Mail Archives: djgpp/2011/09/03/05:53:30
> Date: Sat, 3 Sep 2011 11:23:06 +0300
> From: Ozkan Sezer <sezeroz AT gmail DOT com>
>
> On Sat, Sep 3, 2011 at 6:28 AM, Charles Sandmann <cwsdpmi AT earthlink DOT net> wrote:
> > I think these are related to the ability of some systems to support
> > files > 2GB in size, which are not directly LFN related. There needs
> > to be a different test in the code for this handling before it is used
> > (requires LFN, but LFN does not necessarily provide big file support).
> > This was one of the issues that kept V2.04 in beta.
>
> IIUC, the error would show itself if the accessed file is really >= 2GB,
> however Juan's test was simply running gcc to compile a hello.c and I
> don't think that gcc is trying to access any such large file in the process.
You are probably right. However, Charles's point is worth taking,
because we also fail these two functions with EOVERFLOW when the
registers returned by 71A6 indicate that the file size is > 2GB. So
it could make sense to treat each of these 2 cases differently, where
we currently don't.
Again, running under GDB to see which of these is actually the case
for GCC will allow us to be smarter.
- Raw text -