httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
Subject Re: Can I take down
Date Fri, 16 Feb 2001 16:02:40 GMT
On Fri, 16 Feb 2001, Brian Havard wrote:

> On Thu, 15 Feb 2001 22:09:36 -0800 (PST), Cliff Woolley wrote:
> >
> >--- wrote:
> >> I just requested this file on my server running Linux, through the
> >> mod_include filter, without thd problem that I see when I request the page
> >> from your machine.
> >> 
> >> What confuses me, is that it doesn't look like this is a problem with
> >> sendfile on your machine, because it works without the filter.  I haven't
> >> got a clue what is going on, but it looks like it is OS/2 specific.
> >
> >This part actually makes sense... when he's serving the file static via sendfile,
> >makes it all the way down the filter chain as a file bucket.  When using mod_include,
> >file_read() gets called, meaning no more file bucket.  I might be wrong, but OS/2
> >seem to have MMAP, so it's getting read by apr_read() into heap buckets of a couple
> >each that are getting shuffled around.  Makes sense to me.  Assuming I'm right that
> >doesn't have MMAP, it's interesting to note that OS/2 might be one of the few platforms
> >using that particular code path in file_read().  Still, I wonder if this is a symptom
> >a lower-level problem with some APR function's implementation under OS/2, or if it's
> >symptom of not using MMAP.  Ryan, can you test on your machine with MMAP disabled?
> >yours still work in that case?
> OS/2 doesn't have mmap OR sendfile(). That could be the difference.

Well, I'm not seeing the same thing you are, but there is definately a
problem with the code.  I'm looking into it ASAP.


Ryan Bloom               
406 29th St.
San Francisco, CA 94131

View raw message