There were two chunks of Cortex-A8 code which called the
ARMv7-A CP15 operations; get rid of them, helping prepare
to remove those methods completely:
- post_debug_entry() can use the mrc() method to read
its two registers.
- write_memory() can use dpm->instr_write_data_r0() to
flush the ICache and DCache ... doing it this way is
actually faster since it reduces per-write overhead.
Note that the mrc() method parameters are re-ordered with
respect to the ARM instruction documentation, so that part
can be confusing.
Cleaned up the layout and comments in those areas a bit.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
There is no particular reason to invoke jtag_interface_quit()
on the atexit() handler, it just makes the code more obtuse
and stops other legitimate usage of atexit().
Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>
Ensures that the correct information gets displayed, depending on the
mode of the command being denied. Fixes misreporting all commands as
needing to run "before 'init'".
Clean up two aspects to this routine: bad naming, since it
doesn't restore the context, just the banked registers; and
excess indentation for the bulk of the code.
Also make some of its call sites stash the function's return
code; someday they should use it for error checking.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
This "loop over all registers" routine shared the same mess as
full_context() in terms of dozens of needless number_to_mode()
calls. Fix that, and comments, with related cleanup.
The misnamed xscale_restore_context() had a related bug. It
was restoring the *WRONG REGISTERS* ... always from whatever
the current mode was, instead of using the copy from whichever
register bank it was trying to restore. (But it marked the
intended register as having been restored...) Fixed that.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Finish removing references to the 'interp' global variable from the
command module, encapsulating all reference via command_context.
Eliminates use of the global entirely, so it can be removed. Hurrah!
Adds a log_capture_state structure to pass to the log capture
callback used by the command module. Ensures that the capture occurs
in the proper context.
The 'help' text will become more verbose, so its entire text will be
far more than desired when you only borked your syntax. The usage
still allows the commands to be looked up for more help.
command_done() does not need to return an error, but it needed
Doxygen comment. Provide some for copy_command_context as well.
Note: this audit revealed some potential bugs with the command context
implementation. There was a reason that commands were added at the
end of the list. Shallow copying of command_context means that
the list is shared between them. And commands added at the top-level
before the pre-existing commands will not be available in the shared
context as they were before. Yikes!
Fortunately, this does not seem to occur in general use, as
'add_help_text' gets registered in startup.tcl and claims the first slot
in my own test cases. Thus, it seems that we have been masking the issue
for now, but it shows the need for further architectural improvement in
the core command module.
With the ability to defer 'init', users can access the help system while
still in CONFIG mode. This patch omits commands from the help and usage
list when they cannot be run in the current command mode, making it much
easier to see what can be done at a given time.
Adds 'noinit' command to prevent OpenOCD from running 'init' at the end
up startup, allowing it to be given from telnet or TCL. This provides
the old behavior by default, and users can add this command to their
scripts to get the new behavior.
Moves the telnet and TCL server startup to server_init(), moving their
respective command registration in to server_register_commands().
Adds proper error checking for these particular startup processes.
Moves the core server startup to openocd_main(), improving related error
checking and preparing to defer 'init'.
Rework gdb_init to create flexible APIs (gdb_target_add_{one,all}) and
static helper (gdb_target_start) for starting GDB services. Eliminates
duplicated code and provides general mechanisms for adding GDB services.
The 'init' command is updated to call the new API, and later patches can
decouple its policy of adding all targets therein.
Provides the new capability to use both piped and TCP servers when
multiple targets are defined. The first target fills the pipe, and
others will be started on TCP ports (unless disabled, i.e. gdb_port=0).
Add missing COMMAND_REGISTRATION_DONE.
For now the command syntax for zy1000 needs to be compatible
across 0.3/0.4, the world outside OpenOCD interfaces to
zy1000 using the old syntax. Post 0.4 release(0.4.1 even)
I'll switch to subcommand scheme.
Switch to subcommands post 0.3 lifecycle.
Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>
Streamline the loop by continuing as soon as we know there's no
work to be done; this lets us un-indent almost everything.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
When fetching all the registers, XScale was doing various stupid
things like calling number_to_mode() a few dozen times instead of
just once, and mapping access to each register three times (again,
instead of just once). Stop that.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Use the new mapping interfaces in the debug entry path.
SPSR and the banked registers now have smaller and faster
accessors ... use them.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Code other than main() may invoke "init". When it does so,
customized handlers may need to run ... so make sure the
command context state is updated before they do so.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Those commands presume support for the "classic" set of CPU
modes (FIQ, supervisor, IRQ, etc) ... which aren't supported
by the ARMv7-M or ARMv6-M architectures. They also presume
a "struct arm" base type, which this code doesn't use.
We haven't cleaned up the register handling enough to be able
to share any of those "base" methods.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Eliminate the monolithic tcl_target_func by registering each of its
commands using the new chained command registration mechanism.
Also chains the target's commands under the CPU command, though these
may not work properly without some further modification.
The 'target' command group was implemented using its own command
dispatching, which can be eliminated by using the new chained command
registration mechanism. This patch splits the jim_target() function
into individual handlers, which makes them to be visible to the help and
usage commands. These one-trick handlers are much easier to understand.
Splits bulk of the jtag_tap_configure into jtag_tap_configure_event,
removing three or four levels of indentation in the process.
The resulting code was stylistically improved in other ways, but it
should be functionally identical.
Moves the tertiary jim handlers and required static helpers to the top
of tcl.c, defining them in a new registration array that is chained in
both the top-level context and under the jtag command. The top-level
commands can be removed at some point in the future to reduce clutter.
Explodes the 'jtag' into separate command handlers, which are easier
to understand and extend. Makes the code much easier to understand,
though further simplifications are possible. This patch tries to
minimize the noise when viewed with 'git diff -w'.
Gives these commands improved built-in help and usage information.
Two 'rm' commands were implemented and registered. This removes the
version that would have never been called prior to refactoring the
command registration.
Adds checks for memory allocation failures. Started to use calloc()
instead of malloc()/memset(), but I got carried away. This kind of work
should be done throughout the tree, but it's almost hopeless at present.
Splits the check for a command's ability to run into a helper.
This also fixes a bug whereby commands that specified COMMAND_EXEC
were allowed to run during the configuration stage. This allowed
problematic commands to be called before 'init', defeating the intention
of specifying that command mode. With this change, the run_command()
helper denies access to handlers that should run only after 'init'
during the configuration stage.
Presently, commands registration taks a static handler data pointer.
This patch adds support for commands that require a dynamic pointer,
such as those registered in a dynamic context (e.g. subcommands for a
user-created 'foo.cpu' command). The command_set_handler_data will
update a command (group) to use a new context pointer, while the
CMD_DATA macro allows command handlers to access the value.
Jim handlers should find this value in interp->cmdPrivData.
Updates command registration to provide top-level handlers for all
commands, rather than falling back onto the 'unknown' command. Instead,
that same handler is registered for placeholders, providing the same
functionality under the root verb command name instead. This permits
users to implement their own 'unknown' function, and it resolves some
mind-bending breakage related to function object lookup while recursing.
Changes 'ocd_bounce' to call 'ocd_command' and 'ocd_help' from the
wrapper directly, rather than bouncing through their wrappers. This
prevents endless recursion caused by the above changes, whereby the
'command' wrapper's type check would blow the stack to hell and gone.
Adds 'ocd_bouncer' in startup.tcl that is called as a helper for
all command handlers, shrinking the embedded C wrapper to a mere stub.
Jim handlers are called directly, simple handlers get called with the
wrapper to capture and discard their output on error, and placeholders
call help directly (though the unknown handler still does this too).
It attempts to improve the quality of the error messages as well.
Adds the 'command' group handler, with the 'type' command producing
a string that tells whether the given command is 'native' (for Jim-based
command handlers), 'simple' (for simple built-in commands), 'group'
for command group placeholders, and 'unknown' if not found in the
command registration tables (e.g. core built-ins functions).
There is no DEBUG() macro; don't call one! Always at
least *parse* debug code, to help prevent such errors.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
In target_type.h it's documented that the target must be
halted for add_breakpoint() ... and with slight ambiguity,
also for its add_watchpoint() sibling. So rather than
verifying that constraint in the CPU drivers, do it in the
target_add_{break,watch}point() routines.
Add minor paranoia on the remove_*point() paths too: save
the return value, and print it out in in the LOG_DEBUG message
in case it's nonzero.
Note that with some current cores, like all ARMv7 ones I've
looked at, there's no technical issue preventing watchpoint or
breakpoint add/remove operations on active cores. This model
seems deeply wired into OpenOCD though.
ALSO: the ARM targets were fairly "good" about enforcing that
constraint themselves. The MIPS ones were relied on other code
to catch such stuff, but it's not clear such code existed ...
keep an eye out for new issues on MIPS.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
David Brownell
Øyvind Harboe
Zachary T Welch
Marek Vasut
lementec fabien