[PATCH V2 0/2] Fix kernel32:process tests

[Eric Pouech]

This shall let testbot succeed on kernel32:process tests (fixing evolution
generated by the shell-no-window type of consoles)

Note: the tests had been written at a time where the console in Windows
wasn't a kernel object like the others... (the two lower bits of the handle
where set). And hadn't been updated ever since, so some test was only run
on Win7... This serie also updates those test to focus mainly on Win{>= 8}

Supersede: 232632

Signed-off-by: Eric Pouech 
---

Eric Pouech (2):
      dlls/kernel32/tests: let the console tests pass if current console is shell-no-window
      dlls/kernel32/tests: adapt dup console handle test to non Win7 handles

dlls/kernel32/tests/process.c | 22 ++++++++--------------
 1 file changed, 8 insertions(+), 14 deletions(-)

[PATCH V2 1/2] dlls/kernel32/tests: let the console tests pass if current console is shell-no-window

[Eric Pouech]

This patch shall let this kernel32:process test pass when run from TestBot.

Note: adapted the tests to non Win7 type of handles (the ones with 2 lower bits
set...).
Windows 8 and above use real handles for console.
Wine has been doing so since since 2 years.

The is_console() helper is then not testing what its name pretends. It
actually tests whether we're running the test on W7 :-(
So let's get reed of it.

Signed-off-by: Eric Pouech

---
 dlls/kernel32/tests/process.c |    7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/dlls/kernel32/tests/process.c b/dlls/kernel32/tests/process.c
index 8f18a0d37a7..9629b50a744 100644
--- a/dlls/kernel32/tests/process.c
+++ b/dlls/kernel32/tests/process.c
@@ -1541,8 +1541,13 @@ static void test_Console(void)
     startup.hStdOutput = CreateFileA("CONOUT$", GENERIC_READ|GENERIC_WRITE, 0, &sa, OPEN_EXISTING, 0, 0);

/* first, we need to be sure we're attached to a console */
-    if (!is_console(startup.hStdInput) || !is_console(startup.hStdOutput))
+    if (startup.hStdInput == INVALID_HANDLE_VALUE || startup.hStdOutput == INVALID_HANDLE_VALUE)
     {
+        /* this fails either when this test process is run detached from console
+         * (unlikely, as this very process must be explicitely created with detached flag),
+         * or is attached to a Wine's shell-no-window kind of console (if the later, detach from it)
+         */
+        FreeConsole();
         /* we're not attached to a console, let's do it */
         AllocConsole();
         startup.hStdInput = CreateFileA("CONIN$", GENERIC_READ|GENERIC_WRITE, 0, &sa, OPEN_EXISTING, 0, 0);

[PATCH V2 1/2] dlls/kernel32/tests: let the console tests pass if current console is shell-no-window

[Marvin]

Hi,

While running your changed tests, I think I found new failures.
Being a bot and all I'm not very good at pattern recognition, so I might be
wrong, but could you please double-check?

Full results can be found at:
https://testbot.winehq.org/JobDetails.pl?Key=112924

Your paranoid android.

=== w10pro64_ar (64 bit report) ===

kernel32:
process.c:4980: Test failed: got 0xc0000004
process.c:4998: Test failed: got services.exe SessionId 1

=== debian11 (32 bit Hindi:India report) ===

kernel32:
process.c:1587: Test failed: GetLastError: expecting 87 got 6
process.c:1597: Test failed: GetLastError: expecting 87 got 6

[PATCH V2 2/2] dlls/kernel32/tests: adapt dup console handle test to non Win7 handles

[Eric Pouech]

This test hadn't been changed with console rewrite.
The status before this patch:
- it's not run on Windows except Win7
- it's not run on Wine

Rewrote it so that it run on all cases and inverse the test logic (it was
only testing Win7 behavior, which is no longer what we want).

Signed-off-by: Eric Pouech

---
 dlls/kernel32/tests/process.c |   15 ++-------------
 1 file changed, 2 insertions(+), 13 deletions(-)

diff --git a/dlls/kernel32/tests/process.c b/dlls/kernel32/tests/process.c
index 9629b50a744..ebadc2e8d12 100644
--- a/dlls/kernel32/tests/process.c
+++ b/dlls/kernel32/tests/process.c
@@ -1507,11 +1507,6 @@ static  void    test_DebuggingFlag(void)
     DeleteFileA(resfile);
 }

-static BOOL is_console(HANDLE h)
-{
-    return h != INVALID_HANDLE_VALUE && ((ULONG_PTR)h & 3) == 3;
-}
-
 static void test_Console(void)
 {
     char                buffer[2 * MAX_PATH + 35];
@@ -2500,17 +2495,11 @@ static void test_DuplicateHandle(void)
     DeleteFileA(file_name);

f = CreateFileA("CONIN$", GENERIC_READ|GENERIC_WRITE, 0, NULL, OPEN_EXISTING, 0, 0);
-    if (!is_console(f))
-    {
-        skip("DuplicateHandle on console handle\n");
-        CloseHandle(f);
-        return;
-    }
-
+    ok(f != INVALID_HANDLE_VALUE, "Failed to open CONIN$ %lu\n", GetLastError());
     r = DuplicateHandle(GetCurrentProcess(), f, GetCurrentProcess(), &out,
             0, FALSE, DUPLICATE_SAME_ACCESS | DUPLICATE_CLOSE_SOURCE);
     ok(r, "DuplicateHandle error %lu\n", GetLastError());
-    todo_wine ok(f != out, "f == out\n");
+    ok(f == out || broken(/* Win7 */ (((ULONG_PTR)f & 3) == 3) && (f != out)), "f != out\n");
     CloseHandle(out);
 }

[PATCH V2 2/2] dlls/kernel32/tests: adapt dup console handle test to non Win7 handles

[Marvin]

Hi,

While running your changed tests, I think I found new failures.
Being a bot and all I'm not very good at pattern recognition, so I might be
wrong, but could you please double-check?

Full results can be found at:
https://testbot.winehq.org/JobDetails.pl?Key=112925

Your paranoid android.

=== debian11 (32 bit Hindi:India report) ===

kernel32:
process.c:1582: Test failed: GetLastError: expecting 87 got 6
process.c:1592: Test failed: GetLastError: expecting 87 got 6
process.c:1623: Test failed: Console:CursorX expected 4294955996, but got 0