[all-commits] [llvm/llvm-project] 2a2170: In 92eaad2dd7adb5ee92f397cef85ab11f2612294e I made...

[jimingham via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 2a21700bc5be24532952c2705fbbe784f8ec3164
      https://github.com/llvm/llvm-project/commit/2a21700bc5be24532952c2705fbbe784f8ec3164
  Author: Jim Ingham <jingham at apple.com>
  Date:   2022-05-12 (Thu, 12 May 2022)

  Changed paths:
    M lldb/source/Commands/CommandObjectTarget.cpp
    M lldb/test/API/functionalities/gdb_remote_client/TestNoLocalFile.py

  Log Message:
  -----------
  In 92eaad2dd7adb5ee92f397cef85ab11f2612294e I made it possible to run
a debug session with only a remote path to the file you are debugging
using the SB API's. This patch makes it possible to do this using
target create --remote-file <some_path> without supplying a local file
as well.

Prior to this change we errored out saying that we haven't implemented
copying the binary back from the remote. I didn't implement the copy
back (in the case I'm interested in - iOS debugging - we don't
actually have a way for lldb to do that). This patch doesn't impede
doing that, I just didn't need it. I think for some object file
formats debugging w/o the binary file is hard because of what doesn't
get mapped in. I didn't try to arbitrate that, I'm assuming anybody
who has to do this knows what they are going to get.

If there's a connected platform that can check that the remote file
exists, it will do so, otherwise we trust the user's input - if it
isn't there the process launch is going to fail with no-such-file so
it will be pretty clear what went wrong.

Differential Revision: https://reviews.llvm.org/D124947