Re: Question about mmap wrt compressed files


Jorge Cruz Rodriguez wrote:
> I am working on a system which processes a large number of
> rather large files (>100MB) mostly using mmap -- typically the
> applications read sections of the files sequentially but start and
> stop at random points. From the system end, there is no way
> of predicting which files will be accessed and which sections
> will be examined. Although a lot of data is shipped in and
> out, only a small portion of the files are typically looked at so
> mmap appears to be the only reasonable way to deal with things.
> In any case it is the way things have been done and will be
> modified significantly only with great pain.

If you compress the files you are going to modify your mmap approach,
significantly.
Could you return a pointer to the uncompressed data, and free the
memory instead of unmapping it?

karl m

.

Relevant Pages