netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Simon Horman <horms@kernel.org>
To: Al Viro <viro@zeniv.linux.org.uk>
Cc: linux-fsdevel@vger.kernel.org, brauner@kernel.org,
	cgroups@vger.kernel.org, kvm@vger.kernel.org,
	netdev@vger.kernel.org, torvalds@linux-foundation.org
Subject: Re: [PATCH v3 01/28] net/socket.c: switch to CLASS(fd)
Date: Sat, 2 Nov 2024 12:21:32 +0000	[thread overview]
Message-ID: <20241102122132.GH1838431@kernel.org> (raw)
In-Reply-To: <20241102050827.2451599-1-viro@zeniv.linux.org.uk>

On Sat, Nov 02, 2024 at 05:07:59AM +0000, Al Viro wrote:
> 	The important part in sockfd_lookup_light() is avoiding needless
> file refcount operations, not the marginal reduction of the register
> pressure from not keeping a struct file pointer in the caller.
> 
> 	Switch to use fdget()/fdpu(); with sane use of CLASS(fd) we can
> get a better code generation...
> 
> 	Would be nice if somebody tested it on networking test suites
> (including benchmarks)...
> 
> 	sockfd_lookup_light() does fdget(), uses sock_from_file() to
> get the associated socket and returns the struct socket reference to
> the caller, along with "do we need to fput()" flag.  No matching fdput(),
> the caller does its equivalent manually, using the fact that sock->file
> points to the struct file the socket has come from.
> 
> 	Get rid of that - have the callers do fdget()/fdput() and
> use sock_from_file() directly.  That kills sockfd_lookup_light()
> and fput_light() (no users left).
> 
> 	What's more, we can get rid of explicit fdget()/fdput() by
> switching to CLASS(fd, ...) - code generation does not suffer, since
> now fdput() inserted on "descriptor is not opened" failure exit
> is recognized to be a no-op by compiler.
> 
> Reviewed-by: Christian Brauner <brauner@kernel.org>
> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

...

> diff --git a/net/socket.c b/net/socket.c

...

> @@ -2926,16 +2900,18 @@ static int do_recvmmsg(int fd, struct mmsghdr __user *mmsg,
>  
>  	datagrams = 0;
>  
> -	sock = sockfd_lookup_light(fd, &err, &fput_needed);
> -	if (!sock)
> -		return err;
> +	CLASS(fd, f)(fd);
> +
> +	if (fd_empty(f))
> +		return -EBADF;
> +	sock = sock_from_file(fd_file(f));
> +	if (unlikely(!sock))
> +		return -ENOTSOCK;

Hi Al,

There is an unconditional check on err down on line 2977.
However, with the above change err is now only conditionally
set before we reach that line. Are you sure that it will always
be initialised by the time line 2977 is reached?

...

  parent reply	other threads:[~2024-11-02 12:21 UTC|newest]

Thread overview: 134+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-07-30  5:09 [PATCHSET][RFC] struct fd and memory safety Al Viro
2024-07-30  5:15 ` [PATCH 01/39] memcg_write_event_control(): fix a user-triggerable oops viro
2024-07-30  5:15   ` [PATCH 02/39] introduce fd_file(), convert all accessors to it viro
2024-08-07  9:55     ` Christian Brauner
2024-07-30  5:15   ` [PATCH 03/39] struct fd: representation change viro
2024-07-30 18:10     ` Josef Bacik
2024-08-07 10:07       ` Christian Brauner
2024-08-07 10:03     ` Christian Brauner
2024-07-30  5:15   ` [PATCH 04/39] add struct fd constructors, get rid of __to_fd() viro
2024-08-07 10:09     ` Christian Brauner
2024-07-30  5:15   ` [PATCH 05/39] regularize emptiness checks in fini_module(2) and vfs_dedupe_file_range() viro
2024-08-07 10:10     ` Christian Brauner
2024-07-30  5:15   ` [PATCH 06/39] net/socket.c: switch to CLASS(fd) viro
2024-08-07 10:13     ` Christian Brauner
2024-07-30  5:15   ` [PATCH 07/39] introduce struct fderr, convert overlayfs uses to that viro
2024-07-30  5:15   ` [PATCH 08/39] experimental: convert fs/overlayfs/file.c to CLASS(...) viro
2024-07-30 19:10     ` Josef Bacik
2024-07-30 21:12       ` Al Viro
2024-07-31 21:11         ` Josef Bacik
2024-08-07 10:23     ` Christian Brauner
2024-07-30  5:15   ` [PATCH 09/39] timerfd: switch to CLASS(fd, ...) viro
2024-08-07 10:24     ` Christian Brauner
2024-07-30  5:15   ` [PATCH 10/39] get rid of perf_fget_light(), convert kernel/events/core.c to CLASS(fd) viro
2024-08-07 10:25     ` Christian Brauner
2024-07-30  5:15   ` [PATCH 11/39] switch netlink_getsockbyfilp() to taking descriptor viro
2024-08-07 10:26     ` Christian Brauner
2024-07-30  5:15   ` [PATCH 12/39] do_mq_notify(): saner skb freeing on failures viro
2024-07-30  5:15   ` [PATCH 13/39] do_mq_notify(): switch to CLASS(fd, ...) viro
2024-08-07 10:27     ` Christian Brauner
2024-07-30  5:16   ` [PATCH 14/39] simplify xfs_find_handle() a bit viro
2024-07-30  5:16   ` [PATCH 15/39] convert vmsplice() to CLASS(fd, ...) viro
2024-08-07 10:27     ` Christian Brauner
2024-07-30  5:16   ` [PATCH 16/39] convert __bpf_prog_get() " viro
2024-08-06 21:08     ` Andrii Nakryiko
2024-08-07 10:28     ` Christian Brauner
2024-07-30  5:16   ` [PATCH 17/39] bpf: resolve_pseudo_ldimm64(): take handling of a single ldimm64 insn into helper viro
2024-08-06 22:32     ` Andrii Nakryiko
2024-08-07 10:29       ` Christian Brauner
2024-08-07 15:30         ` Andrii Nakryiko
2024-08-08 16:51           ` Alexei Starovoitov
2024-08-08 20:35             ` Andrii Nakryiko
2024-08-09  1:23               ` Alexei Starovoitov
2024-08-09 17:23                 ` Andrii Nakryiko
2024-08-10  3:29             ` Al Viro
2024-08-12 20:05               ` Andrii Nakryiko
2024-08-13  2:06                 ` Al Viro
2024-08-13  3:32                   ` Andrii Nakryiko
2024-07-30  5:16   ` [PATCH 18/39] bpf maps: switch to CLASS(fd, ...) viro
2024-08-07 10:34     ` Christian Brauner
2024-07-30  5:16   ` [PATCH 19/39] fdget_raw() users: switch to CLASS(fd_raw, ...) viro
2024-08-07 10:35     ` Christian Brauner
2024-07-30  5:16   ` [PATCH 20/39] introduce "fd_pos" class, convert fdget_pos() users to it viro
2024-08-07 10:36     ` Christian Brauner
2024-07-30  5:16   ` [PATCH 21/39] o2hb_region_dev_store(): avoid goto around fdget()/fdput() viro
2024-07-30  5:16   ` [PATCH 22/39] privcmd_ioeventfd_assign(): don't open-code eventfd_ctx_fdget() viro
2024-07-30  5:16   ` [PATCH 23/39] fdget(), trivial conversions viro
2024-08-07 10:37     ` Christian Brauner
2024-07-30  5:16   ` [PATCH 24/39] fdget(), more " viro
2024-08-07 10:39     ` Christian Brauner
2024-07-30  5:16   ` [PATCH 25/39] convert do_preadv()/do_pwritev() viro
2024-08-07 10:39     ` Christian Brauner
2024-07-30  5:16   ` [PATCH 26/39] convert cachestat(2) viro
2024-08-07 10:39     ` Christian Brauner
2024-07-30  5:16   ` [PATCH 27/39] switch spufs_calls_{get,put}() to CLASS() use viro
2024-07-30  5:16   ` [PATCH 28/39] convert spu_run(2) viro
2024-08-07 10:40     ` Christian Brauner
2024-07-30  5:16   ` [PATCH 29/39] convert media_request_get_by_fd() viro
2024-08-07 10:40     ` Christian Brauner
2024-07-30  5:16   ` [PATCH 30/39] convert coda_parse_fd() viro
2024-08-07 10:41     ` Christian Brauner
2024-07-30  5:16   ` [PATCH 31/39] convert cifs_ioctl_copychunk() viro
2024-08-07 10:41     ` Christian Brauner
2024-07-30  5:16   ` [PATCH 32/39] convert vfs_dedupe_file_range() viro
2024-08-07 10:42     ` Christian Brauner
2024-07-30  5:16   ` [PATCH 33/39] convert do_select() viro
2024-08-07 10:42     ` Christian Brauner
2024-07-30  5:16   ` [PATCH 34/39] do_pollfd(): convert to CLASS(fd) viro
2024-08-07 10:43     ` Christian Brauner
2024-07-30  5:16   ` [PATCH 35/39] convert bpf_token_create() viro
2024-08-06 22:42     ` Andrii Nakryiko
2024-08-10  3:46       ` Al Viro
2024-08-12 20:06         ` Andrii Nakryiko
2024-08-07 10:44     ` Christian Brauner
2024-07-30  5:16   ` [PATCH 36/39] assorted variants of irqfd setup: convert to CLASS(fd) viro
2024-08-07 10:46     ` Christian Brauner
2024-08-10  3:53       ` Al Viro
2024-07-30  5:16   ` [PATCH 37/39] memcg_write_event_control(): switch " viro
2024-08-07 10:47     ` Christian Brauner
2024-07-30  5:16   ` [PATCH 38/39] css_set_fork(): switch to CLASS(fd_raw, ...) viro
2024-08-07 10:47     ` Christian Brauner
2024-07-30  5:16   ` [PATCH 39/39] deal with the last remaing boolean uses of fd_file() viro
2024-08-07 10:48     ` Christian Brauner
2024-07-30  7:13   ` [PATCH 01/39] memcg_write_event_control(): fix a user-triggerable oops Michal Hocko
2024-07-30  7:18     ` Al Viro
2024-07-30  7:37       ` Michal Hocko
2024-07-30  5:17 ` [PATCHSET][RFC] struct fd and memory safety Al Viro
2024-07-30 20:02 ` Josef Bacik
2024-07-31  0:43 ` Al Viro
2024-08-06 17:58 ` Jason Gunthorpe
2024-08-06 18:56   ` Al Viro
2024-08-07 10:51 ` Christian Brauner
2024-11-02  5:02 ` [PATCHSET v3] " Al Viro
2024-11-02  5:07   ` [PATCH v3 01/28] net/socket.c: switch to CLASS(fd) Al Viro
2024-11-02  5:08     ` [PATCH v3 02/28] regularize emptiness checks in fini_module(2) and vfs_dedupe_file_range() Al Viro
2024-11-02  5:08     ` [PATCH v3 03/28] timerfd: switch to CLASS(fd) Al Viro
2024-11-02  5:08     ` [PATCH v3 04/28] get rid of perf_fget_light(), convert kernel/events/core.c " Al Viro
2024-11-02  5:08     ` [PATCH v3 05/28] switch netlink_getsockbyfilp() to taking descriptor Al Viro
2024-11-02  5:08     ` [PATCH v3 06/28] do_mq_notify(): saner skb freeing on failures Al Viro
2024-11-02  5:08     ` [PATCH v3 07/28] do_mq_notify(): switch to CLASS(fd) Al Viro
2024-11-02  5:08     ` [PATCH v3 08/28] simplify xfs_find_handle() a bit Al Viro
2024-11-02  5:08     ` [PATCH v3 09/28] convert vmsplice() to CLASS(fd) Al Viro
2024-11-02  5:08     ` [PATCH v3 10/28] fdget_raw() users: switch to CLASS(fd_raw) Al Viro
2024-11-02  5:08     ` [PATCH v3 11/28] introduce "fd_pos" class, convert fdget_pos() users to it Al Viro
2024-11-02  5:08     ` [PATCH v3 12/28] o2hb_region_dev_store(): avoid goto around fdget()/fdput() Al Viro
2024-11-02  5:08     ` [PATCH v3 13/28] privcmd_ioeventfd_assign(): don't open-code eventfd_ctx_fdget() Al Viro
2024-11-02  5:08     ` [PATCH v3 14/28] fdget(), trivial conversions Al Viro
2024-11-11 17:22       ` Francesco Lavra
2024-11-02  5:08     ` [PATCH v3 15/28] fdget(), more " Al Viro
2024-11-02  5:08     ` [PATCH v3 16/28] convert do_preadv()/do_pwritev() Al Viro
2024-11-02  5:08     ` [PATCH v3 17/28] convert cachestat(2) Al Viro
2024-11-02  5:08     ` [PATCH v3 18/28] switch spufs_calls_{get,put}() to CLASS() use Al Viro
2024-11-02  5:08     ` [PATCH v3 19/28] convert spu_run(2) Al Viro
2024-11-02  5:08     ` [PATCH v3 20/28] convert media_request_get_by_fd() Al Viro
2024-11-02  5:08     ` [PATCH v3 21/28] convert cifs_ioctl_copychunk() Al Viro
2024-11-02  5:08     ` [PATCH v3 22/28] convert vfs_dedupe_file_range() Al Viro
2024-11-02  5:08     ` [PATCH v3 23/28] convert do_select() Al Viro
2024-11-02  5:08     ` [PATCH v3 24/28] do_pollfd(): convert to CLASS(fd) Al Viro
2024-11-02  5:08     ` [PATCH v3 25/28] assorted variants of irqfd setup: " Al Viro
2024-11-02  5:08     ` [PATCH v3 26/28] memcg_write_event_control(): switch " Al Viro
2024-11-02  5:08     ` [PATCH v3 27/28] css_set_fork(): switch to CLASS(fd_raw, ...) Al Viro
2024-11-02  5:08     ` [PATCH v3 28/28] deal with the last remaing boolean uses of fd_file() Al Viro
2024-11-02 12:21     ` Simon Horman [this message]
2024-11-03  6:31       ` [PATCH v3 01/28] net/socket.c: switch to CLASS(fd) Al Viro
2024-11-06 10:03         ` Simon Horman

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=20241102122132.GH1838431@kernel.org \
    --to=horms@kernel.org \
    --cc=brauner@kernel.org \
    --cc=cgroups@vger.kernel.org \
    --cc=kvm@vger.kernel.org \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=netdev@vger.kernel.org \
    --cc=torvalds@linux-foundation.org \
    --cc=viro@zeniv.linux.org.uk \
    /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 NNTP newsgroup(s).