> On Mon, 2 Jul 2001, Bill Stoddard wrote:
>
> > >
> > > >cgi on all platforms is broken. Specifically, with the addition of
> > > >filters, we have lost the ability to flush partially written buffers
> > > >received from CGI scripts to the network.
> > > [...]
> > > >Apache 2.0 always does a blocking read (in the content length filter)
> > >
> > > Partial writes to the network from a CGI and content length are
> > > mutually exclusive.
> > >
> >
> > No S**t!! :-) That's why it's broken!
> >
> > Working on a fix now.
>
>
> Bill, there is no fix for this. If the content_length filter determines
> that a C-L is required, then we can't stream CGI's. It looks like the
> problem is simply that we are using the wrong options to apr_bucket_read.
> Switch that to a APR_NONBLOCK_READ, and the problem should go away.
>
> Ryan
I don't think changing the read in the content_length filter to nonblocking is the
right
solution to this problem because we still need to force the network flush. I will
investigate your suggestion though because I might be overlooking something.
IMO the decision to flush or not and how to read from the pipe should be entirely
controlled by mod_cgi. We need a function similar to the old ap_send_fb() code.
To demonstrate, here is some code. YES YES YES I know we don't want to commit this as
written. This code will cause a data copies of all bytes off the stack buffer and into
a
heap buffer. Making this zero copy is easy enough, just wanted to post what I have been
playing with the last 30 minutes to generally show how I think we should solve this
problem (ie, control the byte stream at the top of the filter chain rather than deep in
the filter chain).
I also want to play with making the interface between Apache and CGI scripts full
duplex
as well (to allow a CGI script to read a large POST request and begin responding to
that
POST immediately) which will require controlling the byte stream at the top of the
filter
stack as well.
Bill
Index: mod_cgi.c
===================================================================
RCS file: /home/cvs/httpd-2.0/modules/generators/mod_cgi.c,v
retrieving revision 1.92
diff -u -u -r1.92 mod_cgi.c
--- mod_cgi.c 2001/02/28 15:24:05 1.92
+++ mod_cgi.c 2001/07/02 18:26:53
@@ -768,16 +768,43 @@
}
if (!r->header_only) {
+ #define IOBUFSIZE 8192
+ apr_status_t rv;
+ char buf[IOBUFSIZE];
+ int len = IOBUFSIZE;
+ apr_interval_time_t timeout;
+
bb = apr_brigade_create(r->pool);
- b = apr_bucket_pipe_create(script_in);
- APR_BRIGADE_INSERT_TAIL(bb, b);
- b = apr_bucket_eos_create();
- APR_BRIGADE_INSERT_TAIL(bb, b);
- ap_pass_brigade(r->output_filters, bb);
- }
+
+ /* Set the pipe to non-blocking for the first read */
+ apr_file_pipe_timeout_get(script_in, &timeout);
+ apr_file_pipe_timeout_set(script_in, 0);
- log_script_err(r, script_err);
- apr_file_close(script_err);
+ while (!r->connection->aborted) {
+ len = IOBUFSIZE;
+ rv = apr_file_read(script_in, buf, &len);
+ if (rv == APR_EOF || (rv == APR_SUCCESS && len == 0)) {
+ b = apr_bucket_eos_create();
+ APR_BRIGADE_INSERT_TAIL(bb, b);
+ ap_pass_brigade(r->output_filters, bb);
+ break;
+ }
+ else if (rv == APR_EAGAIN) {
+ /* Set the pipe to blocking and flush the output stream. */
+ apr_file_pipe_timeout_set(script_in, timeout);
+ b = apr_bucket_flush_create();
+ APR_BRIGADE_INSERT_TAIL(bb, b);
+ ap_pass_brigade(r->output_filters, bb);
+ }
+ else {
+ b = apr_bucket_transient_create(buf, len);
+ APR_BRIGADE_INSERT_TAIL(bb, b);
+ ap_pass_brigade(r->output_filters, bb);
+ }
+ }
+ log_script_err(r, script_err);
+ apr_file_close(script_err);
+ }
}
if (script_in && nph) {
