[PATCH] msvcrtd: Allow new to allocated a _CRT_BLOCK block

[Alistair Leslie-Hughes]

Fixes: https://bugs.winehq.org/show_bug.cgi?id=44001

Signed-off-by: Alistair Leslie-Hughes <leslie_alistair(a)hotmail.com>
---
 dlls/msvcrtd/debug.c       | 2 +-
 dlls/msvcrtd/tests/debug.c | 4 ++++
 2 files changed, 5 insertions(+), 1 deletion(-)

diff --git a/dlls/msvcrtd/debug.c b/dlls/msvcrtd/debug.c
index b8496526ae..6bfa730ece 100644
--- a/dlls/msvcrtd/debug.c
+++ b/dlls/msvcrtd/debug.c
@@ -52,6 +52,7 @@ void * CDECL MSVCRTD_operator_new_dbg(MSVCRT_size_t nSize, int nBlockUse,
     switch(_BLOCK_TYPE(nBlockUse))
     {
     case _NORMAL_BLOCK:
+    case _CRT_BLOCK:
         break;
     case _CLIENT_BLOCK:
         FIXME("Unimplemented case for nBlockUse = _CLIENT_BLOCK\n");
@@ -59,7 +60,6 @@ void * CDECL MSVCRTD_operator_new_dbg(MSVCRT_size_t nSize, int nBlockUse,
     case _FREE_BLOCK:
         FIXME("Native code throws an exception here\n");
         return NULL;
-    case _CRT_BLOCK:
     case _IGNORE_BLOCK:
         ERR("Not allowed nBlockUse value: %d\n", _BLOCK_TYPE(nBlockUse));
         return NULL;
diff --git a/dlls/msvcrtd/tests/debug.c b/dlls/msvcrtd/tests/debug.c
index 9055db1382..d234f02f6d 100644
--- a/dlls/msvcrtd/tests/debug.c
+++ b/dlls/msvcrtd/tests/debug.c
@@ -65,6 +65,10 @@ static void test_new(void)
   mem = pMSVCRTD_operator_new_dbg(42, _NORMAL_BLOCK, __FILE__, __LINE__);
   ok(mem != NULL, "memory not allocated\n");
   HeapFree(GetProcessHeap(), 0, mem);
+
+  mem = pMSVCRTD_operator_new_dbg(42, _CRT_BLOCK, __FILE__, __LINE__);
+  ok(mem != NULL, "memory not allocated\n");
+  HeapFree(GetProcessHeap(), 0, mem);
 }
 
 /**********************************************************************/
-- 
2.16.3

[Piotr Caban]

Hi Alistair,

On 04/04/18 02:26, Alistair Leslie-Hughes wrote:
> @@ -52,6 +52,7 @@ void * CDECL MSVCRTD_operator_new_dbg(MSVCRT_size_t nSize, int nBlockUse,
>       switch(_BLOCK_TYPE(nBlockUse))
>       {
>       case _NORMAL_BLOCK:
> +    case _CRT_BLOCK:
>           break;
>       case _CLIENT_BLOCK:
>           FIXME("Unimplemented case for nBlockUse = _CLIENT_BLOCK\n");
> @@ -59,7 +60,6 @@ void * CDECL MSVCRTD_operator_new_dbg(MSVCRT_size_t nSize, int nBlockUse,
>       case _FREE_BLOCK:
>           FIXME("Native code throws an exception here\n");
>           return NULL;
> -    case _CRT_BLOCK:
>       case _IGNORE_BLOCK:
>           ERR("Not allowed nBlockUse value: %d\n", _BLOCK_TYPE(nBlockUse));
>           return NULL;
> Is there a reason for not calling msvcrt.operator_new_dbg instead of 
> implementing the function here? We don't have proper operator_new_dbg 
> implementation anyway. According to my tests it's also possible to 
> allocate _IGNORE_BLOCK. _FREE_BLOCK is terminating the app in msvcrtd 
> case and works in msvcrt (I don't expect any application to depend on 
> that behavior).
> +  mem = pMSVCRTD_operator_new_dbg(42, _CRT_BLOCK, __FILE__, __LINE__);
> +  ok(mem != NULL, "memory not allocated\n");
> +  HeapFree(GetProcessHeap(), 0, mem);
> You can't free the memory that way. Please call operator delete.

Thanks,
Piotr