linux-kernel - Re: [RFC PATCH v2 03/10] perf workqueue: add threadpool start and stop functions

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <CAM9d7cg-AF4u=MPJwBs5QSbT_RgEFXYyum97yu3gy0+2w16d6g@mail.gmail.com>
Date:   Fri, 6 Aug 2021 19:43:09 -0700
From:   Namhyung Kim <namhyung@...nel.org>
To:     Riccardo Mancini <rickyman7@...il.com>
Cc:     Arnaldo Carvalho de Melo <acme@...nel.org>,
        Ian Rogers <irogers@...gle.com>,
        Peter Zijlstra <peterz@...radead.org>,
        Ingo Molnar <mingo@...hat.com>,
        Mark Rutland <mark.rutland@....com>,
        Jiri Olsa <jolsa@...hat.com>,
        linux-kernel <linux-kernel@...r.kernel.org>,
        linux-perf-users <linux-perf-users@...r.kernel.org>,
        Alexey Bayduraev <alexey.v.bayduraev@...ux.intel.com>
Subject: Re: [RFC PATCH v2 03/10] perf workqueue: add threadpool start and
 stop functions

On Fri, Jul 30, 2021 at 8:34 AM Riccardo Mancini <rickyman7@...il.com> wrote:
>
> This patch adds the start and stop functions, alongside the thread
> function.
> Each thread will run until a stop signal is received.
> Furthermore, start and stop are added to the test.
>
> Thread management is based on the prototype from Alexey:
> https://lore.kernel.org/lkml/cover.1625227739.git.alexey.v.bayduraev@linux.intel.com/
>
> Suggested-by: Alexey Bayduraev <alexey.v.bayduraev@...ux.intel.com>
> Signed-off-by: Riccardo Mancini <rickyman7@...il.com>
> ---

[SNIP]
> @@ -93,6 +134,130 @@ static void threadpool_entry__close_pipes(struct threadpool_entry *thread)
>         }
>  }
>
> +/**
> + * threadpool__wait_thread - receive ack from thread
> + *
> + * NB: call only from main thread!
> + */
> +static int threadpool__wait_thread(struct threadpool_entry *thread)

If you wanted to differentiate APIs for main thread, I think it's better
to pass the pool struct (according to the name) and the index like:

  int threadpool__wait_thread(struct threadpool *pool, int idx)

Then it can get a pointer to the entry easily.

> +{
> +       int res;
> +       enum threadpool_msg msg = THREADPOOL_MSG__UNDEFINED;
> +
> +       res = readn(thread->pipes.ack[0], &msg, sizeof(msg));
> +       if (res < 0) {
> +               pr_debug2("threadpool: failed to recv msg from tid=%d: %s\n",
> +                      thread->tid, strerror(errno));
> +               return -THREADPOOL_ERROR__READPIPE;
> +       }
> +       if (msg != THREADPOOL_MSG__ACK) {
> +               pr_debug2("threadpool: received unexpected msg from tid=%d: %s\n",
> +                      thread->tid, threadpool_msg_tags[msg]);
> +               return -THREADPOOL_ERROR__INVALIDMSG;
> +       }
> +
> +       pr_debug2("threadpool: received ack from tid=%d\n", thread->tid);
> +
> +       return 0;
> +}
> +
> +/**
> + * threadpool__terminate_thread - send stop signal to thread and wait for ack
> + *
> + * NB: call only from main thread!
> + */
> +static int threadpool__terminate_thread(struct threadpool_entry *thread)

Ditto.

> +{
> +       int res;
> +       enum threadpool_msg msg = THREADPOOL_MSG__STOP;
> +
> +       res = writen(thread->pipes.cmd[1], &msg, sizeof(msg));
> +       if (res < 0) {
> +               pr_debug2("threadpool: error sending stop msg to tid=%d: %s\n",
> +                       thread->tid, strerror(errno));
> +               return -THREADPOOL_ERROR__WRITEPIPE;
> +       }
> +
> +       return threadpool__wait_thread(thread);
> +}
> +

[SNIP]
> @@ -161,12 +326,30 @@ struct threadpool *threadpool__new(int n_threads)
>   *
>   * Buffer size should be at least THREADPOOL_STRERR_BUFSIZE bytes.
>   */
> -int threadpool__strerror(struct threadpool *pool __maybe_unused, int err, char *buf, size_t size)
> +int threadpool__strerror(struct threadpool *pool, int err, char *buf, size_t size)
>  {
>         char sbuf[STRERR_BUFSIZE], *emsg;
> +       const char *status_str, *errno_str;
>
> -       emsg = str_error_r(err, sbuf, sizeof(sbuf));
> -       return scnprintf(buf, size, "Error: %s.\n", emsg);
> +       status_str = IS_ERR_OR_NULL(pool) ? "error" : threadpool_status_tags[pool->status];
> +
> +       switch (err) {
> +       case -THREADPOOL_ERROR__SIGPROCMASK:
> +       case -THREADPOOL_ERROR__READPIPE:
> +       case -THREADPOOL_ERROR__WRITEPIPE:
> +               emsg = str_error_r(errno, sbuf, sizeof(sbuf));
> +               errno_str = threadpool_errno_str[-err-THREADPOOL_ERROR__OFFSET];
> +               return scnprintf(buf, size, "%s: %s.\n", errno_str, emsg);
> +       case -THREADPOOL_ERROR__INVALIDMSG:
> +               errno_str = threadpool_errno_str[-err-THREADPOOL_ERROR__OFFSET];
> +               return scnprintf(buf, size, "%s.\n", errno_str);
> +       case -THREADPOOL_ERROR__NOTALLOWED:
> +               return scnprintf(buf, size, "%s (%s).\n",
> +                       threadpool_errno_str[-err], status_str);

s/-err/-err-THREADPOOL_ERROR__OFFSET/ ?

It'd be nice if you calculate the index once.

> +       default:
> +               emsg = str_error_r(err, sbuf, sizeof(sbuf));

I'm confused whether the 'err' is negative or positive?

Thanks,
Namhyung


> +               return scnprintf(buf, size, "Error: %s", emsg);
> +       }
>  }
>