Add semihosting and SWO examples to mbed_override_console(). #10553
Conversation
|
@TeroJaasko, here's the header update you suggested. |
|
Added missing |
|
@korjaa, thank you for your changes. |
|
Add missing |
platform/mbed_retarget.h
Outdated
| * | ||
| * Example using SingleWireOutput: | ||
| * @code | ||
| * FileHandle* mbed::mbed_override_console(int) { |
There was a problem hiding this comment.
Our coding standard changed while I was writing this in the first place, and the code got updated, but not the examples. * binds to the right. Fix the existing one too.
There was a problem hiding this comment.
Stars are binding to the right now.
| * static LocalFileSystem fs("host"); | ||
| * if (fileno == STDIN_FILENO) { | ||
| * static FileHandle *in_terminal; | ||
| * static int in_open_result = fs.open(&in_terminal, ":tt", O_RDONLY); |
There was a problem hiding this comment.
That code is a bit RAM-inefficient, as it has two extra "once-only" control structures, and unused open results, but it's probably the simplest way of writing it.
The following would minimise RAM usage, but rather ends up obfuscating it.:
static LocalFileSystem fs("host");
struct TerminalStream {
TerminalStream(FileSystemHandle &fs, int flags) {
if (fs.open(&filehandle, ":tt", flags) != 0) {
filehandle = NULL;
}
}
FileHandle *filehandle;
}
if (fileno == STDIN_FILENO) {
static TerminalStream in_terminal(fs, O_RDONLY):
return in_terminal.filehandle;
} else {
static TerminalStream out_terminal(fs, O_WRONLY);
return out_terminal.filehandle;
}
You could leave it as-is; depends whether we want a simple example or to save 16 bytes of RAM when it's copy-pasted.
There was a problem hiding this comment.
Another possible refinement would be to check if the debugger is attached before attempting the open (LocalFileSystem doesn't), so it doesn't crash.
semihost_connected() should tell you that in principle, but it seems that it's currently conditional on a DEVICE_DEBUG_AWARENESS that I've never heard of and is set on basically no platforms. Not sure why - it's a core CPU feature? All M-class devices should have the bit it's trying to check.
There was a problem hiding this comment.
I think this one is so bluntly visible on the main.cpp, so maybe it would be OK to leave it as it is without the optimal memory consumption. And as you said it's a bit easier to understand.
There was a problem hiding this comment.
This semihost_connected() sounds like a nice enhancement idea. Maybe this could be checked in some future commit when the DEVICE_DEBUG_AWARENESS is more common.
|
We should have a section in our handbook about that as well. |
@bulislaw You mean this one ARMmbed/mbed-os-5-docs#1063? |
|
@korjaa @kjbracey-arm How shall we proceed with this one, has not yet been approved. |
|
I changed the stars and from my point of view this could be merged. @kjbracey-arm ? |
|
Ci started |
Test run: SUCCESSSummary: 11 of 11 test jobs passed |
Description
Add semihosting and SWO examples to mbed_override_console().
Pull request type
Reviewers
@kjbracey-arm
Release Notes