
> - ptid = fork_id_to_ptid (parse_and_eval_long (args));
> - if (ptid == minus_one_ptid)
> - error (_("No such checkpoint id, %s"), args);
> + auto [fi, inf] = parse_checkpoint_id (args);
> + ptid = fi->ptid;
> + gdb_assert (fi != nullptr);
> + pptid = fi->parent_ptid;
>
> - if (ptid == inferior_ptid)
> - error (_("\
> + thread_info *inf_thr = any_thread_of_inferior (inf);
> + gdb_assert (inf_thr != nullptr);
> + if (ptid == inf_thr->ptid)
Feels a bit dangerous to compare the full ptid here, since
any_thread_of_inferior could return a different thread.
I know that since the checkpoints relies on fork, it
only works with single threaded inferiors, but in principle
the child could still spawn other threads after the checkpoint.
We probably mishandle that, but, yeah, it just feels like a breakage
waiting to happen here. Why not simply compare pids instead:
if (ptid.pid () == inf->pid)
then you don't even need to search for a thread at all.
> + {
> + if (inf == current_inferior ())
> + error (_("\
> Please switch to another checkpoint before deleting the current one"));
> + else
> + error (_("\
> +Please switch to another checkpoint before deleting the current one\n\
> +in inferior %d"), inf->num);
The error case in the else branch gave me pause. I think instead of "current"
it may help to say "active" instead. "current" seems confusing, as I'd think
it is referring to the current checkpoint, in the current inferior.
I mean:
Please switch to another checkpoint before deleting the active one\n\
in inferior %d"), inf->num);
But I wonder if we could simplify all this to just say, in either then/else case:
error (_("Cannot delete active checkpoint."));
> + }
>
> if (ptrace (PTRACE_KILL, ptid.pid (), 0, 0))
> error (_("Unable to kill pid %s"), target_pid_to_str (ptid).c_str ());
>
> - fi = find_fork_ptid (ptid);
> - gdb_assert (fi);
> - pptid = fi->parent_ptid;
> -
> if (from_tty)
> gdb_printf (_("Killed %s\n"), target_pid_to_str (ptid).c_str ());
>
> - delete_fork (ptid);
> + delete_fork (ptid, inf);
>
> if (pptid == null_ptid)
> {
> @@ -554,7 +715,7 @@ Please switch to another checkpoint before deleting the current one"));
> If fi->parent_ptid is a part of lwp and it is stopped, waitpid the
> ptid. */
> thread_info *parent = linux_target->find_thread (pptid);
> - if ((parent == NULL && find_fork_ptid (pptid))
> + if ((parent == NULL && find_fork_ptid (pptid).first != nullptr)
> || (parent != NULL && parent->state == THREAD_STOPPED))
> {
> if (inferior_call_waitpid (pptid, ptid.pid ()))
> @@ -571,9 +732,8 @@ detach_checkpoint_command (const char *args, int from_tty)
> if (!args || !*args)
> error (_("Requires argument (checkpoint id to detach)"));
>
> - ptid = fork_id_to_ptid (parse_and_eval_long (args));
> - if (ptid == minus_one_ptid)
> - error (_("No such checkpoint id, %s"), args);
> + auto [fi, inf] = parse_checkpoint_id (args);
> + ptid = fi->ptid;
>
> if (ptid == inferior_ptid)
> error (_("\
> @@ -585,7 +745,7 @@ Please switch to another checkpoint before detaching the current one"));
> if (from_tty)
> gdb_printf (_("Detached %s\n"), target_pid_to_str (ptid).c_str ());
>
> - delete_fork (ptid);
> + delete_fork (ptid, current_inferior ());
> }
>
> /* Print information about currently known checkpoints. */
> @@ -594,72 +754,96 @@ static void
> info_checkpoints_command (const char *arg, int from_tty)
> {
> struct gdbarch *gdbarch = get_current_arch ();
> - int requested = -1;
> + struct inferior *cur_inf = current_inferior ();
> + inferior *req_inf = nullptr;
> + fork_info *req_fi = nullptr;
> bool printed = false;
>
> if (arg && *arg)
> - requested = (int) parse_and_eval_long (arg);
> + std::tie (req_fi, req_inf) = parse_checkpoint_id (arg);
>
> - for (const fork_info &fi : fork_list)
> + /* Figure out whether to print the inferior number in the
> + checkpoint list. */
> + bool print_inf = (!forks_exist_p (current_inferior ())
> + || forks_exist_in_multiple_inferiors_p ());
> +
> + for (inferior *inf : all_inferiors ())
> {
> - if (requested > 0 && fi.num != requested)
> + if (req_inf != nullptr && req_inf != inf)
> continue;
> - printed = true;
> -
> - bool is_current = fi.ptid == inferior_ptid;
> - if (is_current)
> - gdb_printf ("* ");
> - else
> - gdb_printf (" ");
>
> - gdb_printf ("%d %s", fi.num, target_pid_to_str (fi.ptid).c_str ());
> - if (fi.num == 0)
> - gdb_printf (_(" (main process)"));
> + scoped_restore_current_pspace_and_thread restore_pspace_thread;
> + switch_to_program_space_and_thread (inf->pspace);
>
> - if (is_current && inferior_thread ()->state == THREAD_RUNNING)
> + for (const fork_info &fi : fork_list (inf))
> {
> - gdb_printf (_(" <running>\n"));
> - continue;
> + if (req_fi != nullptr && req_fi != &fi)
> + continue;
> + printed = true;
> +
> + thread_info *t = any_thread_of_inferior (inf);
> + bool is_current = fi.ptid == t->ptid;
Similar remark about just comparing the pid.
> + if (is_current && cur_inf == inf)
> + gdb_printf ("*");
> + else
> + gdb_printf (" ");
> +
> + if (print_inf)
> + gdb_printf ("%3d.%-2d", inf->num, fi.num);
> + else
> + gdb_printf ("%3d ", fi.num);
> +
> + /* Print out state associated with current checkpoint. */
> + if (is_current)
> + {
> + gdb_printf ("A");
> + if (t->state == THREAD_RUNNING)
> + gdb_printf ("R");
> + else
> + gdb_printf (" ");
> + gdb_printf (" ");
> + }
> + else
> + gdb_printf (" ");
> +
> + gdb_printf ("%s", target_pid_to_str (fi.ptid).c_str ());
> +
> + gdb_printf (_(" at "));
> + ULONGEST pc
> + = (is_current
> + ? regcache_read_pc (get_thread_regcache (t))
> + : fi.pc);
> + gdb_puts (paddress (gdbarch, pc));
> +
> + symtab_and_line sal = find_pc_line (pc, 0);
> + if (sal.symtab)
> + gdb_printf (_(", file %s"),
> + symtab_to_filename_for_display (sal.symtab));
> + if (sal.line)
> + gdb_printf (_(", line %d"), sal.line);
> + if (!sal.symtab && !sal.line)
> + {
> + struct bound_minimal_symbol msym;
> +
> + msym = lookup_minimal_symbol_by_pc (pc);
> + if (msym.minsym)
> + gdb_printf (", <%s>", msym.minsym->linkage_name ());
> + }
> +
> + gdb_putc ('\n');
> }
> -
> - gdb_printf (_(" at "));
> - ULONGEST pc
> - = (is_current
> - ? regcache_read_pc (get_thread_regcache (inferior_thread ()))
> - : fi.pc);
> - gdb_puts (paddress (gdbarch, pc));
> -
> - symtab_and_line sal = find_pc_line (pc, 0);
> - if (sal.symtab)
> - gdb_printf (_(", file %s"),
> - symtab_to_filename_for_display (sal.symtab));
> - if (sal.line)
> - gdb_printf (_(", line %d"), sal.line);
> - if (!sal.symtab && !sal.line)
> - {
> - struct bound_minimal_symbol msym;
> -
> - msym = lookup_minimal_symbol_by_pc (pc);
> - if (msym.minsym)
> - gdb_printf (", <%s>", msym.minsym->linkage_name ());
> - }
> -
> - gdb_putc ('\n');
> }
>
> if (!printed)
> {
> - if (requested > 0)
> - gdb_printf (_("No checkpoint number %d.\n"), requested);
> - else
> - gdb_printf (_("No checkpoints.\n"));
> + gdb_printf (_("No checkpoints.\n"));
> }
> }
>
> /* The PID of the process we're checkpointing. */
> static int checkpointing_pid = 0;
>
> -int
> +bool
> linux_fork_checkpointing_p (int pid)
> {
> return (checkpointing_pid == pid);
> @@ -689,17 +873,16 @@ checkpoint_command (const char *args, int from_tty)
> struct target_waitstatus last_target_waitstatus;
> ptid_t last_target_ptid;
> struct value *fork_fn = NULL, *ret;
> - struct fork_info *fp;
> pid_t retpid;
>
> - if (!target_has_execution ())
> + if (!target_has_execution ())
> error (_("The program is not being run."));
>
> /* Ensure that the inferior is not multithreaded. */
> update_thread_list ();
> if (inf_has_multiple_threads ())
> error (_("checkpoint: can't checkpoint multiple threads."));
> -
> +
> /* Make the inferior fork, record its (and gdb's) state. */
>
> if (lookup_minimal_symbol ("fork", NULL, NULL).minsym != NULL)
> @@ -727,14 +910,22 @@ checkpoint_command (const char *args, int from_tty)
> retpid = value_as_long (ret);
> get_last_target_status (nullptr, &last_target_ptid, &last_target_waitstatus);
>
> - fp = find_fork_pid (retpid);
> + auto [fp, inf] = find_fork_pid (retpid);
> +
> + if (!fp)
> + error (_("Failed to find new fork"));
>
> if (from_tty)
> {
> int parent_pid;
>
> - gdb_printf (_("checkpoint %d: fork returned pid %ld.\n"),
> - fp != NULL ? fp->num : -1, (long) retpid);
> + gdb_printf (_("checkpoint "));
> + if (forks_exist_in_multiple_inferiors_p ())
> + gdb_printf ("%d.%d: ", inf->num, fp->num);
> + else
> + gdb_printf ("%d: ", fp->num);
> + gdb_printf (_("fork returned pid %ld.\n"), (long) retpid);
For i18n, it's better to keep everything in one sentence. Like:
gdb_printf (_("checkpoint %s: fork returned pid %ld.\n"),
(forks_exist_in_multiple_inferiors_p ()
? string_printf ("%d.%d", inf->num, fp->num).c_str ()
: string_printf ("%d", fp->num).c_str ())
(long) retpid);
> +
> if (info_verbose)
> {
> parent_pid = last_target_ptid.lwp ();
> @@ -745,15 +936,12 @@ checkpoint_command (const char *args, int from_tty)
> }
> }
>
> - if (!fp)
> - error (_("Failed to find new fork"));
> -
> - if (one_fork_p ())
> + if (one_fork_p (inf))
> {
> /* Special case -- if this is the first fork in the list (the
> - list was hitherto empty), then add inferior_ptid first, as a
> - special zeroeth fork id. */
> - fork_list.emplace_front (inferior_ptid.pid ());
> + list was hitherto empty), then add inferior_ptid as a special
> + zeroeth fork id. */
> + fork_list (inf).emplace_front (inferior_ptid.pid (), 0);
> }
>
> fork_save_infrun_state (fp);
> @@ -761,40 +949,53 @@ checkpoint_command (const char *args, int from_tty)
> }
>
> static void
> -linux_fork_context (struct fork_info *newfp, int from_tty)
> +linux_fork_context (struct fork_info *newfp, int from_tty, inferior *newinf)
> {
> - /* Now we attempt to switch processes. */
> - struct fork_info *oldfp;
> + bool inferior_changed = false;
>
> + /* Now we attempt to switch processes. */
> gdb_assert (newfp != NULL);
>
> - oldfp = find_fork_ptid (inferior_ptid);
> - gdb_assert (oldfp != NULL);
> + if (newinf != current_inferior ())
> + {
> + thread_info *tp = any_thread_of_inferior (newinf);
> + switch_to_thread (tp);
> + inferior_changed = true;
> + }
>
> - fork_save_infrun_state (oldfp);
> - remove_breakpoints ();
> - fork_load_infrun_state (newfp);
> - insert_breakpoints ();
> + auto [oldfp, oldinf] = find_fork_ptid (inferior_ptid);
> + gdb_assert (oldfp != NULL);
>
> - gdb_printf (_("Switching to %s\n"),
> - target_pid_to_str (inferior_ptid).c_str ());
> + if (oldfp == newfp)
> + {
> + if (!inferior_changed)
> + error (_("Already using checkpoint %d"), oldfp->num);
I don't see why this is an error. It is not an error to
do "thread 1" when thread 1 is already the selected thread, for example.
> + }
> + else
> + {
> + fork_save_infrun_state (oldfp);
> + remove_breakpoints ();
> + fork_load_infrun_state (newfp);
> + insert_breakpoints ();
> + if (!inferior_changed)
> + gdb_printf (_("Switching to %s\n"),
> + target_pid_to_str (inferior_ptid).c_str ());
> + }
>
> - print_stack_frame (get_selected_frame (NULL), 1, SRC_AND_LOC, 1);
> + notify_user_selected_context_changed
> + (inferior_changed ? (USER_SELECTED_INFERIOR | USER_SELECTED_FRAME)
> + : USER_SELECTED_FRAME);
> }
>
> /* Switch inferior process (checkpoint) context, by checkpoint id. */
> static void
> restart_command (const char *args, int from_tty)
> {
> - struct fork_info *fp;
> -
> if (!args || !*args)
> error (_("Requires argument (checkpoint id to restart)"));
>
> - if ((fp = find_fork_id (parse_and_eval_long (args))) == NULL)
> - error (_("Not found: checkpoint id %s"), args);
> -
> - linux_fork_context (fp, from_tty);
> + auto [fp, inf] = parse_checkpoint_id (args);
> + linux_fork_context (fp, from_tty, inf);
> }
>
> void _initialize_linux_fork ();
> diff --git a/gdb/linux-fork.h b/gdb/linux-fork.h
> index c553aaf0740..0233a6c8cfc 100644
> --- a/gdb/linux-fork.h
> +++ b/gdb/linux-fork.h
> @@ -22,12 +22,13 @@
>
> struct fork_info;
> struct lwp_info;
> -extern void add_fork (pid_t);
> -extern struct fork_info *find_fork_pid (pid_t);
> -extern void linux_fork_killall (void);
> -extern void linux_fork_mourn_inferior (void);
> -extern void linux_fork_detach (int, lwp_info *);
> -extern int forks_exist_p (void);
> -extern int linux_fork_checkpointing_p (int);
> +class inferior;
> +extern void add_fork (pid_t, inferior *inf);
> +extern std::pair<fork_info *, inferior *> find_fork_pid (pid_t);
> +extern void linux_fork_killall (inferior *inf);
> +extern void linux_fork_mourn_inferior (inferior *inf);
> +extern void linux_fork_detach (int, lwp_info *, inferior *inf);
> +extern bool forks_exist_p (inferior *inf);
> +extern bool linux_fork_checkpointing_p (int);
>
> #endif /* LINUX_FORK_H */
> diff --git a/gdb/linux-nat.c b/gdb/linux-nat.c
> index 2602e1f240d..e774691cb20 100644
> --- a/gdb/linux-nat.c
> +++ b/gdb/linux-nat.c
> @@ -1552,13 +1552,13 @@ linux_nat_target::detach (inferior *inf, int from_tty)
> gdb_assert (num_lwps (pid) == 1
> || (target_is_non_stop_p () && num_lwps (pid) == 0));
>
> - if (pid == inferior_ptid.pid () && forks_exist_p ())
> + if (forks_exist_p (inf))
> {
> /* Multi-fork case. The current inferior_ptid is being detached
> from, but there are other viable forks to debug. Detach from
> the current fork, and context-switch to the first
> available. */
> - linux_fork_detach (from_tty, find_lwp_pid (ptid_t (pid)));
> + linux_fork_detach (from_tty, find_lwp_pid (ptid_t (pid)), inf);
> }
> else
> {
> @@ -2076,8 +2076,12 @@ linux_handle_extended_wait (struct lwp_info *lp, int status)
> detach_breakpoints (ptid_t (new_pid, new_pid));
>
> /* Retain child fork in ptrace (stopped) state. */
> - if (!find_fork_pid (new_pid))
> - add_fork (new_pid);
> + if (find_fork_pid (new_pid).first == nullptr)
> + {
> + struct inferior *inf = find_inferior_ptid (linux_target,
> + lp->ptid);
> + add_fork (new_pid, inf);
> + }
>
> /* Report as spurious, so that infrun doesn't want to follow
> this fork. We're actually doing an infcall in
> @@ -3757,8 +3761,8 @@ linux_nat_target::kill ()
> parent will be sleeping if this is a vfork. */
> iterate_over_lwps (pid_ptid, kill_unfollowed_child_callback);
>
> - if (forks_exist_p ())
> - linux_fork_killall ();
> + if (forks_exist_p (current_inferior ()))
> + linux_fork_killall (current_inferior ());
> else
> {
> /* Stop all threads before killing them, since ptrace requires
> @@ -3789,14 +3793,14 @@ linux_nat_target::mourn_inferior ()
>
> close_proc_mem_file (pid);
>
> - if (! forks_exist_p ())
> + if (! forks_exist_p (current_inferior ()))
> /* Normal case, no other forks available. */
> inf_ptrace_target::mourn_inferior ();
> else
> /* Multi-fork case. The current inferior_ptid has exited, but
> there are other viable forks to debug. Delete the exiting
> one and context-switch to the first available. */
> - linux_fork_mourn_inferior ();
> + linux_fork_mourn_inferior (current_inferior ());
>
> /* Let the arch-specific native code know this process is gone. */
> linux_target->low_forget_process (pid);
> diff --git a/gdb/testsuite/gdb.base/checkpoint.exp b/gdb/testsuite/gdb.base/checkpoint.exp
> index f87f528e167..b9a1e3161db 100644
> --- a/gdb/testsuite/gdb.base/checkpoint.exp
> +++ b/gdb/testsuite/gdb.base/checkpoint.exp
> @@ -274,17 +274,17 @@ gdb_test "kill" "" "kill all one" \
> # and confirm that all are gone
> #
>
> -gdb_test "restart 0" "Not found.*" "no more checkpoint 0"
> -gdb_test "restart 1" "Not found.*" "no more checkpoint 1"
> -gdb_test "restart 2" "Not found.*" "no more checkpoint 2"
> -gdb_test "restart 3" "Not found.*" "no more checkpoint 3"
> -gdb_test "restart 4" "Not found.*" "no more checkpoint 4"
> -gdb_test "restart 5" "Not found.*" "no more checkpoint 5"
> -gdb_test "restart 6" "Not found.*" "no more checkpoint 6"
> -gdb_test "restart 7" "Not found.*" "no more checkpoint 7"
> -gdb_test "restart 8" "Not found.*" "no more checkpoint 8"
> -gdb_test "restart 9" "Not found.*" "no more checkpoint 9"
> -gdb_test "restart 10" "Not found.*" "no more checkpoint 10"
> +gdb_test "restart 0" "has no checkpoints" "no more checkpoint 0"
> +gdb_test "restart 1" "has no checkpoints" "no more checkpoint 1"
> +gdb_test "restart 2" "has no checkpoints" "no more checkpoint 2"
> +gdb_test "restart 3" "has no checkpoints" "no more checkpoint 3"
> +gdb_test "restart 4" "has no checkpoints" "no more checkpoint 4"
> +gdb_test "restart 5" "has no checkpoints" "no more checkpoint 5"
> +gdb_test "restart 6" "has no checkpoints" "no more checkpoint 6"
> +gdb_test "restart 7" "has no checkpoints" "no more checkpoint 7"
> +gdb_test "restart 8" "has no checkpoints" "no more checkpoint 8"
> +gdb_test "restart 9" "has no checkpoints" "no more checkpoint 9"
> +gdb_test "restart 10" "has no checkpoints" "no more checkpoint 10"
>
> #
> # Now let's try setting a large number of checkpoints (>600)
> @@ -309,7 +309,7 @@ gdb_test "continue" "breakpoint 2.*" "break2 with many checkpoints"
> set count 0
> set msg "info checkpoints with at least 600 checkpoints"
> gdb_test_multiple "info checkpoints" $msg {
> - -re " $decimal process \[^\r\]*\r\n" {
> + -re " $decimal A? *process \[^\r\]*\r\n" {
> incr count
> exp_continue
> }
> diff --git a/gdb/testsuite/gdb.multi/checkpoint-multi.exp b/gdb/testsuite/gdb.multi/checkpoint-multi.exp
> new file mode 100644
> index 00000000000..929b09b6876
> --- /dev/null
> +++ b/gdb/testsuite/gdb.multi/checkpoint-multi.exp
> @@ -0,0 +1,687 @@
> +# Copyright 2009-2024 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 file tests various scenarios involving multiple inferiors
> +# and the checkpoint command.
> +
> +# Checkpoint support works only on Linux.
> +require {istarget "*-*-linux*"}
> +
> +# Checkpoint support is implemented for the (Linux) native target,
> +# which is indicated when [target_info gdb_protocol] returns the empty
> +# string.
> +if { [target_info gdb_protocol] != "" } {
> + return
> +}
> +
Since very recently, this can now be just:
require gdb_protocol_is_native
> +set testfile "checkpoint-multi"
> +
> +set exec1 "hello"
> +set srcfile1 ${exec1}.c
> +set binfile1 [standard_output_file ${exec1}]
> +
> +set exec2 "goodbye"
> +set srcfile2 ${exec2}.c
> +set binfile2 [standard_output_file ${exec2}]
> +
> +set exec3 "hangout"
> +set srcfile3 ${exec3}.c
> +set binfile3 [standard_output_file ${exec3}]
> +
> +if { [build_executable ${testfile}.exp ${exec1} "${srcfile1}" {debug}] == -1 } {
> + return -1
> +}
> +
> +if { [build_executable ${testfile}.exp ${exec2} "${srcfile2}" {debug}] == -1 } {
> + return -1
> +}
> +
> +if { [build_executable ${testfile}.exp ${exec3} "${srcfile3}" {debug}] == -1 } {
> + return -1
> +}
> +
> +# Start two inferiors, place a checkpoint on inferior 2, but switch
> +# back to inferior 1.
> +proc start_2_inferiors_checkpoint_on_inf_2 {} {
> + clean_restart $::exec1
> +
> + # Start inferior 1.
> + if {[gdb_start_cmd] < 0} {
> + fail "start first inferior"
> + } else {
> + gdb_test "" "main.*" "start first inferior"
> + }
> +
> + # Add a new inferior and exec into it.
> + gdb_test "add-inferior -exec $::binfile2" \
> + "Added inferior 2.*" \
> + "add inferior 2 with -exec $::exec2"
> +
> + # Check that we have multiple inferiors.
> + gdb_test "info inferiors" \
> + "Executable.*$::exec1.*$::exec2.*"
> +
> + # Switch to inferior 2.
> + gdb_test "inferior 2" \
> + "Switching to inferior 2.*$::exec2.*"
> +
> + # Start inferior 2:
> + if {[gdb_start_cmd] < 0} {
> + fail "start second inferior"
> + } else {
> + gdb_test "" "main.*" "start second inferior"
> + }
> +
> + # Set a checkpoint in inferior 2
> + gdb_test "checkpoint" "\[Cc\]heckpoint 1: fork returned pid $::decimal.*"
> +
> + # Step one line in inferior 2.
> + gdb_test "step" "glob = 46;"
> +
> + # Switch back to inferior 1.
> + gdb_test "inferior 1" "Switching to inferior 1.*$::exec1.*"
> +}
> +
> +# Start two inferiors, place a checkpoint on inferior 2, but switch
> +# back to inferior 1. This is like the one above, except that it
> +# swaps the executables loaded into inferior 1 and inferior 2. This
Odd tab instead of two spaces after period.
> +# is important for being able to test "continue to exit". (Because...
> +# hello.c has an infinite loop, but goodbye.c doesn't. In order to
Another one.
> +# test "continue to exit", we need to continue in an executable which
> +# will actually exit.)
> +
> +proc start_2_inferiors_checkpoint_on_inf_2_alt {} {
> + clean_restart $::exec2
> +
> + # Start inferior 1.
> + if {[gdb_start_cmd] < 0} {
> + fail "start first inferior"
> + } else {
> + gdb_test "" "main.*" "start first inferior"
> + }
> +
> + # Add a new inferior and exec exec1 into it.
> + gdb_test "add-inferior -exec $::binfile1" \
> + "Added inferior 2.*" \
> + "add inferior 2 with -exec $::exec1"
> +
> + # Check that we have two inferiors.
> + gdb_test "info inferiors" \
> + "Executable.*$::exec2.*$::exec1.*"
> +
> + # Switch to inferior 2.
> + gdb_test "inferior 2" \
> + "Switching to inferior 2.*$::exec1.*"
> +
> + # Start inferior 2:
> + if {[gdb_start_cmd] < 0} {
> + fail "start second inferior"
> + } else {
> + gdb_test "" "main.*" "start second inferior"
> + }
> +
> + # Set a checkpoint in inferior 2
> + gdb_test "checkpoint" "\[Cc\]heckpoint 1: fork returned pid $::decimal.*"
Why do we need to handle both upper and lowercase? (appears in more places)
> +
> + # next one line in inferior 2.
> + gdb_test "next" "bar\\(\\).*"
> +
> + # Switch back to inferior 1.
> + gdb_test "inferior 1" "Switching to inferior 1.*$::exec2.*"
> +}
> +
> +with_test_prefix "check detach on non-checkpointed inferior" {
> + start_2_inferiors_checkpoint_on_inf_2
> + gdb_test "detach" "Detaching from program.*$::exec1.*Inferior 1.*detached.*"
> +}
> +
> +with_test_prefix "check kill on non-checkpointed inferior" {
> + start_2_inferiors_checkpoint_on_inf_2
> + gdb_test "kill" "" "kill non-checkpointed inferior" \
> + "Kill the program being debugged.*y or n. $" "y"
> +}
> +
> +with_test_prefix "check restart 0 on non-checkpointed inferior" {
> + start_2_inferiors_checkpoint_on_inf_2
> + gdb_test "restart 0" "Inferior 1 has no checkpoints"
> + gdb_test "restart 2.0" "Switching to inferior 2.*?goodbye.*?#0 +mailand .*?glob = 46;.*"
> +}
> +
> +with_test_prefix "check restart 1 on non-checkpointed inferior" {
> + start_2_inferiors_checkpoint_on_inf_2
> + gdb_test "restart 1" "Inferior 1 has no checkpoints"
> + gdb_test "restart 2.1" "Switching to inferior 2.*?goodbye.*?#0 +main .*?mailand\\(\\);.*"
> +}
> +
> +with_test_prefix "check continue to exit on non-checkpointed inferior" {
> + start_2_inferiors_checkpoint_on_inf_2_alt
> + gdb_test "continue" "Inferior 1.*? exited normally.*"
> +}
> +
> +set proc_re "(?:process $::decimal|Thread $::hex \\(LWP $::decimal\\))"
> +set main_proc "\\(main process\\)"
> +set hello_c "hello\\.c"
> +set goodbye_c "goodbye\\.c"
> +set hangout_c "hangout\\.c"
> +
> +with_test_prefix "two inferiors with checkpoints" {
> + start_2_inferiors_checkpoint_on_inf_2
> + gdb_test "info checkpoints" \
> + [multi_line \
> + " 2.0 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
> + " 2.1 $proc_re at $::hex, file.*?$goodbye_c.*?"] \
> + "info checkpoints 1"
> + gdb_test "checkpoint" "\[Cc\]heckpoint 1\\.1: fork returned pid $::decimal.*" \
> + "checkpoint in inferior 1"
> + gdb_test "info checkpoints" \
> + [multi_line \
> + "\\* 1\\.0 A $proc_re at $::hex, file.*?$hello_c.*?" \
> + " 1\\.1 $proc_re at $::hex, file.*?$hello_c.*?" \
> + " 2\\.0 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
> + " 2\\.1 $proc_re at $::hex, file.*?$goodbye_c.*?"] \
> + "info checkpoints 2"
> + gdb_test "restart 0" "Already using checkpoint 0"
> + gdb_test "restart 2.0" \
> + "\\\[Switching to inferior 2.*?mailand.*?glob = 46;.*"
> + gdb_test "next" "\}"
> +
> + gdb_test "restart 1" "^Switching to $proc_re.*?#0 main \\(\\) at.*?$goodbye_c.*mailand\\(\\);"
> + gdb_test "info checkpoints" \
> + [multi_line \
> + " 1\\.0 A $proc_re at $::hex, file.*?$hello_c.*?" \
> + " 1\\.1 $proc_re at $::hex, file.*?$hello_c.*?" \
> + " 2\\.0 $proc_re at $::hex, file.*?$goodbye_c.*?" \
> + "\\* 2\\.1 A $proc_re at $::hex, file.*?$goodbye_c.*?"] \
> + "info checkpoints 3"
> +
> + # Doing "info_checkpoints" twice in a row might seem pointless,
> + # but during work on making the checkpoint code inferior aware,
> + # there was a point at which doing it twice in a row did not
> + # produce the same output.
> + gdb_test "info checkpoints" \
> + [multi_line \
> + " 1\\.0 A $proc_re at $::hex, file.*?$hello_c.*?" \
> + " 1\\.1 $proc_re at $::hex, file.*?$hello_c.*?" \
> + " 2\\.0 $proc_re at $::hex, file.*?$goodbye_c.*?" \
> + "\\* 2\\.1 A $proc_re at $::hex, file.*?$goodbye_c.*?"] \
> + "info checkpoints 4"
> +
> + # Switch back to checkpoint 0; again, there should be no
> + # "Switching to inferior" message.
> + gdb_test "restart 0" "^Switching to $proc_re.*?#0 mailand \\(\\) at.*?$goodbye_c.*\}" \
> + "restart 0 #2"
> +
> + gdb_test "info checkpoints" \
> + [multi_line \
> + " 1\\.0 A $proc_re at $::hex, file.*?$hello_c.*?" \
> + " 1\\.1 $proc_re at $::hex, file.*?$hello_c.*?" \
> + "\\* 2\\.0 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
> + " 2\\.1 $proc_re at $::hex, file.*?$goodbye_c.*?"] \
> + "info checkpoints 5"
> +
> + # Try switching to invalid checkpoints:
> + gdb_test "restart 3" "Invalid checkpoint number 3 for inferior 2"
> + gdb_test "restart 2" "Invalid checkpoint number 2 for inferior 2"
> + gdb_test "restart -1" "Checkpoint number must be a non-negative integer"
> + gdb_test "restart 2.3" "Invalid checkpoint number 3 for inferior 2"
> + gdb_test "restart 3.0" "No inferior number '3'"
> + gdb_test "restart 1.2" "Invalid checkpoint number 2 for inferior 1"
> + gdb_test "restart 1.3" "Invalid checkpoint number 3 for inferior 1"
> + gdb_test "restart 1.-1" "Checkpoint number must be a non-negative integer"
> + gdb_test "restart -1.0" "Inferior number must be a positive integer"
> +
> + # Switch to checkpoint 3; this time, we should see a "Switching to
> + # inferior" message.
> + gdb_test "restart 1.1" \
> + "\\\[Switching to inferior 1.*?main.*?$hello_c.*?alarm \\(240\\);"
> +
> + gdb_test "info checkpoints" \
> + [multi_line \
> + " 1\\.0 $proc_re at $::hex, file.*?$hello_c.*?" \
> + "\\* 1\\.1 A $proc_re at $::hex, file.*?$hello_c.*?" \
> + " 2\\.0 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
> + " 2\\.1 $proc_re at $::hex, file.*?$goodbye_c.*?"] \
> + "info checkpoints 6"
> +
> + gdb_test "restart 2.1" \
> + "Switching to inferior 2.*?#0 main \\(\\) at.*?$goodbye_c.*mailand\\(\\);"
> +
> + gdb_test "info checkpoints" \
> + [multi_line \
> + " 1\\.0 $proc_re at $::hex, file.*?$hello_c.*?" \
> + " 1\\.1 A $proc_re at $::hex, file.*?$hello_c.*?" \
> + " 2\\.0 $proc_re at $::hex, file.*?$goodbye_c.*?" \
> + "\\* 2\\.1 A $proc_re at $::hex, file.*?$goodbye_c.*?"] \
> + "info checkpoints 7"
> +
> + gdb_test "checkpoint" "\[Cc\]heckpoint 2\\.2: fork returned pid $::decimal.*" \
> + "second checkpoint in inferior 2"
> +
> + gdb_test "info checkpoints" \
> + [multi_line \
> + " 1\\.0 $proc_re at $::hex, file.*?$hello_c.*?" \
> + " 1\\.1 A $proc_re at $::hex, file.*?$hello_c.*?" \
> + " 2\\.0 $proc_re at $::hex, file.*?$goodbye_c.*?" \
> + "\\* 2\\.1 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
> + " 2\\.2 $proc_re at $::hex, file.*?$goodbye_c.*?"] \
> + "info checkpoints 8"
> +
> + gdb_test "checkpoint" "\[Cc\]heckpoint 2.3: fork returned pid $::decimal.*" \
> + "third checkpoint in inferior 2"
> +
> + gdb_test "info checkpoints" \
> + [multi_line \
> + " 1\\.0 $proc_re at $::hex, file.*?$hello_c.*?" \
> + " 1\\.1 A $proc_re at $::hex, file.*?$hello_c.*?" \
> + " 2\\.0 $proc_re at $::hex, file.*?$goodbye_c.*?" \
> + "\\* 2\\.1 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
> + " 2\\.2 $proc_re at $::hex, file.*?$goodbye_c.*?" \
> + " 2\\.3 $proc_re at $::hex, file.*?$goodbye_c.*?"] \
> + "info checkpoints 9"
Instead of all the meaningless N in "info checkpoints N", did you consider
wrapping in with_test_prefix, like:
with_test_prefix "third checkpoint in inferior 2" {
gdb_test "checkpoint" "\[Cc\]heckpoint 2.3: fork returned pid $::decimal.*"
gdb_test "info checkpoints" \
[multi_line \
" 1\\.0 $proc_re at $::hex, file.*?$hello_c.*?" \
" 1\\.1 A $proc_re at $::hex, file.*?$hello_c.*?" \
" 2\\.0 $proc_re at $::hex, file.*?$goodbye_c.*?" \
"\\* 2\\.1 A $proc_re at $::hex, file.*?$goodbye_c.*?" \
" 2\\.2 $proc_re at $::hex, file.*?$goodbye_c.*?" \
" 2\\.3 $proc_re at $::hex, file.*?$goodbye_c.*?"]
}
and ...
> +
> + gdb_test "continue" \
> + "Inferior 2 \\(process $decimal\\) exited normally.*?Switching to $proc_re.*?" \
> + "continue to exit in checkpoint 2.1"
> +
> + gdb_test "info checkpoints" \
> + [multi_line \
> + " 1\\.0 $proc_re at $::hex, file.*?$hello_c.*?" \
> + " 1\\.1 A $proc_re at $::hex, file.*?$hello_c.*?" \
> + " 2\\.0 $proc_re at $::hex, file.*?$goodbye_c.*?" \
> + " 2\\.2 $proc_re at $::hex, file.*?$goodbye_c.*?" \
> + "\\* 2\\.3 A $proc_re at $::hex, file.*?$goodbye_c.*?"] \
> + "info checkpoints 10"
with_test_prefix "continue to exit in checkpoint 2.1" {
gdb_test "continue" \
"Inferior 2 \\(process $decimal\\) exited normally.*?Switching to $proc_re.*?"
gdb_test "info checkpoints" \
[multi_line \
" 1\\.0 $proc_re at $::hex, file.*?$hello_c.*?" \
" 1\\.1 A $proc_re at $::hex, file.*?$hello_c.*?" \
" 2\\.0 $proc_re at $::hex, file.*?$goodbye_c.*?" \
" 2\\.2 $proc_re at $::hex, file.*?$goodbye_c.*?" \
"\\* 2\\.3 A $proc_re at $::hex, file.*?$goodbye_c.*?"]
}
... etc., etc.
This would group the commands adding some structure, which helps understand what the commands
are testing, IMO.
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [PATCH v3 2/3] Capitalize output of successful checkpoint command
2024-04-14 19:44 ` [PATCH v3 2/3] Capitalize output of successful checkpoint command Kevin Buettner
@ 2024-05-03 15:54 ` Pedro Alves
0 siblings, 0 replies; 8+ messages in thread
From: Pedro Alves @ 2024-05-03 15:54 UTC (permalink / raw)
To: Kevin Buettner, gdb-patches
On 2024-04-14 20:44, Kevin Buettner wrote:
> This commit causes the output of a "checkpoint" command to be
> changed from:
>
> checkpoint N: fork returned pid DDDD
>
> to:
>
> Checkpoint N: fork returned pid DDDD
>
> This change was made to bring the output of the "checkpoint" command in
> line with that of other commands, e.g.:
>
> (gdb) break main
> Breakpoint 1 at ...
>
> (gdb) catch exec
> Catchpoint 2 (exec)
>
> (gdb) add-inferior
> [New inferior 2]
> Added inferior 2
>
> The tests gdb.base/checkpoint.exp and gdb.base/kill-during-detach.exp
> have been updated to accept either form of output from the
> "checkpoint" command.
Why either instead of just uppercase?
> The test gdb.multi/checkpoint-multi.exp had
> already been written with this change in mind.
Ah, yeah,that answers the question I made there.
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [PATCH v3 3/3] Make thread_db_target::pid_to_str checkpoint-aware
2024-04-14 19:44 ` [PATCH v3 3/3] Make thread_db_target::pid_to_str checkpoint-aware Kevin Buettner
@ 2024-05-03 15:55 ` Pedro Alves
0 siblings, 0 replies; 8+ messages in thread
From: Pedro Alves @ 2024-05-03 15:55 UTC (permalink / raw)
To: Kevin Buettner, gdb-patches
On 2024-04-14 20:44, Kevin Buettner wrote:
> This commit prevents thread_db_target::pid_to_str from considering
> a checkpoint as a thread. The reason for doing this is that pids
> associated with checkpoints can never be a thread due to the fact
> that checkpoints (which are implemented by forking a process) can
> only work with single-threaded processes.
Well, that isn't strictly true, as a checkpoint can spawn a new
thread/clone. Or even exec... But let's not go there...
>
> Without this commit, many of the "info checkpoints" commands
> in gdb.multi/checkpoint-multi.exp will incorrectly show some
> of the checkpoints as threads. E.g...
>
> * 1.0 A Thread 0x7ffff7cd3740 (LWP 128534) at 0x401199, file hello.c, line 51
> 1.2 process 128546 at 0x401199, file hello.c, line 51
> 1.3 process 128547 at 0x401199, file hello.c, line 51
> 2.1 process 128538 at 0x401258, file goodbye.c, line 62
> 2.2 A Thread 0x7ffff7cd3740 (LWP 128542) at 0x401258, file goodbye.c, line 62
> 3.0 A Thread 0x7ffff7cd3740 (LWP 128543) at 0x40115c, file hangout.c, line 31
> 3.2 process 128545 at 0x40115c, file hangout.c, line 31
>
> With this commit in place, the output looks like this instead:
>
> * 1.0 A process 129961 at 0x401199, file hello.c, line 51
> 1.2 process 129974 at 0x401199, file hello.c, line 51
> 1.3 process 129975 at 0x401199, file hello.c, line 51
> 2.1 process 129965 at 0x401258, file goodbye.c, line 62
> 2.2 A process 129969 at 0x401258, file goodbye.c, line 62
> 3.0 A process 129970 at 0x40115c, file hangout.c, line 31
> 3.2 process 129972 at 0x40115c, file hangout.c, line 31
>
> (For brevity, I've removed the directory elements in each of the paths
> above.)
>
> The testcase, gdb.multi/checkpoint-multi.exp, has been updated to
> reflect the fact that only "process" should now appear in output
> from "info checkpoints".
> ---
> gdb/linux-thread-db.c | 4 +++-
> gdb/testsuite/gdb.multi/checkpoint-multi.exp | 2 +-
> 2 files changed, 4 insertions(+), 2 deletions(-)
>
> diff --git a/gdb/linux-thread-db.c b/gdb/linux-thread-db.c
> index 65bf4a79fdf..488a7086acf 100644
> --- a/gdb/linux-thread-db.c
> +++ b/gdb/linux-thread-db.c
> @@ -48,6 +48,7 @@
> #include "gdbsupport/pathstuff.h"
> #include "valprint.h"
> #include "cli/cli-style.h"
> +#include "linux-fork.h"
>
> /* GNU/Linux libthread_db support.
>
> @@ -1657,7 +1658,8 @@ thread_db_target::pid_to_str (ptid_t ptid)
> {
> thread_info *thread_info = current_inferior ()->find_thread (ptid);
>
> - if (thread_info != NULL && thread_info->priv != NULL)
> + if (thread_info != NULL && thread_info->priv != NULL
> + && !forks_exist_p (current_inferior ()))
> {
> thread_db_thread_info *priv = get_thread_db_thread_info (thread_info);
>
I think that it's better to change linux-fork.c, avoid special casing checkpoints
if we can. If linux-fork.c wants to print a process, then it should pass down a
process ptid. Like:
diff --git c/gdb/linux-fork.c w/gdb/linux-fork.c
index e35e438cabc..9003b43dc82 100644
--- c/gdb/linux-fork.c
+++ w/gdb/linux-fork.c
@@ -807,7 +807,8 @@ info_checkpoints_command (const char *arg, int from_tty)
else
gdb_printf (" ");
- gdb_printf ("%s", target_pid_to_str (fi.ptid).c_str ());
+ ptid_t pid_ptid (fi.ptid.pid ());
+ gdb_printf ("%s", target_pid_to_str (pid_ptid).c_str ());
gdb_printf (_(" at "));
ULONGEST pc
^ permalink raw reply [flat|nested] 8+ messages in thread
end of thread, other threads:[~2024-05-03 15:55 UTC | newest]
Thread overview: 8+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2024-04-14 19:44 [PATCH v3 0/3] Make linux checkpoints work with multiple inferiors Kevin Buettner
2024-04-14 19:44 ` [PATCH v3 1/3] " Kevin Buettner
2024-05-03 15:54 ` Pedro Alves
2024-04-14 19:44 ` [PATCH v3 2/3] Capitalize output of successful checkpoint command Kevin Buettner
2024-05-03 15:54 ` Pedro Alves
2024-04-14 19:44 ` [PATCH v3 3/3] Make thread_db_target::pid_to_str checkpoint-aware Kevin Buettner
2024-05-03 15:55 ` Pedro Alves
2024-05-02 1:49 ` [PATCH v3 0/3] Make linux checkpoints work with multiple inferiors Kevin Buettner
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).