Intercept relay_call causes seh:raise_exception code=c0000005

[Jui-Hao Chiang]

Hi, All:

I also find out 2 special things
(1) The address of the array which causes the exception happens to be
the memory-mapped address of comctl32.dll (check the /proc/../smaps)
(2) If the address of the array is pointing to heap (from my malloc),
or some other places which are not occupied by any dll, then "no"
exception will occur

Originally I doubt the exception may be the bug of the RPC library I
am using (libc6-dev package inside Ubuntu).
However, if that's the case, my malloc pointer should also cause
exception (but it doesn't)

Still hope someone can give a hint on how to debug this question

Best wishes,
Jui-Hao

>
> Message: 11
> Date: Wed, 21 Apr 2010 09:37:53 -0400
> From: Jui-Hao Chiang <windtracekimo at gmail.com>
> Subject: Intercept relay_call causes seh:raise_exception code=c0000005
> To: wine-devel at winehq.org
> Message-ID:
>        <i2h52cf60ee1004210637gc7f20e5dq9b958127331b18cf at mail.gmail.com>
> Content-Type: text/plain; charset=ISO-8859-1
>
> Hi, All, I am currently using the WINEDEBUG=relay feature to intercept
> function calls as the following
>
> 1. in dlls/ntdll/relay.c, insert code snippet in
> static LONGLONG WINAPI relay_call( struct relay_descr *descr, unsigned
> int idx, const INT_PTR *stack )
> {
> .....
> my_own_function(entry_point->name, nb_args stack); /* added by myself */
> ret = call_entry_point( entry_point->orig_func, nb_args, stack + 1 );
> /* original Wine code*/
> ....
> }
>
> LONGLONG
> my_own_function(const char *funcname, int nb_args, const INT_PTR *stack)
> {
>  /* CreateBitmap has 5 parameters, the 5th one is pointing to an array
> while others are native type as the following
>    HBITMAP WINAPI CreateBitmap( INT width, INT height, UINT planes,
> UINT bpp, LPCVOID bits )
> */
>    if (strcmp(funcname, "CreateBitmap") == 0)
>             Rpc_CreateBitmap(*(stack+1), *(stack+2), *(stack+3),
> *(stack+4), *(stack+5)); // this function just try to forward all
> parameters to remote RPC server for record
> }
>
> 2. Run program with WINEDEBUG=trace+all,relay wine notepad > dump 2>&1
> 3. If I only pass the parameter 1 to 4 to Rpc_CreateBitmap, then there
> is no problem, but if I add the 5th one (which points to some array),
> then the notepad aborts with he following debug message
>
> 0009:Call gdi32.CreateBitmap(00000008,00000008,00000001,00000001,7e95d822)
> ret=7e8da083
> 0009:trace:seh:raise_exception code=c0000005 flags=0 addr=0xb754a21d
> ip=b754a21d tid=0009
> 0009:trace:seh:raise_exception  info[0]=00000001
> 0009:trace:seh:raise_exception  info[1]=7e95d822
> 0009:trace:seh:raise_exception  eax=00000055 ebx=b7593ff4 ecx=0033fa4c
> edx=00000001 esi=7e95d822 edi=00000001
> 0009:trace:seh:raise_exception  ebp=0033fa54 esp=0033fa44 cs=0073
> ds=007b es=007b fs=0033 gs=003b flags=00010202
> 0009:trace:seh:call_vectored_handlers calling handler at 0x7b83fd50
> code=c0000005 flags=0
> 0009:trace:seh:call_vectored_handlers handler at 0x7b83fd50 returned 0
> 0009:trace:seh:call_stack_handlers calling handler at 0x7bc4a480
> code=c0000005 flags=0
> 0009:trace:seh:__regs_RtlUnwind code=c0000005 flags=2
> 0009:trace:seh:__regs_RtlUnwind calling handler at 0x7bc6bc50
> code=c0000005 flags=2
> 0009:trace:seh:__regs_RtlUnwind handler at 0x7bc6bc50 returned 1
> 0009:exception in PE entry point
> (proc=0x7e95a890,module=0x7e8c0000,reason=PROCESS_ATTACH,res=0x1)
>
> 4. But if I use malloc to copy the content of the array pointed by 5th
> parameter, then there is no problem
> LONGLONG
> my_own_function(const char *funcname, int nb_args, const INT_PTR *stack)
> {
>    char *buffer = NULL;
>    if (strcmp(funcname, "CreateBitmap") == 0) {
>             buffer = (char*)malloc(.....);
>             memcpy(buffer, *(stack+5), ....);
>             Rpc_CreateBitmap(*(stack+1), *(stack+2), *(stack+3),
> *(stack+4), buffer); /* No exception if I do this */
> }
>
> Can anyone gives me a hint that what's happening here? or I am doing
> some illegal operations?
> Appreciate any comments,
> Jui-Hao
>