From: Carlos O'Donell <carlos@redhat.com>
To: Adhemerval Zanella <adhemerval.zanella@linaro.org>,
libc-alpha@sourceware.org
Subject: Re: [PATCH v4 6/6] Linux: optimize clone3 internal usage
Date: Wed, 18 Jan 2023 17:31:35 -0500 [thread overview]
Message-ID: <ed2d2e09-e545-9934-76e9-5aec3c4e17a2@redhat.com> (raw)
In-Reply-To: <20230112135853.3786675-7-adhemerval.zanella@linaro.org>
On 1/12/23 08:58, Adhemerval Zanella wrote:
> Add an optimization to avoid calling clone3 when glibc detects that
> there is no kernel support. It also adds __ASSUME_CLONE3, which allows
> skipping this optimization and issuing the clone3 syscall directly.
>
> It does not handle the the small window between 5.3 and 5.5 for
> posix_spawn (CLONE_CLEAR_SIGHAND was added in 5.5).
>
> Checked on x86_64-linux-gnu.
OK for 2.28.
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
> ---
> include/clone_internal.h | 5 +++++
> sysdeps/unix/sysv/linux/clone-internal.c | 24 ++++++++++++++++++++++-
> sysdeps/unix/sysv/linux/kernel-features.h | 9 +++++++++
> 3 files changed, 37 insertions(+), 1 deletion(-)
>
> diff --git a/include/clone_internal.h b/include/clone_internal.h
> index 3b6cd85f02..dd380f119e 100644
> --- a/include/clone_internal.h
> +++ b/include/clone_internal.h
> @@ -24,6 +24,11 @@ extern int __clone3 (struct clone_args *__cl_args, size_t __size,
> fall back to clone or clone2. */
> extern int __clone_internal (struct clone_args *__cl_args,
> int (*__func) (void *__arg), void *__arg);
> +/* clone3 wrapper with a sticky check to avoid re-issuing the syscall if
> + it fails with ENOSYS. */
> +extern int __clone3_internal (struct clone_args *cl_args,
> + int (*func) (void *args), void *arg)
> + attribute_hidden;
> /* The fallback code which calls clone/clone2 based on clone3 arguments. */
> extern int __clone_internal_fallback (struct clone_args *__cl_args,
> int (*__func) (void *__arg),
> diff --git a/sysdeps/unix/sysv/linux/clone-internal.c b/sysdeps/unix/sysv/linux/clone-internal.c
> index e125c9125e..790739cfce 100644
> --- a/sysdeps/unix/sysv/linux/clone-internal.c
> +++ b/sysdeps/unix/sysv/linux/clone-internal.c
> @@ -76,6 +76,28 @@ __clone_internal_fallback (struct clone_args *cl_args,
> return ret;
> }
>
> +int
> +__clone3_internal (struct clone_args *cl_args, int (*func) (void *args),
> + void *arg)
> +{
> +#ifdef HAVE_CLONE3_WRAPPER
> +# if __ASSUME_CLONE3
> + return __clone3 (cl_args, sizeof (*cl_args), func, arg);
> +# else
> + static int clone3_supported = 1;
> + if (atomic_load_relaxed (&clone3_supported) == 1)
> + {
> + int ret = __clone3 (cl_args, sizeof (*cl_args), func, arg);
> + if (ret != -1 || errno != ENOSYS)
> + return ret;
> +
> + atomic_store_relaxed (&clone3_supported, 0);
> + }
> +# endif
> +#endif
> + __set_errno (ENOSYS);
> + return -1;
> +}
OK.
>
> int
> __clone_internal (struct clone_args *cl_args,
> @@ -83,7 +105,7 @@ __clone_internal (struct clone_args *cl_args,
> {
> #ifdef HAVE_CLONE3_WRAPPER
> int saved_errno = errno;
> - int ret = __clone3 (cl_args, sizeof (*cl_args), func, arg);
> + int ret = __clone3_internal (cl_args, func, arg);
> if (ret != -1 || errno != ENOSYS)
> return ret;
>
> diff --git a/sysdeps/unix/sysv/linux/kernel-features.h b/sysdeps/unix/sysv/linux/kernel-features.h
> index a23d0079a1..4b5c4afbc1 100644
> --- a/sysdeps/unix/sysv/linux/kernel-features.h
> +++ b/sysdeps/unix/sysv/linux/kernel-features.h
> @@ -241,4 +241,13 @@
> # define __ASSUME_FUTEX_LOCK_PI2 0
> #endif
>
> +/* The clone3 system call was introduced across on most architectures in
> + Linux 5.3. Not all ports implements it, so it should be used along
> + HAVE_CLONE3_WRAPPER define. */
> +#if __LINUX_KERNEL_VERSION >= 0x050300
> +# define __ASSUME_CLONE3 1
> +#else
> +# define __ASSUME_CLONE3 0
> +#endif
> +
> #endif /* kernel-features.h */
--
Cheers,
Carlos.
next prev parent reply other threads:[~2023-01-18 22:31 UTC|newest]
Thread overview: 24+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-01-12 13:58 [PATCH v4 0/6] Optimize posix_spawn signal setup with clone3 Adhemerval Zanella
2023-01-12 13:58 ` [PATCH v4 1/6] linux: Do not reset signal handler in posix_spawn if it is already SIG_DFL Adhemerval Zanella
2023-01-18 22:24 ` Carlos O'Donell
2023-01-12 13:58 ` [PATCH v4 2/6] linux: Extend internal clone3 documentation Adhemerval Zanella
2023-01-18 22:25 ` Carlos O'Donell
2023-01-12 13:58 ` [PATCH v4 3/6] Linux: Do not align the stack for __clone3 Adhemerval Zanella
2023-01-18 22:26 ` Carlos O'Donell
2023-01-12 13:58 ` [PATCH v4 4/6] linux: Add clone3 CLONE_CLEAR_SIGHAND optimization to posix_spawn Adhemerval Zanella
2023-01-18 22:28 ` Carlos O'Donell
2023-01-12 13:58 ` [PATCH v4 5/6] aarch64: Add the clone3 wrapper Adhemerval Zanella
2023-01-18 22:29 ` Carlos O'Donell
2023-01-12 13:58 ` [PATCH v4 6/6] Linux: optimize clone3 internal usage Adhemerval Zanella
2023-01-18 22:31 ` Carlos O'Donell [this message]
2023-01-18 17:53 ` [PATCH v4 0/6] Optimize posix_spawn signal setup with clone3 Adhemerval Zanella Netto
2023-01-18 22:24 ` Carlos O'Donell
2023-01-19 14:39 ` Adhemerval Zanella Netto
2023-01-22 23:21 ` Carlos O'Donell
2023-02-28 16:34 ` H.J. Lu
2023-02-28 17:11 ` Adhemerval Zanella Netto
2023-02-28 18:09 ` Florian Weimer
2023-02-28 18:16 ` Adhemerval Zanella Netto
2023-02-28 19:07 ` H.J. Lu
2023-02-28 21:19 ` Andreas Schwab
2023-02-28 21:31 ` H.J. Lu
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=ed2d2e09-e545-9934-76e9-5aec3c4e17a2@redhat.com \
--to=carlos@redhat.com \
--cc=adhemerval.zanella@linaro.org \
--cc=libc-alpha@sourceware.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).