Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

HTM issues from Anton #34

Open
8 tasks
shenki opened this issue May 23, 2018 · 1 comment
Open
8 tasks

HTM issues from Anton #34

shenki opened this issue May 23, 2018 · 1 comment

Comments

@shenki
Copy link
Member

shenki commented May 23, 2018

  • pdbg core trace should check if the tracing core is in ST mode
  • Single issue mode is terribly slow. --start should setup HID0 and –stop should undo it
  • --dump vs --analyse
  • Mikey suggests a –record which works like perf
  • --analyze overwrites an existing file without prompting, but doesn't truncate what is there! A more sane sequence would be to refuse to overwrite unless an -f/–force option is specified, and then we truncate the existing file.
  • htm option seems to be missing help text
  • htm is hiding behind the mysterious --expert option
  • We hardwire the buffers to 1GB at the moment. An option –size argument would be useful, and ideally it would take standard units (–size 4G)
@shenki shenki changed the title HTM issues HTM issues from Anton May 23, 2018
@apopple
Copy link
Collaborator

apopple commented May 23, 2018

htm is hiding behind the mysterious --expert option

It didn't take @mikey long to convince me that --expert was at best ambiguous. It stems from my own ideas around debugging and pdbg were that it shouldn't ever change system state in "unexpected" ways.

So for example getgpr shouldn't stop instructions executing, but instead tell the user what needs to be done to stop instructions executing if they haven't already been. Reason is that generally you don't want your debugging tools screwing with your system in unexpected ways, but they should help a user discover what to do (whether or not we've succeeded with that is certainly debatable).

So the current thinking is we will make user-friendly versions of commands like getgpr and htm which will put the system in the right state if it isn't already and generally just "do what you want" (for certain values of "you" and "want") and a --safe mode which will only "do what you say".

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants