win32utils devel - Jun 2006 - Yet another data structure + pack/unpack question (win32-service)

Hi all,

If you take a look at the service.rb file in the win32-service 
repository (the new one in the toplevel repository path), I''ve got this
bit of code, which succeeds, but I can''t seem to unpack the data 
structure properly.  Did I pack it wrong to begin with?  I should know 
this but I''m spacing out.

proc_status       = [0,0,0,0,0,0,0,0,0,0].pack(''LLLLLLLLLL'')
enum_service      = [0.chr, 0.chr].pack(''pp'') + proc_status
service_buf       = enum_service * 1000
bytes_needed      = [0].pack(''L'')
services_returned = [0].pack(''L'')
resume_handle     = [0].pack(''L'')

bool = EnumServicesStatusEx(
   handle_scm,
   SC_ENUM_PROCESS_INFO,
   SERVICE_WIN32 | SERVICE_DRIVER,
   SERVICE_STATE_ALL,
   service_buf,
   service_buf.size,
   bytes_needed,
   services_returned,
   resume_handle,
   group
)

if bool
   num_services = services_returned.unpack(''L'').first
   index = 0

   1.upto(num_services){ |num|
      info = service_buf[index, enum_service.size-1]
      service_name = info[0,4].pack(''p'') # boom!
      index += enum_service.size
   }
end

Regards,

Dan

Hi,

2006/6/21, Daniel Berger <djberg96 at gmail.com>:
> Hi all,
>
> If you take a look at the service.rb file in the win32-service
> repository (the new one in the toplevel repository path), I''ve got
this
> bit of code, which succeeds, but I can''t seem to unpack the data
> structure properly.  Did I pack it wrong to begin with?  I should know
> this but I''m spacing out.
>
> proc_status       =
[0,0,0,0,0,0,0,0,0,0].pack(''LLLLLLLLLL'')
> enum_service      = [0.chr, 0.chr].pack(''pp'') +
proc_status
> service_buf       = enum_service * 1000
> bytes_needed      = [0].pack(''L'')
> services_returned = [0].pack(''L'')
> resume_handle     = [0].pack(''L'')
>
> bool = EnumServicesStatusEx(
>   handle_scm,
>   SC_ENUM_PROCESS_INFO,
>   SERVICE_WIN32 | SERVICE_DRIVER,
>   SERVICE_STATE_ALL,
>   service_buf,
>   service_buf.size,
>   bytes_needed,
>   services_returned,
>   resume_handle,
>   group
> )
>
> if bool
>   num_services = services_returned.unpack(''L'').first
>   index = 0
>
>   1.upto(num_services){ |num|
>      info = service_buf[index, enum_service.size-1]
>      service_name = info[0,4].pack(''p'') # boom!
>      index += enum_service.size
>   }
> end
>
I guess you still don''t understand the difference of C pointer and
ruby buffer :)

Try this:

Strcpy  = Win32API.new(''msvcrt'', ''strcpy'',
''PL'', ''L'')

 1.upto(num_services){ |num|
     info = service_buf[index, enum_service.size-1]
     service_name = 0.chr * 256
     Strcpy.call(service_name,info[0,4].unpack(''L'').first)
     service_name = service_name.strip
     index += enum_service.size
  }

But, the code still segfault for another bug.

Regards,

Park Heesob

2006/6/21, Heesob Park <phasis at gmail.com>:
> Hi,
>
> 2006/6/21, Daniel Berger <djberg96 at gmail.com>:
> > Hi all,
> >
> > If you take a look at the service.rb file in the win32-service
> > repository (the new one in the toplevel repository path),
I''ve got this
> > bit of code, which succeeds, but I can''t seem to unpack the
data
> > structure properly.  Did I pack it wrong to begin with?  I should know
> > this but I''m spacing out.
> >
> > proc_status       =
[0,0,0,0,0,0,0,0,0,0].pack(''LLLLLLLLLL'')
> > enum_service      = [0.chr, 0.chr].pack(''pp'') +
proc_status
> > service_buf       = enum_service * 1000
> > bytes_needed      = [0].pack(''L'')
> > services_returned = [0].pack(''L'')
> > resume_handle     = [0].pack(''L'')
> >
> > bool = EnumServicesStatusEx(
> >   handle_scm,
> >   SC_ENUM_PROCESS_INFO,
> >   SERVICE_WIN32 | SERVICE_DRIVER,
> >   SERVICE_STATE_ALL,
> >   service_buf,
> >   service_buf.size,
> >   bytes_needed,
> >   services_returned,
> >   resume_handle,
> >   group
> > )
> >
> > if bool
> >   num_services = services_returned.unpack(''L'').first
> >   index = 0
> >
> >   1.upto(num_services){ |num|
> >      info = service_buf[index, enum_service.size-1]
> >      service_name = info[0,4].pack(''p'') # boom!
> >      index += enum_service.size
> >   }
> > end
> >
> I guess you still don''t understand the difference of C pointer and
> ruby buffer :)
>
> Try this:
>
> Strcpy  = Win32API.new(''msvcrt'',
''strcpy'', ''PL'', ''L'')
>
>  1.upto(num_services){ |num|
>     info = service_buf[index, enum_service.size-1]
>     service_name = 0.chr * 256
>     Strcpy.call(service_name,info[0,4].unpack(''L'').first)
>     service_name = service_name.strip
>     index += enum_service.size
>  }
>
> But, the code still segfault for another bug.
>
And I found the bug.

proc_status       = [0,0,0,0,0,0,0,0,0,0].pack(''LLLLLLLLLL'')
should be
proc_status       = [0,0,0,0,0,0,0,0,0].pack(''LLLLLLLLL'')

Regards,

Park Heesob

Heesob Park wrote:
> Hi,
>
> 2006/6/21, Daniel Berger <djberg96 at gmail.com>:
>   
>> Hi all,
>>
>> If you take a look at the service.rb file in the win32-service
>> repository (the new one in the toplevel repository path), I''ve
got this
>> bit of code, which succeeds, but I can''t seem to unpack the
data
>> structure properly.  Did I pack it wrong to begin with?  I should know
>> this but I''m spacing out.
>>
>> proc_status       =
[0,0,0,0,0,0,0,0,0,0].pack(''LLLLLLLLLL'')
>> enum_service      = [0.chr, 0.chr].pack(''pp'') +
proc_status
>> service_buf       = enum_service * 1000
>> bytes_needed      = [0].pack(''L'')
>> services_returned = [0].pack(''L'')
>> resume_handle     = [0].pack(''L'')
>>
>> bool = EnumServicesStatusEx(
>>   handle_scm,
>>   SC_ENUM_PROCESS_INFO,
>>   SERVICE_WIN32 | SERVICE_DRIVER,
>>   SERVICE_STATE_ALL,
>>   service_buf,
>>   service_buf.size,
>>   bytes_needed,
>>   services_returned,
>>   resume_handle,
>>   group
>> )
>>
>> if bool
>>   num_services = services_returned.unpack(''L'').first
>>   index = 0
>>
>>   1.upto(num_services){ |num|
>>      info = service_buf[index, enum_service.size-1]
>>      service_name = info[0,4].pack(''p'') # boom!
>>      index += enum_service.size
>>   }
>> end
>>
>>     
> I guess you still don''t understand the difference of C pointer and
> ruby buffer :)
>
> Try this:
>
> Strcpy  = Win32API.new(''msvcrt'',
''strcpy'', ''PL'', ''L'')
>
>  1.upto(num_services){ |num|
>      info = service_buf[index, enum_service.size-1]
>      service_name = 0.chr * 256
>     
Strcpy.call(service_name,info[0,4].unpack(''L'').first)
>      service_name = service_name.strip
>      index += enum_service.size
>   }
>
> But, the code still segfault for another bug.
>
> Regards,
>
> Park Heesob
> _______________________________________________
> win32utils-devel mailing list
> win32utils-devel at rubyforge.org
> http://rubyforge.org/mailman/listinfo/win32utils-devel
>
>   
Ok, thanks.  I guess part of the reason I got confused as to why you 
can''t do this:

irb(main):016:0> bar = ''hello''
=> "hello"
irb(main):017:0>
bar.unpack("l").pack("l").unpack("p")
ArgumentError: no associated pointer

Is Ruby attaching some metadata to the string that can only be gotten to 
if it''s packed with ''p''?  I mean, why can''t
you go from address to object?

Thanks,

Dan

Hi,

2006/6/21, Daniel Berger <djberg96 at gmail.com>:
> Heesob Park wrote:
> > Hi,
> >
> > 2006/6/21, Daniel Berger <djberg96 at gmail.com>:
> >
> >> Hi all,
> >>
> >> If you take a look at the service.rb file in the win32-service
> >> repository (the new one in the toplevel repository path),
I''ve got this
> >> bit of code, which succeeds, but I can''t seem to unpack
the data
> >> structure properly.  Did I pack it wrong to begin with?  I should
know
> >> this but I''m spacing out.
> >>
> >> proc_status       =
[0,0,0,0,0,0,0,0,0,0].pack(''LLLLLLLLLL'')
> >> enum_service      = [0.chr, 0.chr].pack(''pp'') +
proc_status
> >> service_buf       = enum_service * 1000
> >> bytes_needed      = [0].pack(''L'')
> >> services_returned = [0].pack(''L'')
> >> resume_handle     = [0].pack(''L'')
> >>
> >> bool = EnumServicesStatusEx(
> >>   handle_scm,
> >>   SC_ENUM_PROCESS_INFO,
> >>   SERVICE_WIN32 | SERVICE_DRIVER,
> >>   SERVICE_STATE_ALL,
> >>   service_buf,
> >>   service_buf.size,
> >>   bytes_needed,
> >>   services_returned,
> >>   resume_handle,
> >>   group
> >> )
> >>
> >> if bool
> >>   num_services =
services_returned.unpack(''L'').first
> >>   index = 0
> >>
> >>   1.upto(num_services){ |num|
> >>      info = service_buf[index, enum_service.size-1]
> >>      service_name = info[0,4].pack(''p'') # boom!
> >>      index += enum_service.size
> >>   }
> >> end
> >>
> >>
> > I guess you still don''t understand the difference of C
pointer and
> > ruby buffer :)
> >
> > Try this:
> >
> > Strcpy  = Win32API.new(''msvcrt'',
''strcpy'', ''PL'', ''L'')
> >
> >  1.upto(num_services){ |num|
> >      info = service_buf[index, enum_service.size-1]
> >      service_name = 0.chr * 256
> >     
Strcpy.call(service_name,info[0,4].unpack(''L'').first)
> >      service_name = service_name.strip
> >      index += enum_service.size
> >   }
> >
> > But, the code still segfault for another bug.
> >
> > Regards,
> >
> > Park Heesob
> > _______________________________________________
> > win32utils-devel mailing list
> > win32utils-devel at rubyforge.org
> > http://rubyforge.org/mailman/listinfo/win32utils-devel
> >
> >
> Ok, thanks.  I guess part of the reason I got confused as to why you
> can''t do this:
>
> irb(main):016:0> bar = ''hello''
> => "hello"
> irb(main):017:0>
bar.unpack("l").pack("l").unpack("p")
> ArgumentError: no associated pointer
>
> Is Ruby attaching some metadata to the string that can only be gotten to
> if it''s packed with ''p''?  I mean, why
can''t you go from address to object?
>
Yes, you''re right.
According to Ruby source pack.c and string.c,
Ruby make the string associated with "rb_str_associate".
And when unpacking ''p'', ruby check the string with
"rb_str_associated"
I mean it is almost impossible you go from address to object.

Regards,

Park Heesob