Add ignore command (from #1257)

[erivas]

Tried to follow implementation outline in #1257 , where this was labeled as good-first-issue/help-wanted.

[codecov-commenter]

Codecov Report

Merging #1276 (9bacc93) into dev (fbedf0b) will increase coverage by 0.00%.
The diff coverage is 61.11%.

@@           Coverage Diff           @@
##              dev    #1276   +/-   ##
=======================================
  Coverage   54.87%   54.87%           
=======================================
  Files         182      183    +1     
  Lines       20267    20284   +17     
  Branches     1878     1882    +4     
=======================================
+ Hits        11121    11131   +10     
- Misses       8727     8734    +7     
  Partials      419      419           

Impacted Files	Coverage Δ
pwndbg/commands/ignore.py	56.25% <56.25%> (ø)
pwndbg/commands/__init__.py	69.83% <100.00%> (+0.10%)	⬆️

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[gsingh93]

Thanks! I'll test this out myself, but overall the implementation looks good. Would you also be able to add a test for this command? I think we should test different valid and invalid variants of ignore, where there are no breakpoints, where a non-existent breakpoint number is entered, etc.

[disconnect3d]

How about:

1. Changing this code to:

if bpnum is None:
    bp = max(gdb.breakpoints(), key=lambda bp: bp.number, default=None)
else:
    bp = next((bp for bp in gdb.breakpoints() if bp.number == bpnum), None)

2. Changing the UX to match the original one a bit more:

pwndbg> help ignore
Set ignore-count of breakpoint number N to COUNT.
Usage is `ignore N COUNT'.

pwndbg> ignore 1 123
Will ignore next 123 crossings of breakpoint 1.

pwndbg> ignore 123 123
No breakpoint number 123.

So:

• we can show an example in the description
• display error as 'No breakpoint number .' if it is defined or 'No breakpoints set' if it is not
• for success path, inform user that we did it via "Will ignore next 123 crossings of breakpoint 1."

[disconnect3d]

Fwiw for a multiline description with examples etc. you can use RawDescriptionHelpFormatter as we do in vmmap command: https://github.com/pwndbg/pwndbg/blob/dev/pwndbg/commands/vmmap.py#L50

Also I am wondering if we can grab last bp by just gdb.breakpoints()[-1] but idk if that is always the case so its safer to do it with max(..)

[disconnect3d]

Hm, so we could have:

bps = gdb.breakpoints()

if not bps:
    print(message.error("No breakpoints set."))

if bpnum is None:
    bp = max(bps, key=lambda bp: bp.number)
else:
    bp = next((bp for bp in bps if bp.number == bpnum), None)

    if bp is None:
        print(message.error("No breakpoint number %d." % bpnum))
        return

bp.ignore_count = count
print("Will ignore next %d crossings of breakpoint %d." % (count, bpnum))

[erivas]

Thanks! Applied the code changes.

I didn't add an example in the description, as an usage: ... line was already generated. It looks like this for me:

pwndbg> help ignore
usage: ignore [-h] [N] COUNT

Set ignore-count of breakpoint number N to COUNT.

While the ignore count is positive, execution will not stop on the breakpoint.

By default, if `N' is ommitted, the last breakpoint (i.e. greatest breakpoint number) will be used.

positional arguments:
  N           The breakpoint number N.
  COUNT       The number to set COUNT.

options:
  -h, --help  show this help message and exit

[erivas]

Thanks! I'll test this out myself, but overall the implementation looks good. Would you also be able to add a test for this command? I think we should test different valid and invalid variants of ignore, where there are no breakpoints, where a non-existent breakpoint number is entered, etc.

Thanks, added a few test cases in a file test_command_ignore.py, trying to follow how other commands are tested.

[gsingh93]

The tests look great! Any reason you're calling gdb.execute("file " + REFERENCE_BINARY) instead of start_binary(REFERENCE_BINARY) in some of the tests?

[disconnect3d]

The tests look great! Any reason you're calling gdb.execute("file " + REFERENCE_BINARY) instead of start_binary(REFERENCE_BINARY) in some of the tests?

Seems like an optimization - you don't always have to start the binary but can only load it to gdb