Re: Memory leak with curl_multi_socket_action

On Tue, May 26, 2020 at 12:32 AM Jeffrey Walton <noloader_at_gmail.com> wrote:

> On Mon, May 25, 2020 at 7:16 PM James Read <jamesread5737_at_gmail.com>
> wrote:
> >
> >
> >
> > On Tue, May 26, 2020 at 12:02 AM Jeffrey Walton <noloader_at_gmail.com>
> wrote:
> >>
> >> On Mon, May 25, 2020 at 6:27 PM James Read via curl-library
> >> <curl-library_at_cool.haxx.se> wrote:
> >> >
> >> > ...
> >> >
> >> > Gmail seems to have taken out all the formatting. Apologies. It
> should still compile though.
> >>
> >> I can't speak for others, but... You should probably reduce the code
> >> to a minimal reproducer, and then put it on GitHub or another place
> >> folks can 'git clone' and then 'make'.
> >
> >
> > git clone https://github.com/JamesRead5737/libcurlmemoryleak.git
> >
> > No need to make. Just compile with gcc crawler.c -g -lssl -lcurl
> > Run valgrind with valgrind -v --tool=memcheck --leak-check=full
> --show-reachable=yes --track-origins=yes --log-file=memcheck.log ./a.out
> >
> > This should reproduce what I've been talking about.
>
> The program has been running for about 10 minutes and it still has not
> ended:
>
> Parsed sites: 0, 367 parallel connections, 364 still running
>

Just press ctrl c and the program will exit. That's what the signal handler
function is for.

>
> I think you should reduce the program to a minimal reproducer.
>
> This does not look correct (to me) in new_conn:
>
> conn->easy = curl_easy_init();
>
> I thought you should only call curl_easy_init() once for the
> application, not each thread(?).
>

The application is single threaded. The code follows quite closely the
example at https://curl.haxx.se/libcurl/c/ephiperfifo.html which does the
exact same thing.

James Read

>
> Jeff
>

-------------------------------------------------------------------
Unsubscribe: https://cool.haxx.se/list/listinfo/curl-library
Etiquette: https://curl.haxx.se/mail/etiquette.html
Received on 2020-05-26