Description
Is your feature request related to a problem? Please describe.
I'm always frustrated when an embedded device feel slow. Especially if it's my embedded device. Doubly so if I don't know where my device is spending my cycles.
Describe the solution you'd like
I'd like the ability to record a profile through OpenOCD and analyze it within VSCode
-
We'd need UI to set the target CPU core, then trigger
openocd profile ${profileTime} ${filename}. The UI should have the option to specifyprofileTime,filename, and select a CPU core(if applicable). Iffilenameis not specified, the plugin should choose a temporary file. -
Once the profile data is captured, the UI should load the profile data, look it up in the ELF file according to function names, then display a list of functions that were captured in the profile, along with a percentage of samples per function, sorted by which functions were sampled most often.
-
When a function is chosen for review, the UI should either open a new text window for the function, or take us to the function in the usual text area. The function's lines should be annotated according to the percentage or number of samples spent on each line.
Describe alternatives you've considered
Currently, I use a custom fork of gprof, shell scripts, and a text editor. It's the same solution I've been using for ~20 years, but it's frustrating to need to disconnect VSCode and switch to my old tools just to gather profiling data to switch back to VSCode to edit it.
Additional context
Example of Trace32's a per-function view
Example of ARM Realview's function's per-line view
