* RE: [PATCH v6 1/1] gdb, infcmd: Support jump command in multi-inferior case.
2023-08-16 8:58 99% ` Andrew Burgess
@ 2023-08-16 10:55 99% ` Puputti, Matti
0 siblings, 0 replies; 34+ results
From: Puputti, Matti @ 2023-08-16 10:55 UTC (permalink / raw)
To: Andrew Burgess, gdb-patches
Hi Andrew,
Thank you for the support.
Br,
Matti Puputti
> -----Original Message-----
> From: Andrew Burgess <aburgess@redhat.com>
> Sent: Wednesday, August 16, 2023 10:59 AM
> To: Puputti, Matti <matti.puputti@intel.com>; gdb-patches@sourceware.org
> Subject: Re: [PATCH v6 1/1] gdb, infcmd: Support jump command in multi-
> inferior case.
>
> Matti Puputti <matti.puputti@intel.com> writes:
>
> > From: "Puputti, Matti" <matti.puputti@intel.com>
> >
> > Fixes the issue where jump failed if multiple inferiors run the same
> > source.
>
> Thanks for the latest update.
>
> I took another look through, this looks great, so I have pushed this to
> master.
>
> I did add a new comment to the 'do_test' proc in the test script, and
> extended the description of the fix in the commit message.
>
> Thanks,
> Andrew
>
>
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [relevance 99%]
* Re: [PATCH v6 1/1] gdb, infcmd: Support jump command in multi-inferior case.
2023-07-26 12:29 66% [PATCH v6 " Matti Puputti
@ 2023-08-16 8:58 99% ` Andrew Burgess
2023-08-16 10:55 99% ` Puputti, Matti
0 siblings, 1 reply; 34+ results
From: Andrew Burgess @ 2023-08-16 8:58 UTC (permalink / raw)
To: Matti Puputti, gdb-patches
Matti Puputti <matti.puputti@intel.com> writes:
> From: "Puputti, Matti" <matti.puputti@intel.com>
>
> Fixes the issue where jump failed if multiple inferiors run the same
> source.
Thanks for the latest update.
I took another look through, this looks great, so I have pushed this to
master.
I did add a new comment to the 'do_test' proc in the test script, and
extended the description of the fix in the commit message.
Thanks,
Andrew
>
> See the below example
>
> $ gdb -q ./simple
> Reading symbols from ./simple...
> (gdb) break 2
> Breakpoint 1 at 0x114e: file simple.c, line 2.
> (gdb) run
> Starting program: /temp/simple
>
> Breakpoint 1, main () at simple.c:2
> 2 int a = 42;
> (gdb) add-inferior
> [New inferior 2]
> Added inferior 2 on connection 1 (native)
> (gdb) inferior 2
> [Switching to inferior 2 [<null>] (<noexec>)]
> (gdb) info inferiors
> Num Description Connection Executable
> 1 process 6250 1 (native) /temp/simple
> * 2 <null> 1 (native)
> (gdb) file ./simple
> Reading symbols from ./simple...
> (gdb) run
> Starting program: /temp/simple
>
> Thread 2.1 "simple" hit Breakpoint 1, main () at simple.c:2
> 2 int a = 42;
> (gdb) info inferiors
> Num Description Connection Executable
> 1 process 6250 1 (native) /temp/simple
> * 2 process 6705 1 (native) /temp/simple
> (gdb) jump 3
> Unreasonable jump request
> (gdb)
>
> In this example, jump fails because the debugger finds two different
> locations, one for each inferior.
> Solution is to limit the search to the current program space.
> ---
> gdb/infcmd.c | 2 +-
> gdb/linespec.c | 3 +-
> gdb/testsuite/gdb.base/jump.exp | 180 +++++++++++++++++---------------
> 3 files changed, 98 insertions(+), 87 deletions(-)
>
> diff --git a/gdb/infcmd.c b/gdb/infcmd.c
> index 15702f84894..96c5feafb1a 100644
> --- a/gdb/infcmd.c
> +++ b/gdb/infcmd.c
> @@ -1068,7 +1068,7 @@ jump_command (const char *arg, int from_tty)
> error_no_arg (_("starting address"));
>
> std::vector<symtab_and_line> sals
> - = decode_line_with_last_displayed (arg, DECODE_LINE_FUNFIRSTLINE);
> + = decode_line_with_current_source (arg, DECODE_LINE_FUNFIRSTLINE);
> if (sals.size () != 1)
> {
> /* If multiple sal-objects were found, try dropping those that aren't
> diff --git a/gdb/linespec.c b/gdb/linespec.c
> index 7d969f37fbf..afa9eb4d3ac 100644
> --- a/gdb/linespec.c
> +++ b/gdb/linespec.c
> @@ -3220,7 +3220,8 @@ decode_line_with_current_source (const char *string, int flags)
> location_spec_up locspec = string_to_location_spec (&string,
> current_language);
> std::vector<symtab_and_line> sals
> - = decode_line_1 (locspec.get (), flags, NULL, cursal.symtab, cursal.line);
> + = decode_line_1 (locspec.get (), flags, cursal.pspace, cursal.symtab,
> + cursal.line);
>
> if (*string)
> error (_("Junk at end of line specification: %s"), string);
> diff --git a/gdb/testsuite/gdb.base/jump.exp b/gdb/testsuite/gdb.base/jump.exp
> index 032c4a6279d..77765db15ef 100644
> --- a/gdb/testsuite/gdb.base/jump.exp
> +++ b/gdb/testsuite/gdb.base/jump.exp
> @@ -18,99 +18,109 @@ clear_xfail "*-*-*"
>
> standard_testfile .c
>
> -# Build the test case
> -if { [gdb_compile "${srcdir}/${subdir}/${srcfile}" "${binfile}" executable {debug nowarnings}] != "" } {
> - untested "failed to compile"
> - return -1
> - }
> +if {[prepare_for_testing "failed to prepare" $testfile $srcfile]} {
> + return -1
> +}
>
>
> -# Start with a fresh gdb
> +proc do_tests {} {
> + global decimal srcfile
> +
> + # Set a breakpoint on the statement that we're about to jump to.
> + # The statement doesn't contain a function call.
> + set non_call_line [gdb_get_line_number "bp-on-non-call"]
> + gdb_breakpoint "$non_call_line"
> + set bp_on_non_call \
> + [get_integer_valueof "\$bpnum" "INVALID" "bp_on_non_call"]
> +
> + # Can we jump to the statement? Do we stop there?
> + gdb_test "jump $non_call_line" \
> + "Breakpoint ${bp_on_non_call}(\.${decimal})?, .*${srcfile}:$non_call_line.*" \
> + "jump to non-call"
> +
> + # Set a breakpoint on the statement that we're about to jump to.
> + # The statement does contain a function call.
> + set call_line [gdb_get_line_number "bp-on-call"]
> + gdb_breakpoint "$call_line"
> + set bp_on_call [get_integer_valueof "\$bpnum" "INVALID" "bp_on_call"]
> +
> + # Can we jump to the statement? Do we stop there?
> + gdb_test "jump $call_line" \
> + "Breakpoint ${bp_on_call}(\.${decimal})?, .*${srcfile}:$call_line.*" \
> + "jump to call"
> +
> + # If we disable the breakpoint at the function call, and then
> + # if we jump to that statement, do we not stop there, but at
> + # the following breakpoint?
> + gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
> +
> + gdb_test "jump $call_line" \
> + "Breakpoint ${bp_on_non_call}(\.${decimal})?, .*${srcfile}:$non_call_line.*" \
> + "jump to call with disabled breakpoint"
> +
> + # Disable the breakpoint at the non-function call, so it won't hit
> + # if do_test is called again.
> + gdb_test_no_output "disable ${bp_on_non_call}" "disable bp_on_non_call"
> +
> + # Verify that GDB responds gracefully to the "jump" command without
> + # an argument.
> + gdb_test "jump" "Argument required .starting address.*" \
> + "jump without argument disallowed"
> +
> + # Verify that GDB responds gracefully to the "jump" command with
> + # trailing junk.
> + gdb_test "jump $call_line 100" \
> + "malformed linespec error: unexpected number, \"100\"" \
> + "jump with trailing argument junk"
> +
> + # Verify that GDB responds gracefully to a request to jump out of
> + # the current function. (Note that this will very likely cause the
> + # inferior to die. Be prepared to rerun the inferior, if further
> + # testing is desired.)
> + # Try it both ways: confirming and not confirming the jump.
> + set out_line [gdb_get_line_number "out-of-func"]
> + gdb_test "jump $out_line" \
> + "Not confirmed.*" \
> + "aborted jump out of current function" \
> + "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> + "n"
> +
> + gdb_test "jump $out_line" \
> + "Continuing at.*" \
> + "jump out of current function" \
> + "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> + "y"
> +}
>
> -clean_restart ${binfile}
>
> -if {![runto_main]} {
> - return -1
> +set num_inferiors 1
> +if {![use_gdb_stub]} {
> + set num_inferiors 2
> }
>
> -# Set a breakpoint on the statement that we're about to jump to.
> -# The statement doesn't contain a function call.
> -#
> -set bp_on_non_call 0
> -set non_call_line [gdb_get_line_number "bp-on-non-call"]
> -gdb_test_multiple "break $non_call_line" "break before jump to non-call" {
> - -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line $non_call_line.*$gdb_prompt $" {
> - set bp_on_non_call $expect_out(1,string)
> - pass "break before jump to non-call"
> +# Run to main, add inferiors if needed.
> +if {![runto_main]} {
> + return -1
> +}
> +for {set inf 2} {$inf <= $num_inferiors} {incr inf} {
> + # Start a new inferior, and run it with the same executable.
> + gdb_test "add-inferior -exec ${binfile}" \
> + "Added inferior ${inf}.*" \
> + "add inferior ${inf} with -exec "
> + gdb_test "inferior ${inf}" \
> + "Switching to inferior ${inf} .*" \
> + "switch to inferior ${inf}"
> + if {![runto_main]} {
> + return -1
> }
> }
>
> -# Can we jump to the statement? Do we stop there?
> -#
> -gdb_test "jump $non_call_line" "Breakpoint ${decimal}, .*${srcfile}:$non_call_line.*" \
> - "jump to non-call"
> -
> -# Set a breakpoint on the statement that we're about to jump to.
> -# The statement does contain a function call.
> -#
> -set bp_on_call 0
> -set call_line [gdb_get_line_number "bp-on-call"]
> -gdb_test_multiple "break $call_line" "break before jump to call" {
> - -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line $call_line.*$gdb_prompt $" {
> - set bp_on_call $expect_out(1,string)
> - pass "break before jump to call"
> +# Run tests on all inferiors.
> +for {set inf 1} {$inf <= $num_inferiors} {incr inf} {
> + with_test_prefix "inferior $inf" {
> + # Switch to the target inferior.
> + gdb_test "inferior $inf" ".*Switching to inferior $inf .*"
> + # Run the tests.
> + do_tests
> }
> }
> -
> -# Can we jump to the statement? Do we stop there?
> -#
> -gdb_test "jump $call_line" \
> - "Breakpoint ${decimal}, .*${srcfile}:$call_line.*" \
> - "jump to call"
> -
> -# If we disable the breakpoint at the function call, and then
> -# if we jump to that statement, do we not stop there, but at
> -# the following breakpoint?
> -#
> -gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
> -
> -gdb_test "jump $call_line" "Breakpoint ${decimal}, .*${srcfile}:$non_call_line.*" \
> - "jump to call with disabled breakpoint"
> -
> -# Verify that GDB responds gracefully to the "jump" command without
> -# an argument.
> -#
> -gdb_test "jump" "Argument required .starting address.*" \
> - "jump without argument disallowed"
> -
> -
> -# Verify that GDB responds gracefully to the "jump" command with
> -# trailing junk.
> -#
> -gdb_test "jump $call_line 100" \
> - "malformed linespec error: unexpected number, \"100\"" \
> - "jump with trailing argument junk"
> -
> -
> -# Verify that GDB responds gracefully to a request to jump out of
> -# the current function. (Note that this will very likely cause the
> -# inferior to die. Be prepared to rerun the inferior, if further
> -# testing is desired.)
> -#
> -# Try it both ways: confirming and not confirming the jump.
> -#
> -
> -set out_line [gdb_get_line_number "out-of-func"]
> -gdb_test "jump $out_line" \
> - "Not confirmed.*" \
> - "aborted jump out of current function" \
> - "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> - "n"
> -
> -gdb_test "jump $out_line" \
> - "Continuing at.*" \
> - "jump out of current function" \
> - "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> - "y"
> -
> -gdb_exit
> --
> 2.25.1
>
> Intel Deutschland GmbH
> Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
> Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
> Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
> Chairperson of the Supervisory Board: Nicole Lau
> Registered Office: Munich
> Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [relevance 99%]
* RE: [PATCH v4 1/1] gdb, infcmd: Support jump command in multi-inferior case.
2023-07-26 11:51 99% ` Puputti, Matti
@ 2023-07-26 12:32 99% ` Puputti, Matti
0 siblings, 0 replies; 34+ results
From: Puputti, Matti @ 2023-07-26 12:32 UTC (permalink / raw)
To: Andrew Burgess, gdb-patches
Sincerely sorry for spamming, but I messed with the commit message in patch 5, and had to upload patch 6.
Only difference between patches 5 and 6 is the commit message title.
Patch 6: https://sourceware.org/pipermail/gdb-patches/2023-July/201125.html
Br,
Matti Puputti
> -----Original Message-----
> From: Puputti, Matti
> Sent: Wednesday, July 26, 2023 1:51 PM
> To: Andrew Burgess <aburgess@redhat.com>; gdb-patches@sourceware.org
> Subject: RE: [PATCH v4 1/1] gdb, infcmd: Support jump command in multi-
> inferior case.
>
> Hi Andrew,
>
> I am sorry to take such a long time to react on your latest comments.
>
> I have uploaded patch 5, in which I reverted the comment change, as you
> proposed.
> In addition to that, this patch makes following changes in gdb.base/jump.exp:
> - Remove attribute ‘debug’ from ‘prepare_for_testing’. This attribute is
> passed by default.
> - Rename ‘$inferiors’ to ‘$num_inferiors’. New name is more
> descriptive.
> - Call ‘runto_main’ once before the loop, and start the loop from ‘2’.
> Simplifies code by removing the ‘if {$inf != 1}’
> - Remove ‘gdb_exit’ at the end of the file.
>
> Patch 1: https://sourceware.org/pipermail/gdb-patches/2023-
> February/197269.html
> Comments on patch 1: https://sourceware.org/pipermail/gdb-patches/2023-
> February/197271.html
> Patch 2: https://sourceware.org/pipermail/gdb-patches/2023-
> February/197302.html
> Comments on patch 2: https://sourceware.org/pipermail/gdb-patches/2023-
> April/199055.html
> Patch 3: https://sourceware.org/pipermail/gdb-patches/2023-
> April/199095.html
> Comments on patch 3: https://sourceware.org/pipermail/gdb-patches/2023-
> May/199853.html
> Patch 4: https://sourceware.org/pipermail/gdb-patches/2023-
> June/200524.html
> Comments on patch 4: https://sourceware.org/pipermail/gdb-patches/2023-
> July/200641.html
> Patch 5: https://sourceware.org/pipermail/gdb-patches/2023-
> July/201123.html
>
> Br,
> Matti Puputti
>
> > -----Original Message-----
> > From: Andrew Burgess <aburgess@redhat.com>
> > Sent: Tuesday, July 4, 2023 4:39 PM
> > To: Puputti, Matti <matti.puputti@intel.com>; gdb-
> patches@sourceware.org
> > Subject: Re: [PATCH v4 1/1] gdb, infcmd: Support jump command in multi-
> > inferior case.
> >
> > Matti Puputti <matti.puputti@intel.com> writes:
> >
> > > Fixes the issue where jump failed if multiple inferiors run the same source.
> > >
> > > See the below example
> > >
> > > $ gdb -q ./simple
> > > Reading symbols from ./simple...
> > > (gdb) break 2
> > > Breakpoint 1 at 0x114e: file simple.c, line 2.
> > > (gdb) run
> > > Starting program: /temp/simple
> > >
> > > Breakpoint 1, main () at simple.c:2
> > > 2 int a = 42;
> > > (gdb) add-inferior
> > > [New inferior 2]
> > > Added inferior 2 on connection 1 (native)
> > > (gdb) inferior 2
> > > [Switching to inferior 2 [<null>] (<noexec>)]
> > > (gdb) info inferiors
> > > Num Description Connection Executable
> > > 1 process 6250 1 (native) /temp/simple
> > > * 2 <null> 1 (native)
> > > (gdb) file ./simple
> > > Reading symbols from ./simple...
> > > (gdb) run
> > > Starting program: /temp/simple
> > >
> > > Thread 2.1 "simple" hit Breakpoint 1, main () at simple.c:2
> > > 2 int a = 42;
> > > (gdb) info inferiors
> > > Num Description Connection Executable
> > > 1 process 6250 1 (native) /temp/simple
> > > * 2 process 6705 1 (native) /temp/simple
> > > (gdb) jump 3
> > > Unreasonable jump request
> > > (gdb)
> > >
> > > In this example, jump fails because the debugger finds two different
> > > locations, one for each inferior.
> > > Solution is to limit the search to the current program space.
> >
> > I have one last nit...
> >
> > > ---
> > > gdb/infcmd.c | 2 +-
> > > gdb/linespec.c | 3 +-
> > > gdb/linespec.h | 3 +-
> > > gdb/testsuite/gdb.base/jump.exp | 177 +++++++++++++++++-------------
> --
> > > 4 files changed, 99 insertions(+), 86 deletions(-)
> > >
> > > diff --git a/gdb/infcmd.c b/gdb/infcmd.c
> > > index 15702f84894..96c5feafb1a 100644
> > > --- a/gdb/infcmd.c
> > > +++ b/gdb/infcmd.c
> > > @@ -1068,7 +1068,7 @@ jump_command (const char *arg, int from_tty)
> > > error_no_arg (_("starting address"));
> > >
> > > std::vector<symtab_and_line> sals
> > > - = decode_line_with_last_displayed (arg, DECODE_LINE_FUNFIRSTLINE);
> > > + = decode_line_with_current_source (arg,
> DECODE_LINE_FUNFIRSTLINE);
> > > if (sals.size () != 1)
> > > {
> > > /* If multiple sal-objects were found, try dropping those that aren't
> > > diff --git a/gdb/linespec.c b/gdb/linespec.c
> > > index 7d969f37fbf..afa9eb4d3ac 100644
> > > --- a/gdb/linespec.c
> > > +++ b/gdb/linespec.c
> > > @@ -3220,7 +3220,8 @@ decode_line_with_current_source (const char
> > *string, int flags)
> > > location_spec_up locspec = string_to_location_spec (&string,
> > > current_language);
> > > std::vector<symtab_and_line> sals
> > > - = decode_line_1 (locspec.get (), flags, NULL, cursal.symtab, cursal.line);
> > > + = decode_line_1 (locspec.get (), flags, cursal.pspace, cursal.symtab,
> > > + cursal.line);
> > >
> > > if (*string)
> > > error (_("Junk at end of line specification: %s"), string);
> > > diff --git a/gdb/linespec.h b/gdb/linespec.h
> > > index d5e7334fe2d..c0eef7894d9 100644
> > > --- a/gdb/linespec.h
> > > +++ b/gdb/linespec.h
> > > @@ -139,7 +139,8 @@ extern std::vector<symtab_and_line>
> > decode_line_with_current_source
> > > (const char *, int);
> > >
> > > /* Given a string, return the line specified by it, using the last displayed
> > > - codepoint's values as defaults, or nothing if they aren't valid. */
> > > + codepoint's values as defaults, or nothing if they aren't valid.
> > > + Limit the search to given program space, if specified. */
> > >
> >
> > I think this change to the comment is now incorrect, the arguments to
> > decode_line_with_last_displayed are no longer being changed, so there is
> > never a program space specified.
> >
> > With the comment change reverted:
> >
> > Approved-By: Andrew Burgess <aburgess@redhat.com>
> >
> > Thanks,
> > Andrew
> >
> >
> > > extern std::vector<symtab_and_line> decode_line_with_last_displayed
> > > (const char *, int);
> > > diff --git a/gdb/testsuite/gdb.base/jump.exp
> > b/gdb/testsuite/gdb.base/jump.exp
> > > index 032c4a6279d..4c0d4ed2dbb 100644
> > > --- a/gdb/testsuite/gdb.base/jump.exp
> > > +++ b/gdb/testsuite/gdb.base/jump.exp
> > > @@ -18,99 +18,110 @@ clear_xfail "*-*-*"
> > >
> > > standard_testfile .c
> > >
> > > -# Build the test case
> > > -if { [gdb_compile "${srcdir}/${subdir}/${srcfile}" "${binfile}" executable
> > {debug nowarnings}] != "" } {
> > > - untested "failed to compile"
> > > - return -1
> > > - }
> > > +if {[prepare_for_testing "failed to prepare" $testfile $srcfile debug]} {
> > > + return -1
> > > +}
> > >
> > >
> > > -# Start with a fresh gdb
> > > +proc do_tests {} {
> > > + global decimal srcfile
> > > +
> > > + # Set a breakpoint on the statement that we're about to jump to.
> > > + # The statement doesn't contain a function call.
> > > + set non_call_line [gdb_get_line_number "bp-on-non-call"]
> > > + gdb_breakpoint "$non_call_line"
> > > + set bp_on_non_call \
> > > + [get_integer_valueof "\$bpnum" "INVALID" "bp_on_non_call"]
> > > +
> > > + # Can we jump to the statement? Do we stop there?
> > > + gdb_test "jump $non_call_line" \
> > > + "Breakpoint ${bp_on_non_call}(\.${decimal})?,
> > .*${srcfile}:$non_call_line.*" \
> > > + "jump to non-call"
> > > +
> > > + # Set a breakpoint on the statement that we're about to jump to.
> > > + # The statement does contain a function call.
> > > + set call_line [gdb_get_line_number "bp-on-call"]
> > > + gdb_breakpoint "$call_line"
> > > + set bp_on_call [get_integer_valueof "\$bpnum" "INVALID"
> "bp_on_call"]
> > > +
> > > + # Can we jump to the statement? Do we stop there?
> > > + gdb_test "jump $call_line" \
> > > + "Breakpoint ${bp_on_call}(\.${decimal})?, .*${srcfile}:$call_line.*" \
> > > + "jump to call"
> > > +
> > > + # If we disable the breakpoint at the function call, and then
> > > + # if we jump to that statement, do we not stop there, but at
> > > + # the following breakpoint?
> > > + gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
> > > +
> > > + gdb_test "jump $call_line" \
> > > + "Breakpoint ${bp_on_non_call}(\.${decimal})?,
> > .*${srcfile}:$non_call_line.*" \
> > > + "jump to call with disabled breakpoint"
> > > +
> > > + # Disable the breakpoint at the non-function call, so it won't hit
> > > + # if do_test is called again.
> > > + gdb_test_no_output "disable ${bp_on_non_call}" "disable
> > bp_on_non_call"
> > > +
> > > + # Verify that GDB responds gracefully to the "jump" command without
> > > + # an argument.
> > > + gdb_test "jump" "Argument required .starting address.*" \
> > > + "jump without argument disallowed"
> > > +
> > > + # Verify that GDB responds gracefully to the "jump" command with
> > > + # trailing junk.
> > > + gdb_test "jump $call_line 100" \
> > > + "malformed linespec error: unexpected number, \"100\"" \
> > > + "jump with trailing argument junk"
> > > +
> > > + # Verify that GDB responds gracefully to a request to jump out of
> > > + # the current function. (Note that this will very likely cause the
> > > + # inferior to die. Be prepared to rerun the inferior, if further
> > > + # testing is desired.)
> > > + # Try it both ways: confirming and not confirming the jump.
> > > + set out_line [gdb_get_line_number "out-of-func"]
> > > + gdb_test "jump $out_line" \
> > > + "Not confirmed.*" \
> > > + "aborted jump out of current function" \
> > > + "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> > > + "n"
> > > +
> > > + gdb_test "jump $out_line" \
> > > + "Continuing at.*" \
> > > + "jump out of current function" \
> > > + "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> > > + "y"
> > > +}
> > >
> > > -clean_restart ${binfile}
> > >
> > > -if {![runto_main]} {
> > > - return -1
> > > +set inferiors 1
> > > +if {![use_gdb_stub]} {
> > > + set inferiors 2
> > > }
> > >
> > > -# Set a breakpoint on the statement that we're about to jump to.
> > > -# The statement doesn't contain a function call.
> > > -#
> > > -set bp_on_non_call 0
> > > -set non_call_line [gdb_get_line_number "bp-on-non-call"]
> > > -gdb_test_multiple "break $non_call_line" "break before jump to non-call"
> {
> > > - -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line
> > $non_call_line.*$gdb_prompt $" {
> > > - set bp_on_non_call $expect_out(1,string)
> > > - pass "break before jump to non-call"
> > > +# Run to main, add inferiors if needed.
> > > +for {set inf 1} {$inf <= $inferiors} {incr inf} {
> > > + if {$inf != 1} {
> > > + # Start a new inferior, and run it with the same executable.
> > > + gdb_test "add-inferior -exec ${binfile}" \
> > > + "Added inferior ${inf}.*" \
> > > + "add inferior ${inf} with -exec "
> > > + gdb_test "inferior ${inf}" \
> > > + "Switching to inferior ${inf} .*" \
> > > + "switch to inferior ${inf}"
> > > + }
> > > + if {![runto_main]} {
> > > + return -1
> > > }
> > > }
> > >
> > > -# Can we jump to the statement? Do we stop there?
> > > -#
> > > -gdb_test "jump $non_call_line" "Breakpoint ${decimal},
> > .*${srcfile}:$non_call_line.*" \
> > > - "jump to non-call"
> > > -
> > > -# Set a breakpoint on the statement that we're about to jump to.
> > > -# The statement does contain a function call.
> > > -#
> > > -set bp_on_call 0
> > > -set call_line [gdb_get_line_number "bp-on-call"]
> > > -gdb_test_multiple "break $call_line" "break before jump to call" {
> > > - -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line
> > $call_line.*$gdb_prompt $" {
> > > - set bp_on_call $expect_out(1,string)
> > > - pass "break before jump to call"
> > > +# Run tests on all inferiors.
> > > +for {set inf 1} {$inf <= $inferiors} {incr inf} {
> > > + with_test_prefix "inferior $inf" {
> > > + # Switch to the target inferior.
> > > + gdb_test "inferior $inf" ".*Switching to inferior $inf .*"
> > > + # Run the tests.
> > > + do_tests
> > > }
> > > }
> > >
> > > -# Can we jump to the statement? Do we stop there?
> > > -#
> > > -gdb_test "jump $call_line" \
> > > - "Breakpoint ${decimal}, .*${srcfile}:$call_line.*" \
> > > - "jump to call"
> > > -
> > > -# If we disable the breakpoint at the function call, and then
> > > -# if we jump to that statement, do we not stop there, but at
> > > -# the following breakpoint?
> > > -#
> > > -gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
> > > -
> > > -gdb_test "jump $call_line" "Breakpoint ${decimal},
> > .*${srcfile}:$non_call_line.*" \
> > > - "jump to call with disabled breakpoint"
> > > -
> > > -# Verify that GDB responds gracefully to the "jump" command without
> > > -# an argument.
> > > -#
> > > -gdb_test "jump" "Argument required .starting address.*" \
> > > - "jump without argument disallowed"
> > > -
> > > -
> > > -# Verify that GDB responds gracefully to the "jump" command with
> > > -# trailing junk.
> > > -#
> > > -gdb_test "jump $call_line 100" \
> > > - "malformed linespec error: unexpected number, \"100\"" \
> > > - "jump with trailing argument junk"
> > > -
> > > -
> > > -# Verify that GDB responds gracefully to a request to jump out of
> > > -# the current function. (Note that this will very likely cause the
> > > -# inferior to die. Be prepared to rerun the inferior, if further
> > > -# testing is desired.)
> > > -#
> > > -# Try it both ways: confirming and not confirming the jump.
> > > -#
> > > -
> > > -set out_line [gdb_get_line_number "out-of-func"]
> > > -gdb_test "jump $out_line" \
> > > - "Not confirmed.*" \
> > > - "aborted jump out of current function" \
> > > - "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> > > - "n"
> > > -
> > > -gdb_test "jump $out_line" \
> > > - "Continuing at.*" \
> > > - "jump out of current function" \
> > > - "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> > > - "y"
> > > -
> > > gdb_exit
> > > --
> > > 2.25.1
> > >
> > > Intel Deutschland GmbH
> > > Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
> > > Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
> > > Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
> > > Chairperson of the Supervisory Board: Nicole Lau
> > > Registered Office: Munich
> > > Commercial Register: Amtsgericht Muenchen HRB 186928
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [relevance 99%]
* [PATCH v6 1/1] gdb, infcmd: Support jump command in multi-inferior case.
@ 2023-07-26 12:29 66% Matti Puputti
2023-08-16 8:58 99% ` Andrew Burgess
0 siblings, 1 reply; 34+ results
From: Matti Puputti @ 2023-07-26 12:29 UTC (permalink / raw)
To: gdb-patches, aburgess
From: "Puputti, Matti" <matti.puputti@intel.com>
Fixes the issue where jump failed if multiple inferiors run the same source.
See the below example
$ gdb -q ./simple
Reading symbols from ./simple...
(gdb) break 2
Breakpoint 1 at 0x114e: file simple.c, line 2.
(gdb) run
Starting program: /temp/simple
Breakpoint 1, main () at simple.c:2
2 int a = 42;
(gdb) add-inferior
[New inferior 2]
Added inferior 2 on connection 1 (native)
(gdb) inferior 2
[Switching to inferior 2 [<null>] (<noexec>)]
(gdb) info inferiors
Num Description Connection Executable
1 process 6250 1 (native) /temp/simple
* 2 <null> 1 (native)
(gdb) file ./simple
Reading symbols from ./simple...
(gdb) run
Starting program: /temp/simple
Thread 2.1 "simple" hit Breakpoint 1, main () at simple.c:2
2 int a = 42;
(gdb) info inferiors
Num Description Connection Executable
1 process 6250 1 (native) /temp/simple
* 2 process 6705 1 (native) /temp/simple
(gdb) jump 3
Unreasonable jump request
(gdb)
In this example, jump fails because the debugger finds two different
locations, one for each inferior.
Solution is to limit the search to the current program space.
---
gdb/infcmd.c | 2 +-
gdb/linespec.c | 3 +-
gdb/testsuite/gdb.base/jump.exp | 180 +++++++++++++++++---------------
3 files changed, 98 insertions(+), 87 deletions(-)
diff --git a/gdb/infcmd.c b/gdb/infcmd.c
index 15702f84894..96c5feafb1a 100644
--- a/gdb/infcmd.c
+++ b/gdb/infcmd.c
@@ -1068,7 +1068,7 @@ jump_command (const char *arg, int from_tty)
error_no_arg (_("starting address"));
std::vector<symtab_and_line> sals
- = decode_line_with_last_displayed (arg, DECODE_LINE_FUNFIRSTLINE);
+ = decode_line_with_current_source (arg, DECODE_LINE_FUNFIRSTLINE);
if (sals.size () != 1)
{
/* If multiple sal-objects were found, try dropping those that aren't
diff --git a/gdb/linespec.c b/gdb/linespec.c
index 7d969f37fbf..afa9eb4d3ac 100644
--- a/gdb/linespec.c
+++ b/gdb/linespec.c
@@ -3220,7 +3220,8 @@ decode_line_with_current_source (const char *string, int flags)
location_spec_up locspec = string_to_location_spec (&string,
current_language);
std::vector<symtab_and_line> sals
- = decode_line_1 (locspec.get (), flags, NULL, cursal.symtab, cursal.line);
+ = decode_line_1 (locspec.get (), flags, cursal.pspace, cursal.symtab,
+ cursal.line);
if (*string)
error (_("Junk at end of line specification: %s"), string);
diff --git a/gdb/testsuite/gdb.base/jump.exp b/gdb/testsuite/gdb.base/jump.exp
index 032c4a6279d..77765db15ef 100644
--- a/gdb/testsuite/gdb.base/jump.exp
+++ b/gdb/testsuite/gdb.base/jump.exp
@@ -18,99 +18,109 @@ clear_xfail "*-*-*"
standard_testfile .c
-# Build the test case
-if { [gdb_compile "${srcdir}/${subdir}/${srcfile}" "${binfile}" executable {debug nowarnings}] != "" } {
- untested "failed to compile"
- return -1
- }
+if {[prepare_for_testing "failed to prepare" $testfile $srcfile]} {
+ return -1
+}
-# Start with a fresh gdb
+proc do_tests {} {
+ global decimal srcfile
+
+ # Set a breakpoint on the statement that we're about to jump to.
+ # The statement doesn't contain a function call.
+ set non_call_line [gdb_get_line_number "bp-on-non-call"]
+ gdb_breakpoint "$non_call_line"
+ set bp_on_non_call \
+ [get_integer_valueof "\$bpnum" "INVALID" "bp_on_non_call"]
+
+ # Can we jump to the statement? Do we stop there?
+ gdb_test "jump $non_call_line" \
+ "Breakpoint ${bp_on_non_call}(\.${decimal})?, .*${srcfile}:$non_call_line.*" \
+ "jump to non-call"
+
+ # Set a breakpoint on the statement that we're about to jump to.
+ # The statement does contain a function call.
+ set call_line [gdb_get_line_number "bp-on-call"]
+ gdb_breakpoint "$call_line"
+ set bp_on_call [get_integer_valueof "\$bpnum" "INVALID" "bp_on_call"]
+
+ # Can we jump to the statement? Do we stop there?
+ gdb_test "jump $call_line" \
+ "Breakpoint ${bp_on_call}(\.${decimal})?, .*${srcfile}:$call_line.*" \
+ "jump to call"
+
+ # If we disable the breakpoint at the function call, and then
+ # if we jump to that statement, do we not stop there, but at
+ # the following breakpoint?
+ gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
+
+ gdb_test "jump $call_line" \
+ "Breakpoint ${bp_on_non_call}(\.${decimal})?, .*${srcfile}:$non_call_line.*" \
+ "jump to call with disabled breakpoint"
+
+ # Disable the breakpoint at the non-function call, so it won't hit
+ # if do_test is called again.
+ gdb_test_no_output "disable ${bp_on_non_call}" "disable bp_on_non_call"
+
+ # Verify that GDB responds gracefully to the "jump" command without
+ # an argument.
+ gdb_test "jump" "Argument required .starting address.*" \
+ "jump without argument disallowed"
+
+ # Verify that GDB responds gracefully to the "jump" command with
+ # trailing junk.
+ gdb_test "jump $call_line 100" \
+ "malformed linespec error: unexpected number, \"100\"" \
+ "jump with trailing argument junk"
+
+ # Verify that GDB responds gracefully to a request to jump out of
+ # the current function. (Note that this will very likely cause the
+ # inferior to die. Be prepared to rerun the inferior, if further
+ # testing is desired.)
+ # Try it both ways: confirming and not confirming the jump.
+ set out_line [gdb_get_line_number "out-of-func"]
+ gdb_test "jump $out_line" \
+ "Not confirmed.*" \
+ "aborted jump out of current function" \
+ "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
+ "n"
+
+ gdb_test "jump $out_line" \
+ "Continuing at.*" \
+ "jump out of current function" \
+ "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
+ "y"
+}
-clean_restart ${binfile}
-if {![runto_main]} {
- return -1
+set num_inferiors 1
+if {![use_gdb_stub]} {
+ set num_inferiors 2
}
-# Set a breakpoint on the statement that we're about to jump to.
-# The statement doesn't contain a function call.
-#
-set bp_on_non_call 0
-set non_call_line [gdb_get_line_number "bp-on-non-call"]
-gdb_test_multiple "break $non_call_line" "break before jump to non-call" {
- -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line $non_call_line.*$gdb_prompt $" {
- set bp_on_non_call $expect_out(1,string)
- pass "break before jump to non-call"
+# Run to main, add inferiors if needed.
+if {![runto_main]} {
+ return -1
+}
+for {set inf 2} {$inf <= $num_inferiors} {incr inf} {
+ # Start a new inferior, and run it with the same executable.
+ gdb_test "add-inferior -exec ${binfile}" \
+ "Added inferior ${inf}.*" \
+ "add inferior ${inf} with -exec "
+ gdb_test "inferior ${inf}" \
+ "Switching to inferior ${inf} .*" \
+ "switch to inferior ${inf}"
+ if {![runto_main]} {
+ return -1
}
}
-# Can we jump to the statement? Do we stop there?
-#
-gdb_test "jump $non_call_line" "Breakpoint ${decimal}, .*${srcfile}:$non_call_line.*" \
- "jump to non-call"
-
-# Set a breakpoint on the statement that we're about to jump to.
-# The statement does contain a function call.
-#
-set bp_on_call 0
-set call_line [gdb_get_line_number "bp-on-call"]
-gdb_test_multiple "break $call_line" "break before jump to call" {
- -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line $call_line.*$gdb_prompt $" {
- set bp_on_call $expect_out(1,string)
- pass "break before jump to call"
+# Run tests on all inferiors.
+for {set inf 1} {$inf <= $num_inferiors} {incr inf} {
+ with_test_prefix "inferior $inf" {
+ # Switch to the target inferior.
+ gdb_test "inferior $inf" ".*Switching to inferior $inf .*"
+ # Run the tests.
+ do_tests
}
}
-
-# Can we jump to the statement? Do we stop there?
-#
-gdb_test "jump $call_line" \
- "Breakpoint ${decimal}, .*${srcfile}:$call_line.*" \
- "jump to call"
-
-# If we disable the breakpoint at the function call, and then
-# if we jump to that statement, do we not stop there, but at
-# the following breakpoint?
-#
-gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
-
-gdb_test "jump $call_line" "Breakpoint ${decimal}, .*${srcfile}:$non_call_line.*" \
- "jump to call with disabled breakpoint"
-
-# Verify that GDB responds gracefully to the "jump" command without
-# an argument.
-#
-gdb_test "jump" "Argument required .starting address.*" \
- "jump without argument disallowed"
-
-
-# Verify that GDB responds gracefully to the "jump" command with
-# trailing junk.
-#
-gdb_test "jump $call_line 100" \
- "malformed linespec error: unexpected number, \"100\"" \
- "jump with trailing argument junk"
-
-
-# Verify that GDB responds gracefully to a request to jump out of
-# the current function. (Note that this will very likely cause the
-# inferior to die. Be prepared to rerun the inferior, if further
-# testing is desired.)
-#
-# Try it both ways: confirming and not confirming the jump.
-#
-
-set out_line [gdb_get_line_number "out-of-func"]
-gdb_test "jump $out_line" \
- "Not confirmed.*" \
- "aborted jump out of current function" \
- "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
- "n"
-
-gdb_test "jump $out_line" \
- "Continuing at.*" \
- "jump out of current function" \
- "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
- "y"
-
-gdb_exit
--
2.25.1
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [relevance 66%]
* RE: [PATCH v4 1/1] gdb, infcmd: Support jump command in multi-inferior case.
2023-07-04 14:38 99% ` Andrew Burgess
@ 2023-07-26 11:51 99% ` Puputti, Matti
2023-07-26 12:32 99% ` Puputti, Matti
0 siblings, 1 reply; 34+ results
From: Puputti, Matti @ 2023-07-26 11:51 UTC (permalink / raw)
To: Andrew Burgess, gdb-patches
Hi Andrew,
I am sorry to take such a long time to react on your latest comments.
I have uploaded patch 5, in which I reverted the comment change, as you proposed.
In addition to that, this patch makes following changes in gdb.base/jump.exp:
- Remove attribute ‘debug’ from ‘prepare_for_testing’. This attribute is passed by default.
- Rename ‘$inferiors’ to ‘$num_inferiors’. New name is more descriptive.
- Call ‘runto_main’ once before the loop, and start the loop from ‘2’. Simplifies code by removing the ‘if {$inf != 1}’
- Remove ‘gdb_exit’ at the end of the file.
Patch 1: https://sourceware.org/pipermail/gdb-patches/2023-February/197269.html
Comments on patch 1: https://sourceware.org/pipermail/gdb-patches/2023-February/197271.html
Patch 2: https://sourceware.org/pipermail/gdb-patches/2023-February/197302.html
Comments on patch 2: https://sourceware.org/pipermail/gdb-patches/2023-April/199055.html
Patch 3: https://sourceware.org/pipermail/gdb-patches/2023-April/199095.html
Comments on patch 3: https://sourceware.org/pipermail/gdb-patches/2023-May/199853.html
Patch 4: https://sourceware.org/pipermail/gdb-patches/2023-June/200524.html
Comments on patch 4: https://sourceware.org/pipermail/gdb-patches/2023-July/200641.html
Patch 5: https://sourceware.org/pipermail/gdb-patches/2023-July/201123.html
Br,
Matti Puputti
> -----Original Message-----
> From: Andrew Burgess <aburgess@redhat.com>
> Sent: Tuesday, July 4, 2023 4:39 PM
> To: Puputti, Matti <matti.puputti@intel.com>; gdb-patches@sourceware.org
> Subject: Re: [PATCH v4 1/1] gdb, infcmd: Support jump command in multi-
> inferior case.
>
> Matti Puputti <matti.puputti@intel.com> writes:
>
> > Fixes the issue where jump failed if multiple inferiors run the same source.
> >
> > See the below example
> >
> > $ gdb -q ./simple
> > Reading symbols from ./simple...
> > (gdb) break 2
> > Breakpoint 1 at 0x114e: file simple.c, line 2.
> > (gdb) run
> > Starting program: /temp/simple
> >
> > Breakpoint 1, main () at simple.c:2
> > 2 int a = 42;
> > (gdb) add-inferior
> > [New inferior 2]
> > Added inferior 2 on connection 1 (native)
> > (gdb) inferior 2
> > [Switching to inferior 2 [<null>] (<noexec>)]
> > (gdb) info inferiors
> > Num Description Connection Executable
> > 1 process 6250 1 (native) /temp/simple
> > * 2 <null> 1 (native)
> > (gdb) file ./simple
> > Reading symbols from ./simple...
> > (gdb) run
> > Starting program: /temp/simple
> >
> > Thread 2.1 "simple" hit Breakpoint 1, main () at simple.c:2
> > 2 int a = 42;
> > (gdb) info inferiors
> > Num Description Connection Executable
> > 1 process 6250 1 (native) /temp/simple
> > * 2 process 6705 1 (native) /temp/simple
> > (gdb) jump 3
> > Unreasonable jump request
> > (gdb)
> >
> > In this example, jump fails because the debugger finds two different
> > locations, one for each inferior.
> > Solution is to limit the search to the current program space.
>
> I have one last nit...
>
> > ---
> > gdb/infcmd.c | 2 +-
> > gdb/linespec.c | 3 +-
> > gdb/linespec.h | 3 +-
> > gdb/testsuite/gdb.base/jump.exp | 177 +++++++++++++++++---------------
> > 4 files changed, 99 insertions(+), 86 deletions(-)
> >
> > diff --git a/gdb/infcmd.c b/gdb/infcmd.c
> > index 15702f84894..96c5feafb1a 100644
> > --- a/gdb/infcmd.c
> > +++ b/gdb/infcmd.c
> > @@ -1068,7 +1068,7 @@ jump_command (const char *arg, int from_tty)
> > error_no_arg (_("starting address"));
> >
> > std::vector<symtab_and_line> sals
> > - = decode_line_with_last_displayed (arg, DECODE_LINE_FUNFIRSTLINE);
> > + = decode_line_with_current_source (arg, DECODE_LINE_FUNFIRSTLINE);
> > if (sals.size () != 1)
> > {
> > /* If multiple sal-objects were found, try dropping those that aren't
> > diff --git a/gdb/linespec.c b/gdb/linespec.c
> > index 7d969f37fbf..afa9eb4d3ac 100644
> > --- a/gdb/linespec.c
> > +++ b/gdb/linespec.c
> > @@ -3220,7 +3220,8 @@ decode_line_with_current_source (const char
> *string, int flags)
> > location_spec_up locspec = string_to_location_spec (&string,
> > current_language);
> > std::vector<symtab_and_line> sals
> > - = decode_line_1 (locspec.get (), flags, NULL, cursal.symtab, cursal.line);
> > + = decode_line_1 (locspec.get (), flags, cursal.pspace, cursal.symtab,
> > + cursal.line);
> >
> > if (*string)
> > error (_("Junk at end of line specification: %s"), string);
> > diff --git a/gdb/linespec.h b/gdb/linespec.h
> > index d5e7334fe2d..c0eef7894d9 100644
> > --- a/gdb/linespec.h
> > +++ b/gdb/linespec.h
> > @@ -139,7 +139,8 @@ extern std::vector<symtab_and_line>
> decode_line_with_current_source
> > (const char *, int);
> >
> > /* Given a string, return the line specified by it, using the last displayed
> > - codepoint's values as defaults, or nothing if they aren't valid. */
> > + codepoint's values as defaults, or nothing if they aren't valid.
> > + Limit the search to given program space, if specified. */
> >
>
> I think this change to the comment is now incorrect, the arguments to
> decode_line_with_last_displayed are no longer being changed, so there is
> never a program space specified.
>
> With the comment change reverted:
>
> Approved-By: Andrew Burgess <aburgess@redhat.com>
>
> Thanks,
> Andrew
>
>
> > extern std::vector<symtab_and_line> decode_line_with_last_displayed
> > (const char *, int);
> > diff --git a/gdb/testsuite/gdb.base/jump.exp
> b/gdb/testsuite/gdb.base/jump.exp
> > index 032c4a6279d..4c0d4ed2dbb 100644
> > --- a/gdb/testsuite/gdb.base/jump.exp
> > +++ b/gdb/testsuite/gdb.base/jump.exp
> > @@ -18,99 +18,110 @@ clear_xfail "*-*-*"
> >
> > standard_testfile .c
> >
> > -# Build the test case
> > -if { [gdb_compile "${srcdir}/${subdir}/${srcfile}" "${binfile}" executable
> {debug nowarnings}] != "" } {
> > - untested "failed to compile"
> > - return -1
> > - }
> > +if {[prepare_for_testing "failed to prepare" $testfile $srcfile debug]} {
> > + return -1
> > +}
> >
> >
> > -# Start with a fresh gdb
> > +proc do_tests {} {
> > + global decimal srcfile
> > +
> > + # Set a breakpoint on the statement that we're about to jump to.
> > + # The statement doesn't contain a function call.
> > + set non_call_line [gdb_get_line_number "bp-on-non-call"]
> > + gdb_breakpoint "$non_call_line"
> > + set bp_on_non_call \
> > + [get_integer_valueof "\$bpnum" "INVALID" "bp_on_non_call"]
> > +
> > + # Can we jump to the statement? Do we stop there?
> > + gdb_test "jump $non_call_line" \
> > + "Breakpoint ${bp_on_non_call}(\.${decimal})?,
> .*${srcfile}:$non_call_line.*" \
> > + "jump to non-call"
> > +
> > + # Set a breakpoint on the statement that we're about to jump to.
> > + # The statement does contain a function call.
> > + set call_line [gdb_get_line_number "bp-on-call"]
> > + gdb_breakpoint "$call_line"
> > + set bp_on_call [get_integer_valueof "\$bpnum" "INVALID" "bp_on_call"]
> > +
> > + # Can we jump to the statement? Do we stop there?
> > + gdb_test "jump $call_line" \
> > + "Breakpoint ${bp_on_call}(\.${decimal})?, .*${srcfile}:$call_line.*" \
> > + "jump to call"
> > +
> > + # If we disable the breakpoint at the function call, and then
> > + # if we jump to that statement, do we not stop there, but at
> > + # the following breakpoint?
> > + gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
> > +
> > + gdb_test "jump $call_line" \
> > + "Breakpoint ${bp_on_non_call}(\.${decimal})?,
> .*${srcfile}:$non_call_line.*" \
> > + "jump to call with disabled breakpoint"
> > +
> > + # Disable the breakpoint at the non-function call, so it won't hit
> > + # if do_test is called again.
> > + gdb_test_no_output "disable ${bp_on_non_call}" "disable
> bp_on_non_call"
> > +
> > + # Verify that GDB responds gracefully to the "jump" command without
> > + # an argument.
> > + gdb_test "jump" "Argument required .starting address.*" \
> > + "jump without argument disallowed"
> > +
> > + # Verify that GDB responds gracefully to the "jump" command with
> > + # trailing junk.
> > + gdb_test "jump $call_line 100" \
> > + "malformed linespec error: unexpected number, \"100\"" \
> > + "jump with trailing argument junk"
> > +
> > + # Verify that GDB responds gracefully to a request to jump out of
> > + # the current function. (Note that this will very likely cause the
> > + # inferior to die. Be prepared to rerun the inferior, if further
> > + # testing is desired.)
> > + # Try it both ways: confirming and not confirming the jump.
> > + set out_line [gdb_get_line_number "out-of-func"]
> > + gdb_test "jump $out_line" \
> > + "Not confirmed.*" \
> > + "aborted jump out of current function" \
> > + "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> > + "n"
> > +
> > + gdb_test "jump $out_line" \
> > + "Continuing at.*" \
> > + "jump out of current function" \
> > + "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> > + "y"
> > +}
> >
> > -clean_restart ${binfile}
> >
> > -if {![runto_main]} {
> > - return -1
> > +set inferiors 1
> > +if {![use_gdb_stub]} {
> > + set inferiors 2
> > }
> >
> > -# Set a breakpoint on the statement that we're about to jump to.
> > -# The statement doesn't contain a function call.
> > -#
> > -set bp_on_non_call 0
> > -set non_call_line [gdb_get_line_number "bp-on-non-call"]
> > -gdb_test_multiple "break $non_call_line" "break before jump to non-call" {
> > - -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line
> $non_call_line.*$gdb_prompt $" {
> > - set bp_on_non_call $expect_out(1,string)
> > - pass "break before jump to non-call"
> > +# Run to main, add inferiors if needed.
> > +for {set inf 1} {$inf <= $inferiors} {incr inf} {
> > + if {$inf != 1} {
> > + # Start a new inferior, and run it with the same executable.
> > + gdb_test "add-inferior -exec ${binfile}" \
> > + "Added inferior ${inf}.*" \
> > + "add inferior ${inf} with -exec "
> > + gdb_test "inferior ${inf}" \
> > + "Switching to inferior ${inf} .*" \
> > + "switch to inferior ${inf}"
> > + }
> > + if {![runto_main]} {
> > + return -1
> > }
> > }
> >
> > -# Can we jump to the statement? Do we stop there?
> > -#
> > -gdb_test "jump $non_call_line" "Breakpoint ${decimal},
> .*${srcfile}:$non_call_line.*" \
> > - "jump to non-call"
> > -
> > -# Set a breakpoint on the statement that we're about to jump to.
> > -# The statement does contain a function call.
> > -#
> > -set bp_on_call 0
> > -set call_line [gdb_get_line_number "bp-on-call"]
> > -gdb_test_multiple "break $call_line" "break before jump to call" {
> > - -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line
> $call_line.*$gdb_prompt $" {
> > - set bp_on_call $expect_out(1,string)
> > - pass "break before jump to call"
> > +# Run tests on all inferiors.
> > +for {set inf 1} {$inf <= $inferiors} {incr inf} {
> > + with_test_prefix "inferior $inf" {
> > + # Switch to the target inferior.
> > + gdb_test "inferior $inf" ".*Switching to inferior $inf .*"
> > + # Run the tests.
> > + do_tests
> > }
> > }
> >
> > -# Can we jump to the statement? Do we stop there?
> > -#
> > -gdb_test "jump $call_line" \
> > - "Breakpoint ${decimal}, .*${srcfile}:$call_line.*" \
> > - "jump to call"
> > -
> > -# If we disable the breakpoint at the function call, and then
> > -# if we jump to that statement, do we not stop there, but at
> > -# the following breakpoint?
> > -#
> > -gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
> > -
> > -gdb_test "jump $call_line" "Breakpoint ${decimal},
> .*${srcfile}:$non_call_line.*" \
> > - "jump to call with disabled breakpoint"
> > -
> > -# Verify that GDB responds gracefully to the "jump" command without
> > -# an argument.
> > -#
> > -gdb_test "jump" "Argument required .starting address.*" \
> > - "jump without argument disallowed"
> > -
> > -
> > -# Verify that GDB responds gracefully to the "jump" command with
> > -# trailing junk.
> > -#
> > -gdb_test "jump $call_line 100" \
> > - "malformed linespec error: unexpected number, \"100\"" \
> > - "jump with trailing argument junk"
> > -
> > -
> > -# Verify that GDB responds gracefully to a request to jump out of
> > -# the current function. (Note that this will very likely cause the
> > -# inferior to die. Be prepared to rerun the inferior, if further
> > -# testing is desired.)
> > -#
> > -# Try it both ways: confirming and not confirming the jump.
> > -#
> > -
> > -set out_line [gdb_get_line_number "out-of-func"]
> > -gdb_test "jump $out_line" \
> > - "Not confirmed.*" \
> > - "aborted jump out of current function" \
> > - "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> > - "n"
> > -
> > -gdb_test "jump $out_line" \
> > - "Continuing at.*" \
> > - "jump out of current function" \
> > - "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> > - "y"
> > -
> > gdb_exit
> > --
> > 2.25.1
> >
> > Intel Deutschland GmbH
> > Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
> > Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
> > Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
> > Chairperson of the Supervisory Board: Nicole Lau
> > Registered Office: Munich
> > Commercial Register: Amtsgericht Muenchen HRB 186928
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [relevance 99%]
* Re: [PATCH v4 1/1] gdb, infcmd: Support jump command in multi-inferior case.
2023-06-28 9:00 65% [PATCH v4 " Matti Puputti
@ 2023-07-04 14:38 99% ` Andrew Burgess
2023-07-26 11:51 99% ` Puputti, Matti
0 siblings, 1 reply; 34+ results
From: Andrew Burgess @ 2023-07-04 14:38 UTC (permalink / raw)
To: Matti Puputti, gdb-patches
Matti Puputti <matti.puputti@intel.com> writes:
> Fixes the issue where jump failed if multiple inferiors run the same source.
>
> See the below example
>
> $ gdb -q ./simple
> Reading symbols from ./simple...
> (gdb) break 2
> Breakpoint 1 at 0x114e: file simple.c, line 2.
> (gdb) run
> Starting program: /temp/simple
>
> Breakpoint 1, main () at simple.c:2
> 2 int a = 42;
> (gdb) add-inferior
> [New inferior 2]
> Added inferior 2 on connection 1 (native)
> (gdb) inferior 2
> [Switching to inferior 2 [<null>] (<noexec>)]
> (gdb) info inferiors
> Num Description Connection Executable
> 1 process 6250 1 (native) /temp/simple
> * 2 <null> 1 (native)
> (gdb) file ./simple
> Reading symbols from ./simple...
> (gdb) run
> Starting program: /temp/simple
>
> Thread 2.1 "simple" hit Breakpoint 1, main () at simple.c:2
> 2 int a = 42;
> (gdb) info inferiors
> Num Description Connection Executable
> 1 process 6250 1 (native) /temp/simple
> * 2 process 6705 1 (native) /temp/simple
> (gdb) jump 3
> Unreasonable jump request
> (gdb)
>
> In this example, jump fails because the debugger finds two different
> locations, one for each inferior.
> Solution is to limit the search to the current program space.
I have one last nit...
> ---
> gdb/infcmd.c | 2 +-
> gdb/linespec.c | 3 +-
> gdb/linespec.h | 3 +-
> gdb/testsuite/gdb.base/jump.exp | 177 +++++++++++++++++---------------
> 4 files changed, 99 insertions(+), 86 deletions(-)
>
> diff --git a/gdb/infcmd.c b/gdb/infcmd.c
> index 15702f84894..96c5feafb1a 100644
> --- a/gdb/infcmd.c
> +++ b/gdb/infcmd.c
> @@ -1068,7 +1068,7 @@ jump_command (const char *arg, int from_tty)
> error_no_arg (_("starting address"));
>
> std::vector<symtab_and_line> sals
> - = decode_line_with_last_displayed (arg, DECODE_LINE_FUNFIRSTLINE);
> + = decode_line_with_current_source (arg, DECODE_LINE_FUNFIRSTLINE);
> if (sals.size () != 1)
> {
> /* If multiple sal-objects were found, try dropping those that aren't
> diff --git a/gdb/linespec.c b/gdb/linespec.c
> index 7d969f37fbf..afa9eb4d3ac 100644
> --- a/gdb/linespec.c
> +++ b/gdb/linespec.c
> @@ -3220,7 +3220,8 @@ decode_line_with_current_source (const char *string, int flags)
> location_spec_up locspec = string_to_location_spec (&string,
> current_language);
> std::vector<symtab_and_line> sals
> - = decode_line_1 (locspec.get (), flags, NULL, cursal.symtab, cursal.line);
> + = decode_line_1 (locspec.get (), flags, cursal.pspace, cursal.symtab,
> + cursal.line);
>
> if (*string)
> error (_("Junk at end of line specification: %s"), string);
> diff --git a/gdb/linespec.h b/gdb/linespec.h
> index d5e7334fe2d..c0eef7894d9 100644
> --- a/gdb/linespec.h
> +++ b/gdb/linespec.h
> @@ -139,7 +139,8 @@ extern std::vector<symtab_and_line> decode_line_with_current_source
> (const char *, int);
>
> /* Given a string, return the line specified by it, using the last displayed
> - codepoint's values as defaults, or nothing if they aren't valid. */
> + codepoint's values as defaults, or nothing if they aren't valid.
> + Limit the search to given program space, if specified. */
>
I think this change to the comment is now incorrect, the arguments to
decode_line_with_last_displayed are no longer being changed, so there is
never a program space specified.
With the comment change reverted:
Approved-By: Andrew Burgess <aburgess@redhat.com>
Thanks,
Andrew
> extern std::vector<symtab_and_line> decode_line_with_last_displayed
> (const char *, int);
> diff --git a/gdb/testsuite/gdb.base/jump.exp b/gdb/testsuite/gdb.base/jump.exp
> index 032c4a6279d..4c0d4ed2dbb 100644
> --- a/gdb/testsuite/gdb.base/jump.exp
> +++ b/gdb/testsuite/gdb.base/jump.exp
> @@ -18,99 +18,110 @@ clear_xfail "*-*-*"
>
> standard_testfile .c
>
> -# Build the test case
> -if { [gdb_compile "${srcdir}/${subdir}/${srcfile}" "${binfile}" executable {debug nowarnings}] != "" } {
> - untested "failed to compile"
> - return -1
> - }
> +if {[prepare_for_testing "failed to prepare" $testfile $srcfile debug]} {
> + return -1
> +}
>
>
> -# Start with a fresh gdb
> +proc do_tests {} {
> + global decimal srcfile
> +
> + # Set a breakpoint on the statement that we're about to jump to.
> + # The statement doesn't contain a function call.
> + set non_call_line [gdb_get_line_number "bp-on-non-call"]
> + gdb_breakpoint "$non_call_line"
> + set bp_on_non_call \
> + [get_integer_valueof "\$bpnum" "INVALID" "bp_on_non_call"]
> +
> + # Can we jump to the statement? Do we stop there?
> + gdb_test "jump $non_call_line" \
> + "Breakpoint ${bp_on_non_call}(\.${decimal})?, .*${srcfile}:$non_call_line.*" \
> + "jump to non-call"
> +
> + # Set a breakpoint on the statement that we're about to jump to.
> + # The statement does contain a function call.
> + set call_line [gdb_get_line_number "bp-on-call"]
> + gdb_breakpoint "$call_line"
> + set bp_on_call [get_integer_valueof "\$bpnum" "INVALID" "bp_on_call"]
> +
> + # Can we jump to the statement? Do we stop there?
> + gdb_test "jump $call_line" \
> + "Breakpoint ${bp_on_call}(\.${decimal})?, .*${srcfile}:$call_line.*" \
> + "jump to call"
> +
> + # If we disable the breakpoint at the function call, and then
> + # if we jump to that statement, do we not stop there, but at
> + # the following breakpoint?
> + gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
> +
> + gdb_test "jump $call_line" \
> + "Breakpoint ${bp_on_non_call}(\.${decimal})?, .*${srcfile}:$non_call_line.*" \
> + "jump to call with disabled breakpoint"
> +
> + # Disable the breakpoint at the non-function call, so it won't hit
> + # if do_test is called again.
> + gdb_test_no_output "disable ${bp_on_non_call}" "disable bp_on_non_call"
> +
> + # Verify that GDB responds gracefully to the "jump" command without
> + # an argument.
> + gdb_test "jump" "Argument required .starting address.*" \
> + "jump without argument disallowed"
> +
> + # Verify that GDB responds gracefully to the "jump" command with
> + # trailing junk.
> + gdb_test "jump $call_line 100" \
> + "malformed linespec error: unexpected number, \"100\"" \
> + "jump with trailing argument junk"
> +
> + # Verify that GDB responds gracefully to a request to jump out of
> + # the current function. (Note that this will very likely cause the
> + # inferior to die. Be prepared to rerun the inferior, if further
> + # testing is desired.)
> + # Try it both ways: confirming and not confirming the jump.
> + set out_line [gdb_get_line_number "out-of-func"]
> + gdb_test "jump $out_line" \
> + "Not confirmed.*" \
> + "aborted jump out of current function" \
> + "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> + "n"
> +
> + gdb_test "jump $out_line" \
> + "Continuing at.*" \
> + "jump out of current function" \
> + "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> + "y"
> +}
>
> -clean_restart ${binfile}
>
> -if {![runto_main]} {
> - return -1
> +set inferiors 1
> +if {![use_gdb_stub]} {
> + set inferiors 2
> }
>
> -# Set a breakpoint on the statement that we're about to jump to.
> -# The statement doesn't contain a function call.
> -#
> -set bp_on_non_call 0
> -set non_call_line [gdb_get_line_number "bp-on-non-call"]
> -gdb_test_multiple "break $non_call_line" "break before jump to non-call" {
> - -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line $non_call_line.*$gdb_prompt $" {
> - set bp_on_non_call $expect_out(1,string)
> - pass "break before jump to non-call"
> +# Run to main, add inferiors if needed.
> +for {set inf 1} {$inf <= $inferiors} {incr inf} {
> + if {$inf != 1} {
> + # Start a new inferior, and run it with the same executable.
> + gdb_test "add-inferior -exec ${binfile}" \
> + "Added inferior ${inf}.*" \
> + "add inferior ${inf} with -exec "
> + gdb_test "inferior ${inf}" \
> + "Switching to inferior ${inf} .*" \
> + "switch to inferior ${inf}"
> + }
> + if {![runto_main]} {
> + return -1
> }
> }
>
> -# Can we jump to the statement? Do we stop there?
> -#
> -gdb_test "jump $non_call_line" "Breakpoint ${decimal}, .*${srcfile}:$non_call_line.*" \
> - "jump to non-call"
> -
> -# Set a breakpoint on the statement that we're about to jump to.
> -# The statement does contain a function call.
> -#
> -set bp_on_call 0
> -set call_line [gdb_get_line_number "bp-on-call"]
> -gdb_test_multiple "break $call_line" "break before jump to call" {
> - -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line $call_line.*$gdb_prompt $" {
> - set bp_on_call $expect_out(1,string)
> - pass "break before jump to call"
> +# Run tests on all inferiors.
> +for {set inf 1} {$inf <= $inferiors} {incr inf} {
> + with_test_prefix "inferior $inf" {
> + # Switch to the target inferior.
> + gdb_test "inferior $inf" ".*Switching to inferior $inf .*"
> + # Run the tests.
> + do_tests
> }
> }
>
> -# Can we jump to the statement? Do we stop there?
> -#
> -gdb_test "jump $call_line" \
> - "Breakpoint ${decimal}, .*${srcfile}:$call_line.*" \
> - "jump to call"
> -
> -# If we disable the breakpoint at the function call, and then
> -# if we jump to that statement, do we not stop there, but at
> -# the following breakpoint?
> -#
> -gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
> -
> -gdb_test "jump $call_line" "Breakpoint ${decimal}, .*${srcfile}:$non_call_line.*" \
> - "jump to call with disabled breakpoint"
> -
> -# Verify that GDB responds gracefully to the "jump" command without
> -# an argument.
> -#
> -gdb_test "jump" "Argument required .starting address.*" \
> - "jump without argument disallowed"
> -
> -
> -# Verify that GDB responds gracefully to the "jump" command with
> -# trailing junk.
> -#
> -gdb_test "jump $call_line 100" \
> - "malformed linespec error: unexpected number, \"100\"" \
> - "jump with trailing argument junk"
> -
> -
> -# Verify that GDB responds gracefully to a request to jump out of
> -# the current function. (Note that this will very likely cause the
> -# inferior to die. Be prepared to rerun the inferior, if further
> -# testing is desired.)
> -#
> -# Try it both ways: confirming and not confirming the jump.
> -#
> -
> -set out_line [gdb_get_line_number "out-of-func"]
> -gdb_test "jump $out_line" \
> - "Not confirmed.*" \
> - "aborted jump out of current function" \
> - "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> - "n"
> -
> -gdb_test "jump $out_line" \
> - "Continuing at.*" \
> - "jump out of current function" \
> - "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> - "y"
> -
> gdb_exit
> --
> 2.25.1
>
> Intel Deutschland GmbH
> Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
> Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
> Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
> Chairperson of the Supervisory Board: Nicole Lau
> Registered Office: Munich
> Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [relevance 99%]
* RE: [PATCH v3 1/1] gdb, infcmd: Support jump command in multi-inferior case.
2023-05-24 17:33 91% ` [PATCH " Andrew Burgess
2023-05-25 5:38 99% ` Puputti, Matti
@ 2023-06-28 9:04 99% ` Puputti, Matti
1 sibling, 0 replies; 34+ results
From: Puputti, Matti @ 2023-06-28 9:04 UTC (permalink / raw)
To: Andrew Burgess, gdb-patches
Hi Andrew,
Finally back from my vocation, and coming back to this patch.
I just uploaded patch 4, which adopts all the changes you proposed.
You proposed to call decode_line_with_current_source() instead of decode_line_with_last_displayed().
As you seem to say, it may be matter of taste, which one should be called.
Such change may impact some use-cases - corner cases, one could say.
But there doesn't seem to be any testcase relaying on jump to be relative to what have been listed.
And in the end, it makes sense that jump is relative to the current location.
Patch 1: https://sourceware.org/pipermail/gdb-patches/2023-February/197269.html
Comments on patch 1: https://sourceware.org/pipermail/gdb-patches/2023-February/197271.html
Patch 2: https://sourceware.org/pipermail/gdb-patches/2023-February/197302.html
Comments on patch 2: https://sourceware.org/pipermail/gdb-patches/2023-April/199055.html
Patch 3: https://sourceware.org/pipermail/gdb-patches/2023-April/199095.html
Comments on patch 3: https://sourceware.org/pipermail/gdb-patches/2023-May/199853.html
Patch 4: https://sourceware.org/pipermail/gdb-patches/2023-June/200524.html
Br,
Matti Puputti
> -----Original Message-----
> From: Andrew Burgess <aburgess@redhat.com>
> Sent: Wednesday, May 24, 2023 7:33 PM
> To: Puputti, Matti <matti.puputti@intel.com>; gdb-patches@sourceware.org
> Subject: Re: [PATCH v3 1/1] gdb, infcmd: Support jump command in multi-
> inferior case.
>
> Matti Puputti <matti.puputti@intel.com> writes:
>
> > Fixes the issue where jump failed if multiple inferiors run the same source.
> >
> > See the below example
> >
> > $ gdb -q ./simple
> > Reading symbols from ./simple...
> > (gdb) break 2
> > Breakpoint 1 at 0x114e: file simple.c, line 2.
> > (gdb) run
> > Starting program: /temp/simple
> >
> > Breakpoint 1, main () at simple.c:2
> > 2 int a = 42;
> > (gdb) add-inferior
> > [New inferior 2]
> > Added inferior 2 on connection 1 (native)
> > (gdb) inferior 2
> > [Switching to inferior 2 [<null>] (<noexec>)]
> > (gdb) info inferiors
> > Num Description Connection Executable
> > 1 process 6250 1 (native) /temp/simple
> > * 2 <null> 1 (native)
> > (gdb) file ./simple
> > Reading symbols from ./simple...
> > (gdb) run
> > Starting program: /temp/simple
> >
> > Thread 2.1 "simple" hit Breakpoint 1, main () at simple.c:2
> > 2 int a = 42;
> > (gdb) info inferiors
> > Num Description Connection Executable
> > 1 process 6250 1 (native) /temp/simple
> > * 2 process 6705 1 (native) /temp/simple
> > (gdb) jump 3
> > Unreasonable jump request
> > (gdb)
> >
> > In this example, jump fails because the debugger finds two different
> > locations, one for each inferior.
> > Solution is to limit the search to the current program space.
> > ---
> > gdb/infcmd.c | 3 +-
> > gdb/linespec.c | 7 +-
> > gdb/linespec.h | 5 +-
> > gdb/testsuite/gdb.base/jump.exp | 177 +++++++++++++++++---------------
> > 4 files changed, 103 insertions(+), 89 deletions(-)
> >
> > diff --git a/gdb/infcmd.c b/gdb/infcmd.c
> > index 103899432f7..a79b5d3d4d6 100644
> > --- a/gdb/infcmd.c
> > +++ b/gdb/infcmd.c
> > @@ -1077,7 +1077,8 @@ jump_command (const char *arg, int from_tty)
> > error_no_arg (_("starting address"));
> >
> > std::vector<symtab_and_line> sals
> > - = decode_line_with_last_displayed (arg, DECODE_LINE_FUNFIRSTLINE);
> > + = decode_line_with_last_displayed (arg, DECODE_LINE_FUNFIRSTLINE,
> > + current_program_space);
>
> I wonder if we should stop using decode_line_with_last_displayed here,
> and instead use decode_line_with_current_source? It feels like a jump
> should always be relative to the current location, rather than what we
> happen to have listed.
>
> > if (sals.size () != 1)
> > error (_("Unreasonable jump request"));
> >
> > diff --git a/gdb/linespec.c b/gdb/linespec.c
> > index 7d969f37fbf..920fd3d37b7 100644
> > --- a/gdb/linespec.c
> > +++ b/gdb/linespec.c
> > @@ -3231,7 +3231,8 @@ decode_line_with_current_source (const char
> *string, int flags)
> > /* See linespec.h. */
> >
> > std::vector<symtab_and_line>
> > -decode_line_with_last_displayed (const char *string, int flags)
> > +decode_line_with_last_displayed (const char *string, int flags,
> > + program_space *search_pspace)
> > {
> > if (string == 0)
> > error (_("Empty line specification."));
> > @@ -3240,10 +3241,10 @@ decode_line_with_last_displayed (const char
> *string, int flags)
> > current_language);
> > std::vector<symtab_and_line> sals
> > = (last_displayed_sal_is_valid ()
> > - ? decode_line_1 (locspec.get (), flags, NULL,
> > + ? decode_line_1 (locspec.get (), flags, search_pspace,
> > get_last_displayed_symtab (),
> > get_last_displayed_line ())
>
> This change is why I started looking at the details of this patch again.
> I'm not sure this change makes sense, I would have expected us to use
> get_last_displayed_pspace () here instead of search_pspace, but then it
> becomes harder to describe what search_pspace does as it would only be
> used below.
>
> It we switch jump to use decode_line_with_current_source then I think
> this problem goes away, the changes to decode_line_with_last_displayed
> can be dropped.
>
> At the end of this email you'll find a patch that applies on top of your
> V3, which implements the change I'm suggesting. This passes all the
> 'jump' tests I could find in the testsuite, so I think it's fine, but
> it would be great to hear your thoughts.
>
> Thanks,
> Andrew
>
> > - : decode_line_1 (locspec.get (), flags, NULL, NULL, 0));
> > + : decode_line_1 (locspec.get (), flags, search_pspace, NULL, 0));
> >
> > if (*string)
> > error (_("Junk at end of line specification: %s"), string);
> > diff --git a/gdb/linespec.h b/gdb/linespec.h
> > index d5e7334fe2d..0eb9cb5d9f2 100644
> > --- a/gdb/linespec.h
> > +++ b/gdb/linespec.h
> > @@ -139,10 +139,11 @@ extern std::vector<symtab_and_line>
> decode_line_with_current_source
> > (const char *, int);
> >
> > /* Given a string, return the line specified by it, using the last displayed
> > - codepoint's values as defaults, or nothing if they aren't valid. */
> > + codepoint's values as defaults, or nothing if they aren't valid.
> > + Limit the search to given program space, if specified. */
> >
> > extern std::vector<symtab_and_line> decode_line_with_last_displayed
> > - (const char *, int);
> > + (const char *, int, program_space *search_pspace = nullptr);
> >
> > /* Does P represent one of the keywords? If so, return
> > the keyword. If not, return NULL. */
> > diff --git a/gdb/testsuite/gdb.base/jump.exp
> b/gdb/testsuite/gdb.base/jump.exp
> > index 032c4a6279d..4c0d4ed2dbb 100644
> > --- a/gdb/testsuite/gdb.base/jump.exp
> > +++ b/gdb/testsuite/gdb.base/jump.exp
> > @@ -18,99 +18,110 @@ clear_xfail "*-*-*"
> >
> > standard_testfile .c
> >
> > -# Build the test case
> > -if { [gdb_compile "${srcdir}/${subdir}/${srcfile}" "${binfile}" executable
> {debug nowarnings}] != "" } {
> > - untested "failed to compile"
> > - return -1
> > - }
> > +if {[prepare_for_testing "failed to prepare" $testfile $srcfile debug]} {
> > + return -1
> > +}
> >
> >
> > -# Start with a fresh gdb
> > +proc do_tests {} {
> > + global decimal srcfile
> > +
> > + # Set a breakpoint on the statement that we're about to jump to.
> > + # The statement doesn't contain a function call.
> > + set non_call_line [gdb_get_line_number "bp-on-non-call"]
> > + gdb_breakpoint "$non_call_line"
> > + set bp_on_non_call \
> > + [get_integer_valueof "\$bpnum" "INVALID" "bp_on_non_call"]
> > +
> > + # Can we jump to the statement? Do we stop there?
> > + gdb_test "jump $non_call_line" \
> > + "Breakpoint ${bp_on_non_call}(\.${decimal})?,
> .*${srcfile}:$non_call_line.*" \
> > + "jump to non-call"
> > +
> > + # Set a breakpoint on the statement that we're about to jump to.
> > + # The statement does contain a function call.
> > + set call_line [gdb_get_line_number "bp-on-call"]
> > + gdb_breakpoint "$call_line"
> > + set bp_on_call [get_integer_valueof "\$bpnum" "INVALID" "bp_on_call"]
> > +
> > + # Can we jump to the statement? Do we stop there?
> > + gdb_test "jump $call_line" \
> > + "Breakpoint ${bp_on_call}(\.${decimal})?, .*${srcfile}:$call_line.*" \
> > + "jump to call"
> > +
> > + # If we disable the breakpoint at the function call, and then
> > + # if we jump to that statement, do we not stop there, but at
> > + # the following breakpoint?
> > + gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
> > +
> > + gdb_test "jump $call_line" \
> > + "Breakpoint ${bp_on_non_call}(\.${decimal})?,
> .*${srcfile}:$non_call_line.*" \
> > + "jump to call with disabled breakpoint"
> > +
> > + # Disable the breakpoint at the non-function call, so it won't hit
> > + # if do_test is called again.
> > + gdb_test_no_output "disable ${bp_on_non_call}" "disable
> bp_on_non_call"
> > +
> > + # Verify that GDB responds gracefully to the "jump" command without
> > + # an argument.
> > + gdb_test "jump" "Argument required .starting address.*" \
> > + "jump without argument disallowed"
> > +
> > + # Verify that GDB responds gracefully to the "jump" command with
> > + # trailing junk.
> > + gdb_test "jump $call_line 100" \
> > + "malformed linespec error: unexpected number, \"100\"" \
> > + "jump with trailing argument junk"
> > +
> > + # Verify that GDB responds gracefully to a request to jump out of
> > + # the current function. (Note that this will very likely cause the
> > + # inferior to die. Be prepared to rerun the inferior, if further
> > + # testing is desired.)
> > + # Try it both ways: confirming and not confirming the jump.
> > + set out_line [gdb_get_line_number "out-of-func"]
> > + gdb_test "jump $out_line" \
> > + "Not confirmed.*" \
> > + "aborted jump out of current function" \
> > + "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> > + "n"
> > +
> > + gdb_test "jump $out_line" \
> > + "Continuing at.*" \
> > + "jump out of current function" \
> > + "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> > + "y"
> > +}
> >
> > -clean_restart ${binfile}
> >
> > -if {![runto_main]} {
> > - return -1
> > +set inferiors 1
> > +if {![use_gdb_stub]} {
> > + set inferiors 2
> > }
> >
> > -# Set a breakpoint on the statement that we're about to jump to.
> > -# The statement doesn't contain a function call.
> > -#
> > -set bp_on_non_call 0
> > -set non_call_line [gdb_get_line_number "bp-on-non-call"]
> > -gdb_test_multiple "break $non_call_line" "break before jump to non-call" {
> > - -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line
> $non_call_line.*$gdb_prompt $" {
> > - set bp_on_non_call $expect_out(1,string)
> > - pass "break before jump to non-call"
> > +# Run to main, add inferiors if needed.
> > +for {set inf 1} {$inf <= $inferiors} {incr inf} {
> > + if {$inf != 1} {
> > + # Start a new inferior, and run it with the same executable.
> > + gdb_test "add-inferior -exec ${binfile}" \
> > + "Added inferior ${inf}.*" \
> > + "add inferior ${inf} with -exec "
> > + gdb_test "inferior ${inf}" \
> > + "Switching to inferior ${inf} .*" \
> > + "switch to inferior ${inf}"
> > + }
> > + if {![runto_main]} {
> > + return -1
> > }
> > }
> >
> > -# Can we jump to the statement? Do we stop there?
> > -#
> > -gdb_test "jump $non_call_line" "Breakpoint ${decimal},
> .*${srcfile}:$non_call_line.*" \
> > - "jump to non-call"
> > -
> > -# Set a breakpoint on the statement that we're about to jump to.
> > -# The statement does contain a function call.
> > -#
> > -set bp_on_call 0
> > -set call_line [gdb_get_line_number "bp-on-call"]
> > -gdb_test_multiple "break $call_line" "break before jump to call" {
> > - -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line
> $call_line.*$gdb_prompt $" {
> > - set bp_on_call $expect_out(1,string)
> > - pass "break before jump to call"
> > +# Run tests on all inferiors.
> > +for {set inf 1} {$inf <= $inferiors} {incr inf} {
> > + with_test_prefix "inferior $inf" {
> > + # Switch to the target inferior.
> > + gdb_test "inferior $inf" ".*Switching to inferior $inf .*"
> > + # Run the tests.
> > + do_tests
> > }
> > }
> >
> > -# Can we jump to the statement? Do we stop there?
> > -#
> > -gdb_test "jump $call_line" \
> > - "Breakpoint ${decimal}, .*${srcfile}:$call_line.*" \
> > - "jump to call"
> > -
> > -# If we disable the breakpoint at the function call, and then
> > -# if we jump to that statement, do we not stop there, but at
> > -# the following breakpoint?
> > -#
> > -gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
> > -
> > -gdb_test "jump $call_line" "Breakpoint ${decimal},
> .*${srcfile}:$non_call_line.*" \
> > - "jump to call with disabled breakpoint"
> > -
> > -# Verify that GDB responds gracefully to the "jump" command without
> > -# an argument.
> > -#
> > -gdb_test "jump" "Argument required .starting address.*" \
> > - "jump without argument disallowed"
> > -
> > -
> > -# Verify that GDB responds gracefully to the "jump" command with
> > -# trailing junk.
> > -#
> > -gdb_test "jump $call_line 100" \
> > - "malformed linespec error: unexpected number, \"100\"" \
> > - "jump with trailing argument junk"
> > -
> > -
> > -# Verify that GDB responds gracefully to a request to jump out of
> > -# the current function. (Note that this will very likely cause the
> > -# inferior to die. Be prepared to rerun the inferior, if further
> > -# testing is desired.)
> > -#
> > -# Try it both ways: confirming and not confirming the jump.
> > -#
> > -
> > -set out_line [gdb_get_line_number "out-of-func"]
> > -gdb_test "jump $out_line" \
> > - "Not confirmed.*" \
> > - "aborted jump out of current function" \
> > - "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> > - "n"
> > -
> > -gdb_test "jump $out_line" \
> > - "Continuing at.*" \
> > - "jump out of current function" \
> > - "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> > - "y"
> > -
> > gdb_exit
> > --
> > 2.25.1
> >
> > Intel Deutschland GmbH
> > Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
> > Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
> > Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
> > Chairperson of the Supervisory Board: Nicole Lau
> > Registered Office: Munich
> > Commercial Register: Amtsgericht Muenchen HRB 186928
>
> ---
>
> diff --git a/gdb/infcmd.c b/gdb/infcmd.c
> index 1be72ed09f0..1d10e2908ca 100644
> --- a/gdb/infcmd.c
> +++ b/gdb/infcmd.c
> @@ -1068,8 +1068,7 @@ jump_command (const char *arg, int from_tty)
> error_no_arg (_("starting address"));
>
> std::vector<symtab_and_line> sals
> - = decode_line_with_last_displayed (arg, DECODE_LINE_FUNFIRSTLINE,
> - current_program_space);
> + = decode_line_with_current_source (arg, DECODE_LINE_FUNFIRSTLINE);
> if (sals.size () != 1)
> error (_("Unreasonable jump request"));
>
> diff --git a/gdb/linespec.c b/gdb/linespec.c
> index 920fd3d37b7..9f03a4b3604 100644
> --- a/gdb/linespec.c
> +++ b/gdb/linespec.c
> @@ -3220,7 +3220,8 @@ decode_line_with_current_source (const char
> *string, int flags)
> location_spec_up locspec = string_to_location_spec (&string,
> current_language);
> std::vector<symtab_and_line> sals
> - = decode_line_1 (locspec.get (), flags, NULL, cursal.symtab, cursal.line);
> + = decode_line_1 (locspec.get (), flags, cursal.pspace, cursal.symtab,
> + cursal.line);
>
> if (*string)
> error (_("Junk at end of line specification: %s"), string);
> @@ -3231,8 +3232,7 @@ decode_line_with_current_source (const char
> *string, int flags)
> /* See linespec.h. */
>
> std::vector<symtab_and_line>
> -decode_line_with_last_displayed (const char *string, int flags,
> - program_space *search_pspace)
> +decode_line_with_last_displayed (const char *string, int flags)
> {
> if (string == 0)
> error (_("Empty line specification."));
> @@ -3241,10 +3241,10 @@ decode_line_with_last_displayed (const char
> *string, int flags,
> current_language);
> std::vector<symtab_and_line> sals
> = (last_displayed_sal_is_valid ()
> - ? decode_line_1 (locspec.get (), flags, search_pspace,
> + ? decode_line_1 (locspec.get (), flags, NULL,
> get_last_displayed_symtab (),
> get_last_displayed_line ())
> - : decode_line_1 (locspec.get (), flags, search_pspace, NULL, 0));
> + : decode_line_1 (locspec.get (), flags, NULL, NULL, 0));
>
> if (*string)
> error (_("Junk at end of line specification: %s"), string);
> diff --git a/gdb/linespec.h b/gdb/linespec.h
> index 0eb9cb5d9f2..c0eef7894d9 100644
> --- a/gdb/linespec.h
> +++ b/gdb/linespec.h
> @@ -143,7 +143,7 @@ extern std::vector<symtab_and_line>
> decode_line_with_current_source
> Limit the search to given program space, if specified. */
>
> extern std::vector<symtab_and_line> decode_line_with_last_displayed
> - (const char *, int, program_space *search_pspace = nullptr);
> + (const char *, int);
>
> /* Does P represent one of the keywords? If so, return
> the keyword. If not, return NULL. */
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [relevance 99%]
* [PATCH v4 1/1] gdb, infcmd: Support jump command in multi-inferior case.
@ 2023-06-28 9:00 65% Matti Puputti
2023-07-04 14:38 99% ` Andrew Burgess
0 siblings, 1 reply; 34+ results
From: Matti Puputti @ 2023-06-28 9:00 UTC (permalink / raw)
To: gdb-patches; +Cc: aburgess
Fixes the issue where jump failed if multiple inferiors run the same source.
See the below example
$ gdb -q ./simple
Reading symbols from ./simple...
(gdb) break 2
Breakpoint 1 at 0x114e: file simple.c, line 2.
(gdb) run
Starting program: /temp/simple
Breakpoint 1, main () at simple.c:2
2 int a = 42;
(gdb) add-inferior
[New inferior 2]
Added inferior 2 on connection 1 (native)
(gdb) inferior 2
[Switching to inferior 2 [<null>] (<noexec>)]
(gdb) info inferiors
Num Description Connection Executable
1 process 6250 1 (native) /temp/simple
* 2 <null> 1 (native)
(gdb) file ./simple
Reading symbols from ./simple...
(gdb) run
Starting program: /temp/simple
Thread 2.1 "simple" hit Breakpoint 1, main () at simple.c:2
2 int a = 42;
(gdb) info inferiors
Num Description Connection Executable
1 process 6250 1 (native) /temp/simple
* 2 process 6705 1 (native) /temp/simple
(gdb) jump 3
Unreasonable jump request
(gdb)
In this example, jump fails because the debugger finds two different
locations, one for each inferior.
Solution is to limit the search to the current program space.
---
gdb/infcmd.c | 2 +-
gdb/linespec.c | 3 +-
gdb/linespec.h | 3 +-
gdb/testsuite/gdb.base/jump.exp | 177 +++++++++++++++++---------------
4 files changed, 99 insertions(+), 86 deletions(-)
diff --git a/gdb/infcmd.c b/gdb/infcmd.c
index 15702f84894..96c5feafb1a 100644
--- a/gdb/infcmd.c
+++ b/gdb/infcmd.c
@@ -1068,7 +1068,7 @@ jump_command (const char *arg, int from_tty)
error_no_arg (_("starting address"));
std::vector<symtab_and_line> sals
- = decode_line_with_last_displayed (arg, DECODE_LINE_FUNFIRSTLINE);
+ = decode_line_with_current_source (arg, DECODE_LINE_FUNFIRSTLINE);
if (sals.size () != 1)
{
/* If multiple sal-objects were found, try dropping those that aren't
diff --git a/gdb/linespec.c b/gdb/linespec.c
index 7d969f37fbf..afa9eb4d3ac 100644
--- a/gdb/linespec.c
+++ b/gdb/linespec.c
@@ -3220,7 +3220,8 @@ decode_line_with_current_source (const char *string, int flags)
location_spec_up locspec = string_to_location_spec (&string,
current_language);
std::vector<symtab_and_line> sals
- = decode_line_1 (locspec.get (), flags, NULL, cursal.symtab, cursal.line);
+ = decode_line_1 (locspec.get (), flags, cursal.pspace, cursal.symtab,
+ cursal.line);
if (*string)
error (_("Junk at end of line specification: %s"), string);
diff --git a/gdb/linespec.h b/gdb/linespec.h
index d5e7334fe2d..c0eef7894d9 100644
--- a/gdb/linespec.h
+++ b/gdb/linespec.h
@@ -139,7 +139,8 @@ extern std::vector<symtab_and_line> decode_line_with_current_source
(const char *, int);
/* Given a string, return the line specified by it, using the last displayed
- codepoint's values as defaults, or nothing if they aren't valid. */
+ codepoint's values as defaults, or nothing if they aren't valid.
+ Limit the search to given program space, if specified. */
extern std::vector<symtab_and_line> decode_line_with_last_displayed
(const char *, int);
diff --git a/gdb/testsuite/gdb.base/jump.exp b/gdb/testsuite/gdb.base/jump.exp
index 032c4a6279d..4c0d4ed2dbb 100644
--- a/gdb/testsuite/gdb.base/jump.exp
+++ b/gdb/testsuite/gdb.base/jump.exp
@@ -18,99 +18,110 @@ clear_xfail "*-*-*"
standard_testfile .c
-# Build the test case
-if { [gdb_compile "${srcdir}/${subdir}/${srcfile}" "${binfile}" executable {debug nowarnings}] != "" } {
- untested "failed to compile"
- return -1
- }
+if {[prepare_for_testing "failed to prepare" $testfile $srcfile debug]} {
+ return -1
+}
-# Start with a fresh gdb
+proc do_tests {} {
+ global decimal srcfile
+
+ # Set a breakpoint on the statement that we're about to jump to.
+ # The statement doesn't contain a function call.
+ set non_call_line [gdb_get_line_number "bp-on-non-call"]
+ gdb_breakpoint "$non_call_line"
+ set bp_on_non_call \
+ [get_integer_valueof "\$bpnum" "INVALID" "bp_on_non_call"]
+
+ # Can we jump to the statement? Do we stop there?
+ gdb_test "jump $non_call_line" \
+ "Breakpoint ${bp_on_non_call}(\.${decimal})?, .*${srcfile}:$non_call_line.*" \
+ "jump to non-call"
+
+ # Set a breakpoint on the statement that we're about to jump to.
+ # The statement does contain a function call.
+ set call_line [gdb_get_line_number "bp-on-call"]
+ gdb_breakpoint "$call_line"
+ set bp_on_call [get_integer_valueof "\$bpnum" "INVALID" "bp_on_call"]
+
+ # Can we jump to the statement? Do we stop there?
+ gdb_test "jump $call_line" \
+ "Breakpoint ${bp_on_call}(\.${decimal})?, .*${srcfile}:$call_line.*" \
+ "jump to call"
+
+ # If we disable the breakpoint at the function call, and then
+ # if we jump to that statement, do we not stop there, but at
+ # the following breakpoint?
+ gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
+
+ gdb_test "jump $call_line" \
+ "Breakpoint ${bp_on_non_call}(\.${decimal})?, .*${srcfile}:$non_call_line.*" \
+ "jump to call with disabled breakpoint"
+
+ # Disable the breakpoint at the non-function call, so it won't hit
+ # if do_test is called again.
+ gdb_test_no_output "disable ${bp_on_non_call}" "disable bp_on_non_call"
+
+ # Verify that GDB responds gracefully to the "jump" command without
+ # an argument.
+ gdb_test "jump" "Argument required .starting address.*" \
+ "jump without argument disallowed"
+
+ # Verify that GDB responds gracefully to the "jump" command with
+ # trailing junk.
+ gdb_test "jump $call_line 100" \
+ "malformed linespec error: unexpected number, \"100\"" \
+ "jump with trailing argument junk"
+
+ # Verify that GDB responds gracefully to a request to jump out of
+ # the current function. (Note that this will very likely cause the
+ # inferior to die. Be prepared to rerun the inferior, if further
+ # testing is desired.)
+ # Try it both ways: confirming and not confirming the jump.
+ set out_line [gdb_get_line_number "out-of-func"]
+ gdb_test "jump $out_line" \
+ "Not confirmed.*" \
+ "aborted jump out of current function" \
+ "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
+ "n"
+
+ gdb_test "jump $out_line" \
+ "Continuing at.*" \
+ "jump out of current function" \
+ "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
+ "y"
+}
-clean_restart ${binfile}
-if {![runto_main]} {
- return -1
+set inferiors 1
+if {![use_gdb_stub]} {
+ set inferiors 2
}
-# Set a breakpoint on the statement that we're about to jump to.
-# The statement doesn't contain a function call.
-#
-set bp_on_non_call 0
-set non_call_line [gdb_get_line_number "bp-on-non-call"]
-gdb_test_multiple "break $non_call_line" "break before jump to non-call" {
- -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line $non_call_line.*$gdb_prompt $" {
- set bp_on_non_call $expect_out(1,string)
- pass "break before jump to non-call"
+# Run to main, add inferiors if needed.
+for {set inf 1} {$inf <= $inferiors} {incr inf} {
+ if {$inf != 1} {
+ # Start a new inferior, and run it with the same executable.
+ gdb_test "add-inferior -exec ${binfile}" \
+ "Added inferior ${inf}.*" \
+ "add inferior ${inf} with -exec "
+ gdb_test "inferior ${inf}" \
+ "Switching to inferior ${inf} .*" \
+ "switch to inferior ${inf}"
+ }
+ if {![runto_main]} {
+ return -1
}
}
-# Can we jump to the statement? Do we stop there?
-#
-gdb_test "jump $non_call_line" "Breakpoint ${decimal}, .*${srcfile}:$non_call_line.*" \
- "jump to non-call"
-
-# Set a breakpoint on the statement that we're about to jump to.
-# The statement does contain a function call.
-#
-set bp_on_call 0
-set call_line [gdb_get_line_number "bp-on-call"]
-gdb_test_multiple "break $call_line" "break before jump to call" {
- -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line $call_line.*$gdb_prompt $" {
- set bp_on_call $expect_out(1,string)
- pass "break before jump to call"
+# Run tests on all inferiors.
+for {set inf 1} {$inf <= $inferiors} {incr inf} {
+ with_test_prefix "inferior $inf" {
+ # Switch to the target inferior.
+ gdb_test "inferior $inf" ".*Switching to inferior $inf .*"
+ # Run the tests.
+ do_tests
}
}
-# Can we jump to the statement? Do we stop there?
-#
-gdb_test "jump $call_line" \
- "Breakpoint ${decimal}, .*${srcfile}:$call_line.*" \
- "jump to call"
-
-# If we disable the breakpoint at the function call, and then
-# if we jump to that statement, do we not stop there, but at
-# the following breakpoint?
-#
-gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
-
-gdb_test "jump $call_line" "Breakpoint ${decimal}, .*${srcfile}:$non_call_line.*" \
- "jump to call with disabled breakpoint"
-
-# Verify that GDB responds gracefully to the "jump" command without
-# an argument.
-#
-gdb_test "jump" "Argument required .starting address.*" \
- "jump without argument disallowed"
-
-
-# Verify that GDB responds gracefully to the "jump" command with
-# trailing junk.
-#
-gdb_test "jump $call_line 100" \
- "malformed linespec error: unexpected number, \"100\"" \
- "jump with trailing argument junk"
-
-
-# Verify that GDB responds gracefully to a request to jump out of
-# the current function. (Note that this will very likely cause the
-# inferior to die. Be prepared to rerun the inferior, if further
-# testing is desired.)
-#
-# Try it both ways: confirming and not confirming the jump.
-#
-
-set out_line [gdb_get_line_number "out-of-func"]
-gdb_test "jump $out_line" \
- "Not confirmed.*" \
- "aborted jump out of current function" \
- "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
- "n"
-
-gdb_test "jump $out_line" \
- "Continuing at.*" \
- "jump out of current function" \
- "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
- "y"
-
gdb_exit
--
2.25.1
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [relevance 65%]
* RE: [PATCH v3 1/1] gdb, infcmd: Support jump command in multi-inferior case.
2023-05-24 17:33 91% ` [PATCH " Andrew Burgess
@ 2023-05-25 5:38 99% ` Puputti, Matti
2023-06-28 9:04 99% ` Puputti, Matti
1 sibling, 0 replies; 34+ results
From: Puputti, Matti @ 2023-05-25 5:38 UTC (permalink / raw)
To: Andrew Burgess, gdb-patches
Hi Andrew,
Thank you for these comments.
I'll be addressing them as soon as I will be back in office, in few weeks
Br,
Matti Puputti
> -----Original Message-----
> From: Andrew Burgess <aburgess@redhat.com>
> Sent: Wednesday, May 24, 2023 7:33 PM
> To: Puputti, Matti <matti.puputti@intel.com>; gdb-patches@sourceware.org
> Subject: Re: [PATCH v3 1/1] gdb, infcmd: Support jump command in multi-
> inferior case.
>
> Matti Puputti <matti.puputti@intel.com> writes:
>
> > Fixes the issue where jump failed if multiple inferiors run the same source.
> >
> > See the below example
> >
> > $ gdb -q ./simple
> > Reading symbols from ./simple...
> > (gdb) break 2
> > Breakpoint 1 at 0x114e: file simple.c, line 2.
> > (gdb) run
> > Starting program: /temp/simple
> >
> > Breakpoint 1, main () at simple.c:2
> > 2 int a = 42;
> > (gdb) add-inferior
> > [New inferior 2]
> > Added inferior 2 on connection 1 (native)
> > (gdb) inferior 2
> > [Switching to inferior 2 [<null>] (<noexec>)]
> > (gdb) info inferiors
> > Num Description Connection Executable
> > 1 process 6250 1 (native) /temp/simple
> > * 2 <null> 1 (native)
> > (gdb) file ./simple
> > Reading symbols from ./simple...
> > (gdb) run
> > Starting program: /temp/simple
> >
> > Thread 2.1 "simple" hit Breakpoint 1, main () at simple.c:2
> > 2 int a = 42;
> > (gdb) info inferiors
> > Num Description Connection Executable
> > 1 process 6250 1 (native) /temp/simple
> > * 2 process 6705 1 (native) /temp/simple
> > (gdb) jump 3
> > Unreasonable jump request
> > (gdb)
> >
> > In this example, jump fails because the debugger finds two different
> > locations, one for each inferior.
> > Solution is to limit the search to the current program space.
> > ---
> > gdb/infcmd.c | 3 +-
> > gdb/linespec.c | 7 +-
> > gdb/linespec.h | 5 +-
> > gdb/testsuite/gdb.base/jump.exp | 177 +++++++++++++++++--------------
> -
> > 4 files changed, 103 insertions(+), 89 deletions(-)
> >
> > diff --git a/gdb/infcmd.c b/gdb/infcmd.c
> > index 103899432f7..a79b5d3d4d6 100644
> > --- a/gdb/infcmd.c
> > +++ b/gdb/infcmd.c
> > @@ -1077,7 +1077,8 @@ jump_command (const char *arg, int from_tty)
> > error_no_arg (_("starting address"));
> >
> > std::vector<symtab_and_line> sals
> > - = decode_line_with_last_displayed (arg, DECODE_LINE_FUNFIRSTLINE);
> > + = decode_line_with_last_displayed (arg, DECODE_LINE_FUNFIRSTLINE,
> > + current_program_space);
>
> I wonder if we should stop using decode_line_with_last_displayed here,
> and instead use decode_line_with_current_source? It feels like a jump
> should always be relative to the current location, rather than what we
> happen to have listed.
>
> > if (sals.size () != 1)
> > error (_("Unreasonable jump request"));
> >
> > diff --git a/gdb/linespec.c b/gdb/linespec.c
> > index 7d969f37fbf..920fd3d37b7 100644
> > --- a/gdb/linespec.c
> > +++ b/gdb/linespec.c
> > @@ -3231,7 +3231,8 @@ decode_line_with_current_source (const char
> *string, int flags)
> > /* See linespec.h. */
> >
> > std::vector<symtab_and_line>
> > -decode_line_with_last_displayed (const char *string, int flags)
> > +decode_line_with_last_displayed (const char *string, int flags,
> > + program_space *search_pspace)
> > {
> > if (string == 0)
> > error (_("Empty line specification."));
> > @@ -3240,10 +3241,10 @@ decode_line_with_last_displayed (const char
> *string, int flags)
> > current_language);
> > std::vector<symtab_and_line> sals
> > = (last_displayed_sal_is_valid ()
> > - ? decode_line_1 (locspec.get (), flags, NULL,
> > + ? decode_line_1 (locspec.get (), flags, search_pspace,
> > get_last_displayed_symtab (),
> > get_last_displayed_line ())
>
> This change is why I started looking at the details of this patch again.
> I'm not sure this change makes sense, I would have expected us to use
> get_last_displayed_pspace () here instead of search_pspace, but then it
> becomes harder to describe what search_pspace does as it would only be
> used below.
>
> It we switch jump to use decode_line_with_current_source then I think
> this problem goes away, the changes to decode_line_with_last_displayed
> can be dropped.
>
> At the end of this email you'll find a patch that applies on top of your
> V3, which implements the change I'm suggesting. This passes all the
> 'jump' tests I could find in the testsuite, so I think it's fine, but
> it would be great to hear your thoughts.
>
> Thanks,
> Andrew
>
> > - : decode_line_1 (locspec.get (), flags, NULL, NULL, 0));
> > + : decode_line_1 (locspec.get (), flags, search_pspace, NULL, 0));
> >
> > if (*string)
> > error (_("Junk at end of line specification: %s"), string);
> > diff --git a/gdb/linespec.h b/gdb/linespec.h
> > index d5e7334fe2d..0eb9cb5d9f2 100644
> > --- a/gdb/linespec.h
> > +++ b/gdb/linespec.h
> > @@ -139,10 +139,11 @@ extern std::vector<symtab_and_line>
> decode_line_with_current_source
> > (const char *, int);
> >
> > /* Given a string, return the line specified by it, using the last displayed
> > - codepoint's values as defaults, or nothing if they aren't valid. */
> > + codepoint's values as defaults, or nothing if they aren't valid.
> > + Limit the search to given program space, if specified. */
> >
> > extern std::vector<symtab_and_line> decode_line_with_last_displayed
> > - (const char *, int);
> > + (const char *, int, program_space *search_pspace = nullptr);
> >
> > /* Does P represent one of the keywords? If so, return
> > the keyword. If not, return NULL. */
> > diff --git a/gdb/testsuite/gdb.base/jump.exp
> b/gdb/testsuite/gdb.base/jump.exp
> > index 032c4a6279d..4c0d4ed2dbb 100644
> > --- a/gdb/testsuite/gdb.base/jump.exp
> > +++ b/gdb/testsuite/gdb.base/jump.exp
> > @@ -18,99 +18,110 @@ clear_xfail "*-*-*"
> >
> > standard_testfile .c
> >
> > -# Build the test case
> > -if { [gdb_compile "${srcdir}/${subdir}/${srcfile}" "${binfile}" executable
> {debug nowarnings}] != "" } {
> > - untested "failed to compile"
> > - return -1
> > - }
> > +if {[prepare_for_testing "failed to prepare" $testfile $srcfile debug]} {
> > + return -1
> > +}
> >
> >
> > -# Start with a fresh gdb
> > +proc do_tests {} {
> > + global decimal srcfile
> > +
> > + # Set a breakpoint on the statement that we're about to jump to.
> > + # The statement doesn't contain a function call.
> > + set non_call_line [gdb_get_line_number "bp-on-non-call"]
> > + gdb_breakpoint "$non_call_line"
> > + set bp_on_non_call \
> > + [get_integer_valueof "\$bpnum" "INVALID" "bp_on_non_call"]
> > +
> > + # Can we jump to the statement? Do we stop there?
> > + gdb_test "jump $non_call_line" \
> > + "Breakpoint ${bp_on_non_call}(\.${decimal})?,
> .*${srcfile}:$non_call_line.*" \
> > + "jump to non-call"
> > +
> > + # Set a breakpoint on the statement that we're about to jump to.
> > + # The statement does contain a function call.
> > + set call_line [gdb_get_line_number "bp-on-call"]
> > + gdb_breakpoint "$call_line"
> > + set bp_on_call [get_integer_valueof "\$bpnum" "INVALID"
> "bp_on_call"]
> > +
> > + # Can we jump to the statement? Do we stop there?
> > + gdb_test "jump $call_line" \
> > + "Breakpoint ${bp_on_call}(\.${decimal})?, .*${srcfile}:$call_line.*" \
> > + "jump to call"
> > +
> > + # If we disable the breakpoint at the function call, and then
> > + # if we jump to that statement, do we not stop there, but at
> > + # the following breakpoint?
> > + gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
> > +
> > + gdb_test "jump $call_line" \
> > + "Breakpoint ${bp_on_non_call}(\.${decimal})?,
> .*${srcfile}:$non_call_line.*" \
> > + "jump to call with disabled breakpoint"
> > +
> > + # Disable the breakpoint at the non-function call, so it won't hit
> > + # if do_test is called again.
> > + gdb_test_no_output "disable ${bp_on_non_call}" "disable
> bp_on_non_call"
> > +
> > + # Verify that GDB responds gracefully to the "jump" command without
> > + # an argument.
> > + gdb_test "jump" "Argument required .starting address.*" \
> > + "jump without argument disallowed"
> > +
> > + # Verify that GDB responds gracefully to the "jump" command with
> > + # trailing junk.
> > + gdb_test "jump $call_line 100" \
> > + "malformed linespec error: unexpected number, \"100\"" \
> > + "jump with trailing argument junk"
> > +
> > + # Verify that GDB responds gracefully to a request to jump out of
> > + # the current function. (Note that this will very likely cause the
> > + # inferior to die. Be prepared to rerun the inferior, if further
> > + # testing is desired.)
> > + # Try it both ways: confirming and not confirming the jump.
> > + set out_line [gdb_get_line_number "out-of-func"]
> > + gdb_test "jump $out_line" \
> > + "Not confirmed.*" \
> > + "aborted jump out of current function" \
> > + "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> > + "n"
> > +
> > + gdb_test "jump $out_line" \
> > + "Continuing at.*" \
> > + "jump out of current function" \
> > + "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> > + "y"
> > +}
> >
> > -clean_restart ${binfile}
> >
> > -if {![runto_main]} {
> > - return -1
> > +set inferiors 1
> > +if {![use_gdb_stub]} {
> > + set inferiors 2
> > }
> >
> > -# Set a breakpoint on the statement that we're about to jump to.
> > -# The statement doesn't contain a function call.
> > -#
> > -set bp_on_non_call 0
> > -set non_call_line [gdb_get_line_number "bp-on-non-call"]
> > -gdb_test_multiple "break $non_call_line" "break before jump to non-call"
> {
> > - -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line
> $non_call_line.*$gdb_prompt $" {
> > - set bp_on_non_call $expect_out(1,string)
> > - pass "break before jump to non-call"
> > +# Run to main, add inferiors if needed.
> > +for {set inf 1} {$inf <= $inferiors} {incr inf} {
> > + if {$inf != 1} {
> > + # Start a new inferior, and run it with the same executable.
> > + gdb_test "add-inferior -exec ${binfile}" \
> > + "Added inferior ${inf}.*" \
> > + "add inferior ${inf} with -exec "
> > + gdb_test "inferior ${inf}" \
> > + "Switching to inferior ${inf} .*" \
> > + "switch to inferior ${inf}"
> > + }
> > + if {![runto_main]} {
> > + return -1
> > }
> > }
> >
> > -# Can we jump to the statement? Do we stop there?
> > -#
> > -gdb_test "jump $non_call_line" "Breakpoint ${decimal},
> .*${srcfile}:$non_call_line.*" \
> > - "jump to non-call"
> > -
> > -# Set a breakpoint on the statement that we're about to jump to.
> > -# The statement does contain a function call.
> > -#
> > -set bp_on_call 0
> > -set call_line [gdb_get_line_number "bp-on-call"]
> > -gdb_test_multiple "break $call_line" "break before jump to call" {
> > - -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line
> $call_line.*$gdb_prompt $" {
> > - set bp_on_call $expect_out(1,string)
> > - pass "break before jump to call"
> > +# Run tests on all inferiors.
> > +for {set inf 1} {$inf <= $inferiors} {incr inf} {
> > + with_test_prefix "inferior $inf" {
> > + # Switch to the target inferior.
> > + gdb_test "inferior $inf" ".*Switching to inferior $inf .*"
> > + # Run the tests.
> > + do_tests
> > }
> > }
> >
> > -# Can we jump to the statement? Do we stop there?
> > -#
> > -gdb_test "jump $call_line" \
> > - "Breakpoint ${decimal}, .*${srcfile}:$call_line.*" \
> > - "jump to call"
> > -
> > -# If we disable the breakpoint at the function call, and then
> > -# if we jump to that statement, do we not stop there, but at
> > -# the following breakpoint?
> > -#
> > -gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
> > -
> > -gdb_test "jump $call_line" "Breakpoint ${decimal},
> .*${srcfile}:$non_call_line.*" \
> > - "jump to call with disabled breakpoint"
> > -
> > -# Verify that GDB responds gracefully to the "jump" command without
> > -# an argument.
> > -#
> > -gdb_test "jump" "Argument required .starting address.*" \
> > - "jump without argument disallowed"
> > -
> > -
> > -# Verify that GDB responds gracefully to the "jump" command with
> > -# trailing junk.
> > -#
> > -gdb_test "jump $call_line 100" \
> > - "malformed linespec error: unexpected number, \"100\"" \
> > - "jump with trailing argument junk"
> > -
> > -
> > -# Verify that GDB responds gracefully to a request to jump out of
> > -# the current function. (Note that this will very likely cause the
> > -# inferior to die. Be prepared to rerun the inferior, if further
> > -# testing is desired.)
> > -#
> > -# Try it both ways: confirming and not confirming the jump.
> > -#
> > -
> > -set out_line [gdb_get_line_number "out-of-func"]
> > -gdb_test "jump $out_line" \
> > - "Not confirmed.*" \
> > - "aborted jump out of current function" \
> > - "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> > - "n"
> > -
> > -gdb_test "jump $out_line" \
> > - "Continuing at.*" \
> > - "jump out of current function" \
> > - "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> > - "y"
> > -
> > gdb_exit
> > --
> > 2.25.1
> >
> > Intel Deutschland GmbH
> > Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
> > Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
> > Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
> > Chairperson of the Supervisory Board: Nicole Lau
> > Registered Office: Munich
> > Commercial Register: Amtsgericht Muenchen HRB 186928
>
> ---
>
> diff --git a/gdb/infcmd.c b/gdb/infcmd.c
> index 1be72ed09f0..1d10e2908ca 100644
> --- a/gdb/infcmd.c
> +++ b/gdb/infcmd.c
> @@ -1068,8 +1068,7 @@ jump_command (const char *arg, int from_tty)
> error_no_arg (_("starting address"));
>
> std::vector<symtab_and_line> sals
> - = decode_line_with_last_displayed (arg, DECODE_LINE_FUNFIRSTLINE,
> - current_program_space);
> + = decode_line_with_current_source (arg, DECODE_LINE_FUNFIRSTLINE);
> if (sals.size () != 1)
> error (_("Unreasonable jump request"));
>
> diff --git a/gdb/linespec.c b/gdb/linespec.c
> index 920fd3d37b7..9f03a4b3604 100644
> --- a/gdb/linespec.c
> +++ b/gdb/linespec.c
> @@ -3220,7 +3220,8 @@ decode_line_with_current_source (const char
> *string, int flags)
> location_spec_up locspec = string_to_location_spec (&string,
> current_language);
> std::vector<symtab_and_line> sals
> - = decode_line_1 (locspec.get (), flags, NULL, cursal.symtab, cursal.line);
> + = decode_line_1 (locspec.get (), flags, cursal.pspace, cursal.symtab,
> + cursal.line);
>
> if (*string)
> error (_("Junk at end of line specification: %s"), string);
> @@ -3231,8 +3232,7 @@ decode_line_with_current_source (const char
> *string, int flags)
> /* See linespec.h. */
>
> std::vector<symtab_and_line>
> -decode_line_with_last_displayed (const char *string, int flags,
> - program_space *search_pspace)
> +decode_line_with_last_displayed (const char *string, int flags)
> {
> if (string == 0)
> error (_("Empty line specification."));
> @@ -3241,10 +3241,10 @@ decode_line_with_last_displayed (const char
> *string, int flags,
> current_language);
> std::vector<symtab_and_line> sals
> = (last_displayed_sal_is_valid ()
> - ? decode_line_1 (locspec.get (), flags, search_pspace,
> + ? decode_line_1 (locspec.get (), flags, NULL,
> get_last_displayed_symtab (),
> get_last_displayed_line ())
> - : decode_line_1 (locspec.get (), flags, search_pspace, NULL, 0));
> + : decode_line_1 (locspec.get (), flags, NULL, NULL, 0));
>
> if (*string)
> error (_("Junk at end of line specification: %s"), string);
> diff --git a/gdb/linespec.h b/gdb/linespec.h
> index 0eb9cb5d9f2..c0eef7894d9 100644
> --- a/gdb/linespec.h
> +++ b/gdb/linespec.h
> @@ -143,7 +143,7 @@ extern std::vector<symtab_and_line>
> decode_line_with_current_source
> Limit the search to given program space, if specified. */
>
> extern std::vector<symtab_and_line> decode_line_with_last_displayed
> - (const char *, int, program_space *search_pspace = nullptr);
> + (const char *, int);
>
> /* Does P represent one of the keywords? If so, return
> the keyword. If not, return NULL. */
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [relevance 99%]
* Re: [PATCH v3 1/1] gdb, infcmd: Support jump command in multi-inferior case.
2023-04-25 13:38 64% [PATCH v3 1/1] " Matti Puputti
2023-05-22 8:01 99% ` [PING][PATCH " Puputti, Matti
@ 2023-05-24 17:33 91% ` Andrew Burgess
2023-05-25 5:38 99% ` Puputti, Matti
2023-06-28 9:04 99% ` Puputti, Matti
1 sibling, 2 replies; 34+ results
From: Andrew Burgess @ 2023-05-24 17:33 UTC (permalink / raw)
To: Matti Puputti, gdb-patches
Matti Puputti <matti.puputti@intel.com> writes:
> Fixes the issue where jump failed if multiple inferiors run the same source.
>
> See the below example
>
> $ gdb -q ./simple
> Reading symbols from ./simple...
> (gdb) break 2
> Breakpoint 1 at 0x114e: file simple.c, line 2.
> (gdb) run
> Starting program: /temp/simple
>
> Breakpoint 1, main () at simple.c:2
> 2 int a = 42;
> (gdb) add-inferior
> [New inferior 2]
> Added inferior 2 on connection 1 (native)
> (gdb) inferior 2
> [Switching to inferior 2 [<null>] (<noexec>)]
> (gdb) info inferiors
> Num Description Connection Executable
> 1 process 6250 1 (native) /temp/simple
> * 2 <null> 1 (native)
> (gdb) file ./simple
> Reading symbols from ./simple...
> (gdb) run
> Starting program: /temp/simple
>
> Thread 2.1 "simple" hit Breakpoint 1, main () at simple.c:2
> 2 int a = 42;
> (gdb) info inferiors
> Num Description Connection Executable
> 1 process 6250 1 (native) /temp/simple
> * 2 process 6705 1 (native) /temp/simple
> (gdb) jump 3
> Unreasonable jump request
> (gdb)
>
> In this example, jump fails because the debugger finds two different
> locations, one for each inferior.
> Solution is to limit the search to the current program space.
> ---
> gdb/infcmd.c | 3 +-
> gdb/linespec.c | 7 +-
> gdb/linespec.h | 5 +-
> gdb/testsuite/gdb.base/jump.exp | 177 +++++++++++++++++---------------
> 4 files changed, 103 insertions(+), 89 deletions(-)
>
> diff --git a/gdb/infcmd.c b/gdb/infcmd.c
> index 103899432f7..a79b5d3d4d6 100644
> --- a/gdb/infcmd.c
> +++ b/gdb/infcmd.c
> @@ -1077,7 +1077,8 @@ jump_command (const char *arg, int from_tty)
> error_no_arg (_("starting address"));
>
> std::vector<symtab_and_line> sals
> - = decode_line_with_last_displayed (arg, DECODE_LINE_FUNFIRSTLINE);
> + = decode_line_with_last_displayed (arg, DECODE_LINE_FUNFIRSTLINE,
> + current_program_space);
I wonder if we should stop using decode_line_with_last_displayed here,
and instead use decode_line_with_current_source? It feels like a jump
should always be relative to the current location, rather than what we
happen to have listed.
> if (sals.size () != 1)
> error (_("Unreasonable jump request"));
>
> diff --git a/gdb/linespec.c b/gdb/linespec.c
> index 7d969f37fbf..920fd3d37b7 100644
> --- a/gdb/linespec.c
> +++ b/gdb/linespec.c
> @@ -3231,7 +3231,8 @@ decode_line_with_current_source (const char *string, int flags)
> /* See linespec.h. */
>
> std::vector<symtab_and_line>
> -decode_line_with_last_displayed (const char *string, int flags)
> +decode_line_with_last_displayed (const char *string, int flags,
> + program_space *search_pspace)
> {
> if (string == 0)
> error (_("Empty line specification."));
> @@ -3240,10 +3241,10 @@ decode_line_with_last_displayed (const char *string, int flags)
> current_language);
> std::vector<symtab_and_line> sals
> = (last_displayed_sal_is_valid ()
> - ? decode_line_1 (locspec.get (), flags, NULL,
> + ? decode_line_1 (locspec.get (), flags, search_pspace,
> get_last_displayed_symtab (),
> get_last_displayed_line ())
This change is why I started looking at the details of this patch again.
I'm not sure this change makes sense, I would have expected us to use
get_last_displayed_pspace () here instead of search_pspace, but then it
becomes harder to describe what search_pspace does as it would only be
used below.
It we switch jump to use decode_line_with_current_source then I think
this problem goes away, the changes to decode_line_with_last_displayed
can be dropped.
At the end of this email you'll find a patch that applies on top of your
V3, which implements the change I'm suggesting. This passes all the
'jump' tests I could find in the testsuite, so I think it's fine, but
it would be great to hear your thoughts.
Thanks,
Andrew
> - : decode_line_1 (locspec.get (), flags, NULL, NULL, 0));
> + : decode_line_1 (locspec.get (), flags, search_pspace, NULL, 0));
>
> if (*string)
> error (_("Junk at end of line specification: %s"), string);
> diff --git a/gdb/linespec.h b/gdb/linespec.h
> index d5e7334fe2d..0eb9cb5d9f2 100644
> --- a/gdb/linespec.h
> +++ b/gdb/linespec.h
> @@ -139,10 +139,11 @@ extern std::vector<symtab_and_line> decode_line_with_current_source
> (const char *, int);
>
> /* Given a string, return the line specified by it, using the last displayed
> - codepoint's values as defaults, or nothing if they aren't valid. */
> + codepoint's values as defaults, or nothing if they aren't valid.
> + Limit the search to given program space, if specified. */
>
> extern std::vector<symtab_and_line> decode_line_with_last_displayed
> - (const char *, int);
> + (const char *, int, program_space *search_pspace = nullptr);
>
> /* Does P represent one of the keywords? If so, return
> the keyword. If not, return NULL. */
> diff --git a/gdb/testsuite/gdb.base/jump.exp b/gdb/testsuite/gdb.base/jump.exp
> index 032c4a6279d..4c0d4ed2dbb 100644
> --- a/gdb/testsuite/gdb.base/jump.exp
> +++ b/gdb/testsuite/gdb.base/jump.exp
> @@ -18,99 +18,110 @@ clear_xfail "*-*-*"
>
> standard_testfile .c
>
> -# Build the test case
> -if { [gdb_compile "${srcdir}/${subdir}/${srcfile}" "${binfile}" executable {debug nowarnings}] != "" } {
> - untested "failed to compile"
> - return -1
> - }
> +if {[prepare_for_testing "failed to prepare" $testfile $srcfile debug]} {
> + return -1
> +}
>
>
> -# Start with a fresh gdb
> +proc do_tests {} {
> + global decimal srcfile
> +
> + # Set a breakpoint on the statement that we're about to jump to.
> + # The statement doesn't contain a function call.
> + set non_call_line [gdb_get_line_number "bp-on-non-call"]
> + gdb_breakpoint "$non_call_line"
> + set bp_on_non_call \
> + [get_integer_valueof "\$bpnum" "INVALID" "bp_on_non_call"]
> +
> + # Can we jump to the statement? Do we stop there?
> + gdb_test "jump $non_call_line" \
> + "Breakpoint ${bp_on_non_call}(\.${decimal})?, .*${srcfile}:$non_call_line.*" \
> + "jump to non-call"
> +
> + # Set a breakpoint on the statement that we're about to jump to.
> + # The statement does contain a function call.
> + set call_line [gdb_get_line_number "bp-on-call"]
> + gdb_breakpoint "$call_line"
> + set bp_on_call [get_integer_valueof "\$bpnum" "INVALID" "bp_on_call"]
> +
> + # Can we jump to the statement? Do we stop there?
> + gdb_test "jump $call_line" \
> + "Breakpoint ${bp_on_call}(\.${decimal})?, .*${srcfile}:$call_line.*" \
> + "jump to call"
> +
> + # If we disable the breakpoint at the function call, and then
> + # if we jump to that statement, do we not stop there, but at
> + # the following breakpoint?
> + gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
> +
> + gdb_test "jump $call_line" \
> + "Breakpoint ${bp_on_non_call}(\.${decimal})?, .*${srcfile}:$non_call_line.*" \
> + "jump to call with disabled breakpoint"
> +
> + # Disable the breakpoint at the non-function call, so it won't hit
> + # if do_test is called again.
> + gdb_test_no_output "disable ${bp_on_non_call}" "disable bp_on_non_call"
> +
> + # Verify that GDB responds gracefully to the "jump" command without
> + # an argument.
> + gdb_test "jump" "Argument required .starting address.*" \
> + "jump without argument disallowed"
> +
> + # Verify that GDB responds gracefully to the "jump" command with
> + # trailing junk.
> + gdb_test "jump $call_line 100" \
> + "malformed linespec error: unexpected number, \"100\"" \
> + "jump with trailing argument junk"
> +
> + # Verify that GDB responds gracefully to a request to jump out of
> + # the current function. (Note that this will very likely cause the
> + # inferior to die. Be prepared to rerun the inferior, if further
> + # testing is desired.)
> + # Try it both ways: confirming and not confirming the jump.
> + set out_line [gdb_get_line_number "out-of-func"]
> + gdb_test "jump $out_line" \
> + "Not confirmed.*" \
> + "aborted jump out of current function" \
> + "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> + "n"
> +
> + gdb_test "jump $out_line" \
> + "Continuing at.*" \
> + "jump out of current function" \
> + "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> + "y"
> +}
>
> -clean_restart ${binfile}
>
> -if {![runto_main]} {
> - return -1
> +set inferiors 1
> +if {![use_gdb_stub]} {
> + set inferiors 2
> }
>
> -# Set a breakpoint on the statement that we're about to jump to.
> -# The statement doesn't contain a function call.
> -#
> -set bp_on_non_call 0
> -set non_call_line [gdb_get_line_number "bp-on-non-call"]
> -gdb_test_multiple "break $non_call_line" "break before jump to non-call" {
> - -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line $non_call_line.*$gdb_prompt $" {
> - set bp_on_non_call $expect_out(1,string)
> - pass "break before jump to non-call"
> +# Run to main, add inferiors if needed.
> +for {set inf 1} {$inf <= $inferiors} {incr inf} {
> + if {$inf != 1} {
> + # Start a new inferior, and run it with the same executable.
> + gdb_test "add-inferior -exec ${binfile}" \
> + "Added inferior ${inf}.*" \
> + "add inferior ${inf} with -exec "
> + gdb_test "inferior ${inf}" \
> + "Switching to inferior ${inf} .*" \
> + "switch to inferior ${inf}"
> + }
> + if {![runto_main]} {
> + return -1
> }
> }
>
> -# Can we jump to the statement? Do we stop there?
> -#
> -gdb_test "jump $non_call_line" "Breakpoint ${decimal}, .*${srcfile}:$non_call_line.*" \
> - "jump to non-call"
> -
> -# Set a breakpoint on the statement that we're about to jump to.
> -# The statement does contain a function call.
> -#
> -set bp_on_call 0
> -set call_line [gdb_get_line_number "bp-on-call"]
> -gdb_test_multiple "break $call_line" "break before jump to call" {
> - -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line $call_line.*$gdb_prompt $" {
> - set bp_on_call $expect_out(1,string)
> - pass "break before jump to call"
> +# Run tests on all inferiors.
> +for {set inf 1} {$inf <= $inferiors} {incr inf} {
> + with_test_prefix "inferior $inf" {
> + # Switch to the target inferior.
> + gdb_test "inferior $inf" ".*Switching to inferior $inf .*"
> + # Run the tests.
> + do_tests
> }
> }
>
> -# Can we jump to the statement? Do we stop there?
> -#
> -gdb_test "jump $call_line" \
> - "Breakpoint ${decimal}, .*${srcfile}:$call_line.*" \
> - "jump to call"
> -
> -# If we disable the breakpoint at the function call, and then
> -# if we jump to that statement, do we not stop there, but at
> -# the following breakpoint?
> -#
> -gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
> -
> -gdb_test "jump $call_line" "Breakpoint ${decimal}, .*${srcfile}:$non_call_line.*" \
> - "jump to call with disabled breakpoint"
> -
> -# Verify that GDB responds gracefully to the "jump" command without
> -# an argument.
> -#
> -gdb_test "jump" "Argument required .starting address.*" \
> - "jump without argument disallowed"
> -
> -
> -# Verify that GDB responds gracefully to the "jump" command with
> -# trailing junk.
> -#
> -gdb_test "jump $call_line 100" \
> - "malformed linespec error: unexpected number, \"100\"" \
> - "jump with trailing argument junk"
> -
> -
> -# Verify that GDB responds gracefully to a request to jump out of
> -# the current function. (Note that this will very likely cause the
> -# inferior to die. Be prepared to rerun the inferior, if further
> -# testing is desired.)
> -#
> -# Try it both ways: confirming and not confirming the jump.
> -#
> -
> -set out_line [gdb_get_line_number "out-of-func"]
> -gdb_test "jump $out_line" \
> - "Not confirmed.*" \
> - "aborted jump out of current function" \
> - "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> - "n"
> -
> -gdb_test "jump $out_line" \
> - "Continuing at.*" \
> - "jump out of current function" \
> - "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> - "y"
> -
> gdb_exit
> --
> 2.25.1
>
> Intel Deutschland GmbH
> Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
> Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
> Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
> Chairperson of the Supervisory Board: Nicole Lau
> Registered Office: Munich
> Commercial Register: Amtsgericht Muenchen HRB 186928
---
diff --git a/gdb/infcmd.c b/gdb/infcmd.c
index 1be72ed09f0..1d10e2908ca 100644
--- a/gdb/infcmd.c
+++ b/gdb/infcmd.c
@@ -1068,8 +1068,7 @@ jump_command (const char *arg, int from_tty)
error_no_arg (_("starting address"));
std::vector<symtab_and_line> sals
- = decode_line_with_last_displayed (arg, DECODE_LINE_FUNFIRSTLINE,
- current_program_space);
+ = decode_line_with_current_source (arg, DECODE_LINE_FUNFIRSTLINE);
if (sals.size () != 1)
error (_("Unreasonable jump request"));
diff --git a/gdb/linespec.c b/gdb/linespec.c
index 920fd3d37b7..9f03a4b3604 100644
--- a/gdb/linespec.c
+++ b/gdb/linespec.c
@@ -3220,7 +3220,8 @@ decode_line_with_current_source (const char *string, int flags)
location_spec_up locspec = string_to_location_spec (&string,
current_language);
std::vector<symtab_and_line> sals
- = decode_line_1 (locspec.get (), flags, NULL, cursal.symtab, cursal.line);
+ = decode_line_1 (locspec.get (), flags, cursal.pspace, cursal.symtab,
+ cursal.line);
if (*string)
error (_("Junk at end of line specification: %s"), string);
@@ -3231,8 +3232,7 @@ decode_line_with_current_source (const char *string, int flags)
/* See linespec.h. */
std::vector<symtab_and_line>
-decode_line_with_last_displayed (const char *string, int flags,
- program_space *search_pspace)
+decode_line_with_last_displayed (const char *string, int flags)
{
if (string == 0)
error (_("Empty line specification."));
@@ -3241,10 +3241,10 @@ decode_line_with_last_displayed (const char *string, int flags,
current_language);
std::vector<symtab_and_line> sals
= (last_displayed_sal_is_valid ()
- ? decode_line_1 (locspec.get (), flags, search_pspace,
+ ? decode_line_1 (locspec.get (), flags, NULL,
get_last_displayed_symtab (),
get_last_displayed_line ())
- : decode_line_1 (locspec.get (), flags, search_pspace, NULL, 0));
+ : decode_line_1 (locspec.get (), flags, NULL, NULL, 0));
if (*string)
error (_("Junk at end of line specification: %s"), string);
diff --git a/gdb/linespec.h b/gdb/linespec.h
index 0eb9cb5d9f2..c0eef7894d9 100644
--- a/gdb/linespec.h
+++ b/gdb/linespec.h
@@ -143,7 +143,7 @@ extern std::vector<symtab_and_line> decode_line_with_current_source
Limit the search to given program space, if specified. */
extern std::vector<symtab_and_line> decode_line_with_last_displayed
- (const char *, int, program_space *search_pspace = nullptr);
+ (const char *, int);
/* Does P represent one of the keywords? If so, return
the keyword. If not, return NULL. */
^ permalink raw reply [relevance 91%]
* RE: [PING][PATCH v3 1/1] gdb, infcmd: Support jump command in multi-inferior case.
2023-04-25 13:38 64% [PATCH v3 1/1] " Matti Puputti
@ 2023-05-22 8:01 99% ` Puputti, Matti
2023-05-24 17:33 91% ` [PATCH " Andrew Burgess
1 sibling, 0 replies; 34+ results
From: Puputti, Matti @ 2023-05-22 8:01 UTC (permalink / raw)
To: gdb-patches; +Cc: Andrew Burgess
Kindly asking for review, approval and help to push.
Patch 1: https://sourceware.org/pipermail/gdb-patches/2023-February/197269.html
Andrew’s comments on patch 1: https://sourceware.org/pipermail/gdb-patches/2023-February/197271.html
Patch 2: https://sourceware.org/pipermail/gdb-patches/2023-February/197302.html
Andres’s comments on patch 2: https://sourceware.org/pipermail/gdb-patches/2023-April/199055.html
My msg to Andrew: https://sourceware.org/pipermail/gdb-patches/2023-April/199096.html
Patch 3: https://sourceware.org/pipermail/gdb-patches/2023-April/199095.html
Br,
Matti Puputti
> -----Original Message-----
> From: Gdb-patches <gdb-patches-
> bounces+matti.puputti=intel.com@sourceware.org> On Behalf Of Matti
> Puputti via Gdb-patches
> Sent: Tuesday, April 25, 2023 3:38 PM
> To: gdb-patches@sourceware.org
> Cc: aburgess@redhat.com
> Subject: [PATCH v3 1/1] gdb, infcmd: Support jump command in multi-
> inferior case.
>
> Fixes the issue where jump failed if multiple inferiors run the same source.
>
> See the below example
>
> $ gdb -q ./simple
> Reading symbols from ./simple...
> (gdb) break 2
> Breakpoint 1 at 0x114e: file simple.c, line 2.
> (gdb) run
> Starting program: /temp/simple
>
> Breakpoint 1, main () at simple.c:2
> 2 int a = 42;
> (gdb) add-inferior
> [New inferior 2]
> Added inferior 2 on connection 1 (native)
> (gdb) inferior 2
> [Switching to inferior 2 [<null>] (<noexec>)]
> (gdb) info inferiors
> Num Description Connection Executable
> 1 process 6250 1 (native) /temp/simple
> * 2 <null> 1 (native)
> (gdb) file ./simple
> Reading symbols from ./simple...
> (gdb) run
> Starting program: /temp/simple
>
> Thread 2.1 "simple" hit Breakpoint 1, main () at simple.c:2
> 2 int a = 42;
> (gdb) info inferiors
> Num Description Connection Executable
> 1 process 6250 1 (native) /temp/simple
> * 2 process 6705 1 (native) /temp/simple
> (gdb) jump 3
> Unreasonable jump request
> (gdb)
>
> In this example, jump fails because the debugger finds two different
> locations, one for each inferior.
> Solution is to limit the search to the current program space.
> ---
> gdb/infcmd.c | 3 +-
> gdb/linespec.c | 7 +-
> gdb/linespec.h | 5 +-
> gdb/testsuite/gdb.base/jump.exp | 177 +++++++++++++++++---------------
> 4 files changed, 103 insertions(+), 89 deletions(-)
>
> diff --git a/gdb/infcmd.c b/gdb/infcmd.c
> index 103899432f7..a79b5d3d4d6 100644
> --- a/gdb/infcmd.c
> +++ b/gdb/infcmd.c
> @@ -1077,7 +1077,8 @@ jump_command (const char *arg, int from_tty)
> error_no_arg (_("starting address"));
>
> std::vector<symtab_and_line> sals
> - = decode_line_with_last_displayed (arg, DECODE_LINE_FUNFIRSTLINE);
> + = decode_line_with_last_displayed (arg, DECODE_LINE_FUNFIRSTLINE,
> + current_program_space);
> if (sals.size () != 1)
> error (_("Unreasonable jump request"));
>
> diff --git a/gdb/linespec.c b/gdb/linespec.c
> index 7d969f37fbf..920fd3d37b7 100644
> --- a/gdb/linespec.c
> +++ b/gdb/linespec.c
> @@ -3231,7 +3231,8 @@ decode_line_with_current_source (const char
> *string, int flags)
> /* See linespec.h. */
>
> std::vector<symtab_and_line>
> -decode_line_with_last_displayed (const char *string, int flags)
> +decode_line_with_last_displayed (const char *string, int flags,
> + program_space *search_pspace)
> {
> if (string == 0)
> error (_("Empty line specification."));
> @@ -3240,10 +3241,10 @@ decode_line_with_last_displayed (const char
> *string, int flags)
> current_language);
> std::vector<symtab_and_line> sals
> = (last_displayed_sal_is_valid ()
> - ? decode_line_1 (locspec.get (), flags, NULL,
> + ? decode_line_1 (locspec.get (), flags, search_pspace,
> get_last_displayed_symtab (),
> get_last_displayed_line ())
> - : decode_line_1 (locspec.get (), flags, NULL, NULL, 0));
> + : decode_line_1 (locspec.get (), flags, search_pspace, NULL, 0));
>
> if (*string)
> error (_("Junk at end of line specification: %s"), string);
> diff --git a/gdb/linespec.h b/gdb/linespec.h
> index d5e7334fe2d..0eb9cb5d9f2 100644
> --- a/gdb/linespec.h
> +++ b/gdb/linespec.h
> @@ -139,10 +139,11 @@ extern std::vector<symtab_and_line>
> decode_line_with_current_source
> (const char *, int);
>
> /* Given a string, return the line specified by it, using the last displayed
> - codepoint's values as defaults, or nothing if they aren't valid. */
> + codepoint's values as defaults, or nothing if they aren't valid.
> + Limit the search to given program space, if specified. */
>
> extern std::vector<symtab_and_line> decode_line_with_last_displayed
> - (const char *, int);
> + (const char *, int, program_space *search_pspace = nullptr);
>
> /* Does P represent one of the keywords? If so, return
> the keyword. If not, return NULL. */
> diff --git a/gdb/testsuite/gdb.base/jump.exp
> b/gdb/testsuite/gdb.base/jump.exp
> index 032c4a6279d..4c0d4ed2dbb 100644
> --- a/gdb/testsuite/gdb.base/jump.exp
> +++ b/gdb/testsuite/gdb.base/jump.exp
> @@ -18,99 +18,110 @@ clear_xfail "*-*-*"
>
> standard_testfile .c
>
> -# Build the test case
> -if { [gdb_compile "${srcdir}/${subdir}/${srcfile}" "${binfile}" executable
> {debug nowarnings}] != "" } {
> - untested "failed to compile"
> - return -1
> - }
> +if {[prepare_for_testing "failed to prepare" $testfile $srcfile debug]} {
> + return -1
> +}
>
>
> -# Start with a fresh gdb
> +proc do_tests {} {
> + global decimal srcfile
> +
> + # Set a breakpoint on the statement that we're about to jump to.
> + # The statement doesn't contain a function call.
> + set non_call_line [gdb_get_line_number "bp-on-non-call"]
> + gdb_breakpoint "$non_call_line"
> + set bp_on_non_call \
> + [get_integer_valueof "\$bpnum" "INVALID" "bp_on_non_call"]
> +
> + # Can we jump to the statement? Do we stop there?
> + gdb_test "jump $non_call_line" \
> + "Breakpoint ${bp_on_non_call}(\.${decimal})?,
> .*${srcfile}:$non_call_line.*" \
> + "jump to non-call"
> +
> + # Set a breakpoint on the statement that we're about to jump to.
> + # The statement does contain a function call.
> + set call_line [gdb_get_line_number "bp-on-call"]
> + gdb_breakpoint "$call_line"
> + set bp_on_call [get_integer_valueof "\$bpnum" "INVALID" "bp_on_call"]
> +
> + # Can we jump to the statement? Do we stop there?
> + gdb_test "jump $call_line" \
> + "Breakpoint ${bp_on_call}(\.${decimal})?, .*${srcfile}:$call_line.*" \
> + "jump to call"
> +
> + # If we disable the breakpoint at the function call, and then
> + # if we jump to that statement, do we not stop there, but at
> + # the following breakpoint?
> + gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
> +
> + gdb_test "jump $call_line" \
> + "Breakpoint ${bp_on_non_call}(\.${decimal})?,
> .*${srcfile}:$non_call_line.*" \
> + "jump to call with disabled breakpoint"
> +
> + # Disable the breakpoint at the non-function call, so it won't hit
> + # if do_test is called again.
> + gdb_test_no_output "disable ${bp_on_non_call}" "disable
> bp_on_non_call"
> +
> + # Verify that GDB responds gracefully to the "jump" command without
> + # an argument.
> + gdb_test "jump" "Argument required .starting address.*" \
> + "jump without argument disallowed"
> +
> + # Verify that GDB responds gracefully to the "jump" command with
> + # trailing junk.
> + gdb_test "jump $call_line 100" \
> + "malformed linespec error: unexpected number, \"100\"" \
> + "jump with trailing argument junk"
> +
> + # Verify that GDB responds gracefully to a request to jump out of
> + # the current function. (Note that this will very likely cause the
> + # inferior to die. Be prepared to rerun the inferior, if further
> + # testing is desired.)
> + # Try it both ways: confirming and not confirming the jump.
> + set out_line [gdb_get_line_number "out-of-func"]
> + gdb_test "jump $out_line" \
> + "Not confirmed.*" \
> + "aborted jump out of current function" \
> + "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> + "n"
> +
> + gdb_test "jump $out_line" \
> + "Continuing at.*" \
> + "jump out of current function" \
> + "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> + "y"
> +}
>
> -clean_restart ${binfile}
>
> -if {![runto_main]} {
> - return -1
> +set inferiors 1
> +if {![use_gdb_stub]} {
> + set inferiors 2
> }
>
> -# Set a breakpoint on the statement that we're about to jump to.
> -# The statement doesn't contain a function call.
> -#
> -set bp_on_non_call 0
> -set non_call_line [gdb_get_line_number "bp-on-non-call"]
> -gdb_test_multiple "break $non_call_line" "break before jump to non-call" {
> - -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line
> $non_call_line.*$gdb_prompt $" {
> - set bp_on_non_call $expect_out(1,string)
> - pass "break before jump to non-call"
> +# Run to main, add inferiors if needed.
> +for {set inf 1} {$inf <= $inferiors} {incr inf} {
> + if {$inf != 1} {
> + # Start a new inferior, and run it with the same executable.
> + gdb_test "add-inferior -exec ${binfile}" \
> + "Added inferior ${inf}.*" \
> + "add inferior ${inf} with -exec "
> + gdb_test "inferior ${inf}" \
> + "Switching to inferior ${inf} .*" \
> + "switch to inferior ${inf}"
> + }
> + if {![runto_main]} {
> + return -1
> }
> }
>
> -# Can we jump to the statement? Do we stop there?
> -#
> -gdb_test "jump $non_call_line" "Breakpoint ${decimal},
> .*${srcfile}:$non_call_line.*" \
> - "jump to non-call"
> -
> -# Set a breakpoint on the statement that we're about to jump to.
> -# The statement does contain a function call.
> -#
> -set bp_on_call 0
> -set call_line [gdb_get_line_number "bp-on-call"]
> -gdb_test_multiple "break $call_line" "break before jump to call" {
> - -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line
> $call_line.*$gdb_prompt $" {
> - set bp_on_call $expect_out(1,string)
> - pass "break before jump to call"
> +# Run tests on all inferiors.
> +for {set inf 1} {$inf <= $inferiors} {incr inf} {
> + with_test_prefix "inferior $inf" {
> + # Switch to the target inferior.
> + gdb_test "inferior $inf" ".*Switching to inferior $inf .*"
> + # Run the tests.
> + do_tests
> }
> }
>
> -# Can we jump to the statement? Do we stop there?
> -#
> -gdb_test "jump $call_line" \
> - "Breakpoint ${decimal}, .*${srcfile}:$call_line.*" \
> - "jump to call"
> -
> -# If we disable the breakpoint at the function call, and then
> -# if we jump to that statement, do we not stop there, but at
> -# the following breakpoint?
> -#
> -gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
> -
> -gdb_test "jump $call_line" "Breakpoint ${decimal},
> .*${srcfile}:$non_call_line.*" \
> - "jump to call with disabled breakpoint"
> -
> -# Verify that GDB responds gracefully to the "jump" command without
> -# an argument.
> -#
> -gdb_test "jump" "Argument required .starting address.*" \
> - "jump without argument disallowed"
> -
> -
> -# Verify that GDB responds gracefully to the "jump" command with
> -# trailing junk.
> -#
> -gdb_test "jump $call_line 100" \
> - "malformed linespec error: unexpected number, \"100\"" \
> - "jump with trailing argument junk"
> -
> -
> -# Verify that GDB responds gracefully to a request to jump out of
> -# the current function. (Note that this will very likely cause the
> -# inferior to die. Be prepared to rerun the inferior, if further
> -# testing is desired.)
> -#
> -# Try it both ways: confirming and not confirming the jump.
> -#
> -
> -set out_line [gdb_get_line_number "out-of-func"]
> -gdb_test "jump $out_line" \
> - "Not confirmed.*" \
> - "aborted jump out of current function" \
> - "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> - "n"
> -
> -gdb_test "jump $out_line" \
> - "Continuing at.*" \
> - "jump out of current function" \
> - "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> - "y"
> -
> gdb_exit
> --
> 2.25.1
>
> Intel Deutschland GmbH
> Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
> Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
> Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
> Chairperson of the Supervisory Board: Nicole Lau
> Registered Office: Munich
> Commercial Register: Amtsgericht Muenchen HRB 186928
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [relevance 99%]
* RE: [PATCH v2 1/1] gdb, infcmd: Support jump command in multi-inferior case.
2023-04-24 14:10 99% ` Andrew Burgess
@ 2023-04-25 13:40 99% ` Puputti, Matti
0 siblings, 0 replies; 34+ results
From: Puputti, Matti @ 2023-04-25 13:40 UTC (permalink / raw)
To: Andrew Burgess, gdb-patches
Hi Andrew,
Thank you for your review and (conditional) approval.
I have uploaded new patch to address your comments. In addition I did my
best to clean out some other empty lines and trailing '#' lines, too.
And finally, I updated the commit date.
I hope the new patch is good.
Patch 3: https://sourceware.org/pipermail/gdb-patches/2023-April/199095.html
As you surely have noticed, this is my very first patch series landing in GDB.
If you are ok with the latest patch, can you push it for me, as I don't have
write access?
I plan to upstream more patches in the future and would like to have
"write after approval" access. Can you help me with that as well?
Br,
Matti Puputti
> -----Original Message-----
> From: Andrew Burgess <aburgess@redhat.com>
> Sent: Monday, April 24, 2023 4:11 PM
> To: Puputti, Matti <matti.puputti@intel.com>; gdb-patches@sourceware.org
> Subject: Re: [PATCH v2 1/1] gdb, infcmd: Support jump command in multi-
> inferior case.
>
> Matti Puputti <matti.puputti@intel.com> writes:
>
> > Fixes the issue where jump failed if multiple inferiors run the same source.
> >
> > See the below example
> >
> > $ gdb -q ./simple
> > Reading symbols from ./simple...
> > (gdb) break 2
> > Breakpoint 1 at 0x114e: file simple.c, line 2.
> > (gdb) run
> > Starting program: /temp/simple
> >
> > Breakpoint 1, main () at simple.c:2
> > 2 int a = 42;
> > (gdb) add-inferior
> > [New inferior 2]
> > Added inferior 2 on connection 1 (native)
> > (gdb) inferior 2
> > [Switching to inferior 2 [<null>] (<noexec>)]
> > (gdb) info inferiors
> > Num Description Connection Executable
> > 1 process 6250 1 (native) /temp/simple
> > * 2 <null> 1 (native)
> > (gdb) file ./simple
> > Reading symbols from ./simple...
> > (gdb) run
> > Starting program: /temp/simple
> >
> > Thread 2.1 "simple" hit Breakpoint 1, main () at simple.c:2
> > 2 int a = 42;
> > (gdb) info inferiors
> > Num Description Connection Executable
> > 1 process 6250 1 (native) /temp/simple
> > * 2 process 6705 1 (native) /temp/simple
> > (gdb) jump 3
> > Unreasonable jump request
> > (gdb)
> >
> > In this example, jump fails because the debugger finds two different
> > locations, one for each inferior.
> > Solution is to limit the search to the current program space.
>
> Thanks for the fixes. I have a couple of minor nits but I think this is
> pretty much OK.
>
> > ---
> > gdb/infcmd.c | 3 +-
> > gdb/linespec.c | 7 +-
> > gdb/linespec.h | 5 +-
> > gdb/testsuite/gdb.base/jump.exp | 193 ++++++++++++++++++-------------
> -
> > 4 files changed, 119 insertions(+), 89 deletions(-)
> >
> > diff --git a/gdb/infcmd.c b/gdb/infcmd.c
> > index a851fe1f8c8..18a537847cf 100644
> > --- a/gdb/infcmd.c
> > +++ b/gdb/infcmd.c
> > @@ -1078,7 +1078,8 @@ jump_command (const char *arg, int from_tty)
> > error_no_arg (_("starting address"));
> >
> > std::vector<symtab_and_line> sals
> > - = decode_line_with_last_displayed (arg, DECODE_LINE_FUNFIRSTLINE);
> > + = decode_line_with_last_displayed (arg, DECODE_LINE_FUNFIRSTLINE,
> > + current_program_space);
> > if (sals.size () != 1)
> > error (_("Unreasonable jump request"));
> >
> > diff --git a/gdb/linespec.c b/gdb/linespec.c
> > index 36f2ef46a7c..536636851e7 100644
> > --- a/gdb/linespec.c
> > +++ b/gdb/linespec.c
> > @@ -3231,7 +3231,8 @@ decode_line_with_current_source (const char
> *string, int flags)
> > /* See linespec.h. */
> >
> > std::vector<symtab_and_line>
> > -decode_line_with_last_displayed (const char *string, int flags)
> > +decode_line_with_last_displayed (const char *string, int flags,
> > + program_space *search_pspace)
> > {
> > if (string == 0)
> > error (_("Empty line specification."));
> > @@ -3240,10 +3241,10 @@ decode_line_with_last_displayed (const char
> *string, int flags)
> > current_language);
> > std::vector<symtab_and_line> sals
> > = (last_displayed_sal_is_valid ()
> > - ? decode_line_1 (locspec.get (), flags, NULL,
> > + ? decode_line_1 (locspec.get (), flags, search_pspace,
> > get_last_displayed_symtab (),
> > get_last_displayed_line ())
> > - : decode_line_1 (locspec.get (), flags, NULL, NULL, 0));
> > + : decode_line_1 (locspec.get (), flags, search_pspace, NULL, 0));
> >
> > if (*string)
> > error (_("Junk at end of line specification: %s"), string);
> > diff --git a/gdb/linespec.h b/gdb/linespec.h
> > index d5e7334fe2d..0eb9cb5d9f2 100644
> > --- a/gdb/linespec.h
> > +++ b/gdb/linespec.h
> > @@ -139,10 +139,11 @@ extern std::vector<symtab_and_line>
> decode_line_with_current_source
> > (const char *, int);
> >
> > /* Given a string, return the line specified by it, using the last displayed
> > - codepoint's values as defaults, or nothing if they aren't valid. */
> > + codepoint's values as defaults, or nothing if they aren't valid.
> > + Limit the search to given program space, if specified. */
> >
> > extern std::vector<symtab_and_line> decode_line_with_last_displayed
> > - (const char *, int);
> > + (const char *, int, program_space *search_pspace = nullptr);
> >
> > /* Does P represent one of the keywords? If so, return
> > the keyword. If not, return NULL. */
> > diff --git a/gdb/testsuite/gdb.base/jump.exp
> b/gdb/testsuite/gdb.base/jump.exp
> > index 032c4a6279d..ca214993025 100644
> > --- a/gdb/testsuite/gdb.base/jump.exp
> > +++ b/gdb/testsuite/gdb.base/jump.exp
> > @@ -18,99 +18,126 @@ clear_xfail "*-*-*"
> >
> > standard_testfile .c
> >
> > -# Build the test case
> > -if { [gdb_compile "${srcdir}/${subdir}/${srcfile}" "${binfile}" executable
> {debug nowarnings}] != "" } {
> > - untested "failed to compile"
> > - return -1
> > - }
> > +if {[prepare_for_testing "failed to prepare" $testfile $srcfile debug]} {
> > + return -1
> > +}
> >
> >
> > -# Start with a fresh gdb
> > +proc do_tests {} {
> > + global decimal srcfile
> > +
> > + # Set a breakpoint on the statement that we're about to jump to.
> > + # The statement doesn't contain a function call.
> > + #
> > + set bp_on_non_call 0
>
> This line is now redundant as we unconditionally set a value below.
>
> > + set non_call_line [gdb_get_line_number "bp-on-non-call"]
> > + gdb_breakpoint "$non_call_line"
> > + set bp_on_non_call [get_integer_valueof "\$bpnum" "INVALID"
> "bp_on_non_call"]
>
> This line is a bit long. Please wrap it somewhere.
>
> > +
> > + # Can we jump to the statement? Do we stop there?
> > + #
> > + gdb_test "jump $non_call_line" \
> > + "Breakpoint ${bp_on_non_call}(\.${decimal})?,
> .*${srcfile}:$non_call_line.*" \
> > + "jump to non-call"
> > +
> > + # Set a breakpoint on the statement that we're about to jump to.
> > + # The statement does contain a function call.
> > + #
> > + set bp_on_call 0
>
> Again this line can be removed.
>
> > + set call_line [gdb_get_line_number "bp-on-call"]
> > + gdb_breakpoint "$call_line"
> > + set bp_on_call [get_integer_valueof "\$bpnum" "INVALID"
> "bp_on_call"]
> > +
> > + # Can we jump to the statement? Do we stop there?
> > + #
> > + gdb_test "jump $call_line" \
> > + "Breakpoint ${bp_on_call}(\.${decimal})?, .*${srcfile}:$call_line.*" \
> > + "jump to call"
> > +
> > + # If we disable the breakpoint at the function call, and then
> > + # if we jump to that statement, do we not stop there, but at
> > + # the following breakpoint?
> > + #
> > + gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
> > +
> > + gdb_test "jump $call_line" \
> > + "Breakpoint ${bp_on_non_call}(\.${decimal})?,
> .*${srcfile}:$non_call_line.*" \
> > + "jump to call with disabled breakpoint"
> > +
> > + # Disable the breakpoint at the non-function call, so it won't be
> > + # there if do_test is called again.
>
> I think s/there/hit/ in this comment. When I read this I initially
> expected the code to be deleting the breakpoint rather than just
> disabling it.
>
> > + gdb_test_no_output "disable ${bp_on_non_call}" "disable
> bp_on_non_call"
> > +
> > +
>
> Can delete one of the empty lines added here.
>
> > + # Verify that GDB responds gracefully to the "jump" command without
> > + # an argument.
> > + #
> > + gdb_test "jump" "Argument required .starting address.*" \
> > + "jump without argument disallowed"
> > +
> > +
>
> Again, only need a single blank line here.
>
> > + # Verify that GDB responds gracefully to the "jump" command with
> > + # trailing junk.
> > + #
> > + gdb_test "jump $call_line 100" \
> > + "malformed linespec error: unexpected number, \"100\"" \
> > + "jump with trailing argument junk"
> > +
> > +
>
> And here.
>
> > + # Verify that GDB responds gracefully to a request to jump out of
> > + # the current function. (Note that this will very likely cause the
> > + # inferior to die. Be prepared to rerun the inferior, if further
> > + # testing is desired.)
> > + #
> > + # Try it both ways: confirming and not confirming the jump.
> > + #
>
> Drop this random trailing '#' line.
>
> > +
> > + set out_line [gdb_get_line_number "out-of-func"]
> > + gdb_test "jump $out_line" \
> > + "Not confirmed.*" \
> > + "aborted jump out of current function" \
> > + "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> > + "n"
> > +
> > + gdb_test "jump $out_line" \
> > + "Continuing at.*" \
> > + "jump out of current function" \
> > + "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> > + "y"
> > +}
> >
> > -clean_restart ${binfile}
> >
> > -if {![runto_main]} {
> > - return -1
> > +set inferiors 1
> > +if {![use_gdb_stub]} {
> > + set inferiors 2
> > }
> >
> > -# Set a breakpoint on the statement that we're about to jump to.
> > -# The statement doesn't contain a function call.
> > -#
> > -set bp_on_non_call 0
> > -set non_call_line [gdb_get_line_number "bp-on-non-call"]
> > -gdb_test_multiple "break $non_call_line" "break before jump to non-call"
> {
> > - -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line
> $non_call_line.*$gdb_prompt $" {
> > - set bp_on_non_call $expect_out(1,string)
> > - pass "break before jump to non-call"
> > +# Run to main, add inferiors if needed.
> > +for {set inf 1} {$inf <= $inferiors} {incr inf} {
> > + if {$inf != 1} {
> > + # Start a new inferior, and run it with the same executable.
> > + gdb_test "add-inferior -exec ${binfile}" \
> > + "Added inferior ${inf}.*" \
> > + "add inferior ${inf} with -exec "
> > + gdb_test "inferior ${inf}" \
> > + "Switching to inferior ${inf} .*" \
> > + "switch to inferior ${inf}"
> > + }
> > + if ![runto_main] then {
>
> We don't use the 'then' keyword, this should be written like this:
>
> if {![runto_main]} {
>
>
> > + perror "Couldn't run inferior ${inf} to main"
>
> I don't think this perror call is needed here. runto_main will already
> emit a FAIL if something goes wrong.
>
> > + return -1
> > }
> > }
> >
> > -# Can we jump to the statement? Do we stop there?
> > -#
> > -gdb_test "jump $non_call_line" "Breakpoint ${decimal},
> .*${srcfile}:$non_call_line.*" \
> > - "jump to non-call"
> > -
> > -# Set a breakpoint on the statement that we're about to jump to.
> > -# The statement does contain a function call.
> > -#
> > -set bp_on_call 0
> > -set call_line [gdb_get_line_number "bp-on-call"]
> > -gdb_test_multiple "break $call_line" "break before jump to call" {
> > - -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line
> $call_line.*$gdb_prompt $" {
> > - set bp_on_call $expect_out(1,string)
> > - pass "break before jump to call"
> > +# Run tests on all inferiors.
> > +for {set inf 1} {$inf <= $inferiors} {incr inf} {
> > + with_test_prefix "inferior $inf" {
> > + # Switch to the target inferior.
> > + gdb_test "inferior $inf" ".*Switching to inferior $inf .*"
> > + # Run the tests.
> > + do_tests
> > }
> > }
> >
> > -# Can we jump to the statement? Do we stop there?
> > -#
> > -gdb_test "jump $call_line" \
> > - "Breakpoint ${decimal}, .*${srcfile}:$call_line.*" \
> > - "jump to call"
> > -
> > -# If we disable the breakpoint at the function call, and then
> > -# if we jump to that statement, do we not stop there, but at
> > -# the following breakpoint?
> > -#
> > -gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
> > -
> > -gdb_test "jump $call_line" "Breakpoint ${decimal},
> .*${srcfile}:$non_call_line.*" \
> > - "jump to call with disabled breakpoint"
> > -
> > -# Verify that GDB responds gracefully to the "jump" command without
> > -# an argument.
> > -#
> > -gdb_test "jump" "Argument required .starting address.*" \
> > - "jump without argument disallowed"
> > -
> > -
> > -# Verify that GDB responds gracefully to the "jump" command with
> > -# trailing junk.
> > -#
> > -gdb_test "jump $call_line 100" \
> > - "malformed linespec error: unexpected number, \"100\"" \
> > - "jump with trailing argument junk"
> > -
> > -
> > -# Verify that GDB responds gracefully to a request to jump out of
> > -# the current function. (Note that this will very likely cause the
> > -# inferior to die. Be prepared to rerun the inferior, if further
> > -# testing is desired.)
> > -#
> > -# Try it both ways: confirming and not confirming the jump.
> > -#
> > -
> > -set out_line [gdb_get_line_number "out-of-func"]
> > -gdb_test "jump $out_line" \
> > - "Not confirmed.*" \
> > - "aborted jump out of current function" \
> > - "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> > - "n"
> > -
> > -gdb_test "jump $out_line" \
> > - "Continuing at.*" \
> > - "jump out of current function" \
> > - "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> > - "y"
> > -
> > gdb_exit
> > +return 0
>
> This 'return 0' can be dropped.
>
> With these fixed:
>
> Approved-By: Andrew Burgess <aburgess@redhat.com>
>
>
> Thanks,
> Andrew
>
> > --
> > 2.25.1
> >
> > Intel Deutschland GmbH
> > Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
> > Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
> > Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
> > Chairperson of the Supervisory Board: Nicole Lau
> > Registered Office: Munich
> > Commercial Register: Amtsgericht Muenchen HRB 186928
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [relevance 99%]
* [PATCH v3 1/1] gdb, infcmd: Support jump command in multi-inferior case.
@ 2023-04-25 13:38 64% Matti Puputti
2023-05-22 8:01 99% ` [PING][PATCH " Puputti, Matti
2023-05-24 17:33 91% ` [PATCH " Andrew Burgess
0 siblings, 2 replies; 34+ results
From: Matti Puputti @ 2023-04-25 13:38 UTC (permalink / raw)
To: gdb-patches; +Cc: aburgess
Fixes the issue where jump failed if multiple inferiors run the same source.
See the below example
$ gdb -q ./simple
Reading symbols from ./simple...
(gdb) break 2
Breakpoint 1 at 0x114e: file simple.c, line 2.
(gdb) run
Starting program: /temp/simple
Breakpoint 1, main () at simple.c:2
2 int a = 42;
(gdb) add-inferior
[New inferior 2]
Added inferior 2 on connection 1 (native)
(gdb) inferior 2
[Switching to inferior 2 [<null>] (<noexec>)]
(gdb) info inferiors
Num Description Connection Executable
1 process 6250 1 (native) /temp/simple
* 2 <null> 1 (native)
(gdb) file ./simple
Reading symbols from ./simple...
(gdb) run
Starting program: /temp/simple
Thread 2.1 "simple" hit Breakpoint 1, main () at simple.c:2
2 int a = 42;
(gdb) info inferiors
Num Description Connection Executable
1 process 6250 1 (native) /temp/simple
* 2 process 6705 1 (native) /temp/simple
(gdb) jump 3
Unreasonable jump request
(gdb)
In this example, jump fails because the debugger finds two different
locations, one for each inferior.
Solution is to limit the search to the current program space.
---
gdb/infcmd.c | 3 +-
gdb/linespec.c | 7 +-
gdb/linespec.h | 5 +-
gdb/testsuite/gdb.base/jump.exp | 177 +++++++++++++++++---------------
4 files changed, 103 insertions(+), 89 deletions(-)
diff --git a/gdb/infcmd.c b/gdb/infcmd.c
index 103899432f7..a79b5d3d4d6 100644
--- a/gdb/infcmd.c
+++ b/gdb/infcmd.c
@@ -1077,7 +1077,8 @@ jump_command (const char *arg, int from_tty)
error_no_arg (_("starting address"));
std::vector<symtab_and_line> sals
- = decode_line_with_last_displayed (arg, DECODE_LINE_FUNFIRSTLINE);
+ = decode_line_with_last_displayed (arg, DECODE_LINE_FUNFIRSTLINE,
+ current_program_space);
if (sals.size () != 1)
error (_("Unreasonable jump request"));
diff --git a/gdb/linespec.c b/gdb/linespec.c
index 7d969f37fbf..920fd3d37b7 100644
--- a/gdb/linespec.c
+++ b/gdb/linespec.c
@@ -3231,7 +3231,8 @@ decode_line_with_current_source (const char *string, int flags)
/* See linespec.h. */
std::vector<symtab_and_line>
-decode_line_with_last_displayed (const char *string, int flags)
+decode_line_with_last_displayed (const char *string, int flags,
+ program_space *search_pspace)
{
if (string == 0)
error (_("Empty line specification."));
@@ -3240,10 +3241,10 @@ decode_line_with_last_displayed (const char *string, int flags)
current_language);
std::vector<symtab_and_line> sals
= (last_displayed_sal_is_valid ()
- ? decode_line_1 (locspec.get (), flags, NULL,
+ ? decode_line_1 (locspec.get (), flags, search_pspace,
get_last_displayed_symtab (),
get_last_displayed_line ())
- : decode_line_1 (locspec.get (), flags, NULL, NULL, 0));
+ : decode_line_1 (locspec.get (), flags, search_pspace, NULL, 0));
if (*string)
error (_("Junk at end of line specification: %s"), string);
diff --git a/gdb/linespec.h b/gdb/linespec.h
index d5e7334fe2d..0eb9cb5d9f2 100644
--- a/gdb/linespec.h
+++ b/gdb/linespec.h
@@ -139,10 +139,11 @@ extern std::vector<symtab_and_line> decode_line_with_current_source
(const char *, int);
/* Given a string, return the line specified by it, using the last displayed
- codepoint's values as defaults, or nothing if they aren't valid. */
+ codepoint's values as defaults, or nothing if they aren't valid.
+ Limit the search to given program space, if specified. */
extern std::vector<symtab_and_line> decode_line_with_last_displayed
- (const char *, int);
+ (const char *, int, program_space *search_pspace = nullptr);
/* Does P represent one of the keywords? If so, return
the keyword. If not, return NULL. */
diff --git a/gdb/testsuite/gdb.base/jump.exp b/gdb/testsuite/gdb.base/jump.exp
index 032c4a6279d..4c0d4ed2dbb 100644
--- a/gdb/testsuite/gdb.base/jump.exp
+++ b/gdb/testsuite/gdb.base/jump.exp
@@ -18,99 +18,110 @@ clear_xfail "*-*-*"
standard_testfile .c
-# Build the test case
-if { [gdb_compile "${srcdir}/${subdir}/${srcfile}" "${binfile}" executable {debug nowarnings}] != "" } {
- untested "failed to compile"
- return -1
- }
+if {[prepare_for_testing "failed to prepare" $testfile $srcfile debug]} {
+ return -1
+}
-# Start with a fresh gdb
+proc do_tests {} {
+ global decimal srcfile
+
+ # Set a breakpoint on the statement that we're about to jump to.
+ # The statement doesn't contain a function call.
+ set non_call_line [gdb_get_line_number "bp-on-non-call"]
+ gdb_breakpoint "$non_call_line"
+ set bp_on_non_call \
+ [get_integer_valueof "\$bpnum" "INVALID" "bp_on_non_call"]
+
+ # Can we jump to the statement? Do we stop there?
+ gdb_test "jump $non_call_line" \
+ "Breakpoint ${bp_on_non_call}(\.${decimal})?, .*${srcfile}:$non_call_line.*" \
+ "jump to non-call"
+
+ # Set a breakpoint on the statement that we're about to jump to.
+ # The statement does contain a function call.
+ set call_line [gdb_get_line_number "bp-on-call"]
+ gdb_breakpoint "$call_line"
+ set bp_on_call [get_integer_valueof "\$bpnum" "INVALID" "bp_on_call"]
+
+ # Can we jump to the statement? Do we stop there?
+ gdb_test "jump $call_line" \
+ "Breakpoint ${bp_on_call}(\.${decimal})?, .*${srcfile}:$call_line.*" \
+ "jump to call"
+
+ # If we disable the breakpoint at the function call, and then
+ # if we jump to that statement, do we not stop there, but at
+ # the following breakpoint?
+ gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
+
+ gdb_test "jump $call_line" \
+ "Breakpoint ${bp_on_non_call}(\.${decimal})?, .*${srcfile}:$non_call_line.*" \
+ "jump to call with disabled breakpoint"
+
+ # Disable the breakpoint at the non-function call, so it won't hit
+ # if do_test is called again.
+ gdb_test_no_output "disable ${bp_on_non_call}" "disable bp_on_non_call"
+
+ # Verify that GDB responds gracefully to the "jump" command without
+ # an argument.
+ gdb_test "jump" "Argument required .starting address.*" \
+ "jump without argument disallowed"
+
+ # Verify that GDB responds gracefully to the "jump" command with
+ # trailing junk.
+ gdb_test "jump $call_line 100" \
+ "malformed linespec error: unexpected number, \"100\"" \
+ "jump with trailing argument junk"
+
+ # Verify that GDB responds gracefully to a request to jump out of
+ # the current function. (Note that this will very likely cause the
+ # inferior to die. Be prepared to rerun the inferior, if further
+ # testing is desired.)
+ # Try it both ways: confirming and not confirming the jump.
+ set out_line [gdb_get_line_number "out-of-func"]
+ gdb_test "jump $out_line" \
+ "Not confirmed.*" \
+ "aborted jump out of current function" \
+ "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
+ "n"
+
+ gdb_test "jump $out_line" \
+ "Continuing at.*" \
+ "jump out of current function" \
+ "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
+ "y"
+}
-clean_restart ${binfile}
-if {![runto_main]} {
- return -1
+set inferiors 1
+if {![use_gdb_stub]} {
+ set inferiors 2
}
-# Set a breakpoint on the statement that we're about to jump to.
-# The statement doesn't contain a function call.
-#
-set bp_on_non_call 0
-set non_call_line [gdb_get_line_number "bp-on-non-call"]
-gdb_test_multiple "break $non_call_line" "break before jump to non-call" {
- -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line $non_call_line.*$gdb_prompt $" {
- set bp_on_non_call $expect_out(1,string)
- pass "break before jump to non-call"
+# Run to main, add inferiors if needed.
+for {set inf 1} {$inf <= $inferiors} {incr inf} {
+ if {$inf != 1} {
+ # Start a new inferior, and run it with the same executable.
+ gdb_test "add-inferior -exec ${binfile}" \
+ "Added inferior ${inf}.*" \
+ "add inferior ${inf} with -exec "
+ gdb_test "inferior ${inf}" \
+ "Switching to inferior ${inf} .*" \
+ "switch to inferior ${inf}"
+ }
+ if {![runto_main]} {
+ return -1
}
}
-# Can we jump to the statement? Do we stop there?
-#
-gdb_test "jump $non_call_line" "Breakpoint ${decimal}, .*${srcfile}:$non_call_line.*" \
- "jump to non-call"
-
-# Set a breakpoint on the statement that we're about to jump to.
-# The statement does contain a function call.
-#
-set bp_on_call 0
-set call_line [gdb_get_line_number "bp-on-call"]
-gdb_test_multiple "break $call_line" "break before jump to call" {
- -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line $call_line.*$gdb_prompt $" {
- set bp_on_call $expect_out(1,string)
- pass "break before jump to call"
+# Run tests on all inferiors.
+for {set inf 1} {$inf <= $inferiors} {incr inf} {
+ with_test_prefix "inferior $inf" {
+ # Switch to the target inferior.
+ gdb_test "inferior $inf" ".*Switching to inferior $inf .*"
+ # Run the tests.
+ do_tests
}
}
-# Can we jump to the statement? Do we stop there?
-#
-gdb_test "jump $call_line" \
- "Breakpoint ${decimal}, .*${srcfile}:$call_line.*" \
- "jump to call"
-
-# If we disable the breakpoint at the function call, and then
-# if we jump to that statement, do we not stop there, but at
-# the following breakpoint?
-#
-gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
-
-gdb_test "jump $call_line" "Breakpoint ${decimal}, .*${srcfile}:$non_call_line.*" \
- "jump to call with disabled breakpoint"
-
-# Verify that GDB responds gracefully to the "jump" command without
-# an argument.
-#
-gdb_test "jump" "Argument required .starting address.*" \
- "jump without argument disallowed"
-
-
-# Verify that GDB responds gracefully to the "jump" command with
-# trailing junk.
-#
-gdb_test "jump $call_line 100" \
- "malformed linespec error: unexpected number, \"100\"" \
- "jump with trailing argument junk"
-
-
-# Verify that GDB responds gracefully to a request to jump out of
-# the current function. (Note that this will very likely cause the
-# inferior to die. Be prepared to rerun the inferior, if further
-# testing is desired.)
-#
-# Try it both ways: confirming and not confirming the jump.
-#
-
-set out_line [gdb_get_line_number "out-of-func"]
-gdb_test "jump $out_line" \
- "Not confirmed.*" \
- "aborted jump out of current function" \
- "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
- "n"
-
-gdb_test "jump $out_line" \
- "Continuing at.*" \
- "jump out of current function" \
- "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
- "y"
-
gdb_exit
--
2.25.1
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [relevance 64%]
* Re: [PATCH v2 1/1] gdb, infcmd: Support jump command in multi-inferior case.
2023-02-23 9:54 63% ` [PATCH v2 1/1] " Matti Puputti
@ 2023-04-24 14:10 99% ` Andrew Burgess
2023-04-25 13:40 99% ` Puputti, Matti
0 siblings, 1 reply; 34+ results
From: Andrew Burgess @ 2023-04-24 14:10 UTC (permalink / raw)
To: Matti Puputti, gdb-patches
Matti Puputti <matti.puputti@intel.com> writes:
> Fixes the issue where jump failed if multiple inferiors run the same source.
>
> See the below example
>
> $ gdb -q ./simple
> Reading symbols from ./simple...
> (gdb) break 2
> Breakpoint 1 at 0x114e: file simple.c, line 2.
> (gdb) run
> Starting program: /temp/simple
>
> Breakpoint 1, main () at simple.c:2
> 2 int a = 42;
> (gdb) add-inferior
> [New inferior 2]
> Added inferior 2 on connection 1 (native)
> (gdb) inferior 2
> [Switching to inferior 2 [<null>] (<noexec>)]
> (gdb) info inferiors
> Num Description Connection Executable
> 1 process 6250 1 (native) /temp/simple
> * 2 <null> 1 (native)
> (gdb) file ./simple
> Reading symbols from ./simple...
> (gdb) run
> Starting program: /temp/simple
>
> Thread 2.1 "simple" hit Breakpoint 1, main () at simple.c:2
> 2 int a = 42;
> (gdb) info inferiors
> Num Description Connection Executable
> 1 process 6250 1 (native) /temp/simple
> * 2 process 6705 1 (native) /temp/simple
> (gdb) jump 3
> Unreasonable jump request
> (gdb)
>
> In this example, jump fails because the debugger finds two different
> locations, one for each inferior.
> Solution is to limit the search to the current program space.
Thanks for the fixes. I have a couple of minor nits but I think this is
pretty much OK.
> ---
> gdb/infcmd.c | 3 +-
> gdb/linespec.c | 7 +-
> gdb/linespec.h | 5 +-
> gdb/testsuite/gdb.base/jump.exp | 193 ++++++++++++++++++--------------
> 4 files changed, 119 insertions(+), 89 deletions(-)
>
> diff --git a/gdb/infcmd.c b/gdb/infcmd.c
> index a851fe1f8c8..18a537847cf 100644
> --- a/gdb/infcmd.c
> +++ b/gdb/infcmd.c
> @@ -1078,7 +1078,8 @@ jump_command (const char *arg, int from_tty)
> error_no_arg (_("starting address"));
>
> std::vector<symtab_and_line> sals
> - = decode_line_with_last_displayed (arg, DECODE_LINE_FUNFIRSTLINE);
> + = decode_line_with_last_displayed (arg, DECODE_LINE_FUNFIRSTLINE,
> + current_program_space);
> if (sals.size () != 1)
> error (_("Unreasonable jump request"));
>
> diff --git a/gdb/linespec.c b/gdb/linespec.c
> index 36f2ef46a7c..536636851e7 100644
> --- a/gdb/linespec.c
> +++ b/gdb/linespec.c
> @@ -3231,7 +3231,8 @@ decode_line_with_current_source (const char *string, int flags)
> /* See linespec.h. */
>
> std::vector<symtab_and_line>
> -decode_line_with_last_displayed (const char *string, int flags)
> +decode_line_with_last_displayed (const char *string, int flags,
> + program_space *search_pspace)
> {
> if (string == 0)
> error (_("Empty line specification."));
> @@ -3240,10 +3241,10 @@ decode_line_with_last_displayed (const char *string, int flags)
> current_language);
> std::vector<symtab_and_line> sals
> = (last_displayed_sal_is_valid ()
> - ? decode_line_1 (locspec.get (), flags, NULL,
> + ? decode_line_1 (locspec.get (), flags, search_pspace,
> get_last_displayed_symtab (),
> get_last_displayed_line ())
> - : decode_line_1 (locspec.get (), flags, NULL, NULL, 0));
> + : decode_line_1 (locspec.get (), flags, search_pspace, NULL, 0));
>
> if (*string)
> error (_("Junk at end of line specification: %s"), string);
> diff --git a/gdb/linespec.h b/gdb/linespec.h
> index d5e7334fe2d..0eb9cb5d9f2 100644
> --- a/gdb/linespec.h
> +++ b/gdb/linespec.h
> @@ -139,10 +139,11 @@ extern std::vector<symtab_and_line> decode_line_with_current_source
> (const char *, int);
>
> /* Given a string, return the line specified by it, using the last displayed
> - codepoint's values as defaults, or nothing if they aren't valid. */
> + codepoint's values as defaults, or nothing if they aren't valid.
> + Limit the search to given program space, if specified. */
>
> extern std::vector<symtab_and_line> decode_line_with_last_displayed
> - (const char *, int);
> + (const char *, int, program_space *search_pspace = nullptr);
>
> /* Does P represent one of the keywords? If so, return
> the keyword. If not, return NULL. */
> diff --git a/gdb/testsuite/gdb.base/jump.exp b/gdb/testsuite/gdb.base/jump.exp
> index 032c4a6279d..ca214993025 100644
> --- a/gdb/testsuite/gdb.base/jump.exp
> +++ b/gdb/testsuite/gdb.base/jump.exp
> @@ -18,99 +18,126 @@ clear_xfail "*-*-*"
>
> standard_testfile .c
>
> -# Build the test case
> -if { [gdb_compile "${srcdir}/${subdir}/${srcfile}" "${binfile}" executable {debug nowarnings}] != "" } {
> - untested "failed to compile"
> - return -1
> - }
> +if {[prepare_for_testing "failed to prepare" $testfile $srcfile debug]} {
> + return -1
> +}
>
>
> -# Start with a fresh gdb
> +proc do_tests {} {
> + global decimal srcfile
> +
> + # Set a breakpoint on the statement that we're about to jump to.
> + # The statement doesn't contain a function call.
> + #
> + set bp_on_non_call 0
This line is now redundant as we unconditionally set a value below.
> + set non_call_line [gdb_get_line_number "bp-on-non-call"]
> + gdb_breakpoint "$non_call_line"
> + set bp_on_non_call [get_integer_valueof "\$bpnum" "INVALID" "bp_on_non_call"]
This line is a bit long. Please wrap it somewhere.
> +
> + # Can we jump to the statement? Do we stop there?
> + #
> + gdb_test "jump $non_call_line" \
> + "Breakpoint ${bp_on_non_call}(\.${decimal})?, .*${srcfile}:$non_call_line.*" \
> + "jump to non-call"
> +
> + # Set a breakpoint on the statement that we're about to jump to.
> + # The statement does contain a function call.
> + #
> + set bp_on_call 0
Again this line can be removed.
> + set call_line [gdb_get_line_number "bp-on-call"]
> + gdb_breakpoint "$call_line"
> + set bp_on_call [get_integer_valueof "\$bpnum" "INVALID" "bp_on_call"]
> +
> + # Can we jump to the statement? Do we stop there?
> + #
> + gdb_test "jump $call_line" \
> + "Breakpoint ${bp_on_call}(\.${decimal})?, .*${srcfile}:$call_line.*" \
> + "jump to call"
> +
> + # If we disable the breakpoint at the function call, and then
> + # if we jump to that statement, do we not stop there, but at
> + # the following breakpoint?
> + #
> + gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
> +
> + gdb_test "jump $call_line" \
> + "Breakpoint ${bp_on_non_call}(\.${decimal})?, .*${srcfile}:$non_call_line.*" \
> + "jump to call with disabled breakpoint"
> +
> + # Disable the breakpoint at the non-function call, so it won't be
> + # there if do_test is called again.
I think s/there/hit/ in this comment. When I read this I initially
expected the code to be deleting the breakpoint rather than just
disabling it.
> + gdb_test_no_output "disable ${bp_on_non_call}" "disable bp_on_non_call"
> +
> +
Can delete one of the empty lines added here.
> + # Verify that GDB responds gracefully to the "jump" command without
> + # an argument.
> + #
> + gdb_test "jump" "Argument required .starting address.*" \
> + "jump without argument disallowed"
> +
> +
Again, only need a single blank line here.
> + # Verify that GDB responds gracefully to the "jump" command with
> + # trailing junk.
> + #
> + gdb_test "jump $call_line 100" \
> + "malformed linespec error: unexpected number, \"100\"" \
> + "jump with trailing argument junk"
> +
> +
And here.
> + # Verify that GDB responds gracefully to a request to jump out of
> + # the current function. (Note that this will very likely cause the
> + # inferior to die. Be prepared to rerun the inferior, if further
> + # testing is desired.)
> + #
> + # Try it both ways: confirming and not confirming the jump.
> + #
Drop this random trailing '#' line.
> +
> + set out_line [gdb_get_line_number "out-of-func"]
> + gdb_test "jump $out_line" \
> + "Not confirmed.*" \
> + "aborted jump out of current function" \
> + "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> + "n"
> +
> + gdb_test "jump $out_line" \
> + "Continuing at.*" \
> + "jump out of current function" \
> + "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> + "y"
> +}
>
> -clean_restart ${binfile}
>
> -if {![runto_main]} {
> - return -1
> +set inferiors 1
> +if {![use_gdb_stub]} {
> + set inferiors 2
> }
>
> -# Set a breakpoint on the statement that we're about to jump to.
> -# The statement doesn't contain a function call.
> -#
> -set bp_on_non_call 0
> -set non_call_line [gdb_get_line_number "bp-on-non-call"]
> -gdb_test_multiple "break $non_call_line" "break before jump to non-call" {
> - -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line $non_call_line.*$gdb_prompt $" {
> - set bp_on_non_call $expect_out(1,string)
> - pass "break before jump to non-call"
> +# Run to main, add inferiors if needed.
> +for {set inf 1} {$inf <= $inferiors} {incr inf} {
> + if {$inf != 1} {
> + # Start a new inferior, and run it with the same executable.
> + gdb_test "add-inferior -exec ${binfile}" \
> + "Added inferior ${inf}.*" \
> + "add inferior ${inf} with -exec "
> + gdb_test "inferior ${inf}" \
> + "Switching to inferior ${inf} .*" \
> + "switch to inferior ${inf}"
> + }
> + if ![runto_main] then {
We don't use the 'then' keyword, this should be written like this:
if {![runto_main]} {
> + perror "Couldn't run inferior ${inf} to main"
I don't think this perror call is needed here. runto_main will already
emit a FAIL if something goes wrong.
> + return -1
> }
> }
>
> -# Can we jump to the statement? Do we stop there?
> -#
> -gdb_test "jump $non_call_line" "Breakpoint ${decimal}, .*${srcfile}:$non_call_line.*" \
> - "jump to non-call"
> -
> -# Set a breakpoint on the statement that we're about to jump to.
> -# The statement does contain a function call.
> -#
> -set bp_on_call 0
> -set call_line [gdb_get_line_number "bp-on-call"]
> -gdb_test_multiple "break $call_line" "break before jump to call" {
> - -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line $call_line.*$gdb_prompt $" {
> - set bp_on_call $expect_out(1,string)
> - pass "break before jump to call"
> +# Run tests on all inferiors.
> +for {set inf 1} {$inf <= $inferiors} {incr inf} {
> + with_test_prefix "inferior $inf" {
> + # Switch to the target inferior.
> + gdb_test "inferior $inf" ".*Switching to inferior $inf .*"
> + # Run the tests.
> + do_tests
> }
> }
>
> -# Can we jump to the statement? Do we stop there?
> -#
> -gdb_test "jump $call_line" \
> - "Breakpoint ${decimal}, .*${srcfile}:$call_line.*" \
> - "jump to call"
> -
> -# If we disable the breakpoint at the function call, and then
> -# if we jump to that statement, do we not stop there, but at
> -# the following breakpoint?
> -#
> -gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
> -
> -gdb_test "jump $call_line" "Breakpoint ${decimal}, .*${srcfile}:$non_call_line.*" \
> - "jump to call with disabled breakpoint"
> -
> -# Verify that GDB responds gracefully to the "jump" command without
> -# an argument.
> -#
> -gdb_test "jump" "Argument required .starting address.*" \
> - "jump without argument disallowed"
> -
> -
> -# Verify that GDB responds gracefully to the "jump" command with
> -# trailing junk.
> -#
> -gdb_test "jump $call_line 100" \
> - "malformed linespec error: unexpected number, \"100\"" \
> - "jump with trailing argument junk"
> -
> -
> -# Verify that GDB responds gracefully to a request to jump out of
> -# the current function. (Note that this will very likely cause the
> -# inferior to die. Be prepared to rerun the inferior, if further
> -# testing is desired.)
> -#
> -# Try it both ways: confirming and not confirming the jump.
> -#
> -
> -set out_line [gdb_get_line_number "out-of-func"]
> -gdb_test "jump $out_line" \
> - "Not confirmed.*" \
> - "aborted jump out of current function" \
> - "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> - "n"
> -
> -gdb_test "jump $out_line" \
> - "Continuing at.*" \
> - "jump out of current function" \
> - "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> - "y"
> -
> gdb_exit
> +return 0
This 'return 0' can be dropped.
With these fixed:
Approved-By: Andrew Burgess <aburgess@redhat.com>
Thanks,
Andrew
> --
> 2.25.1
>
> Intel Deutschland GmbH
> Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
> Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
> Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
> Chairperson of the Supervisory Board: Nicole Lau
> Registered Office: Munich
> Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [relevance 99%]
* RE: [PING] [PATCH v2 0/1] gdb, infcmd: Support jump command in multi-inferior case.
2023-04-14 14:24 99% ` Puputti, Matti
@ 2023-04-21 7:24 99% ` Puputti, Matti
0 siblings, 0 replies; 34+ results
From: Puputti, Matti @ 2023-04-21 7:24 UTC (permalink / raw)
To: gdb-patches
*Ping* v6
Kindly asking for review.
Br,
Matti Puputti
Patch 1: https://sourceware.org/pipermail/gdb-patches/2023-February/197269.html
Andrew Burgess' comments on Patch 1: https://sourceware.org/pipermail/gdb-patches/2023-February/197271.html
Patch 2: https://sourceware.org/pipermail/gdb-patches/2023-February/197302.html
>
> *Ping* v5
>
> >
> > *Ping* v4
> >
> > >
> > > *Ping* v3
> > >
> > > >
> > > > *Ping* v2
> > > >
> > > > >
> > > > > Gently pinging for a review.
> > > > >
> > > > > Br,
> > > > > Matti Puputti
> > > > >
> > > > > > -----Original Message-----
> > > > > > From: Gdb-patches <gdb-patches-
> > > > > > bounces+matti.puputti=intel.com@sourceware.org> On Behalf Of
> > > Matti
> > > > > Puputti
> > > > > > via Gdb-patches
> > > > > > Sent: Thursday, February 23, 2023 10:55 AM
> > > > > > To: gdb-patches@sourceware.org
> > > > > > Cc: aburgess@redhat.com
> > > > > > Subject: [PATCH v2 0/1] gdb, infcmd: Support jump command in
> > multi-
> > > > > inferior
> > > > > > case.
> > > > > >
> > > > > > Thanks for the feedback on v1.
> > > > > >
> > > > > > Changes in v2:
> > > > > > - In linespec.h, updated comment to explain the new parameter
> > > > > > search_pspace.
> > > > > >
> > > > > > - In jump.exp, updated to use prepare_for_testing.
> > > > > >
> > > > > > - In jump.exp, updated how bp_on_non_call and bp_on_call are
> set,
> > > > > > updated regexes to use these two variables, and added disabling
> > > > > > of BP $bp_on_non_call.
> > > > > >
> > > > > > Br,
> > > > > > Matti Puputti
> > > > > >
> > > > > > ---
> > > > > >
> > > > > > Matti Puputti (1):
> > > > > > gdb, infcmd: Support jump command in multi-inferior case.
> > > > > >
> > > > > > gdb/infcmd.c | 3 +-
> > > > > > gdb/linespec.c | 7 +-
> > > > > > gdb/linespec.h | 5 +-
> > > > > > gdb/testsuite/gdb.base/jump.exp | 193 ++++++++++++++++++----
> --
> > --
> > > --
> > > > ---
> > > > > -
> > > > > > 4 files changed, 119 insertions(+), 89 deletions(-)
> > > > > >
> > > > > > --
> > > > > > 2.25.1
> > > > > >
> > > > > > Intel Deutschland GmbH
> > > > > > Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
> > > > > > Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
> > > > > > Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany
> Doon
> > > Silva
> > > > > > Chairperson of the Supervisory Board: Nicole Lau
> > > > > > Registered Office: Munich
> > > > > > Commercial Register: Amtsgericht Muenchen HRB 186928
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [relevance 99%]
* RE: [PING] [PATCH v2 0/1] gdb, infcmd: Support jump command in multi-inferior case.
2023-04-06 12:59 99% ` Puputti, Matti
@ 2023-04-14 14:24 99% ` Puputti, Matti
2023-04-21 7:24 99% ` Puputti, Matti
0 siblings, 1 reply; 34+ results
From: Puputti, Matti @ 2023-04-14 14:24 UTC (permalink / raw)
To: gdb-patches
*Ping* v5
>
> *Ping* v4
>
> >
> > *Ping* v3
> >
> > >
> > > *Ping* v2
> > >
> > > >
> > > > Gently pinging for a review.
> > > >
> > > > Br,
> > > > Matti Puputti
> > > >
> > > > > -----Original Message-----
> > > > > From: Gdb-patches <gdb-patches-
> > > > > bounces+matti.puputti=intel.com@sourceware.org> On Behalf Of
> > Matti
> > > > Puputti
> > > > > via Gdb-patches
> > > > > Sent: Thursday, February 23, 2023 10:55 AM
> > > > > To: gdb-patches@sourceware.org
> > > > > Cc: aburgess@redhat.com
> > > > > Subject: [PATCH v2 0/1] gdb, infcmd: Support jump command in
> multi-
> > > > inferior
> > > > > case.
> > > > >
> > > > > Thanks for the feedback on v1.
> > > > >
> > > > > Changes in v2:
> > > > > - In linespec.h, updated comment to explain the new parameter
> > > > > search_pspace.
> > > > >
> > > > > - In jump.exp, updated to use prepare_for_testing.
> > > > >
> > > > > - In jump.exp, updated how bp_on_non_call and bp_on_call are set,
> > > > > updated regexes to use these two variables, and added disabling
> > > > > of BP $bp_on_non_call.
> > > > >
> > > > > Br,
> > > > > Matti Puputti
> > > > >
> > > > > ---
> > > > >
> > > > > Matti Puputti (1):
> > > > > gdb, infcmd: Support jump command in multi-inferior case.
> > > > >
> > > > > gdb/infcmd.c | 3 +-
> > > > > gdb/linespec.c | 7 +-
> > > > > gdb/linespec.h | 5 +-
> > > > > gdb/testsuite/gdb.base/jump.exp | 193 ++++++++++++++++++------
> --
> > --
> > > ---
> > > > -
> > > > > 4 files changed, 119 insertions(+), 89 deletions(-)
> > > > >
> > > > > --
> > > > > 2.25.1
> > > > >
> > > > > Intel Deutschland GmbH
> > > > > Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
> > > > > Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
> > > > > Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon
> > Silva
> > > > > Chairperson of the Supervisory Board: Nicole Lau
> > > > > Registered Office: Munich
> > > > > Commercial Register: Amtsgericht Muenchen HRB 186928
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [relevance 99%]
* RE: [PING] [PATCH v2 0/1] gdb, infcmd: Support jump command in multi-inferior case.
2023-03-31 5:04 99% ` Puputti, Matti
@ 2023-04-06 12:59 99% ` Puputti, Matti
2023-04-14 14:24 99% ` Puputti, Matti
0 siblings, 1 reply; 34+ results
From: Puputti, Matti @ 2023-04-06 12:59 UTC (permalink / raw)
To: gdb-patches
*Ping* v4
> -----Original Message-----
> From: Puputti, Matti
> Sent: Friday, March 31, 2023 7:05 AM
> To: 'gdb-patches@sourceware.org' <gdb-patches@sourceware.org>
> Cc: 'aburgess@redhat.com' <aburgess@redhat.com>
> Subject: RE: [PING] [PATCH v2 0/1] gdb, infcmd: Support jump command in
> multi-inferior case.
>
> *Ping* v3
>
> > -----Original Message-----
> > From: Puputti, Matti
> > Sent: Friday, March 24, 2023 8:19 AM
> > To: gdb-patches@sourceware.org
> > Cc: aburgess@redhat.com
> > Subject: RE: [PING] [PATCH v2 0/1] gdb, infcmd: Support jump command in
> > multi-inferior case.
> >
> > *Ping* v2
> >
> > Br,
> > Matti Puputti
> >
> > > -----Original Message-----
> > > From: Puputti, Matti
> > > Sent: Tuesday, March 7, 2023 5:50 PM
> > > To: gdb-patches@sourceware.org
> > > Cc: aburgess@redhat.com
> > > Subject: [PING] [PATCH v2 0/1] gdb, infcmd: Support jump command in
> > > multi-inferior case.
> > >
> > > Gently pinging for a review.
> > >
> > > Br,
> > > Matti Puputti
> > >
> > > > -----Original Message-----
> > > > From: Gdb-patches <gdb-patches-
> > > > bounces+matti.puputti=intel.com@sourceware.org> On Behalf Of
> Matti
> > > Puputti
> > > > via Gdb-patches
> > > > Sent: Thursday, February 23, 2023 10:55 AM
> > > > To: gdb-patches@sourceware.org
> > > > Cc: aburgess@redhat.com
> > > > Subject: [PATCH v2 0/1] gdb, infcmd: Support jump command in multi-
> > > inferior
> > > > case.
> > > >
> > > > Thanks for the feedback on v1.
> > > >
> > > > Changes in v2:
> > > > - In linespec.h, updated comment to explain the new parameter
> > > > search_pspace.
> > > >
> > > > - In jump.exp, updated to use prepare_for_testing.
> > > >
> > > > - In jump.exp, updated how bp_on_non_call and bp_on_call are set,
> > > > updated regexes to use these two variables, and added disabling
> > > > of BP $bp_on_non_call.
> > > >
> > > > Br,
> > > > Matti Puputti
> > > >
> > > > ---
> > > >
> > > > Matti Puputti (1):
> > > > gdb, infcmd: Support jump command in multi-inferior case.
> > > >
> > > > gdb/infcmd.c | 3 +-
> > > > gdb/linespec.c | 7 +-
> > > > gdb/linespec.h | 5 +-
> > > > gdb/testsuite/gdb.base/jump.exp | 193 ++++++++++++++++++--------
> --
> > ---
> > > -
> > > > 4 files changed, 119 insertions(+), 89 deletions(-)
> > > >
> > > > --
> > > > 2.25.1
> > > >
> > > > Intel Deutschland GmbH
> > > > Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
> > > > Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
> > > > Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon
> Silva
> > > > Chairperson of the Supervisory Board: Nicole Lau
> > > > Registered Office: Munich
> > > > Commercial Register: Amtsgericht Muenchen HRB 186928
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [relevance 99%]
* RE: [PING] [PATCH v2 0/1] gdb, infcmd: Support jump command in multi-inferior case.
2023-03-24 7:19 99% ` Puputti, Matti
@ 2023-03-31 5:04 99% ` Puputti, Matti
2023-04-06 12:59 99% ` Puputti, Matti
0 siblings, 1 reply; 34+ results
From: Puputti, Matti @ 2023-03-31 5:04 UTC (permalink / raw)
To: gdb-patches; +Cc: aburgess
*Ping* v3
> -----Original Message-----
> From: Puputti, Matti
> Sent: Friday, March 24, 2023 8:19 AM
> To: gdb-patches@sourceware.org
> Cc: aburgess@redhat.com
> Subject: RE: [PING] [PATCH v2 0/1] gdb, infcmd: Support jump command in
> multi-inferior case.
>
> *Ping* v2
>
> Br,
> Matti Puputti
>
> > -----Original Message-----
> > From: Puputti, Matti
> > Sent: Tuesday, March 7, 2023 5:50 PM
> > To: gdb-patches@sourceware.org
> > Cc: aburgess@redhat.com
> > Subject: [PING] [PATCH v2 0/1] gdb, infcmd: Support jump command in
> > multi-inferior case.
> >
> > Gently pinging for a review.
> >
> > Br,
> > Matti Puputti
> >
> > > -----Original Message-----
> > > From: Gdb-patches <gdb-patches-
> > > bounces+matti.puputti=intel.com@sourceware.org> On Behalf Of Matti
> > Puputti
> > > via Gdb-patches
> > > Sent: Thursday, February 23, 2023 10:55 AM
> > > To: gdb-patches@sourceware.org
> > > Cc: aburgess@redhat.com
> > > Subject: [PATCH v2 0/1] gdb, infcmd: Support jump command in multi-
> > inferior
> > > case.
> > >
> > > Thanks for the feedback on v1.
> > >
> > > Changes in v2:
> > > - In linespec.h, updated comment to explain the new parameter
> > > search_pspace.
> > >
> > > - In jump.exp, updated to use prepare_for_testing.
> > >
> > > - In jump.exp, updated how bp_on_non_call and bp_on_call are set,
> > > updated regexes to use these two variables, and added disabling
> > > of BP $bp_on_non_call.
> > >
> > > Br,
> > > Matti Puputti
> > >
> > > ---
> > >
> > > Matti Puputti (1):
> > > gdb, infcmd: Support jump command in multi-inferior case.
> > >
> > > gdb/infcmd.c | 3 +-
> > > gdb/linespec.c | 7 +-
> > > gdb/linespec.h | 5 +-
> > > gdb/testsuite/gdb.base/jump.exp | 193 ++++++++++++++++++----------
> ---
> > -
> > > 4 files changed, 119 insertions(+), 89 deletions(-)
> > >
> > > --
> > > 2.25.1
> > >
> > > Intel Deutschland GmbH
> > > Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
> > > Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
> > > Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
> > > Chairperson of the Supervisory Board: Nicole Lau
> > > Registered Office: Munich
> > > Commercial Register: Amtsgericht Muenchen HRB 186928
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [relevance 99%]
* RE: [PING] [PATCH v2 0/1] gdb, infcmd: Support jump command in multi-inferior case.
2023-03-07 16:50 99% ` [PING] [PATCH v2 0/1] " Puputti, Matti
@ 2023-03-24 7:19 99% ` Puputti, Matti
2023-03-31 5:04 99% ` Puputti, Matti
0 siblings, 1 reply; 34+ results
From: Puputti, Matti @ 2023-03-24 7:19 UTC (permalink / raw)
To: gdb-patches; +Cc: aburgess
*Ping* v2
Br,
Matti Puputti
> -----Original Message-----
> From: Puputti, Matti
> Sent: Tuesday, March 7, 2023 5:50 PM
> To: gdb-patches@sourceware.org
> Cc: aburgess@redhat.com
> Subject: [PING] [PATCH v2 0/1] gdb, infcmd: Support jump command in
> multi-inferior case.
>
> Gently pinging for a review.
>
> Br,
> Matti Puputti
>
> > -----Original Message-----
> > From: Gdb-patches <gdb-patches-
> > bounces+matti.puputti=intel.com@sourceware.org> On Behalf Of Matti
> Puputti
> > via Gdb-patches
> > Sent: Thursday, February 23, 2023 10:55 AM
> > To: gdb-patches@sourceware.org
> > Cc: aburgess@redhat.com
> > Subject: [PATCH v2 0/1] gdb, infcmd: Support jump command in multi-
> inferior
> > case.
> >
> > Thanks for the feedback on v1.
> >
> > Changes in v2:
> > - In linespec.h, updated comment to explain the new parameter
> > search_pspace.
> >
> > - In jump.exp, updated to use prepare_for_testing.
> >
> > - In jump.exp, updated how bp_on_non_call and bp_on_call are set,
> > updated regexes to use these two variables, and added disabling
> > of BP $bp_on_non_call.
> >
> > Br,
> > Matti Puputti
> >
> > ---
> >
> > Matti Puputti (1):
> > gdb, infcmd: Support jump command in multi-inferior case.
> >
> > gdb/infcmd.c | 3 +-
> > gdb/linespec.c | 7 +-
> > gdb/linespec.h | 5 +-
> > gdb/testsuite/gdb.base/jump.exp | 193 ++++++++++++++++++-------------
> -
> > 4 files changed, 119 insertions(+), 89 deletions(-)
> >
> > --
> > 2.25.1
> >
> > Intel Deutschland GmbH
> > Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
> > Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
> > Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
> > Chairperson of the Supervisory Board: Nicole Lau
> > Registered Office: Munich
> > Commercial Register: Amtsgericht Muenchen HRB 186928
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [relevance 99%]
* [PING] [PATCH v2 0/1] gdb, infcmd: Support jump command in multi-inferior case.
2023-02-23 9:54 99% [PATCH v2 0/1] " Matti Puputti
2023-02-23 9:54 63% ` [PATCH v2 1/1] " Matti Puputti
@ 2023-03-07 16:50 99% ` Puputti, Matti
2023-03-24 7:19 99% ` Puputti, Matti
1 sibling, 1 reply; 34+ results
From: Puputti, Matti @ 2023-03-07 16:50 UTC (permalink / raw)
To: gdb-patches; +Cc: aburgess
Gently pinging for a review.
Br,
Matti Puputti
> -----Original Message-----
> From: Gdb-patches <gdb-patches-
> bounces+matti.puputti=intel.com@sourceware.org> On Behalf Of Matti Puputti
> via Gdb-patches
> Sent: Thursday, February 23, 2023 10:55 AM
> To: gdb-patches@sourceware.org
> Cc: aburgess@redhat.com
> Subject: [PATCH v2 0/1] gdb, infcmd: Support jump command in multi-inferior
> case.
>
> Thanks for the feedback on v1.
>
> Changes in v2:
> - In linespec.h, updated comment to explain the new parameter
> search_pspace.
>
> - In jump.exp, updated to use prepare_for_testing.
>
> - In jump.exp, updated how bp_on_non_call and bp_on_call are set,
> updated regexes to use these two variables, and added disabling
> of BP $bp_on_non_call.
>
> Br,
> Matti Puputti
>
> ---
>
> Matti Puputti (1):
> gdb, infcmd: Support jump command in multi-inferior case.
>
> gdb/infcmd.c | 3 +-
> gdb/linespec.c | 7 +-
> gdb/linespec.h | 5 +-
> gdb/testsuite/gdb.base/jump.exp | 193 ++++++++++++++++++--------------
> 4 files changed, 119 insertions(+), 89 deletions(-)
>
> --
> 2.25.1
>
> Intel Deutschland GmbH
> Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
> Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
> Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
> Chairperson of the Supervisory Board: Nicole Lau
> Registered Office: Munich
> Commercial Register: Amtsgericht Muenchen HRB 186928
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [relevance 99%]
* [PATCH v2 1/1] gdb, infcmd: Support jump command in multi-inferior case.
2023-02-23 9:54 99% [PATCH v2 0/1] " Matti Puputti
@ 2023-02-23 9:54 63% ` Matti Puputti
2023-04-24 14:10 99% ` Andrew Burgess
2023-03-07 16:50 99% ` [PING] [PATCH v2 0/1] " Puputti, Matti
1 sibling, 1 reply; 34+ results
From: Matti Puputti @ 2023-02-23 9:54 UTC (permalink / raw)
To: gdb-patches; +Cc: aburgess
Fixes the issue where jump failed if multiple inferiors run the same source.
See the below example
$ gdb -q ./simple
Reading symbols from ./simple...
(gdb) break 2
Breakpoint 1 at 0x114e: file simple.c, line 2.
(gdb) run
Starting program: /temp/simple
Breakpoint 1, main () at simple.c:2
2 int a = 42;
(gdb) add-inferior
[New inferior 2]
Added inferior 2 on connection 1 (native)
(gdb) inferior 2
[Switching to inferior 2 [<null>] (<noexec>)]
(gdb) info inferiors
Num Description Connection Executable
1 process 6250 1 (native) /temp/simple
* 2 <null> 1 (native)
(gdb) file ./simple
Reading symbols from ./simple...
(gdb) run
Starting program: /temp/simple
Thread 2.1 "simple" hit Breakpoint 1, main () at simple.c:2
2 int a = 42;
(gdb) info inferiors
Num Description Connection Executable
1 process 6250 1 (native) /temp/simple
* 2 process 6705 1 (native) /temp/simple
(gdb) jump 3
Unreasonable jump request
(gdb)
In this example, jump fails because the debugger finds two different
locations, one for each inferior.
Solution is to limit the search to the current program space.
---
gdb/infcmd.c | 3 +-
gdb/linespec.c | 7 +-
gdb/linespec.h | 5 +-
gdb/testsuite/gdb.base/jump.exp | 193 ++++++++++++++++++--------------
4 files changed, 119 insertions(+), 89 deletions(-)
diff --git a/gdb/infcmd.c b/gdb/infcmd.c
index a851fe1f8c8..18a537847cf 100644
--- a/gdb/infcmd.c
+++ b/gdb/infcmd.c
@@ -1078,7 +1078,8 @@ jump_command (const char *arg, int from_tty)
error_no_arg (_("starting address"));
std::vector<symtab_and_line> sals
- = decode_line_with_last_displayed (arg, DECODE_LINE_FUNFIRSTLINE);
+ = decode_line_with_last_displayed (arg, DECODE_LINE_FUNFIRSTLINE,
+ current_program_space);
if (sals.size () != 1)
error (_("Unreasonable jump request"));
diff --git a/gdb/linespec.c b/gdb/linespec.c
index 36f2ef46a7c..536636851e7 100644
--- a/gdb/linespec.c
+++ b/gdb/linespec.c
@@ -3231,7 +3231,8 @@ decode_line_with_current_source (const char *string, int flags)
/* See linespec.h. */
std::vector<symtab_and_line>
-decode_line_with_last_displayed (const char *string, int flags)
+decode_line_with_last_displayed (const char *string, int flags,
+ program_space *search_pspace)
{
if (string == 0)
error (_("Empty line specification."));
@@ -3240,10 +3241,10 @@ decode_line_with_last_displayed (const char *string, int flags)
current_language);
std::vector<symtab_and_line> sals
= (last_displayed_sal_is_valid ()
- ? decode_line_1 (locspec.get (), flags, NULL,
+ ? decode_line_1 (locspec.get (), flags, search_pspace,
get_last_displayed_symtab (),
get_last_displayed_line ())
- : decode_line_1 (locspec.get (), flags, NULL, NULL, 0));
+ : decode_line_1 (locspec.get (), flags, search_pspace, NULL, 0));
if (*string)
error (_("Junk at end of line specification: %s"), string);
diff --git a/gdb/linespec.h b/gdb/linespec.h
index d5e7334fe2d..0eb9cb5d9f2 100644
--- a/gdb/linespec.h
+++ b/gdb/linespec.h
@@ -139,10 +139,11 @@ extern std::vector<symtab_and_line> decode_line_with_current_source
(const char *, int);
/* Given a string, return the line specified by it, using the last displayed
- codepoint's values as defaults, or nothing if they aren't valid. */
+ codepoint's values as defaults, or nothing if they aren't valid.
+ Limit the search to given program space, if specified. */
extern std::vector<symtab_and_line> decode_line_with_last_displayed
- (const char *, int);
+ (const char *, int, program_space *search_pspace = nullptr);
/* Does P represent one of the keywords? If so, return
the keyword. If not, return NULL. */
diff --git a/gdb/testsuite/gdb.base/jump.exp b/gdb/testsuite/gdb.base/jump.exp
index 032c4a6279d..ca214993025 100644
--- a/gdb/testsuite/gdb.base/jump.exp
+++ b/gdb/testsuite/gdb.base/jump.exp
@@ -18,99 +18,126 @@ clear_xfail "*-*-*"
standard_testfile .c
-# Build the test case
-if { [gdb_compile "${srcdir}/${subdir}/${srcfile}" "${binfile}" executable {debug nowarnings}] != "" } {
- untested "failed to compile"
- return -1
- }
+if {[prepare_for_testing "failed to prepare" $testfile $srcfile debug]} {
+ return -1
+}
-# Start with a fresh gdb
+proc do_tests {} {
+ global decimal srcfile
+
+ # Set a breakpoint on the statement that we're about to jump to.
+ # The statement doesn't contain a function call.
+ #
+ set bp_on_non_call 0
+ set non_call_line [gdb_get_line_number "bp-on-non-call"]
+ gdb_breakpoint "$non_call_line"
+ set bp_on_non_call [get_integer_valueof "\$bpnum" "INVALID" "bp_on_non_call"]
+
+ # Can we jump to the statement? Do we stop there?
+ #
+ gdb_test "jump $non_call_line" \
+ "Breakpoint ${bp_on_non_call}(\.${decimal})?, .*${srcfile}:$non_call_line.*" \
+ "jump to non-call"
+
+ # Set a breakpoint on the statement that we're about to jump to.
+ # The statement does contain a function call.
+ #
+ set bp_on_call 0
+ set call_line [gdb_get_line_number "bp-on-call"]
+ gdb_breakpoint "$call_line"
+ set bp_on_call [get_integer_valueof "\$bpnum" "INVALID" "bp_on_call"]
+
+ # Can we jump to the statement? Do we stop there?
+ #
+ gdb_test "jump $call_line" \
+ "Breakpoint ${bp_on_call}(\.${decimal})?, .*${srcfile}:$call_line.*" \
+ "jump to call"
+
+ # If we disable the breakpoint at the function call, and then
+ # if we jump to that statement, do we not stop there, but at
+ # the following breakpoint?
+ #
+ gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
+
+ gdb_test "jump $call_line" \
+ "Breakpoint ${bp_on_non_call}(\.${decimal})?, .*${srcfile}:$non_call_line.*" \
+ "jump to call with disabled breakpoint"
+
+ # Disable the breakpoint at the non-function call, so it won't be
+ # there if do_test is called again.
+ gdb_test_no_output "disable ${bp_on_non_call}" "disable bp_on_non_call"
+
+
+ # Verify that GDB responds gracefully to the "jump" command without
+ # an argument.
+ #
+ gdb_test "jump" "Argument required .starting address.*" \
+ "jump without argument disallowed"
+
+
+ # Verify that GDB responds gracefully to the "jump" command with
+ # trailing junk.
+ #
+ gdb_test "jump $call_line 100" \
+ "malformed linespec error: unexpected number, \"100\"" \
+ "jump with trailing argument junk"
+
+
+ # Verify that GDB responds gracefully to a request to jump out of
+ # the current function. (Note that this will very likely cause the
+ # inferior to die. Be prepared to rerun the inferior, if further
+ # testing is desired.)
+ #
+ # Try it both ways: confirming and not confirming the jump.
+ #
+
+ set out_line [gdb_get_line_number "out-of-func"]
+ gdb_test "jump $out_line" \
+ "Not confirmed.*" \
+ "aborted jump out of current function" \
+ "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
+ "n"
+
+ gdb_test "jump $out_line" \
+ "Continuing at.*" \
+ "jump out of current function" \
+ "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
+ "y"
+}
-clean_restart ${binfile}
-if {![runto_main]} {
- return -1
+set inferiors 1
+if {![use_gdb_stub]} {
+ set inferiors 2
}
-# Set a breakpoint on the statement that we're about to jump to.
-# The statement doesn't contain a function call.
-#
-set bp_on_non_call 0
-set non_call_line [gdb_get_line_number "bp-on-non-call"]
-gdb_test_multiple "break $non_call_line" "break before jump to non-call" {
- -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line $non_call_line.*$gdb_prompt $" {
- set bp_on_non_call $expect_out(1,string)
- pass "break before jump to non-call"
+# Run to main, add inferiors if needed.
+for {set inf 1} {$inf <= $inferiors} {incr inf} {
+ if {$inf != 1} {
+ # Start a new inferior, and run it with the same executable.
+ gdb_test "add-inferior -exec ${binfile}" \
+ "Added inferior ${inf}.*" \
+ "add inferior ${inf} with -exec "
+ gdb_test "inferior ${inf}" \
+ "Switching to inferior ${inf} .*" \
+ "switch to inferior ${inf}"
+ }
+ if ![runto_main] then {
+ perror "Couldn't run inferior ${inf} to main"
+ return -1
}
}
-# Can we jump to the statement? Do we stop there?
-#
-gdb_test "jump $non_call_line" "Breakpoint ${decimal}, .*${srcfile}:$non_call_line.*" \
- "jump to non-call"
-
-# Set a breakpoint on the statement that we're about to jump to.
-# The statement does contain a function call.
-#
-set bp_on_call 0
-set call_line [gdb_get_line_number "bp-on-call"]
-gdb_test_multiple "break $call_line" "break before jump to call" {
- -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line $call_line.*$gdb_prompt $" {
- set bp_on_call $expect_out(1,string)
- pass "break before jump to call"
+# Run tests on all inferiors.
+for {set inf 1} {$inf <= $inferiors} {incr inf} {
+ with_test_prefix "inferior $inf" {
+ # Switch to the target inferior.
+ gdb_test "inferior $inf" ".*Switching to inferior $inf .*"
+ # Run the tests.
+ do_tests
}
}
-# Can we jump to the statement? Do we stop there?
-#
-gdb_test "jump $call_line" \
- "Breakpoint ${decimal}, .*${srcfile}:$call_line.*" \
- "jump to call"
-
-# If we disable the breakpoint at the function call, and then
-# if we jump to that statement, do we not stop there, but at
-# the following breakpoint?
-#
-gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
-
-gdb_test "jump $call_line" "Breakpoint ${decimal}, .*${srcfile}:$non_call_line.*" \
- "jump to call with disabled breakpoint"
-
-# Verify that GDB responds gracefully to the "jump" command without
-# an argument.
-#
-gdb_test "jump" "Argument required .starting address.*" \
- "jump without argument disallowed"
-
-
-# Verify that GDB responds gracefully to the "jump" command with
-# trailing junk.
-#
-gdb_test "jump $call_line 100" \
- "malformed linespec error: unexpected number, \"100\"" \
- "jump with trailing argument junk"
-
-
-# Verify that GDB responds gracefully to a request to jump out of
-# the current function. (Note that this will very likely cause the
-# inferior to die. Be prepared to rerun the inferior, if further
-# testing is desired.)
-#
-# Try it both ways: confirming and not confirming the jump.
-#
-
-set out_line [gdb_get_line_number "out-of-func"]
-gdb_test "jump $out_line" \
- "Not confirmed.*" \
- "aborted jump out of current function" \
- "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
- "n"
-
-gdb_test "jump $out_line" \
- "Continuing at.*" \
- "jump out of current function" \
- "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
- "y"
-
gdb_exit
+return 0
--
2.25.1
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [relevance 63%]
* [PATCH v2 0/1] gdb, infcmd: Support jump command in multi-inferior case.
@ 2023-02-23 9:54 99% Matti Puputti
2023-02-23 9:54 63% ` [PATCH v2 1/1] " Matti Puputti
2023-03-07 16:50 99% ` [PING] [PATCH v2 0/1] " Puputti, Matti
0 siblings, 2 replies; 34+ results
From: Matti Puputti @ 2023-02-23 9:54 UTC (permalink / raw)
To: gdb-patches; +Cc: aburgess
Thanks for the feedback on v1.
Changes in v2:
- In linespec.h, updated comment to explain the new parameter
search_pspace.
- In jump.exp, updated to use prepare_for_testing.
- In jump.exp, updated how bp_on_non_call and bp_on_call are set,
updated regexes to use these two variables, and added disabling
of BP $bp_on_non_call.
Br,
Matti Puputti
---
Matti Puputti (1):
gdb, infcmd: Support jump command in multi-inferior case.
gdb/infcmd.c | 3 +-
gdb/linespec.c | 7 +-
gdb/linespec.h | 5 +-
gdb/testsuite/gdb.base/jump.exp | 193 ++++++++++++++++++--------------
4 files changed, 119 insertions(+), 89 deletions(-)
--
2.25.1
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [relevance 99%]
* RE: [PATCH 1/1] gdb, infcmd: Support jump command in multi-inferior case.
2023-02-21 16:56 99% ` Andrew Burgess
@ 2023-02-22 17:43 99% ` Puputti, Matti
0 siblings, 0 replies; 34+ results
From: Puputti, Matti @ 2023-02-22 17:43 UTC (permalink / raw)
To: Andrew Burgess, gdb-patches
Hi Andrew,
Thank four for the review and comments.
> > --- a/gdb/linespec.h
> > +++ b/gdb/linespec.h
> > @@ -142,7 +142,7 @@ extern std::vector<symtab_and_line>
> decode_line_with_current_source
> > codepoint's values as defaults, or nothing if they aren't valid. */
> >
> > extern std::vector<symtab_and_line> decode_line_with_last_displayed
> > - (const char *, int);
> > + (const char *, int, program_space *search_pspace = nullptr);
>
> The comment above this declaration should be updated to mention the new
> parameter and explain how to use it.
>
I will update the comment to explain the new parameter.
> I notice there's only two uses of decode_line_with_last_displayed, the
> other being in the 'info line' function. I don't think it's something
> to do in this patch; but I wonder if it would make sense to update that
> call to restrict to the current program space too? If I say:
>
> (gdb) info line 3
>
> I'm probably asking about the current inferior..
>
> Have you thought about this at all?
>
In this commit, I focused on case where requested command cannot be executed.
If more than one locations are found, jump_command prints an error.
But info_line_command prints data on each found location.
I would propose to keep this commit focused on jump_command only.
> > --- a/gdb/testsuite/gdb.base/jump.exp
> > +++ b/gdb/testsuite/gdb.base/jump.exp
> > @@ -25,92 +25,124 @@ if { [gdb_compile "${srcdir}/${subdir}/${srcfile}"
> "${binfile}" executable {deb
> > }
> >
> >
> > -# Start with a fresh gdb
> > -
> > -clean_restart ${binfile}
>
> I know this is not your fault; the test is clearly pretty old, but it
> would be great, if instead of moving this clean_restart around we
> instead updated the test to use prepare_for_testing.
>
I will update my commit to address this.
> > + # Set a breakpoint on the statement that we're about to jump to.
> > + # The statement doesn't contain a function call.
> > + #
> > + set bp_on_non_call 0
> > + set non_call_line [gdb_get_line_number "bp-on-non-call"]
> > + gdb_test_multiple "break $non_call_line" "break before jump to non-call" {
> > + -re "\[Bb\]reakpoint (${decimal}) at ${hex}:
> .*${srcfile}.*$non_call_line.*$gdb_prompt $" {
> > + set bp_on_non_call $expect_out(1,string)
> > + pass "break before jump to non-call"
>
> Similarly, here, instead of tweaking the pattern, this can all be
> replaced with:
>
> # Set a breakpoint on the statement that we're about to jump to.
> # The statement doesn't contain a function call.
> #
> set non_call_line [gdb_get_line_number "bp-on-non-call"]
> gdb_breakpoint "$non_call_line"
> set bp_on_non_call [get_integer_valueof "\$bpnum" "INVALID"]
>
> which I think is much clearer.
>
>
I will update my commit to address this.
> > + # Can we jump to the statement? Do we stop there?
> > + #
> > + gdb_test "jump $non_call_line" "Breakpoint ${decimal}(\.${decimal})?,
> .*${srcfile}:$non_call_line.*" \
> > + "jump to non-call"
>
> We collected bp_on_non_call above, but we never actually use it. Again,
> this is not your fault, this was just not a great test to begin with,
> but we could change this pattern here to replace the first $decimal with
> $bp_on_non_call.
>
I will update my commit to address this.
> > + set call_line [gdb_get_line_number "bp-on-call"]
> > + gdb_test_multiple "break $call_line" "break before jump to call" {
> > + -re "\[Bb\]reakpoint (${decimal}) at ${hex}:
> .*${srcfile}.*$call_line.*$gdb_prompt $" {
> > + set bp_on_call $expect_out(1,string)
> > + pass "break before jump to call"
>
> This can be cleaned up like the earlier one.
>
Will do.
> > + # Can we jump to the statement? Do we stop there?
> > + #
> > + gdb_test "jump $call_line" \
> > + "Breakpoint ${decimal}(\.${decimal})?, .*${srcfile}:$call_line.*" \
> > + "jump to call"
>
> And we can use $bp_on_call to replace the first $decimal here.
Will do.
>
> Thanks,
> Andrew
>
I will work on a v2.
Br,
Matti Puputti
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [relevance 99%]
* Re: [PATCH 1/1] gdb, infcmd: Support jump command in multi-inferior case.
2023-02-21 15:13 64% [PATCH 1/1] gdb, infcmd: Support jump command in multi-inferior case Matti Puputti
@ 2023-02-21 16:56 99% ` Andrew Burgess
2023-02-22 17:43 99% ` Puputti, Matti
0 siblings, 1 reply; 34+ results
From: Andrew Burgess @ 2023-02-21 16:56 UTC (permalink / raw)
To: Matti Puputti, gdb-patches
Matti Puputti via Gdb-patches <gdb-patches@sourceware.org> writes:
> Fixes the issue where jump failed if multiple inferiors run the same source.
>
> See the below example
>
> $ gdb -q ./simple
> Reading symbols from ./simple...
> (gdb) break 2
> Breakpoint 1 at 0x114e: file simple.c, line 2.
> (gdb) run
> Starting program: /temp/simple
>
> Breakpoint 1, main () at simple.c:2
> 2 int a = 42;
> (gdb) add-inferior
> [New inferior 2]
> Added inferior 2 on connection 1 (native)
> (gdb) inferior 2
> [Switching to inferior 2 [<null>] (<noexec>)]
> (gdb) info inferiors
> Num Description Connection Executable
> 1 process 6250 1 (native) /temp/simple
> * 2 <null> 1 (native)
> (gdb) file ./simple
> Reading symbols from ./simple...
> (gdb) run
> Starting program: /temp/simple
>
> Thread 2.1 "simple" hit Breakpoint 1, main () at simple.c:2
> 2 int a = 42;
> (gdb) info inferiors
> Num Description Connection Executable
> 1 process 6250 1 (native) /temp/simple
> * 2 process 6705 1 (native) /temp/simple
> (gdb) jump 3
> Unreasonable jump request
> (gdb)
>
> In this example, jump fails because the debugger finds two different
> locations, one for each inferior.
> Solution is to limit the search to the current program space.
> ---
> gdb/infcmd.c | 3 +-
> gdb/linespec.c | 7 +-
> gdb/linespec.h | 2 +-
> gdb/testsuite/gdb.base/jump.exp | 186 +++++++++++++++++++-------------
> 4 files changed, 116 insertions(+), 82 deletions(-)
> mode change 100644 => 100755 gdb/testsuite/gdb.base/jump.exp
>
> diff --git a/gdb/infcmd.c b/gdb/infcmd.c
> index fd88b8ca328..2f981cc895f 100644
> --- a/gdb/infcmd.c
> +++ b/gdb/infcmd.c
> @@ -1078,7 +1078,8 @@ jump_command (const char *arg, int from_tty)
> error_no_arg (_("starting address"));
>
> std::vector<symtab_and_line> sals
> - = decode_line_with_last_displayed (arg, DECODE_LINE_FUNFIRSTLINE);
> + = decode_line_with_last_displayed (arg, DECODE_LINE_FUNFIRSTLINE,
> + current_program_space);
> if (sals.size () != 1)
> error (_("Unreasonable jump request"));
>
> diff --git a/gdb/linespec.c b/gdb/linespec.c
> index d3def7ae070..5969c46e685 100644
> --- a/gdb/linespec.c
> +++ b/gdb/linespec.c
> @@ -3231,7 +3231,8 @@ decode_line_with_current_source (const char *string, int flags)
> /* See linespec.h. */
>
> std::vector<symtab_and_line>
> -decode_line_with_last_displayed (const char *string, int flags)
> +decode_line_with_last_displayed (const char *string, int flags,
> + program_space *search_pspace)
> {
> if (string == 0)
> error (_("Empty line specification."));
> @@ -3240,10 +3241,10 @@ decode_line_with_last_displayed (const char *string, int flags)
> current_language);
> std::vector<symtab_and_line> sals
> = (last_displayed_sal_is_valid ()
> - ? decode_line_1 (locspec.get (), flags, NULL,
> + ? decode_line_1 (locspec.get (), flags, search_pspace,
> get_last_displayed_symtab (),
> get_last_displayed_line ())
> - : decode_line_1 (locspec.get (), flags, NULL, NULL, 0));
> + : decode_line_1 (locspec.get (), flags, search_pspace, NULL, 0));
>
> if (*string)
> error (_("Junk at end of line specification: %s"), string);
> diff --git a/gdb/linespec.h b/gdb/linespec.h
> index d5e7334fe2d..70e04cbbb21 100644
> --- a/gdb/linespec.h
> +++ b/gdb/linespec.h
> @@ -142,7 +142,7 @@ extern std::vector<symtab_and_line> decode_line_with_current_source
> codepoint's values as defaults, or nothing if they aren't valid. */
>
> extern std::vector<symtab_and_line> decode_line_with_last_displayed
> - (const char *, int);
> + (const char *, int, program_space *search_pspace = nullptr);
The comment above this declaration should be updated to mention the new
parameter and explain how to use it.
I notice there's only two uses of decode_line_with_last_displayed, the
other being in the 'info line' function. I don't think it's something
to do in this patch; but I wonder if it would make sense to update that
call to restrict to the current program space too? If I say:
(gdb) info line 3
I'm probably asking about the current inferior..
Have you thought about this at all?
>
> /* Does P represent one of the keywords? If so, return
> the keyword. If not, return NULL. */
> diff --git a/gdb/testsuite/gdb.base/jump.exp b/gdb/testsuite/gdb.base/jump.exp
> old mode 100644
> new mode 100755
> index 032c4a6279d..1040f6355a5
> --- a/gdb/testsuite/gdb.base/jump.exp
> +++ b/gdb/testsuite/gdb.base/jump.exp
> @@ -25,92 +25,124 @@ if { [gdb_compile "${srcdir}/${subdir}/${srcfile}" "${binfile}" executable {deb
> }
>
>
> -# Start with a fresh gdb
> -
> -clean_restart ${binfile}
I know this is not your fault; the test is clearly pretty old, but it
would be great, if instead of moving this clean_restart around we
instead updated the test to use prepare_for_testing.
> -
> -if {![runto_main]} {
> - return -1
> -}
> -
> -# Set a breakpoint on the statement that we're about to jump to.
> -# The statement doesn't contain a function call.
> -#
> -set bp_on_non_call 0
> -set non_call_line [gdb_get_line_number "bp-on-non-call"]
> -gdb_test_multiple "break $non_call_line" "break before jump to non-call" {
> - -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line $non_call_line.*$gdb_prompt $" {
> - set bp_on_non_call $expect_out(1,string)
> - pass "break before jump to non-call"
> +proc do_tests {} {
> + global decimal hex srcfile gdb_prompt
> + # Set a breakpoint on the statement that we're about to jump to.
> + # The statement doesn't contain a function call.
> + #
> + set bp_on_non_call 0
> + set non_call_line [gdb_get_line_number "bp-on-non-call"]
> + gdb_test_multiple "break $non_call_line" "break before jump to non-call" {
> + -re "\[Bb\]reakpoint (${decimal}) at ${hex}: .*${srcfile}.*$non_call_line.*$gdb_prompt $" {
> + set bp_on_non_call $expect_out(1,string)
> + pass "break before jump to non-call"
Similarly, here, instead of tweaking the pattern, this can all be
replaced with:
# Set a breakpoint on the statement that we're about to jump to.
# The statement doesn't contain a function call.
#
set non_call_line [gdb_get_line_number "bp-on-non-call"]
gdb_breakpoint "$non_call_line"
set bp_on_non_call [get_integer_valueof "\$bpnum" "INVALID"]
which I think is much clearer.
> + }
> }
> -}
>
> -# Can we jump to the statement? Do we stop there?
> -#
> -gdb_test "jump $non_call_line" "Breakpoint ${decimal}, .*${srcfile}:$non_call_line.*" \
> - "jump to non-call"
> -
> -# Set a breakpoint on the statement that we're about to jump to.
> -# The statement does contain a function call.
> -#
> -set bp_on_call 0
> -set call_line [gdb_get_line_number "bp-on-call"]
> -gdb_test_multiple "break $call_line" "break before jump to call" {
> - -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line $call_line.*$gdb_prompt $" {
> - set bp_on_call $expect_out(1,string)
> - pass "break before jump to call"
> + # Can we jump to the statement? Do we stop there?
> + #
> + gdb_test "jump $non_call_line" "Breakpoint ${decimal}(\.${decimal})?, .*${srcfile}:$non_call_line.*" \
> + "jump to non-call"
We collected bp_on_non_call above, but we never actually use it. Again,
this is not your fault, this was just not a great test to begin with,
but we could change this pattern here to replace the first $decimal with
$bp_on_non_call.
> +
> + # Set a breakpoint on the statement that we're about to jump to.
> + # The statement does contain a function call.
> + #
> + set bp_on_call 0
> + set call_line [gdb_get_line_number "bp-on-call"]
> + gdb_test_multiple "break $call_line" "break before jump to call" {
> + -re "\[Bb\]reakpoint (${decimal}) at ${hex}: .*${srcfile}.*$call_line.*$gdb_prompt $" {
> + set bp_on_call $expect_out(1,string)
> + pass "break before jump to call"
This can be cleaned up like the earlier one.
> + }
> }
> -}
>
> -# Can we jump to the statement? Do we stop there?
> -#
> -gdb_test "jump $call_line" \
> - "Breakpoint ${decimal}, .*${srcfile}:$call_line.*" \
> - "jump to call"
> -
> -# If we disable the breakpoint at the function call, and then
> -# if we jump to that statement, do we not stop there, but at
> -# the following breakpoint?
> -#
> -gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
> -
> -gdb_test "jump $call_line" "Breakpoint ${decimal}, .*${srcfile}:$non_call_line.*" \
> - "jump to call with disabled breakpoint"
> + # Can we jump to the statement? Do we stop there?
> + #
> + gdb_test "jump $call_line" \
> + "Breakpoint ${decimal}(\.${decimal})?, .*${srcfile}:$call_line.*" \
> + "jump to call"
And we can use $bp_on_call to replace the first $decimal here.
Thanks,
Andrew
> +
> + # If we disable the breakpoint at the function call, and then
> + # if we jump to that statement, do we not stop there, but at
> + # the following breakpoint?
> + #
> + gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
> +
> + gdb_test "jump $call_line" "Breakpoint ${decimal}(\.${decimal})?, .*${srcfile}:$non_call_line.*" \
> + "jump to call with disabled breakpoint"
> +
> + # Verify that GDB responds gracefully to the "jump" command without
> + # an argument.
> + #
> + gdb_test "jump" "Argument required .starting address.*" \
> + "jump without argument disallowed"
> +
> +
> + # Verify that GDB responds gracefully to the "jump" command with
> + # trailing junk.
> + #
> + gdb_test "jump $call_line 100" \
> + "malformed linespec error: unexpected number, \"100\"" \
> + "jump with trailing argument junk"
> +
> +
> + # Verify that GDB responds gracefully to a request to jump out of
> + # the current function. (Note that this will very likely cause the
> + # inferior to die. Be prepared to rerun the inferior, if further
> + # testing is desired.)
> + #
> + # Try it both ways: confirming and not confirming the jump.
> + #
> +
> + set out_line [gdb_get_line_number "out-of-func"]
> + gdb_test "jump $out_line" \
> + "Not confirmed.*" \
> + "aborted jump out of current function" \
> + "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> + "n"
> +
> + gdb_test "jump $out_line" \
> + "Continuing at.*" \
> + "jump out of current function" \
> + "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> + "y"
> +}
>
> -# Verify that GDB responds gracefully to the "jump" command without
> -# an argument.
> -#
> -gdb_test "jump" "Argument required .starting address.*" \
> - "jump without argument disallowed"
>
> +# Start with a fresh gdb.
>
> -# Verify that GDB responds gracefully to the "jump" command with
> -# trailing junk.
> -#
> -gdb_test "jump $call_line 100" \
> - "malformed linespec error: unexpected number, \"100\"" \
> - "jump with trailing argument junk"
> +clean_restart ${binfile}
>
> +set inferiors 1
> +if {![use_gdb_stub]} {
> + set inferiors 2
> +}
>
> -# Verify that GDB responds gracefully to a request to jump out of
> -# the current function. (Note that this will very likely cause the
> -# inferior to die. Be prepared to rerun the inferior, if further
> -# testing is desired.)
> -#
> -# Try it both ways: confirming and not confirming the jump.
> -#
> +# Run to main, add inferiors if needed.
> +for {set inf 1} {$inf <= $inferiors} {incr inf} {
> + if {$inf != 1} {
> + # Start a new inferior, and run it with the same executable.
> + gdb_test "add-inferior -exec ${binfile}" \
> + "Added inferior ${inf}.*" \
> + "add inferior ${inf} with -exec "
> + gdb_test "inferior ${inf}" \
> + "Switching to inferior ${inf} .*" \
> + "switch to inferior ${inf}"
> + }
> + if ![runto_main] then {
> + perror "Couldn't run inferior ${inf} to main"
> + return -1
> + }
> +}
>
> -set out_line [gdb_get_line_number "out-of-func"]
> -gdb_test "jump $out_line" \
> - "Not confirmed.*" \
> - "aborted jump out of current function" \
> - "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> - "n"
> -
> -gdb_test "jump $out_line" \
> - "Continuing at.*" \
> - "jump out of current function" \
> - "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
> - "y"
> +# Run tests on all inferiors.
> +for {set inf 1} {$inf <= $inferiors} {incr inf} {
> + with_test_prefix "inferior $inf" {
> + # Switch to the target inferior.
> + gdb_test "inferior $inf" ".*Switching to inferior $inf .*"
> + # Run the tests.
> + do_tests
> + }
> +}
>
> gdb_exit
> --
> 2.25.1
>
> Intel Deutschland GmbH
> Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
> Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
> Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
> Chairperson of the Supervisory Board: Nicole Lau
> Registered Office: Munich
> Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [relevance 99%]
* [PATCH 1/1] gdb, infcmd: Support jump command in multi-inferior case.
@ 2023-02-21 15:13 64% Matti Puputti
2023-02-21 16:56 99% ` Andrew Burgess
0 siblings, 1 reply; 34+ results
From: Matti Puputti @ 2023-02-21 15:13 UTC (permalink / raw)
To: gdb-patches
Fixes the issue where jump failed if multiple inferiors run the same source.
See the below example
$ gdb -q ./simple
Reading symbols from ./simple...
(gdb) break 2
Breakpoint 1 at 0x114e: file simple.c, line 2.
(gdb) run
Starting program: /temp/simple
Breakpoint 1, main () at simple.c:2
2 int a = 42;
(gdb) add-inferior
[New inferior 2]
Added inferior 2 on connection 1 (native)
(gdb) inferior 2
[Switching to inferior 2 [<null>] (<noexec>)]
(gdb) info inferiors
Num Description Connection Executable
1 process 6250 1 (native) /temp/simple
* 2 <null> 1 (native)
(gdb) file ./simple
Reading symbols from ./simple...
(gdb) run
Starting program: /temp/simple
Thread 2.1 "simple" hit Breakpoint 1, main () at simple.c:2
2 int a = 42;
(gdb) info inferiors
Num Description Connection Executable
1 process 6250 1 (native) /temp/simple
* 2 process 6705 1 (native) /temp/simple
(gdb) jump 3
Unreasonable jump request
(gdb)
In this example, jump fails because the debugger finds two different
locations, one for each inferior.
Solution is to limit the search to the current program space.
---
gdb/infcmd.c | 3 +-
gdb/linespec.c | 7 +-
gdb/linespec.h | 2 +-
gdb/testsuite/gdb.base/jump.exp | 186 +++++++++++++++++++-------------
4 files changed, 116 insertions(+), 82 deletions(-)
mode change 100644 => 100755 gdb/testsuite/gdb.base/jump.exp
diff --git a/gdb/infcmd.c b/gdb/infcmd.c
index fd88b8ca328..2f981cc895f 100644
--- a/gdb/infcmd.c
+++ b/gdb/infcmd.c
@@ -1078,7 +1078,8 @@ jump_command (const char *arg, int from_tty)
error_no_arg (_("starting address"));
std::vector<symtab_and_line> sals
- = decode_line_with_last_displayed (arg, DECODE_LINE_FUNFIRSTLINE);
+ = decode_line_with_last_displayed (arg, DECODE_LINE_FUNFIRSTLINE,
+ current_program_space);
if (sals.size () != 1)
error (_("Unreasonable jump request"));
diff --git a/gdb/linespec.c b/gdb/linespec.c
index d3def7ae070..5969c46e685 100644
--- a/gdb/linespec.c
+++ b/gdb/linespec.c
@@ -3231,7 +3231,8 @@ decode_line_with_current_source (const char *string, int flags)
/* See linespec.h. */
std::vector<symtab_and_line>
-decode_line_with_last_displayed (const char *string, int flags)
+decode_line_with_last_displayed (const char *string, int flags,
+ program_space *search_pspace)
{
if (string == 0)
error (_("Empty line specification."));
@@ -3240,10 +3241,10 @@ decode_line_with_last_displayed (const char *string, int flags)
current_language);
std::vector<symtab_and_line> sals
= (last_displayed_sal_is_valid ()
- ? decode_line_1 (locspec.get (), flags, NULL,
+ ? decode_line_1 (locspec.get (), flags, search_pspace,
get_last_displayed_symtab (),
get_last_displayed_line ())
- : decode_line_1 (locspec.get (), flags, NULL, NULL, 0));
+ : decode_line_1 (locspec.get (), flags, search_pspace, NULL, 0));
if (*string)
error (_("Junk at end of line specification: %s"), string);
diff --git a/gdb/linespec.h b/gdb/linespec.h
index d5e7334fe2d..70e04cbbb21 100644
--- a/gdb/linespec.h
+++ b/gdb/linespec.h
@@ -142,7 +142,7 @@ extern std::vector<symtab_and_line> decode_line_with_current_source
codepoint's values as defaults, or nothing if they aren't valid. */
extern std::vector<symtab_and_line> decode_line_with_last_displayed
- (const char *, int);
+ (const char *, int, program_space *search_pspace = nullptr);
/* Does P represent one of the keywords? If so, return
the keyword. If not, return NULL. */
diff --git a/gdb/testsuite/gdb.base/jump.exp b/gdb/testsuite/gdb.base/jump.exp
old mode 100644
new mode 100755
index 032c4a6279d..1040f6355a5
--- a/gdb/testsuite/gdb.base/jump.exp
+++ b/gdb/testsuite/gdb.base/jump.exp
@@ -25,92 +25,124 @@ if { [gdb_compile "${srcdir}/${subdir}/${srcfile}" "${binfile}" executable {deb
}
-# Start with a fresh gdb
-
-clean_restart ${binfile}
-
-if {![runto_main]} {
- return -1
-}
-
-# Set a breakpoint on the statement that we're about to jump to.
-# The statement doesn't contain a function call.
-#
-set bp_on_non_call 0
-set non_call_line [gdb_get_line_number "bp-on-non-call"]
-gdb_test_multiple "break $non_call_line" "break before jump to non-call" {
- -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line $non_call_line.*$gdb_prompt $" {
- set bp_on_non_call $expect_out(1,string)
- pass "break before jump to non-call"
+proc do_tests {} {
+ global decimal hex srcfile gdb_prompt
+ # Set a breakpoint on the statement that we're about to jump to.
+ # The statement doesn't contain a function call.
+ #
+ set bp_on_non_call 0
+ set non_call_line [gdb_get_line_number "bp-on-non-call"]
+ gdb_test_multiple "break $non_call_line" "break before jump to non-call" {
+ -re "\[Bb\]reakpoint (${decimal}) at ${hex}: .*${srcfile}.*$non_call_line.*$gdb_prompt $" {
+ set bp_on_non_call $expect_out(1,string)
+ pass "break before jump to non-call"
+ }
}
-}
-# Can we jump to the statement? Do we stop there?
-#
-gdb_test "jump $non_call_line" "Breakpoint ${decimal}, .*${srcfile}:$non_call_line.*" \
- "jump to non-call"
-
-# Set a breakpoint on the statement that we're about to jump to.
-# The statement does contain a function call.
-#
-set bp_on_call 0
-set call_line [gdb_get_line_number "bp-on-call"]
-gdb_test_multiple "break $call_line" "break before jump to call" {
- -re "\[Bb\]reakpoint (${decimal}) at ${hex}: file .*${srcfile}, line $call_line.*$gdb_prompt $" {
- set bp_on_call $expect_out(1,string)
- pass "break before jump to call"
+ # Can we jump to the statement? Do we stop there?
+ #
+ gdb_test "jump $non_call_line" "Breakpoint ${decimal}(\.${decimal})?, .*${srcfile}:$non_call_line.*" \
+ "jump to non-call"
+
+ # Set a breakpoint on the statement that we're about to jump to.
+ # The statement does contain a function call.
+ #
+ set bp_on_call 0
+ set call_line [gdb_get_line_number "bp-on-call"]
+ gdb_test_multiple "break $call_line" "break before jump to call" {
+ -re "\[Bb\]reakpoint (${decimal}) at ${hex}: .*${srcfile}.*$call_line.*$gdb_prompt $" {
+ set bp_on_call $expect_out(1,string)
+ pass "break before jump to call"
+ }
}
-}
-# Can we jump to the statement? Do we stop there?
-#
-gdb_test "jump $call_line" \
- "Breakpoint ${decimal}, .*${srcfile}:$call_line.*" \
- "jump to call"
-
-# If we disable the breakpoint at the function call, and then
-# if we jump to that statement, do we not stop there, but at
-# the following breakpoint?
-#
-gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
-
-gdb_test "jump $call_line" "Breakpoint ${decimal}, .*${srcfile}:$non_call_line.*" \
- "jump to call with disabled breakpoint"
+ # Can we jump to the statement? Do we stop there?
+ #
+ gdb_test "jump $call_line" \
+ "Breakpoint ${decimal}(\.${decimal})?, .*${srcfile}:$call_line.*" \
+ "jump to call"
+
+ # If we disable the breakpoint at the function call, and then
+ # if we jump to that statement, do we not stop there, but at
+ # the following breakpoint?
+ #
+ gdb_test_no_output "disable $bp_on_call" "disable breakpoint on call"
+
+ gdb_test "jump $call_line" "Breakpoint ${decimal}(\.${decimal})?, .*${srcfile}:$non_call_line.*" \
+ "jump to call with disabled breakpoint"
+
+ # Verify that GDB responds gracefully to the "jump" command without
+ # an argument.
+ #
+ gdb_test "jump" "Argument required .starting address.*" \
+ "jump without argument disallowed"
+
+
+ # Verify that GDB responds gracefully to the "jump" command with
+ # trailing junk.
+ #
+ gdb_test "jump $call_line 100" \
+ "malformed linespec error: unexpected number, \"100\"" \
+ "jump with trailing argument junk"
+
+
+ # Verify that GDB responds gracefully to a request to jump out of
+ # the current function. (Note that this will very likely cause the
+ # inferior to die. Be prepared to rerun the inferior, if further
+ # testing is desired.)
+ #
+ # Try it both ways: confirming and not confirming the jump.
+ #
+
+ set out_line [gdb_get_line_number "out-of-func"]
+ gdb_test "jump $out_line" \
+ "Not confirmed.*" \
+ "aborted jump out of current function" \
+ "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
+ "n"
+
+ gdb_test "jump $out_line" \
+ "Continuing at.*" \
+ "jump out of current function" \
+ "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
+ "y"
+}
-# Verify that GDB responds gracefully to the "jump" command without
-# an argument.
-#
-gdb_test "jump" "Argument required .starting address.*" \
- "jump without argument disallowed"
+# Start with a fresh gdb.
-# Verify that GDB responds gracefully to the "jump" command with
-# trailing junk.
-#
-gdb_test "jump $call_line 100" \
- "malformed linespec error: unexpected number, \"100\"" \
- "jump with trailing argument junk"
+clean_restart ${binfile}
+set inferiors 1
+if {![use_gdb_stub]} {
+ set inferiors 2
+}
-# Verify that GDB responds gracefully to a request to jump out of
-# the current function. (Note that this will very likely cause the
-# inferior to die. Be prepared to rerun the inferior, if further
-# testing is desired.)
-#
-# Try it both ways: confirming and not confirming the jump.
-#
+# Run to main, add inferiors if needed.
+for {set inf 1} {$inf <= $inferiors} {incr inf} {
+ if {$inf != 1} {
+ # Start a new inferior, and run it with the same executable.
+ gdb_test "add-inferior -exec ${binfile}" \
+ "Added inferior ${inf}.*" \
+ "add inferior ${inf} with -exec "
+ gdb_test "inferior ${inf}" \
+ "Switching to inferior ${inf} .*" \
+ "switch to inferior ${inf}"
+ }
+ if ![runto_main] then {
+ perror "Couldn't run inferior ${inf} to main"
+ return -1
+ }
+}
-set out_line [gdb_get_line_number "out-of-func"]
-gdb_test "jump $out_line" \
- "Not confirmed.*" \
- "aborted jump out of current function" \
- "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
- "n"
-
-gdb_test "jump $out_line" \
- "Continuing at.*" \
- "jump out of current function" \
- "Line $out_line is not in `main'. Jump anyway.*y or n. $" \
- "y"
+# Run tests on all inferiors.
+for {set inf 1} {$inf <= $inferiors} {incr inf} {
+ with_test_prefix "inferior $inf" {
+ # Switch to the target inferior.
+ gdb_test "inferior $inf" ".*Switching to inferior $inf .*"
+ # Run the tests.
+ do_tests
+ }
+}
gdb_exit
--
2.25.1
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [relevance 64%]
* RE: [PING][PATCH 0/2] Some patches for multi inferior case
2022-08-25 12:54 99% ` Thiago Jung Bauermann
@ 2022-08-25 14:55 99% ` Sargsyan, Eduard
0 siblings, 0 replies; 34+ results
From: Sargsyan, Eduard @ 2022-08-25 14:55 UTC (permalink / raw)
To: Thiago Jung Bauermann; +Cc: gdb-patches
Hi!
Here are changes I was talking about:
https://sourceware.org/pipermail/gdb-patches/2022-May/189337.html
https://sourceware.org/pipermail/gdb-patches/2022-May/189338.html
Thanks.
-----Original Message-----
From: Thiago Jung Bauermann <thiago.bauermann@linaro.org>
Sent: Thursday, August 25, 2022 2:55 PM
To: Sargsyan, Eduard <eduard.sargsyan@intel.com>
Cc: gdb-patches@sourceware.org
Subject: Re: [PING][PATCH 0/2] Some patches for multi inferior case
Hello Eduard,
"Sargsyan, Eduard via Gdb-patches" <gdb-patches@sourceware.org> writes:
> Hi all,
>
> Do we have any ideas/comments on the patches I would like to merge? I haven't got any feedback since May...
Did you post the patches? I only have the cover letter in my mail folder.
I also checked the new public-inbox archive (thanks Mark Wielaard!):
https://inbox.sourceware.org/gdb-patches/?q=s%3A%22multi+inferior+case%22
And I only see the cover letter there as well.
--
Thiago
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [relevance 99%]
* Re: [PING][PATCH 0/2] Some patches for multi inferior case
2022-08-25 9:04 96% ` Sargsyan, Eduard
2022-08-25 9:04 95% ` Sargsyan, Eduard
@ 2022-08-25 12:54 99% ` Thiago Jung Bauermann
2022-08-25 14:55 99% ` Sargsyan, Eduard
1 sibling, 1 reply; 34+ results
From: Thiago Jung Bauermann @ 2022-08-25 12:54 UTC (permalink / raw)
To: Sargsyan, Eduard; +Cc: gdb-patches
Hello Eduard,
"Sargsyan, Eduard via Gdb-patches" <gdb-patches@sourceware.org> writes:
> Hi all,
>
> Do we have any ideas/comments on the patches I would like to merge? I haven't got any feedback since May...
Did you post the patches? I only have the cover letter in my mail
folder.
I also checked the new public-inbox archive (thanks Mark Wielaard!):
https://inbox.sourceware.org/gdb-patches/?q=s%3A%22multi+inferior+case%22
And I only see the cover letter there as well.
--
Thiago
^ permalink raw reply [relevance 99%]
* RE: [PING][PATCH 0/2] Some patches for multi inferior case
2022-08-01 7:06 99% ` Sargsyan, Eduard
@ 2022-08-25 9:04 96% ` Sargsyan, Eduard
2022-08-25 9:04 95% ` Sargsyan, Eduard
2022-08-25 12:54 99% ` Thiago Jung Bauermann
0 siblings, 2 replies; 34+ results
From: Sargsyan, Eduard @ 2022-08-25 9:04 UTC (permalink / raw)
To: gdb-patches
Hi all,
Do we have any ideas/comments on the patches I would like to merge? I haven't got any feedback since May...
From: Sargsyan, Eduard
Sent: Monday, August 1, 2022 9:06 AM
To: gdb-patches@sourceware.org
Subject: RE: [PING][PATCH 0/2] Some patches for multi inferior case
Hello.
Guys do we have any updates, comments on patches suggested to integrate?
From: Sargsyan, Eduard <eduard.sargsyan@intel.com<mailto:eduard.sargsyan@intel.com>>
Sent: Thursday, July 7, 2022 3:46 PM
To: gdb-patches@sourceware.org<mailto:gdb-patches@sourceware.org>
Subject: [PING][PATCH 0/2] Some patches for multi inferior case
Hi all.
Any updates/thoughts/comments?
________________________________
From: Sargsyan, Eduard <eduard.sargsyan@intel.com<mailto:eduard.sargsyan@intel.com>>
Sent: Monday, May 23, 2022 7:56 PM
To: gdb-patches@sourceware.org<mailto:gdb-patches@sourceware.org> <gdb-patches@sourceware.org<mailto:gdb-patches@sourceware.org>>
Cc: Sargsyan, Eduard <eduard.sargsyan@intel.com<mailto:eduard.sargsyan@intel.com>>
Subject: [PATCH 0/2] Some patches for multi inferior case
Hi all,
this is a set of patches originally written by Alexandar Paunovic.
I just rebased them onto current master.
Regards,
Eduard
Aleksandar Paunovic (2):
gdb: Fix deleted thread when issuing next command
gdb: Improve the resuming of the stepped thread
gdb/infrun.c | 17 +++-
.../gdb.base/breakpoint-running-inferior-1.c | 39 ++++++++
.../gdb.base/breakpoint-running-inferior-2.c | 39 ++++++++
.../gdb.base/breakpoint-running-inferior.exp | 89 +++++++++++++++++++
4 files changed, 183 insertions(+), 1 deletion(-)
create mode 100644 gdb/testsuite/gdb.base/breakpoint-running-inferior-1.c
create mode 100644 gdb/testsuite/gdb.base/breakpoint-running-inferior-2.c
create mode 100644 gdb/testsuite/gdb.base/breakpoint-running-inferior.exp
--
2.25.1
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [relevance 96%]
* RE: [PING][PATCH 0/2] Some patches for multi inferior case
2022-08-25 9:04 96% ` Sargsyan, Eduard
@ 2022-08-25 9:04 95% ` Sargsyan, Eduard
2022-08-25 12:54 99% ` Thiago Jung Bauermann
1 sibling, 0 replies; 34+ results
From: Sargsyan, Eduard @ 2022-08-25 9:04 UTC (permalink / raw)
To: gdb-patches
[-- Attachment #1: Type: text/plain, Size: 2282 bytes --]
Hi all,
Do we have any ideas/comments on the patches I would like to merge? I haven't got any feedback since May...
From: Sargsyan, Eduard
Sent: Monday, August 1, 2022 9:06 AM
To: gdb-patches@sourceware.org
Subject: RE: [PING][PATCH 0/2] Some patches for multi inferior case
Hello.
Guys do we have any updates, comments on patches suggested to integrate?
From: Sargsyan, Eduard <eduard.sargsyan@intel.com<mailto:eduard.sargsyan@intel.com>>
Sent: Thursday, July 7, 2022 3:46 PM
To: gdb-patches@sourceware.org<mailto:gdb-patches@sourceware.org>
Subject: [PING][PATCH 0/2] Some patches for multi inferior case
Hi all.
Any updates/thoughts/comments?
________________________________
From: Sargsyan, Eduard <eduard.sargsyan@intel.com<mailto:eduard.sargsyan@intel.com>>
Sent: Monday, May 23, 2022 7:56 PM
To: gdb-patches@sourceware.org<mailto:gdb-patches@sourceware.org> <gdb-patches@sourceware.org<mailto:gdb-patches@sourceware.org>>
Cc: Sargsyan, Eduard <eduard.sargsyan@intel.com<mailto:eduard.sargsyan@intel.com>>
Subject: [PATCH 0/2] Some patches for multi inferior case
Hi all,
this is a set of patches originally written by Alexandar Paunovic.
I just rebased them onto current master.
Regards,
Eduard
Aleksandar Paunovic (2):
gdb: Fix deleted thread when issuing next command
gdb: Improve the resuming of the stepped thread
gdb/infrun.c | 17 +++-
.../gdb.base/breakpoint-running-inferior-1.c | 39 ++++++++
.../gdb.base/breakpoint-running-inferior-2.c | 39 ++++++++
.../gdb.base/breakpoint-running-inferior.exp | 89 +++++++++++++++++++
4 files changed, 183 insertions(+), 1 deletion(-)
create mode 100644 gdb/testsuite/gdb.base/breakpoint-running-inferior-1.c
create mode 100644 gdb/testsuite/gdb.base/breakpoint-running-inferior-2.c
create mode 100644 gdb/testsuite/gdb.base/breakpoint-running-inferior.exp
--
2.25.1
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [relevance 95%]
* RE: [PING][PATCH 0/2] Some patches for multi inferior case
2022-07-07 13:46 99% ` [PING][PATCH " Sargsyan, Eduard
@ 2022-08-01 7:06 99% ` Sargsyan, Eduard
2022-08-25 9:04 96% ` Sargsyan, Eduard
0 siblings, 1 reply; 34+ results
From: Sargsyan, Eduard @ 2022-08-01 7:06 UTC (permalink / raw)
To: gdb-patches
Hello.
Guys do we have any updates, comments on patches suggested to integrate?
From: Sargsyan, Eduard <eduard.sargsyan@intel.com>
Sent: Thursday, July 7, 2022 3:46 PM
To: gdb-patches@sourceware.org
Subject: [PING][PATCH 0/2] Some patches for multi inferior case
Hi all.
Any updates/thoughts/comments?
________________________________
From: Sargsyan, Eduard <eduard.sargsyan@intel.com<mailto:eduard.sargsyan@intel.com>>
Sent: Monday, May 23, 2022 7:56 PM
To: gdb-patches@sourceware.org<mailto:gdb-patches@sourceware.org> <gdb-patches@sourceware.org<mailto:gdb-patches@sourceware.org>>
Cc: Sargsyan, Eduard <eduard.sargsyan@intel.com<mailto:eduard.sargsyan@intel.com>>
Subject: [PATCH 0/2] Some patches for multi inferior case
Hi all,
this is a set of patches originally written by Alexandar Paunovic.
I just rebased them onto current master.
Regards,
Eduard
Aleksandar Paunovic (2):
gdb: Fix deleted thread when issuing next command
gdb: Improve the resuming of the stepped thread
gdb/infrun.c | 17 +++-
.../gdb.base/breakpoint-running-inferior-1.c | 39 ++++++++
.../gdb.base/breakpoint-running-inferior-2.c | 39 ++++++++
.../gdb.base/breakpoint-running-inferior.exp | 89 +++++++++++++++++++
4 files changed, 183 insertions(+), 1 deletion(-)
create mode 100644 gdb/testsuite/gdb.base/breakpoint-running-inferior-1.c
create mode 100644 gdb/testsuite/gdb.base/breakpoint-running-inferior-2.c
create mode 100644 gdb/testsuite/gdb.base/breakpoint-running-inferior.exp
--
2.25.1
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [relevance 99%]
* [PING][PATCH 0/2] Some patches for multi inferior case
2022-05-23 17:56 99% [PATCH 0/2] Some patches for multi inferior case Eduard Sargsyan
@ 2022-07-07 13:46 99% ` Sargsyan, Eduard
2022-08-01 7:06 99% ` Sargsyan, Eduard
0 siblings, 1 reply; 34+ results
From: Sargsyan, Eduard @ 2022-07-07 13:46 UTC (permalink / raw)
To: gdb-patches
Hi all.
Any updates/thoughts/comments?
________________________________
From: Sargsyan, Eduard <eduard.sargsyan@intel.com>
Sent: Monday, May 23, 2022 7:56 PM
To: gdb-patches@sourceware.org <gdb-patches@sourceware.org>
Cc: Sargsyan, Eduard <eduard.sargsyan@intel.com>
Subject: [PATCH 0/2] Some patches for multi inferior case
Hi all,
this is a set of patches originally written by Alexandar Paunovic.
I just rebased them onto current master.
Regards,
Eduard
Aleksandar Paunovic (2):
gdb: Fix deleted thread when issuing next command
gdb: Improve the resuming of the stepped thread
gdb/infrun.c | 17 +++-
.../gdb.base/breakpoint-running-inferior-1.c | 39 ++++++++
.../gdb.base/breakpoint-running-inferior-2.c | 39 ++++++++
.../gdb.base/breakpoint-running-inferior.exp | 89 +++++++++++++++++++
4 files changed, 183 insertions(+), 1 deletion(-)
create mode 100644 gdb/testsuite/gdb.base/breakpoint-running-inferior-1.c
create mode 100644 gdb/testsuite/gdb.base/breakpoint-running-inferior-2.c
create mode 100644 gdb/testsuite/gdb.base/breakpoint-running-inferior.exp
--
2.25.1
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
^ permalink raw reply [relevance 99%]
* RE: [PATCH 0/2] Some patches for multi inferior case
2022-05-23 18:00 99% [PATCH " Eduard Sargsyan
@ 2022-06-22 8:22 99% ` Sargsyan, Eduard
0 siblings, 0 replies; 34+ results
From: Sargsyan, Eduard @ 2022-06-22 8:22 UTC (permalink / raw)
To: gdb-patches
Hello.
Any ideas about the patches?
-----Original Message-----
From: Sargsyan, Eduard <eduard.sargsyan@intel.com>
Sent: Monday, May 23, 2022 8:01 PM
To: gdb-patches@sourceware.org
Cc: Sargsyan, Eduard <eduard.sargsyan@intel.com>
Subject: [PATCH 0/2] Some patches for multi inferior case
Hi all,
this is a set of patches originally written by Alexandar Paunovic.
I just rebased them onto current master.
Regards,
Eduard
Aleksandar Paunovic (2):
gdb: Fix deleted thread when issuing next command
gdb: Improve the resuming of the stepped thread
gdb/infrun.c | 17 +++-
.../gdb.base/breakpoint-running-inferior-1.c | 39 ++++++++ .../gdb.base/breakpoint-running-inferior-2.c | 39 ++++++++ .../gdb.base/breakpoint-running-inferior.exp | 89 +++++++++++++++++++
4 files changed, 183 insertions(+), 1 deletion(-) create mode 100644 gdb/testsuite/gdb.base/breakpoint-running-inferior-1.c
create mode 100644 gdb/testsuite/gdb.base/breakpoint-running-inferior-2.c
create mode 100644 gdb/testsuite/gdb.base/breakpoint-running-inferior.exp
--
2.25.1
^ permalink raw reply [relevance 99%]
* [PATCH 0/2] Some patches for multi inferior case
@ 2022-05-23 18:00 99% Eduard Sargsyan
2022-06-22 8:22 99% ` Sargsyan, Eduard
0 siblings, 1 reply; 34+ results
From: Eduard Sargsyan @ 2022-05-23 18:00 UTC (permalink / raw)
To: gdb-patches; +Cc: Eduard Sargsyan
Hi all,
this is a set of patches originally written by Alexandar Paunovic.
I just rebased them onto current master.
Regards,
Eduard
Aleksandar Paunovic (2):
gdb: Fix deleted thread when issuing next command
gdb: Improve the resuming of the stepped thread
gdb/infrun.c | 17 +++-
.../gdb.base/breakpoint-running-inferior-1.c | 39 ++++++++
.../gdb.base/breakpoint-running-inferior-2.c | 39 ++++++++
.../gdb.base/breakpoint-running-inferior.exp | 89 +++++++++++++++++++
4 files changed, 183 insertions(+), 1 deletion(-)
create mode 100644 gdb/testsuite/gdb.base/breakpoint-running-inferior-1.c
create mode 100644 gdb/testsuite/gdb.base/breakpoint-running-inferior-2.c
create mode 100644 gdb/testsuite/gdb.base/breakpoint-running-inferior.exp
--
2.25.1
^ permalink raw reply [relevance 99%]
* [PATCH 0/2] Some patches for multi inferior case
@ 2022-05-23 17:56 99% Eduard Sargsyan
2022-07-07 13:46 99% ` [PING][PATCH " Sargsyan, Eduard
0 siblings, 1 reply; 34+ results
From: Eduard Sargsyan @ 2022-05-23 17:56 UTC (permalink / raw)
To: gdb-patches; +Cc: Eduard Sargsyan
Hi all,
this is a set of patches originally written by Alexandar Paunovic.
I just rebased them onto current master.
Regards,
Eduard
Aleksandar Paunovic (2):
gdb: Fix deleted thread when issuing next command
gdb: Improve the resuming of the stepped thread
gdb/infrun.c | 17 +++-
.../gdb.base/breakpoint-running-inferior-1.c | 39 ++++++++
.../gdb.base/breakpoint-running-inferior-2.c | 39 ++++++++
.../gdb.base/breakpoint-running-inferior.exp | 89 +++++++++++++++++++
4 files changed, 183 insertions(+), 1 deletion(-)
create mode 100644 gdb/testsuite/gdb.base/breakpoint-running-inferior-1.c
create mode 100644 gdb/testsuite/gdb.base/breakpoint-running-inferior-2.c
create mode 100644 gdb/testsuite/gdb.base/breakpoint-running-inferior.exp
--
2.25.1
^ permalink raw reply [relevance 99%]
Results 1-34 of 34 | reverse | sort options + mbox downloads above
-- links below jump to the message on this page --
2022-05-23 17:56 99% [PATCH 0/2] Some patches for multi inferior case Eduard Sargsyan
2022-07-07 13:46 99% ` [PING][PATCH " Sargsyan, Eduard
2022-08-01 7:06 99% ` Sargsyan, Eduard
2022-08-25 9:04 96% ` Sargsyan, Eduard
2022-08-25 9:04 95% ` Sargsyan, Eduard
2022-08-25 12:54 99% ` Thiago Jung Bauermann
2022-08-25 14:55 99% ` Sargsyan, Eduard
2022-05-23 18:00 99% [PATCH " Eduard Sargsyan
2022-06-22 8:22 99% ` Sargsyan, Eduard
2023-02-21 15:13 64% [PATCH 1/1] gdb, infcmd: Support jump command in multi-inferior case Matti Puputti
2023-02-21 16:56 99% ` Andrew Burgess
2023-02-22 17:43 99% ` Puputti, Matti
2023-02-23 9:54 99% [PATCH v2 0/1] " Matti Puputti
2023-02-23 9:54 63% ` [PATCH v2 1/1] " Matti Puputti
2023-04-24 14:10 99% ` Andrew Burgess
2023-04-25 13:40 99% ` Puputti, Matti
2023-03-07 16:50 99% ` [PING] [PATCH v2 0/1] " Puputti, Matti
2023-03-24 7:19 99% ` Puputti, Matti
2023-03-31 5:04 99% ` Puputti, Matti
2023-04-06 12:59 99% ` Puputti, Matti
2023-04-14 14:24 99% ` Puputti, Matti
2023-04-21 7:24 99% ` Puputti, Matti
2023-04-25 13:38 64% [PATCH v3 1/1] " Matti Puputti
2023-05-22 8:01 99% ` [PING][PATCH " Puputti, Matti
2023-05-24 17:33 91% ` [PATCH " Andrew Burgess
2023-05-25 5:38 99% ` Puputti, Matti
2023-06-28 9:04 99% ` Puputti, Matti
2023-06-28 9:00 65% [PATCH v4 " Matti Puputti
2023-07-04 14:38 99% ` Andrew Burgess
2023-07-26 11:51 99% ` Puputti, Matti
2023-07-26 12:32 99% ` Puputti, Matti
2023-07-26 12:29 66% [PATCH v6 " Matti Puputti
2023-08-16 8:58 99% ` Andrew Burgess
2023-08-16 10:55 99% ` Puputti, Matti
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).