GNU bug report logs - #78773
[PATCH] Speedup url-retrieve-synchronously for low-latency connections

Previous Next

Full log

Message #29 received at 78773 <at> debbugs.gnu.org (full text, mbox):

From: Steven Allen <steven <at> stebalien.com>
To: Eli Zaretskii <eliz <at> gnu.org>
Cc: 78773 <at> debbugs.gnu.org, rpluim <at> gmail.com, larsi <at> gnus.org
Subject: Re: bug#78773: [PATCH] Speedup url-retrieve-synchronously for
 low-latency connections
Date: Thu, 12 Jun 2025 13:54:01 -0700

[Message part 1 (text/plain, inline)]

Steven Allen <steven <at> stebalien.com> writes:

> Eli Zaretskii <eliz <at> gnu.org> writes:
>
>>> From: Steven Allen <steven <at> stebalien.com>
>>> Cc: 78773 <at> debbugs.gnu.org, larsi <at> gnus.org
>>> Date: Thu, 12 Jun 2025 11:02:41 -0700
>>>
>>> > Perhaps instrumenting wait_reading_process_output with printf's would
>>> > help to understand the control flow in the case of nil and non-nil
>>> > PROCESS argument?
>>>
>>> I'll do that today. I have a suspicion that fast requests waiting on a
>>> single process exit early here:
>>>
>>> https://https.git.savannah.gnu.org/cgit/git/emacs.git/tree/src/process.c?h=81a3e4e51167be51c63eae682331210bc62f7280#n5562
>>
>> The condition for that block is
>>
>>       if (wait_proc
>> 	  && ! EQ (wait_proc->status, Qrun)
>> 	  && ! connecting_status (wait_proc->status))
>>
>> And the comment says "Don't wait for output from a non-running
>> process."  Is the case here that the network connection was already
>> closed when we read from the sub-process?  I thought that was not the
>> case.
>
> With my patch, it does hit that case, but on the third loop through so
> it's not exiting early.
>
> I also tried setting the timeout in `accept-process-output' to nil and
> that got me the same speedup as my fix but ended up looping 23 times so
> there's something really strange going on with the timeout here.

Ok, I think I've figured it out. We loop twice and wait on a timeout
(READ_OUTPUT_DELAY_INCREMENT) the second loop because we're already done
reading.

The first time through, we hit the following, recording that we've
gotten some output:

  https://cgit.git.savannah.gnu.org/cgit/emacs.git/tree/src/process.c?h=81a3e4e51167be51c63eae682331210bc62f7280#n5974


We hit the following, setting our timeout to READ_OUTPUT_DELAY_INCREMENT:

  https://cgit.git.savannah.gnu.org/cgit/emacs.git/tree/src/process.c?h=81a3e4e51167be51c63eae682331210bc62f7280#n5679

Then we hit the select call and wait the full timeout (nothing to read)
and we finally exit here (because our timeout has passed):

  https://cgit.git.savannah.gnu.org/cgit/emacs.git/tree/src/process.c?h=81a3e4e51167be51c63eae682331210bc62f7280#n5832

***

I think what we need to do is modify:

  https://cgit.git.savannah.gnu.org/cgit/emacs.git/tree/src/process.c?h=81a3e4e51167be51c63eae682331210bc62f7280#n5978

from

    		  if (wait_proc == XPROCESS (proc))
		    wait = MINIMUM;

to

    		  if (!wait_proc || wait_proc == XPROCESS (proc))
		    wait = MINIMUM;

So that we set the timeout to 0 here:

  https://cgit.git.savannah.gnu.org/cgit/emacs.git/tree/src/process.c?h=81a3e4e51167be51c63eae682331210bc62f7280#n5439

And exit early.

[0001-Exit-accept-process-output-early-when-we-get-output-.patch (text/x-patch, attachment)]

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.