[PATCH 0/1] MR5123: kernel32/tests: Fix console test with odd-sized consoles.

[Jinoh Kang (＠iamahuman)]

-- 
https://gitlab.winehq.org/wine/wine/-/merge_requests/5123

[PATCH 1/1] kernel32/tests: Fix console test with odd-sized consoles.

[Jinoh Kang]

From: Jinoh Kang <jinoh.kang.kr(a)gmail.com>

---
 dlls/kernel32/tests/process.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/dlls/kernel32/tests/process.c b/dlls/kernel32/tests/process.c
index 2904a31d014..364d0ae2561 100644
--- a/dlls/kernel32/tests/process.c
+++ b/dlls/kernel32/tests/process.c
@@ -545,8 +545,8 @@ static void     doChild(const char* file, const char* option)
         ok( ret, "Setting mode (%ld)\n", GetLastError());
         ret = SetConsoleMode(hConOut, modeOut ^ 1);
         ok( ret, "Setting mode (%ld)\n", GetLastError());
-        sbi.dwCursorPosition.X ^= 1;
-        sbi.dwCursorPosition.Y ^= 1;
+        sbi.dwCursorPosition.X = (sbi.dwCursorPosition.X ^ 1) % sbi.dwSize.X;
+        sbi.dwCursorPosition.Y = (sbi.dwCursorPosition.Y ^ 1) % sbi.dwSize.Y;
         ret = SetConsoleCursorPosition(hConOut, sbi.dwCursorPosition);
         ok( ret, "Setting cursor position (%ld)\n", GetLastError());
     }
-- 
GitLab

https://gitlab.winehq.org/wine/wine/-/merge_requests/5123

[eric pouech (＠epo)]

eric pouech (@epo) commented about dlls/kernel32/tests/process.c:
> ok( ret, "Setting mode (%ld)\n", GetLastError());
>          ret = SetConsoleMode(hConOut, modeOut ^ 1);
>          ok( ret, "Setting mode (%ld)\n", GetLastError());
> -        sbi.dwCursorPosition.X ^= 1;
> -        sbi.dwCursorPosition.Y ^= 1;
> +        sbi.dwCursorPosition.X = (sbi.dwCursorPosition.X ^ 1) % sbi.dwSize.X;
> +        sbi.dwCursorPosition.Y = (sbi.dwCursorPosition.Y ^ 1) % sbi.dwSize.Y;

afaict, there's no explicit check to avoid 0 sized screen buffers in either dimension (right click inside a console let you change window&sb size to eg. (0,1))
so I'm not 100% confident that the modular operation will not yield a division by zero exception

perhaps something like (untested)
```
sbi.dwCursorPosition.X = !sbi.dwCursorPosition.X;
```
would be sufficient (the test only checks in parent that it gets same position as the one set by child)

-- 
https://gitlab.winehq.org/wine/wine/-/merge_requests/5123#note_62052

[Jinoh Kang (＠iamahuman)]

> 0 sized screen buffers in either dimension

In this case, `SetConsoleCursorPosition` will alwaya fail since even `(0, 0)` is out of bounds.

Maybe just skip this test?

-- 
https://gitlab.winehq.org/wine/wine/-/merge_requests/5123#note_62059

[Jinoh Kang (＠iamahuman)]

On Tue Feb 20 14:40:17 2024 +0000, Jinoh Kang wrote:
> > 0 sized screen buffers in either dimension
> > In this case, `SetConsoleCursorPosition` will alwaya fail since even
> > `(0, 0)` is out of bounds.
> > Maybe just skip this test?
> > Actually I don't think this is a problem. If size is 0 (either dim), the test already fails w/o this patch. The modulo just turns test failure into a div exception.

-- 
https://gitlab.winehq.org/wine/wine/-/merge_requests/5123#note_62060