[OpenAFS-port-darwin] Compiler running off the end of file

[RD Schaffer]

Hi Ragnar,

   Thanks for the response. You are the first to answer this message.  
How might one get an idea of a timescale for a fix here? Should I 
report this problem elsewhere?

		thanks, RD Schaffer


On Wednesday, Dec 11, 2002, at 19:35 Europe/Paris, Ragnar Sundblad 
wrote:

>
> (Did this get answered?)
>
> --On den 6 december 2002 15:16 +0100 RD Schaffer 
> <R.D.Schaffer@cern.ch> wrote:
>
>> Hi there,
>>
>>    I am using a powerbook G4 running 10.2.2 having loaded the afs 10.2
>> version. This has been working fine for editing/copying/moving files. 
>> I
>> have now tried to build some of our software which is commonly kept in
>> afs and normally build on linux. I am using the 10.2 mac developer's 
>> kit
>> with gcc 3.2. What I see is that some, but not all C++ files "run off 
>> the
>> end" of the file with errors:
> ...
>
> It is probably the same problem that Arla had some year or two ago:
> gcc mmaps the file and assumes it ends with a zero byte, the
> file system has to zero out the rest of the block for gcc to work.
>
> I'd say it is a bug in gcc, but it is also a security hole that
> you get to see data from someone elses file or from some random
> memory page.
>
> I have no idea if and/or when this is fixed in the OpenAFS client.
>
> /ragge
>
>
>
Email:    R.D.Schaffer@cern.ch
Address:  LAL BAT 200        tel(Orsay): 33-1 64 46 8378
                   BP 34                     tel(cern) : 41-22 76 71267
                  91898 ORSAY
                  France