win32utils devel - Nov 2007 - Service.services bug (pure Ruby version)

Hi all,

I''ve hit a curious bug in the Service.services method. This is in the 
pure Ruby version in CVS. Check it out and run this snippet:

Service.services{ |s|
    p s.service_name
    p s.display_name
}

That will segfault after about 14 entries for me. It appears to be a 
problem with memcpy on line 658, but I couldn''t tell you why.

The interesting thing is that the more I use ''puts'', the less
likely
this error is to occur. To show you what I mean, try this snippet:

Service.services{ |s|
    puts s.service_name
    p s.display_name
}

That gets further, but will still segfault. Now try this:

Service.services{ |s|
    puts s.service_name
    puts s.display_name
}

That one gets me through all services without segfaulting.

I know I''ve seen various reports over the years of a
''puts'' that made
bugs mysteriously disappear, so I''m wondering if I''ve hit upon
something. Can anyone tell me what''s going on?

Thanks,

Dan

Hi,

2007/11/3, Daniel Berger <djberg96 at gmail.com>:
>
> Hi all,
>
> I''ve hit a curious bug in the Service.services method. This is in
the
> pure Ruby version in CVS. Check it out and run this snippet:
>
> Service.services{ |s|
>    p s.service_name
>    p s.display_name
> }
>
> That will segfault after about 14 entries for me. It appears to be a
> problem with memcpy on line 658, but I couldn''t tell you why.
>
> The interesting thing is that the more I use ''puts'', the
less likely
> this error is to occur. To show you what I mean, try this snippet:
>
> Service.services{ |s|
>    puts s.service_name
>    p s.display_name
> }
>
> That gets further, but will still segfault. Now try this:
>
> Service.services{ |s|
>    puts s.service_name
>    puts s.display_name
> }
>
> That one gets me through all services without segfaulting.
>
> I know I''ve seen various reports over the years of a
''puts'' that made
> bugs mysteriously disappear, so I''m wondering if I''ve hit
upon
> something. Can anyone tell me what''s going on?
>
> Thanks,
>
> Dan

The segfaulting is due to the memcpy in service.rb

            dep_buf = 0.chr * 260
            memcpy(dep_buf,
config_buf[24,4].unpack(''L'').first,
dep_buf.size)
            dependencies =
dep_buf.split("\000\000").first.split(0.chr)

 Here is an agly but working code:

            dep_buf = ""
            psz_depend = config_buf[24,4].unpack(''L'').first
            while psz_depend!=0
              char_buf = 0.chr
              memcpy(char_buf, psz_depend, 1)
              psz_depend += 1
              dep_buf += char_buf
              break if dep_buf[-2,2]=="\0\0"
            end
            dependencies = []
            if dep_buf != "\0\0"
              dependencies =
dep_buf.split("\000\000").first.split(0.chr)
            end

Regards,

Park Heesob
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
http://rubyforge.org/pipermail/win32utils-devel/attachments/20071104/58155fa9/attachment.html

Heesob Park wrote:

<snip>

> The segfaulting is due to the memcpy in service.rb 
>  
>             dep_buf = 0.chr * 260
>             memcpy(dep_buf,
config_buf[24,4].unpack(''L'').first,
> dep_buf.size)
>             dependencies =
dep_buf.split("\000\000").first.split(0.chr)
>  
> Here is an agly but working code:
>  
>             dep_buf = "" 
>             psz_depend =
config_buf[24,4].unpack(''L'').first
>             while psz_depend!=0
>               char_buf = 0.chr
>               memcpy(char_buf, psz_depend, 1)
>               psz_depend += 1             
>               dep_buf += char_buf
>               break if dep_buf[-2,2]=="\0\0"             
>             end
>             dependencies = []
>             if dep_buf != "\0\0"
>               dependencies =
dep_buf.split("\000\000").first.split(0.chr)
>             end
>  
> Regards,
>  
> Park Heesob
Excellent, that works, thanks! It does seem like there ought to be a 
more efficient way, but we can worry about that later.

Thanks,

Dan