[Mono-dev] Syscall.mmap problem

[Jonathan Pryor]

On Sun, 2007-10-28 at 18:20 -0300, Mauricio Henriquez wrote:
> in my mmap test the "/pr.txt" file exist so the "error handling" is only 
> for test porpuses and to watch if any of the mmap paremeter is wronge 
> (this is a test of the mmap not the final working with the mmap on the 
> video4linus.net), in my case the file descriptor is correct (not < 0), 
> in any case thak you Jonathan for your test code, if I run your code I 
> get the same Console.WriteLine(start.ToString()) = -1 and this message 
> with the error handling:
> 
> mmap: Value too large for defined data type //I don't know what means

The only ideas I have left are:

1. What are the permissions of `/pr.txt'?

2. What platform is this on?  x86?  PPC?  Which operating system?

3. What version of Mono?

4. Are you compiling Mono yourself?

The mere fact you got "mmap: Value too large for defined data type" as
output implies that you're running svn (as perror() used to always write
"Success"), and still suggests the problem is in
Mono_Posix_Syscall_mmap().

If this is the case, then please run within gdb (see input after (gdb)
prompts):

$ gdb `which mono`
(gdb) b Mono_Posix_Syscall_mmap
Function "Mono_Posix_Syscall_mmap" not defined.
Make breakpoint pending on future shared library load? (y or [n]) y
Breakpoint 1 (Mono_Posix_Syscall_mmap) pending.
(gdb) r mmap.exe /pr.txt
Starting program: /home/jon/Development/mono-HEAD/install/bin/mono
mmap.exe /pr.txt
warning: Lowest section in system-supplied DSO at 0xffffe000 is .hash at
ffffe0b4
[Thread debugging using libthread_db enabled]
[New Thread 4158057136 (LWP 20303)]
[New Thread 4149185440 (LWP 20306)]
[New Thread 4149037984 (LWP 20307)]
Breakpoint 2 at 0xf73a0da0: file sys-mman.c, line 27.
Pending breakpoint "Mono_Posix_Syscall_mmap" resolved
fstat ok
length = 6
[Switching to Thread 4158057136 (LWP 20303)]

Breakpoint 2, Mono_Posix_Syscall_mmap (start=0x0, length=6, prot=1,
flags=2, fd=6, offset=0) at sys-mman.c:27
27              mph_return_val_if_size_t_overflow (length, MAP_FAILED);
(gdb) n
24      {
(gdb) n
27              mph_return_val_if_size_t_overflow (length, MAP_FAILED);
(gdb) n
30              if (Mono_Posix_FromMmapProts (prot, &_prot) == -1)
(gdb) n
32              if (Mono_Posix_FromMmapFlags (flags, &_flags) == -1)
(gdb) n
35              return mmap (start, (size_t) length, _prot, _flags, fd,
(off_t) offset);
(gdb) n
36      }
(gdb) c
Continuing.

The point is to see where Mono_Posix_Syscall_mmap() returns -- does it
actually call mmap()?  Or does it exit early because of an overflow or
translation error?  If it does exit early, where does it ext?

Thanks,
 - Jon