[PATCH v6 0/1] MR155: ntdll: Fix reading stale Information from IOSB.

[Jinoh Kang (＠iamahuman)]

Enforce proper atomic update so that other threads do not read stale
data from IO_STATUS_BLOCK.

Signed-off-by: Jinoh Kang

--
  v6: ntdll: Fix reading stale Information from IOSB.

https://gitlab.winehq.org/wine/wine/-/merge_requests/155

[PATCH v6 1/1] ntdll: Fix reading stale Information from IOSB.

[Jinoh Kang]

From: Jinoh Kang

Ensure proper atomic update, so that the application do not read stale
information from IO_STATUS_BLOCK.

Also, ensure IOSB Status loads are not reordered with Information loads
from the consumer side as well, specifically in the following functions:

- GetOverlappedResultEx
- WSAGetOverlappedResult
---
 dlls/kernelbase/file.c         |  2 +-
 dlls/ntdll/unix/unix_private.h | 25 ++++++++++++++++++++++---
 dlls/ws2_32/socket.c           |  2 +-
 3 files changed, 24 insertions(+), 5 deletions(-)

diff --git a/dlls/kernelbase/file.c b/dlls/kernelbase/file.c
index 5ba7e0be419..3d4989ebeed 100644
--- a/dlls/kernelbase/file.c
+++ b/dlls/kernelbase/file.c
@@ -3133,7 +3133,7 @@ BOOL WINAPI DECLSPEC_HOTPATCH GetOverlappedResultEx( HANDLE file, OVERLAPPED *ov

TRACE( "(%p %p %p %lu %d)\n", file, overlapped, result, timeout, alertable );

-    status = overlapped->Internal;
+    status = ReadAcquire( (volatile LONG *)&overlapped->Internal );
     if (status == STATUS_PENDING)
     {
         if (!timeout)
diff --git a/dlls/ntdll/unix/unix_private.h b/dlls/ntdll/unix/unix_private.h
index 47f0f9c56a9..3939186ad05 100644
--- a/dlls/ntdll/unix/unix_private.h
+++ b/dlls/ntdll/unix/unix_private.h
@@ -361,6 +361,8 @@ static inline BOOL in_wow64_call(void)

static inline void set_async_iosb( client_ptr_t iosb, NTSTATUS status, ULONG_PTR info )
 {
+    NTSTATUS *status_ptr;
+
     if (!iosb) return;

if (in_wow64_call())
@@ -370,19 +372,36 @@ static inline void set_async_iosb( client_ptr_t iosb, NTSTATUS status, ULONG_PTR
             NTSTATUS Status;
             ULONG    Information;
         } *io = wine_server_get_ptr( iosb );
-        io->Status = status;
+        status_ptr = &io->Status;
         io->Information = info;
     }
     else
     {
         IO_STATUS_BLOCK *io = wine_server_get_ptr( iosb );
 #ifdef NONAMELESSUNION
-        io->u.Status = status;
+        status_ptr = &io->u.Status;
 #else
-        io->Status = status;
+        status_ptr = &io->Status;
 #endif
         io->Information = info;
     }
+
+    /* GetOverlappedResultEx() skips waiting for the event/file handle if the
+     * Status field indicates completion, and returns the result from the
+     * Information field.
+     *
+     * Hence, we have to ensure that writes to the Information field are seen
+     * from the other threads *before* writes to the Status field.
+     *
+     * Otherwise, a race condition results: if writing the Status field has
+     * been completed but the subsequent update to the Information field has
+     * not, the application may end up reading stale value from it.
+     *
+     * The race condition can be especially problematic with applications that
+     * use the HasOverlappedIoCompleted() macro in a tight loop to poll for
+     * completion of overlapped I/O.
+     */
+    WriteRelease( (volatile LONG *)status_ptr, status );
 }

static inline client_ptr_t iosb_client_ptr( IO_STATUS_BLOCK *io )
diff --git a/dlls/ws2_32/socket.c b/dlls/ws2_32/socket.c
index 2f115f02e48..3489c437dac 100644
--- a/dlls/ws2_32/socket.c
+++ b/dlls/ws2_32/socket.c
@@ -3702,7 +3702,7 @@ BOOL WINAPI WSAGetOverlappedResult( SOCKET s, LPWSAOVERLAPPED lpOverlapped,
         return FALSE;
     }

-    status = lpOverlapped->Internal;
+    status = ReadAcquire( (volatile LONG *)&lpOverlapped->Internal );
     if (status == STATUS_PENDING)
     {
         if (!fWait)

[Jinoh Kang (＠iamahuman)]

(Originally I intended to test for the existence of the barrier in ARM64 Windows, but decided that I'll never really get to it, so I pushed the code. Most bad code probably are in x86 side anyway...)