[PATCH 5/5] ntoskrnl.exe: Make user shared data pointers volatile (GCC 11).

[Zebediah Figura]

On 9/27/21 11:43 AM, Rémi Bernon wrote:
> On 9/27/21 6:33 PM, Zebediah Figura wrote:
>> On 9/27/21 3:58 AM, Rémi Bernon wrote:
>>> So that GCC 11 stops warning about reading from a 0-size memory region.
>>>
>>> Signed-off-by: Rémi Bernon <rbernon at codeweavers.com>
>>> ---
>>>    dlls/ntoskrnl.exe/instr.c | 4 ++--
>>>    1 file changed, 2 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/dlls/ntoskrnl.exe/instr.c b/dlls/ntoskrnl.exe/instr.c
>>> index f197570db0c..fbcd376dbc1 100644
>>> --- a/dlls/ntoskrnl.exe/instr.c
>>> +++ b/dlls/ntoskrnl.exe/instr.c
>>> @@ -498,8 +498,8 @@ WINE_DEFAULT_DEBUG_CHANNEL(int);
>>>    #define SIB_BASE( sib, rex )    (((sib) & 7) | (((rex) & REX_B) ? 8
>>> : 0))
>>>    /* keep in sync with dlls/ntdll/thread.c:thread_init */
>>> -static const BYTE *wine_user_shared_data = (BYTE *)0x7ffe0000;
>>> -static const BYTE *user_shared_data      = (BYTE *)0xfffff78000000000;
>>> +static const BYTE *volatile wine_user_shared_data = (BYTE *)0x7ffe0000;
>>> +static const BYTE *volatile user_shared_data      = (BYTE
>>> *)0xfffff78000000000;
>>>    static inline DWORD64 *get_int_reg( CONTEXT *context, int index )
>>>    {
>>>
>>
>> This looks wrong. It should presumably be "const volatile BYTE *"
>> (actually: "const volatile BYTE *const"), but I'm guessing that doesn't
>> actually fix the warning. Granted, there's an open GCC bug for this [1],
>> and marking the variable volatile is suggested as a workaround...
>>
>> Perhaps at least we should mark that we're working around a GCC bug in
>> the code, since otherwise it looks like "volatile" is in the wrong place.
>>
>> [1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99578
>>
> 
> No, the contents of the memory don't need to be volatile, the pointer
> does. This way GCC cannot assume its fixed value (the warning triggers
> when accessing non-NULL pointers).
> 

Well, right, to hide the GCC warning, that's true. But the contents of 
the memory *should* be volatile (they're arbitrarily rewritten by 
another process), and in theory the value of the "user_shared_data" 
pointer should never change. Knowing that, "const BYTE *volatile" looks 
like a mistake.