Buy commercial curl support. We
help you work out your issues, debug your libcurl applications, use the API,
port to new platforms, add new features and more. With a team lead by the
curl founder Daniel himself.
Re: transactions stall after March 6 commit to improve wakeup and wait code in multi
- Contemporary messages sorted: [ by date ] [ by thread ] [ by subject ] [ by author ] [ by messages with attachments ]
From: Daniel Stenberg via curl-library <curl-library_at_lists.haxx.se>
Date: Sun, 14 Jun 2026 23:54:42 +0200 (CEST)
On Tue, 9 Jun 2026, Bryan Henderson via curl-library wrote:
> In old code, and for the past twenty years, this finds the file
> descriptor either already ready to read or ready very soon after. After
> that, my code calls curl_multi_perform to advance the HTTP transaction.
>
> But after the subject commit, the file descriptor never becomes ready to
> read. My code eventually times out as a failsafe and calls
> curl_multi_perform anyway, and the transaction does in fact advance.
> After that, the pselect/curl_multi_perform loop properly completes the
> transaction.
>
> Is this a bug, or is the algorithm above not supposed to work?
This sounds like it could be a possible bug.
Any chance you can for example modify an existing example source code to
reproduce this behavior? Then we would have a good reproducer to work with!
Date: Sun, 14 Jun 2026 23:54:42 +0200 (CEST)
On Tue, 9 Jun 2026, Bryan Henderson via curl-library wrote:
> In old code, and for the past twenty years, this finds the file
> descriptor either already ready to read or ready very soon after. After
> that, my code calls curl_multi_perform to advance the HTTP transaction.
>
> But after the subject commit, the file descriptor never becomes ready to
> read. My code eventually times out as a failsafe and calls
> curl_multi_perform anyway, and the transaction does in fact advance.
> After that, the pselect/curl_multi_perform loop properly completes the
> transaction.
>
> Is this a bug, or is the algorithm above not supposed to work?
This sounds like it could be a possible bug.
Any chance you can for example modify an existing example source code to
reproduce this behavior? Then we would have a good reproducer to work with!
-- / daniel.haxx.se || https://rock-solid.curl.dev -- Unsubscribe: https://lists.haxx.se/mailman/listinfo/curl-library Etiquette: https://curl.se/mail/etiquette.htmlReceived on 2026-06-14