Patchwork [PATCHv2,2/2] gdb: Change how frames are selected for 'frame' and 'info frame'.

login
register
mail settings
Submitter Andrew Burgess
Date May 8, 2018, 4:58 p.m.
Message ID <63020671a997f926cd747677cd4e614e51e81f8d.1525797846.git.andrew.burgess@embecosm.com>
Download mbox | patch
Permalink /patch/27156/
State New
Headers show

Comments

Andrew Burgess - May 8, 2018, 4:58 p.m.
The 'frame' command, and thanks to code reuse the 'info frame' and
'select-frame' commands, currently have an overloaded mechanism for
selecting a frame.

These commands take one or two parameters, if it's one parameter then
we first try to use the parameter as an integer to select a frame by
level (or depth in the stack).  If that fails then we treat the
parameter as an address and try to select a stack frame by
stack-address.  If we still have not selected a stack frame, or we
initially had two parameters, then we create a new stack frame and
select that.

The result of this is that a typo by the user, entering the wrong stack
frame level for example, can result in a brand new frame being created
rather than an error.

The purpose of this commit is to remove this overloading, while still
offering the same functionality through some new sub-commands.  By
making the default behaviour of 'frame' (and friends) be to select a
stack frame by level number, it is hoped that enough
backwards-compatibility is maintained that users will not be overly
inconvenienced.

The 'frame', 'select-frame', and 'info frame' commands now all take a
frame specification string as an argument, this string can be any of the
following:

  (1) An integer.  This is treated as a frame number.  If a frame for
  that number does not exist then the user gets an error.

  (2) A string like 'level <NUMBER>', where <NUMBER> is a frame number
  as in option (1) above.

  (3) A string like 'address <ADDRESS>', where <ADDRESS> is a
  stack-frame address.  If there is no frame for this address then the
  user gets an error.

  (4) A string like 'function <NAME>', where <NAME> is a function name,
  the inner most frame for function <NAME> is selected.  If there is no
  frame for function <NAME> then the user gets an error.

  (5) A string like 'create <STACK-ADDRESS>', this creates a new frame
  with stack address <STACK-ADDRESS>.

  (6) A string like 'create <STACK-ADDRESS> <PC-ADDRESS>', this creates
  a new frame with stack address <STACK-ADDRESS> an the pc <PC-ADDRESS>.

This change assumes that the most common use of the commands like
'frame' is to select a frame by frame level number, it is for this
reason that this is the behaviour that is kept for backwards
compatibility.  Any of the alternative behaviours, which are assumed to
be less used, now require a change in user behaviour.

The MI command '-stack-select-frame' has also changed in the same way.
The default behaviour is to select a frame by level number, but the
other selection methods are also available.

gdb/ChangeLog:

	(NEWS): Mention changes to frame related commands.
	* cli/cli-decode.c (add_cmd_suppress_notification): New function.
	(add_prefix_cmd_suppress_notification): New function.
	(add_com_suppress_notification): Call
	add_cmd_suppress_notification.
	* command.h (add_cmd_suppress_notification): Declare.
	(add_prefix_cmd_suppress_notification): Declare.
	* mi/mi-cmd-stack.c (mi_cmd_stack_select_frame): Use new function
	select_frame_from_spec.
	* stack.c (find_frame_for_function): Add declaration.
	(find_frame_for_address): New function.
	(parse_frame_specification): Delete function.
	(frame_selection_by_function_completer): New function.
	(info_frame_command): Rename to...
	(info_frame_command_core): ...this, and update parameter types.
	(select_frame_command): Rename to...
	(select_frame_command_core): ...this, and update parameter types.
	(frame_command): Rename to...
	(frame_command_core): ...this, and update parameter types.
	(class frame_command_helper): New class to wrap implementations of
	frame related sub-commands.
	(select_frame_from_spec): New function.
	(frame_cmd_list): New global for sub-commands.
	(select_frame_cmd_list): New global for sub-commands.
	(info_frame_cmd_list): New global for sub-commands.
	(_initialize_stack): Register sub-commands for 'frame',
	'select-frame', and 'info frame'.
	* stack.h (select_frame_command): Delete declarationn.
	(select_frame_from_spec): Declare new function.

gdb/doc/ChangeLog:

	* gdb.texinfo (Selection): Rewrite documentation for 'frame' and
	'select-frame' commands.
	(Frame Info): Rewrite documentation for 'info frame' command.
	(GDB/MI Stack Manipulation): Update description of
	'-stack-select-frame'.

gdb/testsuite/ChangeLog:

	* gdb.base/frame-selection.exp: New file.
	* gdb.base/frame-selection.c: New file.
	* gdb.mi/mi-frame-selection.exp: New file.
	* gdb.mi/mi-frame-selection.c: New file.
---
 gdb/ChangeLog                               |  32 ++
 gdb/NEWS                                    |  29 ++
 gdb/cli/cli-decode.c                        |  44 ++-
 gdb/command.h                               |  14 +
 gdb/doc/ChangeLog                           |   8 +
 gdb/doc/gdb.texinfo                         |  91 +++--
 gdb/mi/mi-cmd-stack.c                       |   4 +-
 gdb/stack.c                                 | 518 +++++++++++++++++++---------
 gdb/stack.h                                 |   2 +-
 gdb/testsuite/ChangeLog                     |   7 +
 gdb/testsuite/gdb.base/frame-selection.c    |  52 +++
 gdb/testsuite/gdb.base/frame-selection.exp  | 156 +++++++++
 gdb/testsuite/gdb.mi/mi-frame-selection.c   |  34 ++
 gdb/testsuite/gdb.mi/mi-frame-selection.exp |  90 +++++
 14 files changed, 884 insertions(+), 197 deletions(-)
 create mode 100644 gdb/testsuite/gdb.base/frame-selection.c
 create mode 100644 gdb/testsuite/gdb.base/frame-selection.exp
 create mode 100644 gdb/testsuite/gdb.mi/mi-frame-selection.c
 create mode 100644 gdb/testsuite/gdb.mi/mi-frame-selection.exp
Eli Zaretskii - May 11, 2018, 1:46 p.m.
> From: Andrew Burgess <andrew.burgess@embecosm.com>
> Cc: Andrew Burgess <andrew.burgess@embecosm.com>
> Date: Tue,  8 May 2018 17:58:45 +0100
> 
> The 'frame' command, and thanks to code reuse the 'info frame' and
> 'select-frame' commands, currently have an overloaded mechanism for
> selecting a frame.
> 
> These commands take one or two parameters, if it's one parameter then
> we first try to use the parameter as an integer to select a frame by
> level (or depth in the stack).  If that fails then we treat the
> parameter as an address and try to select a stack frame by
> stack-address.  If we still have not selected a stack frame, or we
> initially had two parameters, then we create a new stack frame and
> select that.
> 
> The result of this is that a typo by the user, entering the wrong stack
> frame level for example, can result in a brand new frame being created
> rather than an error.
> 
> The purpose of this commit is to remove this overloading, while still
> offering the same functionality through some new sub-commands.

An alternative could be to request confirmation before creating a new
frame.

> diff --git a/gdb/NEWS b/gdb/NEWS
> index 46f6635dda0..fe91887cde4 100644
> --- a/gdb/NEWS
> +++ b/gdb/NEWS
> @@ -12,6 +12,35 @@
>  * C expressions can now use _Alignof, and C++ expressions can now use
>    alignof.
>  
> +* Changes to the "frame", "select-frame", and "info frame" CLI
> +  commands, as well as the "-stack-select-frame" MI command.
> +  Selecting frames by number remains unchanged, however, selecting
> +  frames by stack-address, or creating new frames now requires the use
> +  of a sub-command.  Various sub-commands now exist for the various
> +  methods of selecting a frame; level, address, function, and create.
> +  As an example, here are the variants of "frame" that are now available:
> +  - frame <number>
> +  - frame level <number>
> +    Both of these select frame at level <number>.
> +  - frame address <stack-address>
> +    Select frame for <stack-address>.
> +  - frame function <name>
> +    Select inner most frame for function <name>.
> +  - frame create <stack-address>
> +    Create a frame for <stack-address>.
> +  - frame create <stack-address> <pc-address>
> +    Create a frame for <stack-address> <pc-address>.
> +  There are similar variants for the "select-frame" and "info frame"
> +  commands.  The MI command "-stack-select-frame" places the
> +  sub-command name after the command name as a keyword, the available
> +  variants are:
> +  -stack-select-frame <number>
> +  -stack-select-frame level <number>
> +  -stack-select-frame address <stack-address>
> +  -stack-select-frame function <name>
> +  -stack-select-frame create <stack-address>
> +  -stack-select-frame create <stack-address> <pc-address>

This is IMO too wordy for NEWS.  I think it's enough to mention that
new sub-command and tell that creation of new frames now requires that
sub-command.

> -@item frame @var{stack-addr} [ @var{pc-addr} ]
> -@itemx f @var{stack-addr} [ @var{pc-addr} ]
> -Select the frame at address @var{stack-addr}.  This is useful mainly if the
> -chaining of stack frames has been damaged by a bug, making it
> -impossible for @value{GDBN} to assign numbers properly to all frames.  In
> -addition, this can be useful when your program has multiple stacks and
> -switches between them.  The optional @var{pc-addr} can also be given to
> -specify the value of PC for the stack frame.
> +@kindex frame address
> +@item address @var{stack-address}
> +Select the frame with stack address @var{stack-address}.  Recall that
> +each stack frame has a stack frame address, which roughly corresponds
> +to the location on the stack where the stack frame local variables are
> +stored.

The text in the "Recall" sentence IMO doesn't convey anything useful,
it just says that each frame has a stack address which can be used to
select it, i.e. repeats what the previous sentence already said.

> +@kindex frame create
> +@item create @var{stack-address} @r{[} @var{pc-addr} @r{]}
> +Create and then select a new frame at stack address @var{stack-addr}.
> +This is useful mainly if the chaining of stack frames has been damaged
> +by a bug, making it impossible for @value{GDBN} to assign numbers
> +properly to all frames.  In addition, this can be useful when your
> +program has multiple stacks and switches between them.  The optional
> +@var{pc-addr} can also be given to specify the value of PC for the
> +stack frame.

I'm surprised: is it true that PC-ADDR is used only for creating new
stack frames?  The description refers to programs with multiple
stacks, in which case I'd expect to be able to refer to existing stack
frames on another stack.  The original text seemed to allow that, at
least implicitly.  Also, is it true that damaged stack is only
relevant when creating new frames?

Thanks.
Andrew Burgess - May 21, 2018, 11:53 a.m.
Eli,

Thanks for your feedback.  I've addressed two of your comments, but
wanted to better understand your third point so that I could address
is correctly...

* Eli Zaretskii <eliz@gnu.org> [2018-05-11 16:46:30 +0300]:

> > From: Andrew Burgess <andrew.burgess@embecosm.com>
> > Cc: Andrew Burgess <andrew.burgess@embecosm.com>
> > Date: Tue,  8 May 2018 17:58:45 +0100
> > 
> > The 'frame' command, and thanks to code reuse the 'info frame' and
> > 'select-frame' commands, currently have an overloaded mechanism for
> > selecting a frame.
> > 
> > These commands take one or two parameters, if it's one parameter then
> > we first try to use the parameter as an integer to select a frame by
> > level (or depth in the stack).  If that fails then we treat the
> > parameter as an address and try to select a stack frame by
> > stack-address.  If we still have not selected a stack frame, or we
> > initially had two parameters, then we create a new stack frame and
> > select that.
> > 
> > The result of this is that a typo by the user, entering the wrong stack
> > frame level for example, can result in a brand new frame being created
> > rather than an error.
> > 
> > The purpose of this commit is to remove this overloading, while still
> > offering the same functionality through some new sub-commands.
> 
> An alternative could be to request confirmation before creating a new
> frame.
> 
> > diff --git a/gdb/NEWS b/gdb/NEWS
> > index 46f6635dda0..fe91887cde4 100644
> > --- a/gdb/NEWS
> > +++ b/gdb/NEWS
> > @@ -12,6 +12,35 @@
> >  * C expressions can now use _Alignof, and C++ expressions can now use
> >    alignof.
> >  
> > +* Changes to the "frame", "select-frame", and "info frame" CLI
> > +  commands, as well as the "-stack-select-frame" MI command.
> > +  Selecting frames by number remains unchanged, however, selecting
> > +  frames by stack-address, or creating new frames now requires the use
> > +  of a sub-command.  Various sub-commands now exist for the various
> > +  methods of selecting a frame; level, address, function, and create.
> > +  As an example, here are the variants of "frame" that are now available:
> > +  - frame <number>
> > +  - frame level <number>
> > +    Both of these select frame at level <number>.
> > +  - frame address <stack-address>
> > +    Select frame for <stack-address>.
> > +  - frame function <name>
> > +    Select inner most frame for function <name>.
> > +  - frame create <stack-address>
> > +    Create a frame for <stack-address>.
> > +  - frame create <stack-address> <pc-address>
> > +    Create a frame for <stack-address> <pc-address>.
> > +  There are similar variants for the "select-frame" and "info frame"
> > +  commands.  The MI command "-stack-select-frame" places the
> > +  sub-command name after the command name as a keyword, the available
> > +  variants are:
> > +  -stack-select-frame <number>
> > +  -stack-select-frame level <number>
> > +  -stack-select-frame address <stack-address>
> > +  -stack-select-frame function <name>
> > +  -stack-select-frame create <stack-address>
> > +  -stack-select-frame create <stack-address> <pc-address>
> 
> This is IMO too wordy for NEWS.  I think it's enough to mention that
> new sub-command and tell that creation of new frames now requires that
> sub-command.
> 
> > -@item frame @var{stack-addr} [ @var{pc-addr} ]
> > -@itemx f @var{stack-addr} [ @var{pc-addr} ]
> > -Select the frame at address @var{stack-addr}.  This is useful mainly if the
> > -chaining of stack frames has been damaged by a bug, making it
> > -impossible for @value{GDBN} to assign numbers properly to all frames.  In
> > -addition, this can be useful when your program has multiple stacks and
> > -switches between them.  The optional @var{pc-addr} can also be given to
> > -specify the value of PC for the stack frame.
> > +@kindex frame address
> > +@item address @var{stack-address}
> > +Select the frame with stack address @var{stack-address}.  Recall that
> > +each stack frame has a stack frame address, which roughly corresponds
> > +to the location on the stack where the stack frame local variables are
> > +stored.
> 
> The text in the "Recall" sentence IMO doesn't convey anything useful,
> it just says that each frame has a stack address which can be used to
> select it, i.e. repeats what the previous sentence already said.
> 
> > +@kindex frame create
> > +@item create @var{stack-address} @r{[} @var{pc-addr} @r{]}
> > +Create and then select a new frame at stack address @var{stack-addr}.
> > +This is useful mainly if the chaining of stack frames has been damaged
> > +by a bug, making it impossible for @value{GDBN} to assign numbers
> > +properly to all frames.  In addition, this can be useful when your
> > +program has multiple stacks and switches between them.  The optional
> > +@var{pc-addr} can also be given to specify the value of PC for the
> > +stack frame.
> 
> I'm surprised: is it true that PC-ADDR is used only for creating new
> stack frames?  The description refers to programs with multiple
> stacks, in which case I'd expect to be able to refer to existing stack
> frames on another stack.  The original text seemed to allow that, at
> least implicitly.  Also, is it true that damaged stack is only
> relevant when creating new frames?

I don't really understand what it is you're asking here, but I think
it might be related to overloading of the word "frame", and possibly
the keyword "create" not being a good choice.

Sure, within the inferior there are a set of frames, some of these
form the current stack, and some might exist on some alternative
stack.  Those frames exist regardless of GDB's ability to visualise
them.

However, there's a second use of the word frame, which we use for
GDB's representation of a stack-frame.  Under this second meaning the
only frames that exist are those GDB can derive from the current
machine state.  So, if the user asks for a backtrace and is told about
#0 A, #1 B, #2 C, then GDB only knows about those 3 frames.

If the user knows of #3 D that called C (but the unwind failed for
some reason) then they can use: 'frame create STACK-ADDR PC-ADDR' to
"create" a suitable frame and examine the machine state.  The frame
being "created" here is really a GDB frame, that is, a representation
of a preexisting inferior frame.

Similarly, if there's a second stack the user can "create" frames to
represent the frames on the second stack.

What we /can't/ do at the moment is backtrace out of these newly
created frames, which is a bit of a shame.  I do have a plan to work
on this, but that's for another day...

Anyway, I'm happy to rework the text if you can suggest some
improvements.  Alternatively, maybe it was my choice of "create" as
the keyword that was confusing... again, if you have any better
suggestions I'm happy to change it, I'm not tied to "create".

Thanks,

Andrew
Eli Zaretskii - May 21, 2018, 4:32 p.m.
> Date: Mon, 21 May 2018 12:53:57 +0100
> From: Andrew Burgess <andrew.burgess@embecosm.com>
> Cc: gdb-patches@sourceware.org
> 
> I don't really understand what it is you're asking here, but I think
> it might be related to overloading of the word "frame", and possibly
> the keyword "create" not being a good choice.
> 
> Sure, within the inferior there are a set of frames, some of these
> form the current stack, and some might exist on some alternative
> stack.  Those frames exist regardless of GDB's ability to visualise
> them.
> 
> However, there's a second use of the word frame, which we use for
> GDB's representation of a stack-frame.  Under this second meaning the
> only frames that exist are those GDB can derive from the current
> machine state.  So, if the user asks for a backtrace and is told about
> #0 A, #1 B, #2 C, then GDB only knows about those 3 frames.
> 
> If the user knows of #3 D that called C (but the unwind failed for
> some reason) then they can use: 'frame create STACK-ADDR PC-ADDR' to
> "create" a suitable frame and examine the machine state.  The frame
> being "created" here is really a GDB frame, that is, a representation
> of a preexisting inferior frame.

The old text was this:

> > > -@item frame @var{stack-addr} [ @var{pc-addr} ]
> > > -@itemx f @var{stack-addr} [ @var{pc-addr} ]
> > > -Select the frame at address @var{stack-addr}.  This is useful mainly if the
> > > -chaining of stack frames has been damaged by a bug, making it
> > > -impossible for @value{GDBN} to assign numbers properly to all frames.  In
> > > -addition, this can be useful when your program has multiple stacks and
> > > -switches between them.  The optional @var{pc-addr} can also be given to
> > > -specify the value of PC for the stack frame.

This seems to imply that if the frame #3 D existed, but the unwind
failed to find it, the user could say "frame STACK-ADDR PC-ADDR" to
refer to that frame.

By contrast, your new text:

> > > +@kindex frame create
> > > +@item create @var{stack-address} @r{[} @var{pc-addr} @r{]}
> > > +Create and then select a new frame at stack address @var{stack-addr}.
> > > +This is useful mainly if the chaining of stack frames has been damaged
> > > +by a bug, making it impossible for @value{GDBN} to assign numbers
> > > +properly to all frames.  In addition, this can be useful when your
> > > +program has multiple stacks and switches between them.  The optional
> > > +@var{pc-addr} can also be given to specify the value of PC for the
> > > +stack frame.

forces the use of "create", which is confusing, since the frame does
exist on the stack, albeit unbeknownst to GDB.

IOW, I always thought of stack frames as existing or not independently
of whether the GDB unwinder succeeded to find them, so it's strange
for me to talk about "creating" a frame in this use case.  For that
matter, I don't understand why we need to force the user to type
"create" explicitly.

> Anyway, I'm happy to rework the text if you can suggest some
> improvements.  Alternatively, maybe it was my choice of "create" as
> the keyword that was confusing... again, if you have any better
> suggestions I'm happy to change it, I'm not tied to "create".

Why do we need an extra keyword, again?

If we do need a keyword, how about "frame add"?
Andrew Burgess - June 5, 2018, 6:53 p.m.
Eli,

Sorry for the delay in replying, this dropped off my radar for a while
with other things...

* Eli Zaretskii <eliz@gnu.org> [2018-05-21 19:32:33 +0300]:

> > Date: Mon, 21 May 2018 12:53:57 +0100
> > From: Andrew Burgess <andrew.burgess@embecosm.com>
> > Cc: gdb-patches@sourceware.org
> > 
> > I don't really understand what it is you're asking here, but I think
> > it might be related to overloading of the word "frame", and possibly
> > the keyword "create" not being a good choice.
> > 
> > Sure, within the inferior there are a set of frames, some of these
> > form the current stack, and some might exist on some alternative
> > stack.  Those frames exist regardless of GDB's ability to visualise
> > them.
> > 
> > However, there's a second use of the word frame, which we use for
> > GDB's representation of a stack-frame.  Under this second meaning the
> > only frames that exist are those GDB can derive from the current
> > machine state.  So, if the user asks for a backtrace and is told about
> > #0 A, #1 B, #2 C, then GDB only knows about those 3 frames.
> > 
> > If the user knows of #3 D that called C (but the unwind failed for
> > some reason) then they can use: 'frame create STACK-ADDR PC-ADDR' to
> > "create" a suitable frame and examine the machine state.  The frame
> > being "created" here is really a GDB frame, that is, a representation
> > of a preexisting inferior frame.
> 
> The old text was this:
> 
> > > > -@item frame @var{stack-addr} [ @var{pc-addr} ]
> > > > -@itemx f @var{stack-addr} [ @var{pc-addr} ]
> > > > -Select the frame at address @var{stack-addr}.  This is useful mainly if the
> > > > -chaining of stack frames has been damaged by a bug, making it
> > > > -impossible for @value{GDBN} to assign numbers properly to all frames.  In
> > > > -addition, this can be useful when your program has multiple stacks and
> > > > -switches between them.  The optional @var{pc-addr} can also be given to
> > > > -specify the value of PC for the stack frame.
> 
> This seems to imply that if the frame #3 D existed, but the unwind
> failed to find it, the user could say "frame STACK-ADDR PC-ADDR" to
> refer to that frame.
> 
> By contrast, your new text:
> 
> > > > +@kindex frame create
> > > > +@item create @var{stack-address} @r{[} @var{pc-addr} @r{]}
> > > > +Create and then select a new frame at stack address @var{stack-addr}.
> > > > +This is useful mainly if the chaining of stack frames has been damaged
> > > > +by a bug, making it impossible for @value{GDBN} to assign numbers
> > > > +properly to all frames.  In addition, this can be useful when your
> > > > +program has multiple stacks and switches between them.  The optional
> > > > +@var{pc-addr} can also be given to specify the value of PC for the
> > > > +stack frame.
> 
> forces the use of "create", which is confusing, since the frame does
> exist on the stack, albeit unbeknownst to GDB.
> 
> IOW, I always thought of stack frames as existing or not independently
> of whether the GDB unwinder succeeded to find them, so it's strange
> for me to talk about "creating" a frame in this use case.

Sure. I think our mental models for stack frames and GDB are pretty
much in sync.  Like I previously tried to explain the "create" here is
only "creating" within GDB, not within the inferior....

>                                                            For that
> matter, I don't understand why we need to force the user to type
> "create" explicitly.

Because, whether we call it "creating" or "viewing an existing frame
that is in the inferior, but not in the backtrace", I do think that
there are two _very similar_ but different actions here.

One action is that the user wants to select a frame that GDB is aware
of, and is part of the backtrace.

The second action is that the user wants to use a $sp and $pc value to
visualise a stack frame that is not part of GDB's backtrace.

My assumption going into this work is that the first case (selecting
by number) is what most users do, most of the time, and that the
second case is much rarer.

By forcing the second case to live behind a new keyword we prevent the
user from accidentally visualising a new stack frame in the case where
they miss-type the stack frame number.  Under the new model,
miss-typing a frame number would give an error.

> 
> > Anyway, I'm happy to rework the text if you can suggest some
> > improvements.  Alternatively, maybe it was my choice of "create" as
> > the keyword that was confusing... again, if you have any better
> > suggestions I'm happy to change it, I'm not tied to "create".
> 
> Why do we need an extra keyword, again?
> 
> If we do need a keyword, how about "frame add"?

Personally, I think 'add' is worse than 'create' - what's the frame
being added too?  But I do acknowledge that 'create' is not ideal
either.

I wonder if 'new' is better than 'create', maybe implies less "making
something in the inferior"?  Or how about, 'for' instead, like this:

  (gdb) frame for STACK-ADDR PC-ADDR

Alternatively, could this problem be solved by a better description in
the manual, maybe something like this:

    Create within GDB a representation of a stack frame that is not
    part of the current backtrace.  The newly created frame has a
    stack address @var{stack-addr}, and an optional program counter
    address @var{pc-addr}.

    This is useful mainly if the chaining of stack frames has been
    damaged by a bug, making it impossible for @value{GDBN} to assign
    numbers properly to all frames.  In addition, this can be useful
    when your program has multiple stacks and switches between them.

Thanks for all your help reviewing this patch.

Andrew
Philippe Waroquiers - June 5, 2018, 9:15 p.m.
On Tue, 2018-06-05 at 19:53 +0100, Andrew Burgess wrote:
> > If we do need a keyword, how about "frame add"?
> 
> Personally, I think 'add' is worse than 'create' - what's the frame
> being added too?  But I do acknowledge that 'create' is not ideal
> either.
> 
> I wonder if 'new' is better than 'create', maybe implies less "making
> something in the inferior"?  Or how about, 'for' instead, like this:
> 
>   (gdb) frame for STACK-ADDR PC-ADDR

If nothing is added or created or ...,
maybe you could also use one of:
    (gdb) frame interpret STACK-ADDR PC-ADDR
or
    (gdb) frame look STACK-ADDR PC-ADDR
or
    (gdb) frame view STACK-ADDR PC-ADDR
?

Philippe
Andrew Burgess - June 6, 2018, 8:22 a.m.
* Philippe Waroquiers <philippe.waroquiers@skynet.be> [2018-06-05 23:15:16 +0200]:

> On Tue, 2018-06-05 at 19:53 +0100, Andrew Burgess wrote:
> > > If we do need a keyword, how about "frame add"?
> > 
> > Personally, I think 'add' is worse than 'create' - what's the frame
> > being added too?  But I do acknowledge that 'create' is not ideal
> > either.
> > 
> > I wonder if 'new' is better than 'create', maybe implies less "making
> > something in the inferior"?  Or how about, 'for' instead, like this:
> > 
> >   (gdb) frame for STACK-ADDR PC-ADDR
> 
> If nothing is added or created or ...,
> maybe you could also use one of:
>     (gdb) frame interpret STACK-ADDR PC-ADDR
> or
>     (gdb) frame look STACK-ADDR PC-ADDR
> or
>     (gdb) frame view STACK-ADDR PC-ADDR

I like 'view', the docs could read:

    View a frame that is not part of GDB backtrace.  The frame viewed
    has stack address @var{stack-addr}, and optionally, a program
    counter address of @var{pc-addr}.

    This is useful mainly if the chaining of stack frames has been
    damaged by a bug, making it impossible for @value{GDBN} to assign
    numbers properly to all frames.  In addition, this can be useful
    when your program has multiple stacks and switches between them.

Eli, how does this sound?

Thanks,
Andrew
Eli Zaretskii - June 6, 2018, 2:56 p.m.
> Date: Wed, 6 Jun 2018 09:22:11 +0100
> From: Andrew Burgess <andrew.burgess@embecosm.com>
> Cc: Philippe Waroquiers <philippe.waroquiers@skynet.be>,
> 	gdb-patches@sourceware.org
> 
>     View a frame that is not part of GDB backtrace.  The frame viewed
>     has stack address @var{stack-addr}, and optionally, a program
>     counter address of @var{pc-addr}.
> 
>     This is useful mainly if the chaining of stack frames has been
>     damaged by a bug, making it impossible for @value{GDBN} to assign
>     numbers properly to all frames.  In addition, this can be useful
>     when your program has multiple stacks and switches between them.
> 
> Eli, how does this sound?

Sounds good (modulo "@value{GDBN}" instead of "GDB").

Thanks.

Patch

diff --git a/gdb/NEWS b/gdb/NEWS
index 46f6635dda0..fe91887cde4 100644
--- a/gdb/NEWS
+++ b/gdb/NEWS
@@ -12,6 +12,35 @@ 
 * C expressions can now use _Alignof, and C++ expressions can now use
   alignof.
 
+* Changes to the "frame", "select-frame", and "info frame" CLI
+  commands, as well as the "-stack-select-frame" MI command.
+  Selecting frames by number remains unchanged, however, selecting
+  frames by stack-address, or creating new frames now requires the use
+  of a sub-command.  Various sub-commands now exist for the various
+  methods of selecting a frame; level, address, function, and create.
+  As an example, here are the variants of "frame" that are now available:
+  - frame <number>
+  - frame level <number>
+    Both of these select frame at level <number>.
+  - frame address <stack-address>
+    Select frame for <stack-address>.
+  - frame function <name>
+    Select inner most frame for function <name>.
+  - frame create <stack-address>
+    Create a frame for <stack-address>.
+  - frame create <stack-address> <pc-address>
+    Create a frame for <stack-address> <pc-address>.
+  There are similar variants for the "select-frame" and "info frame"
+  commands.  The MI command "-stack-select-frame" places the
+  sub-command name after the command name as a keyword, the available
+  variants are:
+  -stack-select-frame <number>
+  -stack-select-frame level <number>
+  -stack-select-frame address <stack-address>
+  -stack-select-frame function <name>
+  -stack-select-frame create <stack-address>
+  -stack-select-frame create <stack-address> <pc-address>
+
 * New commands
 
 set debug fbsd-nat
diff --git a/gdb/cli/cli-decode.c b/gdb/cli/cli-decode.c
index 0afe36aa4e1..3ec20346b08 100644
--- a/gdb/cli/cli-decode.c
+++ b/gdb/cli/cli-decode.c
@@ -278,6 +278,23 @@  add_cmd (const char *name, enum command_class theclass,
   return result;
 }
 
+/* Add an element with a suppress notification to the LIST of commands.  */
+
+struct cmd_list_element *
+add_cmd_suppress_notification (const char *name, enum command_class theclass,
+			       cmd_const_cfunc_ftype *fun, const char *doc,
+			       struct cmd_list_element **list,
+			       int *suppress_notification)
+{
+  struct cmd_list_element *element;
+
+  element = add_cmd (name, theclass, fun, doc, list);
+  element->suppress_notification = suppress_notification;
+
+  return element;
+}
+
+
 /* Deprecates a command CMD.
    REPLACEMENT is the name of the command which should be used in
    place of this command, or NULL if no such command exists.
@@ -389,6 +406,25 @@  add_prefix_cmd (const char *name, enum command_class theclass,
   return c;
 }
 
+/* Like ADD_PREFIX_CMD but sets the suppress_notification pointer on the
+   new command list element.  */
+
+struct cmd_list_element *
+add_prefix_cmd_suppress_notification
+               (const char *name, enum command_class theclass,
+		cmd_const_cfunc_ftype *fun,
+		const char *doc, struct cmd_list_element **prefixlist,
+		const char *prefixname, int allow_unknown,
+		struct cmd_list_element **list,
+		int *suppress_notification)
+{
+  struct cmd_list_element *element
+    = add_prefix_cmd (name, theclass, fun, doc, prefixlist,
+		      prefixname, allow_unknown, list);
+  element->suppress_notification = suppress_notification;
+  return element;
+}
+
 /* Like add_prefix_cmd but sets the abbrev_flag on the new command.  */
 
 struct cmd_list_element *
@@ -922,12 +958,8 @@  add_com_suppress_notification (const char *name, enum command_class theclass,
 			       cmd_const_cfunc_ftype *fun, const char *doc,
 			       int *suppress_notification)
 {
-  struct cmd_list_element *element;
-
-  element = add_cmd (name, theclass, fun, doc, &cmdlist);
-  element->suppress_notification = suppress_notification;
-
-  return element;
+  return add_cmd_suppress_notification (name, theclass, fun, doc,
+					&cmdlist, suppress_notification);
 }
 
 /* Recursively walk the commandlist structures, and print out the
diff --git a/gdb/command.h b/gdb/command.h
index 3dde2475cb1..e3d55c2dcba 100644
--- a/gdb/command.h
+++ b/gdb/command.h
@@ -148,6 +148,12 @@  extern struct cmd_list_element *add_cmd (const char *, enum command_class,
 					 const char *,
 					 struct cmd_list_element **);
 
+extern struct cmd_list_element *add_cmd_suppress_notification
+			(const char *name, enum command_class theclass,
+			 cmd_const_cfunc_ftype *fun, const char *doc,
+			 struct cmd_list_element **list,
+			 int *suppress_notification);
+
 extern struct cmd_list_element *add_alias_cmd (const char *, const char *,
 					       enum command_class, int,
 					       struct cmd_list_element **);
@@ -165,6 +171,14 @@  extern struct cmd_list_element *add_prefix_cmd (const char *, enum command_class
 						const char *, int,
 						struct cmd_list_element **);
 
+extern struct cmd_list_element *add_prefix_cmd_suppress_notification
+			(const char *name, enum command_class theclass,
+			 cmd_const_cfunc_ftype *fun,
+			 const char *doc, struct cmd_list_element **prefixlist,
+			 const char *prefixname, int allow_unknown,
+			 struct cmd_list_element **list,
+			 int *suppress_notification);
+
 extern struct cmd_list_element *add_abbrev_prefix_cmd (const char *,
 						       enum command_class,
 						       cmd_const_cfunc_ftype *fun,
diff --git a/gdb/doc/gdb.texinfo b/gdb/doc/gdb.texinfo
index 28f083f96e3..61471d32e51 100644
--- a/gdb/doc/gdb.texinfo
+++ b/gdb/doc/gdb.texinfo
@@ -7573,21 +7573,58 @@ 
 @table @code
 @kindex frame@r{, selecting}
 @kindex f @r{(@code{frame})}
-@item frame @var{n}
-@itemx f @var{n}
-Select frame number @var{n}.  Recall that frame zero is the innermost
-(currently executing) frame, frame one is the frame that called the
-innermost one, and so on.  The highest-numbered frame is the one for
-@code{main}.
+@item frame @r{[} @var{frame-selection-spec} @r{]}
+@item f @r{[} @var{frame-selection-spec} @r{]}
+The @command{frame} command allows different stack frames to be
+selected.  The @var{frame-selection-spec} can be any of the following:
+
+@table @code
+@kindex frame level
+@item @var{num}
+@item level @var{num}
+Select frame number @var{num}.  Recall that frame zero is the
+innermost (currently executing) frame, frame one is the frame that
+called the innermost one, and so on.  The highest-numbered frame is
+the one for @code{main}.
+
+As this is the most common method of navigating the frame stack then
+the string @command{level} can be dropped, the following two commands
+are equivalent:
+
+@smallexample
+(@value{GDBP}) frame 3
+(@value{GDBP}) frame level 3
+@end smallexample
 
-@item frame @var{stack-addr} [ @var{pc-addr} ]
-@itemx f @var{stack-addr} [ @var{pc-addr} ]
-Select the frame at address @var{stack-addr}.  This is useful mainly if the
-chaining of stack frames has been damaged by a bug, making it
-impossible for @value{GDBN} to assign numbers properly to all frames.  In
-addition, this can be useful when your program has multiple stacks and
-switches between them.  The optional @var{pc-addr} can also be given to
-specify the value of PC for the stack frame.
+@kindex frame address
+@item address @var{stack-address}
+Select the frame with stack address @var{stack-address}.  Recall that
+each stack frame has a stack frame address, which roughly corresponds
+to the location on the stack where the stack frame local variables are
+stored.
+
+@kindex frame function
+@item function @var{function-name}
+Select the stack frame for function @var{function-name}.  If there are
+multiple stack frames for function @var{function-name} then the inner
+most stack frame is selected.
+
+@kindex frame create
+@item create @var{stack-address} @r{[} @var{pc-addr} @r{]}
+Create and then select a new frame at stack address @var{stack-addr}.
+This is useful mainly if the chaining of stack frames has been damaged
+by a bug, making it impossible for @value{GDBN} to assign numbers
+properly to all frames.  In addition, this can be useful when your
+program has multiple stacks and switches between them.  The optional
+@var{pc-addr} can also be given to specify the value of PC for the
+stack frame.
+
+When a frame has been created and selected uing @command{frame create}
+then you can always return to the original stack using one of the
+previous stack frame selection instructions, for example
+@command{frame level 0}.
+
+@end table
 
 @kindex up
 @item up @var{n}
@@ -7630,11 +7667,13 @@ 
 
 @table @code
 @kindex select-frame
-@item select-frame
+@item select-frame @r{[} @var{frame-selection-spec} @r{]}
 The @code{select-frame} command is a variant of @code{frame} that does
 not display the new frame after selecting it.  This command is
 intended primarily for use in @value{GDBN} command scripts, where the
-output might be unnecessary and distracting.
+output might be unnecessary and distracting.  The
+@var{frame-selection-spec} is as for the @command{frame} command
+described in @ref{Selection, ,Selecting a Frame}.
 
 @kindex down-silently
 @kindex up-silently
@@ -7692,13 +7731,12 @@ 
 something has gone wrong that has made the stack format fail to fit
 the usual conventions.
 
-@item info frame @var{addr}
-@itemx info f @var{addr}
-Print a verbose description of the frame at address @var{addr}, without
-selecting that frame.  The selected frame remains unchanged by this
-command.  This requires the same kind of address (more than one for some
-architectures) that you specify in the @code{frame} command.
-@xref{Selection, ,Selecting a Frame}.
+@item info frame @r{[} @var{frame-selection-spec} @r{]}
+@itemx info f @r{[} @var{frame-selection-spec} @r{]}
+Print a verbose description of the frame selected by
+@var{frame-selection-spec}.  The @var{frame-selection-spec} is the
+same as for the @command{frame} command (@pxref{Selection, ,Selecting
+a Frame}).  The selected frame remains unchanged by this command.
 
 @kindex info args
 @item info args
@@ -29996,11 +30034,12 @@ 
 @subsubheading Synopsis
 
 @smallexample
- -stack-select-frame @var{framenum}
+ -stack-select-frame @var{frame-selection-spec}
 @end smallexample
 
-Change the selected frame.  Select a different frame @var{framenum} on
-the stack.
+Change the selected frame, the @var{frame-selection-spec} is as for
+the @command{frame} command described in @ref{Selection, ,Selecting a
+Frame}.
 
 This command in deprecated in favor of passing the @samp{--frame}
 option to every command.
diff --git a/gdb/mi/mi-cmd-stack.c b/gdb/mi/mi-cmd-stack.c
index 443d1ed4183..c79d7a95e59 100644
--- a/gdb/mi/mi-cmd-stack.c
+++ b/gdb/mi/mi-cmd-stack.c
@@ -680,10 +680,8 @@  list_args_or_locals (enum what_to_list what, enum print_values values,
 void
 mi_cmd_stack_select_frame (const char *command, char **argv, int argc)
 {
-  if (argc == 0 || argc > 1)
+  if (!select_frame_from_spec (argv, argc, 0))
     error (_("-stack-select-frame: Usage: FRAME_SPEC"));
-
-  select_frame_command (argv[0], 1 /* not used */ );
 }
 
 void
diff --git a/gdb/stack.c b/gdb/stack.c
index 5e99a42053c..6f122395558 100644
--- a/gdb/stack.c
+++ b/gdb/stack.c
@@ -102,6 +102,9 @@  static void set_last_displayed_sal (int valid,
 				    struct symtab *symtab,
 				    int line);
 
+static struct frame_info *find_frame_for_function (const char *);
+static struct frame_info *find_frame_for_address (CORE_ADDR);
+
 /* Zero means do things normally; we are interacting directly with the
    user.  One means print the full filename and linenumber when a
    frame is printed, and do so in a format emacs18/emacs19.22 can
@@ -1255,134 +1258,34 @@  print_frame (struct frame_info *frame, int print_level,
 }
 
 
-/* Read a frame specification in whatever the appropriate format is from
-   FRAME_EXP.  Call error() if the specification is in any way invalid (so
-   this function never returns NULL).  When SELECTED_FRAME_P is non-NULL
-   set its target to indicate that the default selected frame was used.  */
+/* Completion function for "frame function", "info frame function", and
+   "select-frame function" commands.  */
 
-static struct frame_info *
-parse_frame_specification (const char *frame_exp, int *selected_frame_p)
+void
+frame_selection_by_function_completer (struct cmd_list_element *ignore,
+				       completion_tracker &tracker,
+				       const char *text, const char *word)
 {
-  int numargs;
-  struct value *args[4];
-  CORE_ADDR addrs[ARRAY_SIZE (args)];
-
-  if (frame_exp == NULL)
-    numargs = 0;
-  else
-    {
-      numargs = 0;
-      while (1)
-	{
-	  const char *p;
-
-	  /* Skip leading white space, bail of EOL.  */
-	  frame_exp = skip_spaces (frame_exp);
-	  if (!*frame_exp)
-	    break;
-
-	  /* Parse the argument, extract it, save it.  */
-	  for (p = frame_exp;
-	       *p && !ISSPACE (*p);
-	       p++);
-	  std::string addr_string (frame_exp, p - frame_exp);
-	  frame_exp = p;
-	  
-	  /* NOTE: Parse and evaluate expression, but do not use
-	     functions such as parse_and_eval_long or
-	     parse_and_eval_address to also extract the value.
-	     Instead value_as_long and value_as_address are used.
-	     This avoids problems with expressions that contain
-	     side-effects.  */
-	  if (numargs >= ARRAY_SIZE (args))
-	    error (_("Too many args in frame specification"));
-	  args[numargs++] = parse_and_eval (addr_string.c_str ());
-	}
-    }
-
-  /* If no args, default to the selected frame.  */
-  if (numargs == 0)
-    {
-      if (selected_frame_p != NULL)
-	(*selected_frame_p) = 1;
-      return get_selected_frame (_("No stack."));
-    }
-
-  /* None of the remaining use the selected frame.  */
-  if (selected_frame_p != NULL)
-    (*selected_frame_p) = 0;
-
-  /* Assume the single arg[0] is an integer, and try using that to
-     select a frame relative to current.  */
-  if (numargs == 1)
-    {
-      struct frame_info *fid;
-      int level = value_as_long (args[0]);
-
-      fid = find_relative_frame (get_current_frame (), &level);
-      if (level == 0)
-	/* find_relative_frame was successful.  */
-	return fid;
-    }
-
-  /* Convert each value into a corresponding address.  */
-  {
-    int i;
-
-    for (i = 0; i < numargs; i++)
-      addrs[i] = value_as_address (args[i]);
-  }
-
-  /* Assume that the single arg[0] is an address, use that to identify
-     a frame with a matching ID.  Should this also accept stack/pc or
-     stack/pc/special.  */
-  if (numargs == 1)
-    {
-      struct frame_id id = frame_id_build_wild (addrs[0]);
-      struct frame_info *fid;
-
-      /* If (s)he specifies the frame with an address, he deserves
-	 what (s)he gets.  Still, give the highest one that matches.
-	 (NOTE: cagney/2004-10-29: Why highest, or outer-most, I don't
-	 know).  */
-      for (fid = get_current_frame ();
-	   fid != NULL;
-	   fid = get_prev_frame (fid))
-	{
-	  if (frame_id_eq (id, get_frame_id (fid)))
-	    {
-	      struct frame_info *prev_frame;
-
-	      while (1)
-		{
-		  prev_frame = get_prev_frame (fid);
-		  if (!prev_frame
-		      || !frame_id_eq (id, get_frame_id (prev_frame)))
-		    break;
-		  fid = prev_frame;
-		}
-	      return fid;
-	    }
-	}
-      }
-
-  /* We couldn't identify the frame as an existing frame, but
-     perhaps we can create one with a single argument.  */
-  if (numargs == 1)
-    return create_new_frame (addrs[0], 0);
-  else if (numargs == 2)
-    return create_new_frame (addrs[0], addrs[1]);
-  else
-    error (_("Too many args in frame specification"));
+  /* This is used to complete function names within a stack.  It would be
+     nice if instead of offering all available function names, we only
+     offered functions that were actually in the stack.  However, this
+     would probably mean unwinding the stack to completion, which could
+     take too long (or on a corrupted stack, possibly not end).  For now I
+     offer all symbol names as a safer choice.  */
+  collect_symbol_completion_matches (tracker,
+				     complete_symbol_mode::EXPRESSION,
+				     symbol_name_match_type::EXPRESSION,
+				     text, word);
 }
 
-/* Print verbosely the selected frame or the frame at address
-   ADDR_EXP.  Absolutely all information in the frame is printed.  */
+/* Core of all the "info frame" sub-commands.  Print verbosely the frame FI
+   if SELECTED_FRAME_P is true then frame FI is the current frame, which
+   was selected as a default due to the user not providing any arguments
+   to select some other frame.  */
 
 static void
-info_frame_command (const char *addr_exp, int from_tty)
+info_frame_command_core (struct frame_info *fi, bool selected_frame_p)
 {
-  struct frame_info *fi;
   struct symbol *func;
   struct symtab *s;
   struct frame_info *calling_frame_info;
@@ -1390,7 +1293,6 @@  info_frame_command (const char *addr_exp, int from_tty)
   const char *funname = 0;
   enum language funlang = language_unknown;
   const char *pc_regname;
-  int selected_frame_p;
   struct gdbarch *gdbarch;
   CORE_ADDR frame_pc;
   int frame_pc_p;
@@ -1398,7 +1300,6 @@  info_frame_command (const char *addr_exp, int from_tty)
   CORE_ADDR caller_pc = 0;
   int caller_pc_p = 0;
 
-  fi = parse_frame_specification (addr_exp, &selected_frame_p);
   gdbarch = get_frame_arch (fi);
 
   /* Name of the value returned by get_frame_pc().  Per comments, "pc"
@@ -1687,6 +1588,140 @@  info_frame_command (const char *addr_exp, int from_tty)
   }
 }
 
+/* The core of all the "select-frame" sub-commands.  Just wraps a call to
+   SELECT_FRAME.  */
+
+static void
+select_frame_command_core (struct frame_info *fi, bool ignored)
+{
+  struct frame_info *prev_frame = get_selected_frame_if_set ();
+  select_frame (fi);
+  if (get_selected_frame_if_set () != prev_frame)
+    gdb::observers::user_selected_context_changed.notify (USER_SELECTED_FRAME);
+}
+
+/* The core of all the "frame" sub-commands.  Select frame FI, and if this
+   means we change frame send out a change notification (otherwise, just
+   reprint the current frame summary).   */
+
+static void
+frame_command_core (struct frame_info *fi, bool ignored)
+{
+  struct frame_info *prev_frame = get_selected_frame_if_set ();
+
+  select_frame (fi);
+  if (get_selected_frame_if_set () != prev_frame)
+    gdb::observers::user_selected_context_changed.notify (USER_SELECTED_FRAME);
+  else
+    print_selected_thread_frame (current_uiout, USER_SELECTED_FRAME);
+}
+
+/* The three commands 'frame', 'select-frame', and 'info frame' all have a
+   common set of sub-commands that allow a specific frame to be selected.
+   All of the sub-command functions are static methods within this class
+   template which is then instantiated below.  */
+
+template <void (*FPTR) (struct frame_info *, bool)>
+class frame_command_helper
+{
+public:
+
+  /* The "frame level" family of commands.  The ARG is a number that is
+     the frame's level in the stack.  */
+  static void
+  level (const char *arg, int from_tty)
+  {
+    int level = value_as_long (parse_and_eval (arg));
+    struct frame_info *fid
+      = find_relative_frame (get_current_frame (), &level);
+    if (level != 0)
+      error (_("No frame at level %s."), arg);
+    FPTR (fid, false);
+  }
+
+  /* The "frame address" family of commands.  ARG is a stack-pointer
+     address for an existing frame.  This command does not allow new
+     frames to be created.  */
+
+  static void
+  address (const char *arg, int from_tty)
+  {
+    CORE_ADDR addr = value_as_address (parse_and_eval (arg));
+    struct frame_info *fid = find_frame_for_address (addr);
+    if (fid == NULL)
+      error (_("No frame at address %s."), arg);
+    FPTR (fid, false);
+  }
+
+  /* The "frame create" family of commands.  ARG is one or two addresses
+     and is used to create a frame.  The addresses are stack-pointer
+     address, and (optional) pc-address.  */
+
+  static void
+  create (const char *args, int from_tty)
+  {
+    struct frame_info *fid;
+
+    if (args == NULL)
+    error (_("Missing address argument to create a frame"));
+
+    gdb_argv argv (args);
+
+    if (argv.count () == 2)
+      {
+	CORE_ADDR addr[2];
+
+	addr [0] = value_as_address (parse_and_eval (argv[0]));
+	addr [1] = value_as_address (parse_and_eval (argv[1]));
+	fid = create_new_frame (addr[0], addr[1]);
+      }
+    else
+      {
+	CORE_ADDR addr = value_as_address (parse_and_eval (argv[0]));
+	fid = create_new_frame (addr, false);
+      }
+    FPTR (fid, false);
+  }
+
+  /* The "frame function" family of commands.  ARG is the name of a
+     function within the stack, the first function (searching from frame
+     0) with that name will be selected.  */
+
+  static void
+  function (const char *arg, int from_tty)
+  {
+    if (arg == NULL)
+      error (_("Missing function name argument"));
+    struct frame_info *fid = find_frame_for_function (arg);
+    if (fid == NULL)
+      error (_("No frame for function \"%s\"."), arg);
+    FPTR (fid, false);
+  }
+
+  /* The "frame" base command, that is, when no sub-command is specified.
+     If one argument is provided then we assume that this is a frame level
+     number as historically, this was the supported command syntax that was
+     used most often.
+
+     If no argument is provided, then the current frame is selected.  */
+
+  static void
+  base_command (const char *arg, int from_tty)
+  {
+    if (arg == NULL)
+      FPTR (get_selected_frame (_("No stack.")), true);
+    else
+      level (arg, from_tty);
+  }
+};
+
+/* Create three FRAME_COMMAND_HELPER instances to implement the
+   sub-commands for 'info frame', 'frame', and 'select-frame' commands.  */
+
+static frame_command_helper <info_frame_command_core> info_frame_cmd;
+static frame_command_helper <frame_command_core> frame_cmd;
+static frame_command_helper <select_frame_command_core> select_frame_cmd;
+
 /* Print briefly all stack frames or just the innermost COUNT_EXP
    frames.  */
 
@@ -2212,37 +2247,43 @@  find_relative_frame (struct frame_info *frame, int *level_offset_ptr)
   return frame;
 }
 
-/* The "select_frame" command.  With no argument this is a NOP.
-   Select the frame at level LEVEL_EXP if it is a valid level.
-   Otherwise, treat LEVEL_EXP as an address expression and select it.
-
-   See parse_frame_specification for more info on proper frame
-   expressions.  */
+/* Select a frame using ARGV.  This is used from the MI -stack-select-frame
+   to provide the same frame specification mechanism that the CLI has for
+   commands like 'frame'.  The return value is true if the contents of
+   ARGV looked like a sensible attempt to change the frame (an error might
+   still be thrown though), or false if the contents of ARGV are not a
+   correct frame specification.  */
 
-void
-select_frame_command (const char *level_exp, int from_tty)
+bool
+select_frame_from_spec (char **argv, int argc, int from_tty)
 {
-  struct frame_info *prev_frame = get_selected_frame_if_set ();
-
-  select_frame (parse_frame_specification (level_exp, NULL));
-  if (get_selected_frame_if_set () != prev_frame)
-    gdb::observers::user_selected_context_changed.notify (USER_SELECTED_FRAME);
-}
-
-/* The "frame" command.  With no argument, print the selected frame
-   briefly.  With an argument, behave like select_frame and then print
-   the selected frame.  */
+  if (argc == 1)
+    select_frame_cmd.level (argv[0], from_tty);
+  else if (argc == 2
+	   && strncasecmp ("level", argv[0], strlen (argv[0])) == 0)
+    select_frame_cmd.level (argv[1], from_tty);
+  else if (argc == 2
+	   && strncasecmp ("address", argv[0], strlen (argv[0])) == 0)
+    select_frame_cmd.address (argv[1], from_tty);
+  else if (argc == 2
+	   && strncasecmp ("function", argv[0], strlen (argv[0])) == 0)
+    select_frame_cmd.function (argv[1], from_tty);
+  else if ((argc == 2 || argc == 3)
+	   && strncasecmp ("create", argv[0], strlen (argv[0])) == 0)
+    {
+      std::string arg;
 
-static void
-frame_command (const char *level_exp, int from_tty)
-{
-  struct frame_info *prev_frame = get_selected_frame_if_set ();
+      if (argc == 2)
+	arg = string_printf ("%s", argv[1]);
+      else
+	arg = string_printf ("%s %s", argv[1], argv[2]);
 
-  select_frame (parse_frame_specification (level_exp, NULL));
-  if (get_selected_frame_if_set () != prev_frame)
-    gdb::observers::user_selected_context_changed.notify (USER_SELECTED_FRAME);
+      select_frame_cmd.create (arg.c_str (), from_tty);
+    }
   else
-    print_selected_thread_frame (current_uiout, USER_SELECTED_FRAME);
+    return false;
+
+  return true;
 }
 
 /* Select the frame up one or COUNT_EXP stack levels from the
@@ -2516,9 +2557,59 @@  func_command (const char *arg, int from_tty)
     }
 }
 
+/* Find inner-mode frame with frame address ADDRESS.  Return NULL if no
+   matching frame can be found.  */
+
+static struct frame_info *
+find_frame_for_address (CORE_ADDR address)
+{
+  struct frame_id id;
+  struct frame_info *fid;
+
+  id = frame_id_build_wild (address);
+
+  /* If (s)he specifies the frame with an address, he deserves
+     what (s)he gets.  Still, give the highest one that matches.
+     (NOTE: cagney/2004-10-29: Why highest, or outer-most, I don't
+     know).  */
+  for (fid = get_current_frame ();
+       fid != NULL;
+       fid = get_prev_frame (fid))
+    {
+      if (frame_id_eq (id, get_frame_id (fid)))
+	{
+	  struct frame_info *prev_frame;
+
+	  while (1)
+	    {
+	      prev_frame = get_prev_frame (fid);
+	      if (!prev_frame
+		  || !frame_id_eq (id, get_frame_id (prev_frame)))
+		break;
+	      fid = prev_frame;
+	    }
+	  return fid;
+	}
+    }
+  return NULL;
+}
+
+
+
+/* Commands with a prefix of `frame'.  */
+static struct cmd_list_element *frame_cmd_list = NULL;
+
+/* Commands with a prefix of `frame'.  */
+static struct cmd_list_element *select_frame_cmd_list = NULL;
+
+/* Commands with a prefix of `info frame'.  */
+static struct cmd_list_element *info_frame_cmd_list = NULL;
+
 void
 _initialize_stack (void)
 {
+  struct cmd_list_element *cmd;
+
   add_com ("return", class_stack, return_command, _("\
 Make selected stack frame return to its caller.\n\
 Control remains in the debugger, but when you continue\n\
@@ -2541,20 +2632,89 @@  An argument says how many frames down to go."));
 Same as the `down' command, but does not print anything.\n\
 This is useful in command scripts."));
 
-  add_com ("frame", class_stack, frame_command, _("\
-Select and print a stack frame.\nWith no argument, \
-print the selected stack frame.  (See also \"info frame\").\n\
-An argument specifies the frame to select.\n\
-It can be a stack frame number or the address of the frame."));
-
+  add_prefix_cmd ("frame", class_stack,
+		  &frame_cmd.base_command, _("\
+Select and print a stack frame.\n\
+With no argument, print the selected stack frame.  (See also \"info frame\").\n\
+A single numerical argument specifies the frame to select."),
+		  &frame_cmd_list, "frame ", 1, &cmdlist);
   add_com_alias ("f", "frame", class_stack, 1);
 
-  add_com_suppress_notification ("select-frame", class_stack, select_frame_command, _("\
+  add_cmd ("address", class_stack, &frame_cmd.address,
+	   _("\
+Select and print a stack frame by stack address\n\
+\n\
+Usage: frame address STACK-ADDRESS"),
+	   &frame_cmd_list);
+
+  add_cmd ("create", class_stack, &frame_cmd.create,
+	   _("\
+Create, select, and print a new stack frame.\n\
+\n\
+Usage: frame create STACK-ADDRESS\n\
+       frame create STACK-ADDRESS PC-ADDRESS"),
+	   &frame_cmd_list);
+
+  cmd = add_cmd ("function", class_stack, &frame_cmd.function,
+	   _("\
+Select and print a stack frame by function name.\n\
+\n\
+Usage: frame function NAME\n\
+\n\
+The innermost frame that visited function NAME is selected."),
+	   &frame_cmd_list);
+  set_cmd_completer (cmd, frame_selection_by_function_completer);
+
+
+  add_cmd ("level", class_stack, &frame_cmd.level,
+	   _("\
+Select and print a stack frame by level.\n\
+\n\
+Usage: frame level NUMBER"),
+	   &frame_cmd_list);
+
+  cmd = add_prefix_cmd_suppress_notification ("select-frame", class_stack,
+		      &select_frame_cmd.base_command, _("\
 Select a stack frame without printing anything.\n\
-An argument specifies the frame to select.\n\
-It can be a stack frame number or the address of the frame."),
+A single numerical argument specifies the frame to select."),
+		      &select_frame_cmd_list, "select-frame ", 1, &cmdlist,
+		      &cli_suppress_notification.user_selected_context);
+
+  add_cmd_suppress_notification ("address", class_stack,
+			 &select_frame_cmd.address, _("\
+Select a stack frame by stack address.\n\
+\n\
+Usage: frame address STACK-ADDRESS"),
+			 &select_frame_cmd_list,
+			 &cli_suppress_notification.user_selected_context);
+
+
+  add_cmd_suppress_notification ("create", class_stack,
+		 &select_frame_cmd.create, _("\
+Create and select a new stack frame.\n\
+\n\
+Usage: frame create STACK-ADDRESS\n\
+       frame create STACK-ADDRESS PC-ADDRESS"),
+		 &select_frame_cmd_list,
 		 &cli_suppress_notification.user_selected_context);
 
+  cmd = add_cmd_suppress_notification ("function", class_stack,
+	       &select_frame_cmd.function, _("\
+Select a stack frame by function name.\n\
+\n\
+Usage: frame function NAME"),
+	       &select_frame_cmd_list,
+	       &cli_suppress_notification.user_selected_context);
+  set_cmd_completer (cmd, frame_selection_by_function_completer);
+
+  add_cmd_suppress_notification ("level", class_stack,
+			 &select_frame_cmd.level, _("\
+Select a stack frame by level.\n\
+\n\
+Usage: frame level NUMBER"),
+			 &select_frame_cmd_list,
+			 &cli_suppress_notification.user_selected_context);
+
   add_com ("backtrace", class_stack, backtrace_command, _("\
 Print backtrace of all stack frames, or innermost COUNT frames.\n\
 Usage: backtrace [QUALIFIERS]... [COUNT]\n\
@@ -2568,9 +2728,45 @@  on this backtrace."));
   add_info ("stack", backtrace_command,
 	    _("Backtrace of the stack, or innermost COUNT frames."));
   add_info_alias ("s", "stack", 1);
-  add_info ("frame", info_frame_command,
-	    _("All about selected stack frame, or frame at ADDR."));
-  add_info_alias ("f", "frame", 1);
+
+  add_prefix_cmd ("frame", class_info, &info_frame_cmd.base_command,
+		  _("All about the selected stack frame.\n\
+With no arguments, displays information about the currently selected stack\n\
+frame.  Alternatively a frame specification may be provided (See \"frame\")\n\
+the information is then printed about the specified frame."),
+		  &info_frame_cmd_list, "info frame ", 1, &infolist);
+  add_com_alias ("f", "frame", class_stack, 1);
+
+  add_cmd ("address", class_stack, &info_frame_cmd.address,
+	   _("\
+Print information about a stack frame selected by stack address.\n\
+\n\
+Usage: frame address STACK-ADDRESS"),
+	   &info_frame_cmd_list);
+
+  add_cmd ("create", class_stack, &info_frame_cmd.create,
+	   _("\
+Create a new stack frame, and print information about it.\n\
+\n\
+Usage: frame create STACK-ADDRESS\n\
+       frame create STACK-ADDRESS PC-ADDRESS"),
+	   &info_frame_cmd_list);
+
+  cmd = add_cmd ("function", class_stack, &info_frame_cmd.function,
+	   _("\
+Print information about a stack frame selected by function name.\n\
+\n\
+Usage: frame function NAME"),
+	   &info_frame_cmd_list);
+  set_cmd_completer (cmd, frame_selection_by_function_completer);
+
+  add_cmd ("level", class_stack, &info_frame_cmd.level,
+	   _("\
+Print information about a stack frame selected by level.\n\
+\n\
+Usage: frame level NUMBER"),
+	   &info_frame_cmd_list);
+
   add_info ("locals", info_locals_command,
 	    _("Local variables of current stack frame."));
   add_info ("args", info_args_command,
diff --git a/gdb/stack.h b/gdb/stack.h
index ca190efa9c3..29794872d70 100644
--- a/gdb/stack.h
+++ b/gdb/stack.h
@@ -20,7 +20,7 @@ 
 #ifndef STACK_H
 #define STACK_H
 
-void select_frame_command (const char *level_exp, int from_tty);
+bool select_frame_from_spec (char **argv, int argc, int from_tty);
 
 gdb::unique_xmalloc_ptr<char> find_frame_funname (struct frame_info *frame,
 						  enum language *funlang,
diff --git a/gdb/testsuite/gdb.base/frame-selection.c b/gdb/testsuite/gdb.base/frame-selection.c
new file mode 100644
index 00000000000..f3d81f223e0
--- /dev/null
+++ b/gdb/testsuite/gdb.base/frame-selection.c
@@ -0,0 +1,52 @@ 
+/* Copyright 2018 Free Software Foundation, Inc.
+
+   This file is part of GDB.
+
+   This program is free software; you can redistribute it and/or modify
+   it under the terms of the GNU General Public License as published by
+   the Free Software Foundation; either version 3 of the License, or
+   (at your option) any later version.
+
+   This program is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+   GNU General Public License for more details.
+
+   You should have received a copy of the GNU General Public License
+   along with this program.  If not, see <http://www.gnu.org/licenses/>.  */
+
+int
+frame_2 (void)
+{
+  return 0;
+}
+
+int
+frame_1 (void)
+{
+  return frame_2 ();
+}
+
+int
+recursive (int arg)
+{
+  int v;
+
+  if (arg < 2)
+    v = recursive (arg + 1);
+  else
+    v = frame_2 ();
+
+  return v;
+}
+
+int
+main (void)
+{
+  int i, j;
+
+  i = frame_1 ();
+  j = recursive (0);
+
+  return i + j;
+}
diff --git a/gdb/testsuite/gdb.base/frame-selection.exp b/gdb/testsuite/gdb.base/frame-selection.exp
new file mode 100644
index 00000000000..6df85494839
--- /dev/null
+++ b/gdb/testsuite/gdb.base/frame-selection.exp
@@ -0,0 +1,156 @@ 
+# Copyright 2018 Free Software Foundation, Inc.
+
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 3 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+if { [prepare_for_testing frame-cmds.exp "frame-cmds" {frame-cmds.c} {debug nowarnings}] } {
+    return -1
+}
+set srcfile frame-cmds.c
+
+runto_main
+gdb_breakpoint frame_2
+gdb_continue_to_breakpoint frame_2
+
+gdb_test "bt" "#0  frame_2.*#1  $hex in frame_1.*#2  $hex in main.*" "backtrace at breakpoint"
+
+# Perform "info frame" to extract the frames address.
+proc get_frame_address { {testname ""} } {
+    global hex gdb_prompt
+
+    set testname "get_frame_address: ${testname}"
+    set frame_address "unknown"
+
+    send_gdb "info frame\n"
+    gdb_expect {
+	-re ", frame at ($hex):\r\n.*\r\n$gdb_prompt $" {
+	    set frame_address $expect_out(1,string)
+	    pass $testname
+	}
+	-re ".*$gdb_prompt $" { fail $testname }
+	timeout { fail "$testname (timeout)" }
+    }
+
+    return $frame_address
+}
+
+# Passes a list of addresses.  Return a new address that is not in the
+# list.
+proc get_new_address { {addresses {}} } {
+    return [format "%#x" [expr [lindex $addresses [llength addresses]] + 0x10 ]]
+}
+
+proc check_frame { level address function } {
+    global hex gdb_prompt
+
+    set testname "check frame level ${level}"
+    send_gdb "info frame\n"
+    gdb_expect {
+	-re "Stack level ${level}, frame at ($address):\r\n rip = $hex in ${function} \(\[^\r\n\]*\); saved rip = $hex\r\n.*\r\n$gdb_prompt $" {
+	    pass $testname
+	}
+	-re ".*$gdb_prompt $" { fail $testname }
+	timeout { fail "$testname (timeout)" }
+    }
+}
+
+# Select frame using level number, but relying on this being the
+# default action, so "frame 0" performs "frame level 0".
+gdb_test "frame 0" "#0  frame_2.*"
+set frame_0_address [ get_frame_address "frame 0" ]
+gdb_test "frame 1" "#1  $hex in frame_1.*"
+set frame_1_address [ get_frame_address "frame 1" ]
+gdb_test "frame 2" "#2  $hex in main.*"
+set frame_2_address [ get_frame_address "frame 2" ]
+gdb_test "frame 3" "No frame at level 3\."
+
+# Find an address that matches no frame
+set no_frame_address [ get_new_address [list $frame_0_address \
+					     $frame_1_address \
+					     $frame_2_address] ]
+
+# Select frame using 'level' specification.
+gdb_test "frame level 0" "#0  frame_2.*"
+gdb_test "frame level 1" "#1  $hex in frame_1.*"
+gdb_test "frame level 2" "#2  $hex in main.*"
+gdb_test "frame level 3" "No frame at level 3\."
+
+# Select frame by address.
+gdb_test "frame address ${frame_0_address}" "#0  frame_2.*"
+gdb_test "frame address ${frame_1_address}" "#1  $hex in frame_1.*"
+gdb_test "frame address ${frame_2_address}" "#2  $hex in main.*"
+gdb_test "frame address ${no_frame_address}" \
+    "No frame at address ${no_frame_address}\."
+
+# Select frame by function.
+gdb_test "frame function frame_2" "#0  frame_2.*"
+gdb_test "frame function frame_1" "#1  $hex in frame_1.*"
+gdb_test "frame function main" "#2  $hex in main.*"
+# Check for a distinction between a known function not in the stack
+# trace, and an unknown function.
+gdb_test "frame function recursive" "No frame for function \"recursive\"."
+gdb_test "frame function foo" "Function \"foo\" not defined."
+
+
+gdb_test_no_output "select-frame 0"
+check_frame "0" "${frame_0_address}" "frame_2"
+gdb_test_no_output "select-frame 1"
+check_frame "1" "${frame_1_address}" "frame_1"
+gdb_test_no_output "select-frame 2"
+check_frame "2" "${frame_2_address}" "main"
+gdb_test "select-frame 3" "No frame at level 3\."
+
+gdb_test_no_output "select-frame level 0"
+check_frame "0" "${frame_0_address}" "frame_2"
+gdb_test_no_output "select-frame level 1"
+check_frame "1" "${frame_1_address}" "frame_1"
+gdb_test_no_output "select-frame level 2"
+check_frame "2" "${frame_2_address}" "main"
+gdb_test "select-frame level 3" "No frame at level 3\."
+
+gdb_test_no_output "select-frame address ${frame_0_address}"
+check_frame "0" "${frame_0_address}" "frame_2"
+gdb_test_no_output "select-frame address ${frame_1_address}"
+check_frame "1" "${frame_1_address}" "frame_1"
+gdb_test_no_output "select-frame address ${frame_2_address}"
+check_frame "2" "${frame_2_address}" "main"
+gdb_test "select-frame address ${no_frame_address}" \
+    "No frame at address ${no_frame_address}\."
+
+gdb_test_no_output "select-frame function frame_2"
+check_frame "0" "${frame_0_address}" "frame_2"
+gdb_test_no_output "select-frame function frame_1"
+check_frame "1" "${frame_1_address}" "frame_1"
+gdb_test_no_output "select-frame function main"
+check_frame "2" "${frame_2_address}" "main"
+# Check for a distinction between a known function not in the stack
+# trace, and an unknown function.
+gdb_test "select-frame function recursive" \
+    "No frame for function \"recursive\"."
+gdb_test "select-frame function foo" \
+    "Function \"foo\" not defined."
+
+# Now continue until we hit the breakpoint again.
+gdb_continue_to_breakpoint frame_2
+gdb_test "bt" \
+    "#0  frame_2.*#1  $hex in recursive.*#2  $hex in recursive.*#3  $hex in recursive.*#4  $hex in main.*" \
+    "backtrace at breakpoint with recursive frames"
+
+# Check "frame function" when a function name occurs multiple times in
+# the stack.  The inner most (lowest level number) should always be
+# selected.
+gdb_test "frame function frame_2" "#0  frame_2.*"
+gdb_test "frame function recursive" "#1  $hex in recursive.*"
+gdb_test "frame function recursive" "#1  $hex in recursive.*"
+gdb_test "frame function main" "#4  $hex in main.*"
+gdb_test "frame function recursive" "#1  $hex in recursive.*"
diff --git a/gdb/testsuite/gdb.mi/mi-frame-selection.c b/gdb/testsuite/gdb.mi/mi-frame-selection.c
new file mode 100644
index 00000000000..70b992c9149
--- /dev/null
+++ b/gdb/testsuite/gdb.mi/mi-frame-selection.c
@@ -0,0 +1,34 @@ 
+/* Copyright 2018 Free Software Foundation, Inc.
+
+   This file is part of GDB.
+
+   This program is free software; you can redistribute it and/or modify
+   it under the terms of the GNU General Public License as published by
+   the Free Software Foundation; either version 3 of the License, or
+   (at your option) any later version.
+
+   This program is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+   GNU General Public License for more details.
+
+   You should have received a copy of the GNU General Public License
+   along with this program.  If not, see <http://www.gnu.org/licenses/>.  */
+
+int
+frame_2 (void)
+{
+  return 0;
+}
+
+int
+frame_1 (void)
+{
+  return frame_2 ();
+}
+
+int
+main (void)
+{
+  return frame_1 ();
+}
diff --git a/gdb/testsuite/gdb.mi/mi-frame-selection.exp b/gdb/testsuite/gdb.mi/mi-frame-selection.exp
new file mode 100644
index 00000000000..d51b0ae9315
--- /dev/null
+++ b/gdb/testsuite/gdb.mi/mi-frame-selection.exp
@@ -0,0 +1,90 @@ 
+# Copyright 2018 Free Software Foundation, Inc.
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 3 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+# This tests an issue relating to updating var objects in user created
+# frames.
+
+load_lib mi-support.exp
+set MIFLAGS "-i=mi"
+
+gdb_exit
+if [mi_gdb_start] {
+    continue
+}
+
+standard_testfile
+
+if { [gdb_compile "${srcdir}/${subdir}/${srcfile}" "${binfile}" executable {debug}] != "" } {
+    untested $testfile
+    return -1
+}
+
+mi_delete_breakpoints
+mi_gdb_reinitialize_dir $srcdir/$subdir
+mi_gdb_load ${binfile}
+
+mi_create_breakpoint "frame_2" \
+    "break-insert into frame_2" \
+    -number 1 -func frame_2 -file ".*${srcfile}"
+
+mi_run_cmd
+mi_expect_stop "breakpoint-hit" "frame_2" "" ".*${srcfile}" \
+	".*" { "" "disp=\"keep\"" } "run to breakpoint"
+
+mi_gdb_test "-stack-info-depth" \
+    "\\^done,depth=\"3\"" \
+    "stack info-depth"
+
+for {set i 0} {$i < 5} {incr i} {
+
+    if { $i < 3 } then {
+	mi_gdb_test "-stack-select-frame $i" \
+	    {\^done} \
+	    "-stack-select-frame $i"
+
+	mi_gdb_test "-stack-info-frame" \
+	    "\\^done,frame=\\\{level=\"$i\"\[^\}\]+\\\}" \
+	    "check level has has changed to $i"
+
+    } else {
+	mi_gdb_test "-stack-select-frame create $i" \
+	    {\^done} \
+	    "-stack-select-frame create frame at $i"
+
+	mi_gdb_test "-stack-info-frame" \
+	    "\\^done,frame=\\\{level=\"0\"\[^\}\]+\\\}" \
+	    "check level has has changed to 0, when creating a frame at sp=$i"
+    }
+}
+
+mi_gdb_test "-stack-select-frame 5" \
+    {\^error,msg="No frame at level 5."} \
+    "-stack-select-frame 5, with no keyword"
+
+mi_gdb_test "-stack-select-frame level 5" \
+    {\^error,msg="No frame at level 5."} \
+    "-stack-select-frame 5, using 'level' keyword"
+
+mi_gdb_test "-stack-select-frame function main" \
+    {\^done} \
+    "-stack-select-frame select frame for function main"
+
+mi_gdb_test "-stack-info-frame" \
+    "\\^done,frame=\\\{level=\"2\"\[^\}\]+\\\}" \
+    "check level has has changed to 2 for function main"
+
+mi_gdb_test "-stack-select-frame function unknown_function" \
+    {\^error,msg=\"Function \\\"unknown_function\\\" not defined.\"} \
+    "-stack-select-frame check error on undefined function name"