__wine_spec_relay_entry_points - need more info please

[DieselMachine]

Hi, Wine experts.
The program I'm trying to run in Wine crashes with SIGSEGV in the __wine_spec_relay_entry_points. Here is disassembled code that I got from gdb

Code: Select all

Dump of assembler code for function __wine_spec_relay_entry_points:
   0x68d97bbd <+0>:     nop
   0x68d97bbe <+1>:     xchg   %ax,%ax
   0x68d97bc0 <+3>:     push   %esp
   0x68d97bc1 <+4>:     push   $0x30000
   0x68d97bc6 <+9>:     call   0x68d98790 <__wine_spec_get_pc_thunk_eax>
   0x68d97bcb <+14>:    lea    0x17845(%eax),%eax
   0x68d97bd1 <+20>:    push   %eax
   0x68d97bd2 <+21>:    call   *0x4(%eax)
   0x68d97bd5 <+24>:    ret    $0xc
   0x68d97bd8 <+27>:    push   %esp
   0x68d97bd9 <+28>:    push   $0x30001
   0x68d97bde <+33>:    call   0x68d98790 <__wine_spec_get_pc_thunk_eax>
   0x68d97be3 <+38>:    lea    0x1782d(%eax),%eax
   0x68d97be9 <+44>:    push   %eax
   0x68d97bea <+45>:    call   *0x4(%eax)
   0x68d97bed <+48>:    ret    $0xc
   0x68d97bf0 <+51>:    push   %esp
   0x68d97bf1 <+52>:    push   $0x10002
   0x68d97bf6 <+57>:    call   0x68d98790 <__wine_spec_get_pc_thunk_eax>
   0x68d97bfb <+62>:    lea    0x17815(%eax),%eax
   0x68d97c01 <+68>:    push   %eax
   0x68d97c02 <+69>:    call   *0x4(%eax)
   0x68d97c05 <+72>:    ret    $0x4
   0x68d97c08 <+75>:    mov    %edi,%edi
   0x68d97c0a <+77>:    push   %ebp
   0x68d97c0b <+78>:    mov    %esp,%ebp
   0x68d97c0d <+80>:    add    %ch,%al
   0x68d97c0f <+82>:    jge    0x68d97c1c <__wine_spec_relay_entry_points+95>
   0x68d97c11 <+84>:    add    %al,(%eax)
   0x68d97c13 <+86>:    lea    0x177fd(%eax),%eax
   0x68d97c19 <+92>:    push   %eax
   0x68d97c1a <+93>:    call   *0x4(%eax)
   0x68d97c1d <+96>:    ret    $0xc
   0x68d97c20 <+99>:    push   %esp
   0x68d97c21 <+100>:   push   $0x30004
   0x68d97c26 <+105>:   call   0x68d98790 <__wine_spec_get_pc_thunk_eax>
   0x68d97c2b <+110>:   lea    0x177e5(%eax),%eax
   0x68d97c31 <+116>:   push   %eax
   0x68d97c32 <+117>:   call   *0x4(%eax)
   0x68d97c35 <+120>:   ret    $0xc
...............................................................................
   0x68d985f5 <+2616>:  lea    0x0(%esi),%esi
   0x68d985f8 <+2619>:  push   %esp
   0x68d985f9 <+2620>:  push   $0x20096
   0x68d985fe <+2625>:  call   0x68d98790 <__wine_spec_get_pc_thunk_eax>
   0x68d98603 <+2630>:  lea    0x16e0d(%eax),%eax
   0x68d98609 <+2636>:  push   %eax
   0x68d9860a <+2637>:  call   *0x4(%eax)
   0x68d9860d <+2640>:  ret    $0x8
End of assembler dump.

Look at the part 0x68d97c08 <+75>-0x68d97c1d <+96> - it is different from all others which look the same. My program crashes here 0x68d97c1a <+93>: call *0x4(%eax) because memory at this address couldn't be read.
So, I have several questions.
1. What is the place where this code is generated? When DLL is loading into the address space?
2. Is the part I mentioned above broken? And how could it be generated like this?
All that I found is the code in output_relay_debug in /wine-1.1.43/tools/winebuild/spec32.c. Is this the source of the generated assembler code? But I didn't see "broken" code here

Thanks for your help.

[James McKenzie]

DieselMachine wrote:

Hi, Wine experts.
The program I'm trying to run in Wine crashes with SIGSEGV in the __wine_spec_relay_entry_points. Here is disassembled code that I got from gdb

[Disassembled code removed]

Look at the part 0x68d97c08 <+75>-0x68d97c1d <+96> - it is different from all others which look the same. My program crashes here 0x68d97c1a <+93>: call *0x4(%eax) because memory at this address couldn't be read.
So, I have several questions.
1. What is the place where this code is generated? When DLL is loading into the address space?

Look for the call in the code.

2. Is the part I mentioned above broken? And how could it be generated like this?

No the code is not 'broken'. You are trying to write into space not
owned by you. Now that is a bug.

All that I found is the code in output_relay_debug in /wine-1.1.43/tools/winebuild/spec32.c. Is this the source of the generated assembler code? But I didn't see "broken" code here

Try using wine-1.1.44 and advise if the problem is fixed or not.

BTW, what program are you trying to run?

James McKenzie

[DieselMachine]

Thanks for the response.

I think that memory at 0x68d97c08-0x68d97c0f was corrupted by someone because <__wine_spec_relay_entry_points+95> is not a valid assembler instruction but there is a jump to it

Code: Select all

0x68d97c0f <+82>:    jge    0x68d97c1c <__wine_spec_relay_entry_points+95>

Is there a way I can get log or debug everything that happens with memory related to __wine_spec_relay_entry_points after it has been loaded?

I'm trying to run Lineage 2 Gracia Epilogue at one of the private servers.
http://appdb.winehq.org/objectManager.p ... &iId=18866.
Wine 1.1.44 packages are not ready (no updates for me). I'll try as soon as they will be available

[DieselMachine]

And one more question. Why is breakpoint into __wine_spec_relay_entry_points sets to the different location from the current execution?

Code: Select all

(gdb) bt
#0  0x68df5c08 in __wine_spec_relay_entry_points () from /usr/bin/../lib/wine/ws2_32.dll.so
#1  0x00c08aa1 in ?? ()
#2  0x00cb47d6 in ?? ()
#3  0x203cd97a in ?? ()
#4  0x05e80001 in ?? ()
Backtrace stopped: previous frame inner to this frame (corrupt stack?)
(gdb) break __wine_spec_relay_entry_points
Breakpoint 15 at 0x21b32e70
(gdb) info break
Num     Type           Disp Enb Address    What
7       breakpoint     keep y   0x7b870897 in VirtualProtect at virtual.c:215
        breakpoint already hit 4 times
15      breakpoint     keep y   0x21b32e70 <__wine_spec_relay_entry_points>

[Charles Davis]

On 5/9/10 2:16 PM, DieselMachine wrote:

And one more question. Why is breakpoint into __wine_spec_relay_entry_points sets to the different location from the current execution?

Because there are many symbols called '__wine_spec_relay_entry_points'.
There's one in every Wine module that exports symbols. Winedbg doesn't
know which one you mean, but instead of telling you this, it picks the
first one it finds.

Chip

[DieselMachine]

Thanks for you help.

I've found that __wine_spec_relay_entry_points from ws2_32.dll loads uncorrupted and then becomes corrupted. I'll debug further to see what's happened.

I think this thread is closed