Re: [PATCH v3 4/7] Per-inferior/Inferior-qualified thread IDs

[Simon Marchi <simon dot marchi at ericsson dot com>]

A few comments:

On 16-01-06 08:03 AM, Pedro Alves wrote:
> This commit changes GDB to track thread numbers per-inferior.  Then,
> if you're debugging multiple inferiors, GDB displays
> "inferior-num.thread-num" instead of just "thread-num" whenever it
> needs to display a thread:
> 
>  (gdb) info inferiors
>    Num  Description       Executable
>    1    process 6022     /home/pedro/gdb/tests/threads
>  * 2    process 6037     /home/pedro/gdb/tests/threads
>  (gdb) info threads
>    Id   Target Id         Frame
>    1.1  Thread 0x7ffff7fc2740 (LWP 6022) "threads" (running)
>    1.2  Thread 0x7ffff77c0700 (LWP 6028) "threads" (running)
>    1.3  Thread 0x7ffff7fc2740 (LWP 6032) "threads" (running)
>    2.1  Thread 0x7ffff7fc1700 (LWP 6037) "threads" (running)
>    2.2  Thread 0x7ffff77c0700 (LWP 6038) "threads" (running)
>  * 2.3  Thread 0x7ffff7fc2740 (LWP 6039) "threads" (running)
>  (gdb)
> ...
>  (gdb) thread 1.1
>  [Switching to thread 1.1 (Thread 0x7ffff7fc2740 (LWP 8155))]
>  (gdb)
> ...

It was already the case before your patch, but I just noticed that
when switching to a running thread, there should probably be a
space before (running):

(gdb) thread 1.1
[Switching to thread 1.1 (Thread 0x7ffff7fad9c0 (LWP 15027))](running)

> diff --git a/gdb/testsuite/gdb.multi/tids.c b/gdb/testsuite/gdb.multi/tids.c
> new file mode 100644
> index 0000000..00a8298
> --- /dev/null
> +++ b/gdb/testsuite/gdb.multi/tids.c
> @@ -0,0 +1,52 @@
> +/* This testcase is part of GDB, the GNU debugger.
> +
> +   Copyright 2015 Free Software Foundation, Inc.
> +
> +   This program is free software; you can redistribute it and/or modify
> +   it under the terms of the GNU General Public License as published by
> +   the Free Software Foundation; either version 3 of the License, or
> +   (at your option) any later version.
> +
> +   This program is distributed in the hope that it will be useful,
> +   but WITHOUT ANY WARRANTY; without even the implied warranty of
> +   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> +   GNU General Public License for more details.
> +
> +   You should have received a copy of the GNU General Public License
> +   along with this program.  If not, see <http://www.gnu.org/licenses/>.  */
> +
> +#include <unistd.h>
> +#include <pthread.h>
> +
> +pthread_t child_thread[2];
> +
> +void *
> +thread_function2 (void *arg)
> +{
> +  while (1)
> +    sleep (1);

Nit: this should probably return NULL (thread_function1 as well).

> diff --git a/gdb/testsuite/gdb.multi/tids.exp b/gdb/testsuite/gdb.multi/tids.exp
> new file mode 100644
> index 0000000..7b51c80
> --- /dev/null
> +++ b/gdb/testsuite/gdb.multi/tids.exp
> @@ -0,0 +1,296 @@
> +# Copyright 2015 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/>.
> +
> +# Test thread ID parsing and display.
> +
> +load_lib gdb-python.exp
> +
> +standard_testfile
> +
> +# Multiple inferiors are needed, therefore both native and extended
> +# gdbserver modes are supported.  Only non-extended gdbserver is not
> +# supported.
> +if [target_info exists use_gdb_stub] {
> +    untested ${testfile}.exp
> +    return
> +}
> +
> +if { [prepare_for_testing "failed to prepare" ${testfile} ${srcfile} {pthreads debug}] } {
> +    return -1
> +}
> +
> +clean_restart ${testfile}
> +
> +if { ![runto_main] } then {
> +    return -1
> +}
> +
> +# Issue "thread apply TID_LIST p 1" and expect EXP_TID_LIST (a list of

Nit: in the code it's "p 1234".

> +# thread ids) to be displayed.
> +proc thread_apply {tid_list exp_tid_list {message ""}} {
> +    global decimal
> +    set any "\[^\r\n\]*"
> +    set expected [string_to_regexp $exp_tid_list]
> +
> +    set r ""
> +    foreach tid $expected {
> +	append r "\[\r\n\]+"
> +	append r "Thread $tid $any:\r\n"
> +	append r "\\$$decimal = 1234"
> +    }
> +
> +    set cmd "thread apply $tid_list"
> +    if {$message == ""} {
> +	set message $cmd
> +    }
> +    gdb_test "$cmd p 1234" $r $message
> +}
> +
> +# Issue "info threads TID_LIST" and expect EXP_TID_LIST (a list of
> +# thread ids) to be displayed.
> +proc info_threads {tid_list exp_tid_list {message ""}} {
> +    set any "\[^\r\n\]*"
> +    set expected [string_to_regexp $exp_tid_list]
> +    set r [join $expected " ${any}\r\n${any} "]
> +    set r "${any} $r ${any}"
> +    set cmd "info threads $tid_list"
> +    if {$message == ""} {
> +	set message $cmd
> +    }
> +    gdb_test $cmd $r $message
> +}
> +
> +# Issue both "info threads TID_LIST" and expect INFO_THR output.  Then

"Issue both" does not make sense in this sentence.

> +# issue "thread apply TID_LIST" and expect THR_APPLY output.  If
> +# THR_APPLY is omitted, INFO_THR is expected instead.
> +proc thr_apply_info_thr {tid_list info_thr {thr_apply ""}} {
> +    if {$thr_apply == ""} {
> +	set thr_apply $info_thr
> +    }
> +
> +    info_threads $tid_list $info_thr
> +    thread_apply $tid_list $thr_apply
> +}
> +
> +# Issue both "info threads TID_LIST" and "thread apply TID_LIST" and
> +# expect both commands to error out with EXP_ERROR.
> +proc thr_apply_info_thr_error {tid_list exp_error}  {
> +    gdb_test "info threads $tid_list" \
> +	$exp_error
> +
> +    gdb_test "thread apply $tid_list p 1234" \
> +	$exp_error \
> +	"thread apply $tid_list"
> +}
> +
> +# Issue both "info threads TID_LIST" and "thread apply TID_LIST" and
> +# expect the command to error out with "Invalid thread ID: $EXPECTED".
> +# EXPECTED is a literal string, not a regexp.
> +proc thr_apply_info_thr_invalid {tid_list expected} {
> +    set expected [string_to_regexp $expected]
> +    gdb_test "info threads $tid_list" \
> +	"Invalid thread ID: $expected"
> +
> +    gdb_test "thread apply $tid_list p 1234" \
> +	"Invalid thread ID: $expected p 1234" \
> +	"thread apply $tid_list"
> +}
> +
> +# "info threads" while there's only inferior 1 should show
> +# single-number thread IDs.
> +with_test_prefix "single inferior" {
> +    info_threads "" "1"
> +
> +    gdb_test "thread" "Current thread is 1 .*"
> +}
> +
> +# "info threads" while there are multiple inferiors should show
> +# qualified thread IDs.
> +with_test_prefix "two inferiors" {
> +    # Add another inferior.
> +    gdb_test "add-inferior" "Added inferior 2.*" "add empty inferior 2"
> +
> +    # Now that we'd added another inferior, thread IDs now show the

"Now that we've added" ?

> +    # inferior number.
> +    info_threads "" "1.1"
> +
> +    gdb_test "thread" "Current thread is 1\.1 .*"
> +
> +    gdb_test "inferior 2" "Switching to inferior 2 .*" "switch to inferior 2"
> +    gdb_test "file ${binfile}" ".*" "load file in inferior 2"
> +
> +    runto_main
> +
> +    # Now that we'd added another inferior, thread IDs now show the

Again.

> diff --git a/gdb/tid-parse.h b/gdb/tid-parse.h
> new file mode 100644
> index 0000000..3c79c08
> --- /dev/null
> +++ b/gdb/tid-parse.h
> @@ -0,0 +1,148 @@
> +/* TID parsing for GDB, the GNU debugger.
> +   Copyright (C) 2015-2016 Free Software Foundation, Inc.
> +
> +   This file is part of GDB.
> +
> +   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/>.  */
> +
> +#ifndef TID_PARSE_H
> +#define TID_PARSE_H
> +
> +#include "cli/cli-utils.h"
> +
> +struct thread_info;
> +
> +/* Issue an invalid thread ID error, pointing at STRING, the invalid
> +   ID.  */
> +extern void ATTRIBUTE_NORETURN invalid_thread_id_error (const char *string);
> +
> +/* Parse TIDSTR as a per-inferior thread ID, in either INF_NUM.THR_NUM
> +   or THR_NUM form.  In the latter case, the missing INF_NUM is filled
> +   in from the current inferior.  If ENDPTR is not NULL,
> +   parse_thread_id stores the address of the first character after the
> +   thread ID.  Either a valid thread is returned, or an error is
> +   thrown.  */
> +struct thread_info *parse_thread_id (const char *tidstr, const char **end);
> +
> +/* The possible states of the tid range parser's state machine.  */
> +enum tid_range_state
> +{
> +  /* Parsing the inferior number.  */
> +  TID_RANGE_STATE_INFERIOR,
> +
> +  /* Parsing the thread number or thread number range.  */
> +  TID_RANGE_STATE_THREAD_RANGE,
> +};
> +
> +/* An object of this type is passed to tid_range_parser_get_tid.  It
> +   must be initialized by calling tid_range_parser_init.  This type is
> +   defined here so that it can be stack-allocated, but all members
> +   should be treated as opaque.  */
> +struct tid_range_parser
> +{
> +  /* What sub-component are we expecting.  */
> +  enum tid_range_state state;
> +
> +  /* The string being parsed.  When parsing has finished, this points
> +     past the last parsed token.  */
> +  const char *string;
> +
> +  /* The range parser state when we're parsing the thread number
> +     sub-component.  */
> +  struct get_number_or_range_state range_parser;
> +
> +  /* Last inferior number returned.  */
> +  int inf_num;
> +
> +  /* True if the TID last parsed was explicitly inferior-qualified.
> +     IOW, whether the spec specified an inferior number
> +     explicitly.  */
> +  int qualified;
> +
> +  /* The inferior number to assume if the TID is not qualified.  */
> +  int default_inferior;
> +};
> +
> +/* Initialize a tid_range_parser for use with
> +   tid_range_parser_get_tid.  TIDLIST is the string to be parsed.
> +   DEFAULT_INFERIOR is the inferior number to assume if a
> +   non-qualified thread ID is found.  */
> +extern void tid_range_parser_init (struct tid_range_parser *parser,
> +				   const char *tidlist,
> +				   int default_inferior);
> +
> +/* Parse a thread ID or a thread range list.
> +
> +   A range will be of the form
> +   <inferior_num>.<thread_number1>-<thread_number1> and will represent

- <inferior_num>.<thread_number1>-<thread_number1> and will represent
+ <inferior_num>.<thread_number1>-<thread_number2> and will represent

> +   all the threads of inferior inferior_num with number between
> +   thread_number1 and thread_number2, inclusive.  <inferior_num> can
> +   also be omitted, as in <thread_number1>-<thread_number1>, in which

Again.

> +   case GDB infers the inferior number from the current inferior.
> +
> +   While processing a thread ID range list, this function is called
> +   iteratively; at each call it will return (in the INF_NUM and
> +   THR_NUM output parameters) the next thread in the range.

Would it be good to precise that it's unrelated to the threads that
exist in gdb?  It iterates solely on the numbers and doesn't care
if that thread really exists.

Also, it might be good to mention what the stop condition for the iteration
is.  Maybe as a simple usage example?

> +
> +   At the beginning of parsing a thread range, the char pointer
> +   PARSER->string will be advanced past <thread_number1> and left
> +   pointing at the '-' token.  Subsequent calls will not advance the
> +   pointer until the range is completed.  The call that completes the
> +   range will advance the pointer past <thread_number2>.  */


> +extern void tid_range_parser_get_tid (struct tid_range_parser *parser,
> +				      int *inf_num, int *thr_num);

Is there a reason why _get_tid_range returns success/failure and _get_tid
doesn't return anything?  Can't it fail the same way, since the later is
implemented with the former?

> +
> +/* Like tid_range_parser_get_tid, but return a thread range per call,
> +   rather then a single thread.
> +
> +   If the next element in the list is a single thread ID, then
> +   *THR_START and *THR_END are set to the same value.  E.g.:
> +
> +     list: "1.2 3.4-6"
> +     1st call: *INF_NUM=1; *THR_START=2; *THR_END=2
> +     2nd call: *INF_NUM=3; *THR_START=4; *THR_END=4
> +
> +
> +   Returns true if parsed a thread/range successfully, false
> +   otherwise.  */

IIUC, the value of thr_end defines if you want the to get a single thread id,
or a range.  Should it be mentioned here?

> +extern int tid_range_parser_get_tid_range (struct tid_range_parser *parser,
> +					   int *inf_num,
> +					   int *thr_start, int *thr_end);
> +
> +/* Returns non-zero if parsing has completed.  */
> +extern int tid_range_parser_finished (struct tid_range_parser *parser);
> +
> +/* Return the string being parsed.  When parsing has finished, this
> +   points past the last parsed token.  */
> +const char *tid_range_parser_string (struct tid_range_parser *parser);
> +
> +/* When parsing a range, advance past the final token in the range.  */
> +extern void tid_range_parser_skip (struct tid_range_parser *parser);
> +
> +/* True if the TID last parsed was explicitly inferior-qualified.
> +   IOW, whether the spec specified an inferior number explicitly.  */
> +extern int tid_range_parser_qualified (struct tid_range_parser *parser);
> +
> +/* Accept a string-form list of thread IDs such as is accepted by
> +   tid_range_parser_get_tid.  Return TRUE if the INF_NUM.THR.NUM

TRUE -> true

> +   thread is in the list.  DEFAULT_INFERIOR is the inferior number to
> +   assume if a non-qualified thread ID is found in the list.
> +
> +   By definition, an empty list includes all threads.  This is to be
> +   interpreted as typing a command such as "info threads" with no
> +   arguments.  */
> +extern int tid_is_in_list (const char *list, int default_inferior,
> +			   int inf_num, int thr_num);
> +
> +#endif /* TID_PARSE_H */
> diff --git a/gdb/varobj.c b/gdb/varobj.c
> index d47bdc2..8116a5b 100644
> --- a/gdb/varobj.c
> +++ b/gdb/varobj.c
> @@ -78,7 +78,7 @@ struct varobj_root
>       not NULL.  */
>    struct frame_id frame;
>  
> -  /* The thread ID that this varobj_root belong to.  This field
> +  /* The global thread ID that this varobj_root belongs to.  This field
>       is only valid if valid_block is not NULL.
>       When not 0, indicates which thread 'frame' belongs to.
>       When 0, indicates that the thread list was empty when the varobj_root
> @@ -380,7 +380,7 @@ varobj_create (char *objname,
>  	    error (_("Failed to find the specified frame"));
>  
>  	  var->root->frame = get_frame_id (fi);
> -	  var->root->thread_id = pid_to_thread_id (inferior_ptid);
> +	  var->root->thread_id = ptid_to_global_thread_id (inferior_ptid);
>  	  old_id = get_frame_id (get_selected_frame (NULL));
>  	  select_frame (fi);	 
>  	}
> @@ -2363,8 +2363,9 @@ value_of_root_1 (struct varobj **var_handle)
>      }
>    else
>      {
> -      ptid_t ptid = thread_id_to_pid (var->root->thread_id);
> -      if (in_thread_list (ptid))
> +      ptid_t ptid = global_thread_id_to_ptid (var->root->thread_id);
> +
> +      if (!ptid_equal (minus_one_ptid, ptid))
>  	{
>  	  switch_to_thread (ptid);
>  	  within_scope = check_scope (var);
> 

windows-tdep.c fails to build with this patch.  There is still one usage of find_thread_id:

/home/emaisin/src/binutils-gdb/gdb/windows-tdep.c: In function ‘display_tib’:
/home/emaisin/src/binutils-gdb/gdb/windows-tdep.c:378:7: error: implicit declaration of function ‘find_thread_id’ [-Werror=implicit-function-declaration]
       tp = find_thread_id (gdb_id);
       ^
/home/emaisin/src/binutils-gdb/gdb/windows-tdep.c:378:10: error: assignment makes pointer from integer without a cast [-Werror]
       tp = find_thread_id (gdb_id);
          ^
cc1: all warnings being treated as errors
make: *** [windows-tdep.o] Error 1