Re: [patch] TUX 2.4.2-L8

[daniel a fulton delphiauto com]

sorry, forgot to send below to the list.

Dainel

---------------------- Forwarded by Daniel A Fulton/cn/delphiauto on 03/07/2001
08:13 AM ---------------------------


To:   mingo@elte.hu@delphiauto
cc:

Subject:  Re: [patch] TUX 2.4.2-L8  (Document link: Daniel A Fulton)

L8 looks good.  wwwget, lynx requests are returned correctly.  ab succeeds in
testing for static content.

however, if I use ab to send concurrent requests or just a huge number of
requests that are redirected or handled by user space module,  ab still gives
tests "aborted" failure.

do you think maybe still same problem?  because ab would only abort if its still
getting incorrect buffer like zeros from the server or server doesn't respond to
repeated requests, right?  i could be totally wrong.  i can send you tcpdump if
you would like.

one thing too.  i have a user space module that handles requests for redirects /
caching application server pages.  i haven't done much testing but it looks like
i found a problem.

now that under 2.0.12, if I use user space module called (module_1) and redirect
to page_1_html which is a page that doesn't exist.

http://localhost/module_1?page_1_html

tux will crash as in TUX_thread 0 stopping

again, i could be totally wrong and this may be correct behavior but it would
make more sense to me that if you make a request to a user space and to a page
that doesn't exist, there would be a more graceful exit than server crash.

thanks again.
Daniel



Ingo Molnar <mingo@elte.hu> on 03/06/2001 08:34:18 PM

Please respond to tux-list@redhat.com

To:   Oleg Drokin <green@iXcelerator.com>
cc:   Daniel A Fulton/cn/delphiauto@DELPHI ASIA, TUX Development Mailing List
      <tux-list@redhat.com>

Subject:  [patch] TUX 2.4.2-L8




Oleg, Daniel,

i found and fixed the bug, please try -L8, does this fix your problems
too?:

  http://people.redhat.com/mingo/TUX-patches/tux2-full-2.4.2-L8.bz2

The bug was a thinko in abuf.c, introduced in -L6. req->abuf.buf needs to
be page_address(page)+offset, not page_address(page). This bug caused the
incorrect buffer to be sent to the browser, typically an all-zeroes reply.

     Ingo



_______________________________________________
tux-list mailing list
tux-list@redhat.com
https://listman.redhat.com/mailman/listinfo/tux-list