[PORT] BeOS port problem and suggestion
Nick Clifton
nickc@redhat.com
Thu Sep 19 14:49:00 GMT 2002
More information about the Binutils mailing list
Thu Sep 19 14:49:00 GMT 2002
- Previous message (by thread): [PORT] BeOS port problem and suggestion
- Next message (by thread): alpha reloc tweaks
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Hi Thierry, > One is a problem with the BeOS fseek that rely on BFS filesystem : > on BFS when you fseek on a new create file, none of offset is set > with zeros. So if you fseek/seek and you don't use the pass area, > when you will write your file, you will have some garbage inside. > > I would like to handle this problem like this : > > First modify the configure tools chain and add a test to verify how > fseek/seek work on the host platform. If I detect a problem (such the > BeOS one), use a bfd_seek()/bfd_fseek() functions in place of seek()/ > fseek(). Internally bfd already uses a bfd_seek function. There should be no (significant) places inside the BFD code where seek or fseek is used instead of bfd_seek. It is not really BFD's job to provide a replacement for seek/fseek in a more general context. In fact this ought to be something that BeOS provides as some kind of compatibility library. There has already been a patch submission to address this problem. Please have a look at this: http://sources.redhat.com/ml/binutils/2000-08/msg00437.html Note - this patch was rejected, for very good reasons (an fseek past the end of the file would fill it with zeroes, even if there were no writes past the end of the file). Unless you can convince us that whatever patch you develop does not suffer from this problem it will be rejected again. > Since bfd is common to binutils/gcc, where I will send my patches ? bfd is not common to gcc. It is used by GDB, and its behavior is influenced by GCC, but it is certainly not tied to it. Anyway the answer to your question is that you should submit any patch developed to: binutils@sources.redhat.com > Since the focus is set on gcc 3.x does my patch could be include > even in the 2.95.3 tree ? BFD is not in the 2.95.3 tree or the 3.x tree. It is in the binutils releases. The binutils release usually associated with gcc 2.95.3 is 2.11. And no, any patch you develop will not be back ported into the 2.11 binutils release. Cheers Nick
- Previous message (by thread): [PORT] BeOS port problem and suggestion
- Next message (by thread): alpha reloc tweaks
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Binutils mailing list