[dotest] Add the ability to set environment variables for the inferior.

[JDevlieghere]

Summary

This patch adds a dotest flag for setting environment variables for the inferior. This is different from the current --env flag, which sets variables in the debugger's environment. This allows us to set things like LD_LIBRARY_PATH for testing.

[amccarth]

lldb/packages/Python/lldbsuite/test/lldbtest.py

1861 ↗

(On Diff #206523)

env or settings set target.env-vars? On Windows, where spaces in paths are common, this works:

(lldb) settings set target.env-vars "MY_PATH=C:\Program Files\stuff"

but this doesn't:

(lldb) env "MY_PATH=C:\Program Files\stuff"

[JDevlieghere]

lldb/packages/Python/lldbsuite/test/lldbtest.py

1861 ↗

(On Diff #206523)

I've updated the diff, but this sounds more like a bug in the alias?

[amccarth]

The LGTM, but I wasn't nominated as a reviewer, and I was mostly looking at it from the point of Windows compatibility.

lldb/trunk/packages/Python/lldbsuite/test/dotest.py

def parseOptionsAndInitTestdirs():
     if args.set_env_vars:
         for env_var in args.set_env_vars:
             parts = env_var.split('=', 1)
             if len(parts) == 1:
                 os.environ[parts[0]] = ""
             else:
                 os.environ[parts[0]] = parts[1]
 
+    if args.set_inferior_env_vars:
+        lldbtest_config.inferior_env = ' '.join(args.set_inferior_env_vars)
+
     # only print the args if being verbose (and parsable is off)
     if args.v and not args.q:
         print(sys.argv)
 
     if args.h:
         do_help = True
 
     if args.compiler:

lldb/trunk/packages/Python/lldbsuite/test/dotest_args.py

group.add_argument(
         action='append',
         help='Specify an environment variable to unset before running the test cases. e.g., -u DYLD_INSERT_LIBRARIES -u MallocScribble')
     group.add_argument(
         '--env',
         dest='set_env_vars',
         metavar='variable',
         action='append',
         help='Specify an environment variable to set to the given value before running the test cases e.g.: --env CXXFLAGS=-O3 --env DYLD_INSERT_LIBRARIES')
+    group.add_argument(
+        '--inferior-env',
+        dest='set_inferior_env_vars',
+        metavar='variable',
+        action='append',
+        help='Specify an environment variable to set to the given value for the inferior.')
     X('-v', 'Do verbose mode of unittest framework (print out each test case invocation)')
     group.add_argument(
         '--enable-crash-dialog',
         dest='disable_crash_dialog',
         action='store_false',
         help='(Windows only) When LLDB crashes, display the Windows crash dialog.')
     group.set_defaults(disable_crash_dialog=True)
 

lldb/trunk/packages/Python/lldbsuite/test/lldbtest.py

class TestBase(Base):
     def setUp(self):
         #import traceback
         # traceback.print_stack()
 
         # Works with the test driver to conditionally skip tests via
         # decorators.
         Base.setUp(self)
 
+        if lldbtest_config.inferior_env:
+            self.runCmd('settings set target.env-vars {}'.format(lldbtest_config.inferior_env))
+
         # Set the clang modules cache path used by LLDB.
         mod_cache = os.path.join(os.environ["LLDB_BUILD"], "module-cache-lldb")
         self.runCmd('settings set symbols.clang-modules-cache-path "%s"'
                     % mod_cache)
 
         # Disable Spotlight lookup. The testsuite creates
         # different binaries with the same UUID, because they only
         # differ in the debug info, which is not being hashed.

lldb/trunk/packages/Python/lldbsuite/test/lldbtest_config.py

 # leave logs/traces even for successful test runs
 log_success = False
 
 # Indicate whether we're testing with an out-of-tree debugserver
 out_of_tree_debugserver = False
 
 # path to the lldb command line executable tool
 lldbExec = None
+
+# Environment variables for the inferior
+inferior_env = None