Linux filesystem development
 help / color / mirror / Atom feed
From: Christian Brauner <brauner@kernel.org>
To: NeilBrown <neil@brown.name>, Oleg Nesterov <oleg@redhat.com>,
	 linux-fsdevel@vger.kernel.org
Cc: Alexander Viro <viro@zeniv.linux.org.uk>, Jan Kara <jack@suse.cz>,
	 Xin Zhao <jackzxcui1989@163.com>,
	Mateusz Guzik <mjguzik@gmail.com>,
	 Jeff Layton <jlayton@kernel.org>, Jens Axboe <axboe@kernel.dk>,
	 "Christian Brauner (Amutable)" <brauner@kernel.org>
Subject: [PATCH RFC v2 10/16] fs: make close_range() synchronous
Date: Wed, 02 Sep 2026 17:58:28 +0200	[thread overview]
Message-ID: <20260902-work-coredump-unlock-self-v2-10-1bece368cbb1@kernel.org> (raw)
In-Reply-To: <20260902-work-coredump-unlock-self-v2-0-1bece368cbb1@kernel.org>

__range_close() closes through filp_close() so every file the caller
held the last reference to is punted to task work. That costs one
cmpxchg per file plus a list entry for any later task_work_cancel() to
search under ->pi_lock. close_range(2) exists to close many descriptors
in one go fast. So convert it to the same synchronous treatment as
close(2) and close_files().

We collect the last references through filp_close_list() and run any
final __fput()s through fput_list() right before returning.
close_range(2) now behaves like close(2).

Signed-off-by: Christian Brauner (Amutable) <brauner@kernel.org>
---
 fs/file.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/fs/file.c b/fs/file.c
index 47ce844be95b..608e6a1fa69b 100644
--- a/fs/file.c
+++ b/fs/file.c
@@ -798,7 +798,7 @@ static inline void __range_cloexec(struct files_struct *cur_fds,
 }
 
 static inline void __range_close(struct files_struct *files, unsigned int fd,
-				 unsigned int max_fd)
+				 unsigned int max_fd, struct llist_head *list)
 {
 	struct file *file;
 	struct fdtable *fdt;
@@ -815,7 +815,8 @@ static inline void __range_close(struct files_struct *files, unsigned int fd,
 		file = file_close_fd_locked(files, fd);
 		if (file) {
 			spin_unlock(&files->file_lock);
-			filp_close(file, files);
+			filp_flush(file, files);
+			fput_close_list(file, list);
 			cond_resched();
 			spin_lock(&files->file_lock);
 			fdt = files_fdtable(files);
@@ -845,6 +846,7 @@ SYSCALL_DEFINE3(close_range, unsigned int, fd, unsigned int, max_fd,
 {
 	struct task_struct *me = current;
 	struct files_struct *cur_fds = me->files, *fds = NULL;
+	LLIST_HEAD(to_close);
 
 	if (flags & ~(CLOSE_RANGE_UNSHARE | CLOSE_RANGE_CLOEXEC))
 		return -EINVAL;
@@ -876,7 +878,7 @@ SYSCALL_DEFINE3(close_range, unsigned int, fd, unsigned int, max_fd,
 	if (flags & CLOSE_RANGE_CLOEXEC)
 		__range_cloexec(cur_fds, fd, max_fd);
 	else
-		__range_close(cur_fds, fd, max_fd);
+		__range_close(cur_fds, fd, max_fd, &to_close);
 
 	if (fds) {
 		/*
@@ -886,6 +888,7 @@ SYSCALL_DEFINE3(close_range, unsigned int, fd, unsigned int, max_fd,
 		put_files_struct(switch_files_struct(me, cur_fds));
 	}
 
+	fput_list(&to_close);
 	return 0;
 }
 

-- 
2.53.0


  parent reply	other threads:[~2026-09-02 15:59 UTC|newest]

Thread overview: 24+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-09-02 15:58 [PATCH RFC v2 00/16] coredump, files: exit files on request Christian Brauner
2026-09-02 15:58 ` [PATCH RFC v2 01/16] fs: don't open-code file_close_fd() in close_fd() Christian Brauner
2026-09-02 15:58 ` [PATCH RFC v2 02/16] fs: add fput_close_list() and fput_list() Christian Brauner
2026-09-02 17:41   ` Mateusz Guzik
2026-09-02 17:58     ` Mateusz Guzik
2026-09-04  8:22       ` Christian Brauner
2026-09-02 15:58 ` [PATCH RFC v2 03/16] fs: rename do_close_on_exec() to close_cloexec_files() Christian Brauner
2026-09-02 17:41   ` Mateusz Guzik
2026-09-02 15:58 ` [PATCH RFC v2 04/16] fs: defer the final fput of close-on-exec files past the exec locks Christian Brauner
2026-09-02 17:50   ` Mateusz Guzik
2026-09-02 15:58 ` [PATCH RFC v2 05/16] fs: add switch_files_struct() Christian Brauner
2026-09-02 15:58 ` [PATCH RFC v2 06/16] fs: move unshare_fd() to fs/file.c Christian Brauner
2026-09-02 17:51   ` Mateusz Guzik
2026-09-02 15:58 ` [PATCH RFC v2 07/16] fs: remove unshare_files() Christian Brauner
2026-09-02 17:52   ` Mateusz Guzik
2026-09-02 15:58 ` [PATCH RFC v2 08/16] exec: release fdtable after exec locks drop Christian Brauner
2026-09-02 15:58 ` [PATCH RFC v2 09/16] fs: make close_files() synchronous Christian Brauner
2026-09-02 15:58 ` Christian Brauner [this message]
2026-09-02 15:58 ` [PATCH RFC v2 11/16] coredump: s/startup/done/g Christian Brauner
2026-09-02 15:58 ` [PATCH RFC v2 12/16] coredump: factor out coredump_wait_inactive() Christian Brauner
2026-09-02 15:58 ` [PATCH RFC v2 13/16] fs: add alloc_files_struct() Christian Brauner
2026-09-02 15:58 ` [PATCH RFC v2 14/16] coredump: add COREDUMP_CLOSE_FILES Christian Brauner
2026-09-02 15:58 ` [PATCH RFC v2 15/16] tools: sync coredump.h header Christian Brauner
2026-09-02 15:58 ` [PATCH RFC v2 16/16] selftests/coredump: test COREDUMP_CLOSE_FILES Christian Brauner

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=20260902-work-coredump-unlock-self-v2-10-1bece368cbb1@kernel.org \
    --to=brauner@kernel.org \
    --cc=axboe@kernel.dk \
    --cc=jack@suse.cz \
    --cc=jackzxcui1989@163.com \
    --cc=jlayton@kernel.org \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=mjguzik@gmail.com \
    --cc=neil@brown.name \
    --cc=oleg@redhat.com \
    --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