[lit] Test /dev/null support on Windows.

[zturner]

This might be a dumb question, but in Windows you can use NUL instead of /dev/null. Instead of all this hackery around temp files, why don't we just replace /dev/null with NUL?

[rnk]

In D44771#1045875, @zturner wrote:

This might be a dumb question, but in Windows you can use NUL instead of /dev/null. Instead of all this hackery around temp files, why don't we just replace /dev/null with NUL?

We should do that, but I consider it future work. It will require more testing. I have a feeling if you just say open("NUL", mode) you will create a file called "NUL" in the cwd. It's not something you can feed to generic filesystem APIs.

utils/lit/tests/unit/TestRunner.py

105 ↗

(On Diff #139417)

This should match the 4 space indent.

[amccarth]

In D44771#1046003, @rnk wrote:

I have a feeling if you just say open("NUL", mode) you will create a file called "NUL" in the cwd. It's not something you can feed to generic filesystem APIs.

No, it won't create a file named NUL. The special file names are handled at a relatively low level in the Windows stack (below _open). You would have to use a namespace escape like \\?\ to avoid having Windows treat your special file name as a regular name. The special names are: CON, PRN, AUX, NUL, COM1, COM2, COM3, COM4, COM5, COM6, COM7, COM8, COM9, LPT1, LPT2, LPT3, LPT4, LPT5, LPT6, LPT7, LPT8, and LPT9.

[mtrofin]

Spaces (2->4)

[rnk]

I ran the tests, and the unit tests don't pass because the /dev/null replacement happens much later during test execution, not command parsing. Unit testing the text executor would require much more faking/mocking than is really worth it. I would be comfortable making this change include a revert of rL328157. That will exercise the new replacement enough.

[mtrofin]

Revert "[llvm-profdata] Use "-o /dev/null" in invalid-profdata.test"

[mtrofin]

• [lit] windows dev null shshell test

[rnk]

Looks good after reverting the unit test that didn't work.

utils/lit/tests/Inputs/shtest-shell/dev-null.txt

1 ↗

(On Diff #139728)

New test looks good, thanks!

utils/lit/tests/unit/TestRunner.py

108 ↗

(On Diff #139728)

This test still fails because the parser doesn't do the replacement. You need to revert changes to this file and probably testrunner-custom-parsers/test.txt before committing or tests will fail on Windows.

[mtrofin]

Revert old test changes.

llvm/trunk/test/tools/llvm-profdata/invalid-profdata.test

 RUN: echo "# NumValueSites:"                             >> %t.input
 RUN: echo "2"                                            >> %t.input
 RUN: echo "2"                                            >> %t.input
 RUN: echo "f1:10"                                        >> %t.input
 RUN: echo "f2:0"                                         >> %t.input
 RUN: echo "1"                                            >> %t.input
 RUN: echo ":10"                                          >> %t.input
 
-RUN: not llvm-profdata merge %t.input -text -o /dev/null 2>&1  | FileCheck %s --check-prefix=BROKEN
+RUN: not llvm-profdata merge %t.input -text -output=/dev/null 2>&1  | FileCheck %s --check-prefix=BROKEN
 BROKEN: Malformed instrumentation profile data
 
 RUN: echo ":ir"                     > %t.input
 RUN: echo "_ZN6Thread5StartEv"      >> %t.input
 RUN: echo "# Func Hash:"            >> %t.input
 RUN: echo "288793635542036872"      >> %t.input
 RUN: echo "# Num Counters:"         >> %t.input
 RUN: echo "3"                       >> %t.input

llvm/trunk/utils/lit/lit/TestRunner.py

 
 kIsWindows = platform.system() == 'Windows'
 
 # Don't use close_fds on Windows.
 kUseCloseFDs = not kIsWindows
 
 # Use temporary files to replace /dev/null on Windows.
 kAvoidDevNull = kIsWindows
+kDevNull = "/dev/null"
 
 class ShellEnvironment(object):
 
     """Mutable shell environment containing things like CWD and env vars.
 
     Environment variables are not implemented, but cwd tracking is.
     """
 
for (index, r) in enumerate(redirects):
             continue
 
         redir_filename = None
         name = expand_glob(filename, cmd_shenv.cwd)
         if len(name) != 1:
            raise InternalShellError(cmd, "Unsupported: glob in "
                                     "redirect expanded to multiple files")
         name = name[0]
-        if kAvoidDevNull and name == '/dev/null':
+        if kAvoidDevNull and name == kDevNull:
             fd = tempfile.TemporaryFile(mode=mode)
         elif kIsWindows and name == '/dev/tty':
             # Simulate /dev/tty on Windows.
             # "CON" is a special filename for the console.
             fd = open("CON", mode)
         else:
             # Make sure relative paths are relative to the cwd.
             redir_filename = os.path.join(cmd_shenv.cwd, name)
for i,j in enumerate(cmd.commands):
             if not executable:
                 executable = lit.util.which(args[0], cmd_shenv.env['PATH'])
             if not executable:
                 raise InternalShellError(j, '%r: command not found' % j.args[0])
 
         # Replace uses of /dev/null with temporary files.
         if kAvoidDevNull:
             for i,arg in enumerate(args):
-                if arg == "/dev/null":
+                if kDevNull in arg:
                     f = tempfile.NamedTemporaryFile(delete=False)
                     f.close()
                     named_temp_files.append(f.name)
-                    args[i] = f.name
+                    args[i] = arg.replace(kDevNull, f.name)
 
         # Expand all glob expressions
         args = expand_glob_expressions(args, cmd_shenv.cwd)
         if is_builtin_cmd:
             args.insert(0, "python")
             args[1] = os.path.join(builtin_commands_dir ,args[1] + ".py")
 
         # On Windows, do our own command line quoting for better compatibility

llvm/trunk/utils/lit/tests/Inputs/shtest-shell/check_args.py

+#!/usr/bin/env python
+
+import argparse
+import platform
+
+parser = argparse.ArgumentParser()
+parser.add_argument("--my_arg", "-a")
+
+args = parser.parse_args()
+
+answer = (platform.system() == "Windows" and
+          args.my_arg == "/dev/null" and "ERROR") or "OK"
+
+print(answer)

llvm/trunk/utils/lit/tests/Inputs/shtest-shell/dev-null.txt

+# Check handling of /dev/null in command line options
+# On windows, it should be redirected to a temp file.
+#
+# RUN: "%{python}" %S/check_args.py --my_arg /dev/null | FileCheck %s --check-prefix=CHECK1
+# CHECK1: OK
+
+# RUN: "%{python}" %S/check_args.py --my_arg=/dev/null | FileCheck %s --check-prefix=CHECK2
+# CHECK2: OK
+
+# RUN: "%{python}" %S/check_args.py -a /dev/null | FileCheck %s --check-prefix=CHECK3
+# CHECK3: OK
+
+# RUN: "%{python}" %S/check_args.py -a=/dev/null | FileCheck %s --check-prefix=CHECK4
+# CHECK4: OK