From: Jens Axboe <axboe@kernel.dk>
To: io-uring@vger.kernel.org
Subject: [PATCHSET RFC v4] Inherited restrictions and BPF filtering
Date: Fri, 16 Jan 2026 15:38:37 -0700 [thread overview]
Message-ID: <20260116224356.399361-1-axboe@kernel.dk> (raw)
Hi,
Followup to v3 here:
https://lore.kernel.org/io-uring/20260115165244.1037465-1-axboe@kernel.dk/
which has all the gory details. The main new thing in v4 is proper
support for stacking of BPF filters, when a task goes through fork.
These are referenced, and COW'ed if new filters are attempted installed
in the referenced filter.
Performance and functionality is otherwise the same. See the changelog
for more details.
Comments welcome! Kernel branch can be found here:
https://git.kernel.org/pub/scm/linux/kernel/git/axboe/linux.git/log/?h=io_uring-bpf-restrictions
and sits on top of for-7.0/io_uring.
Changes since v3:
- Move fork.c logic to an io_uring helper, io_uring_fork(), rather than
open code it.
- Sort out the filter and filter table references, and drop the patch
that made io_restriction dynamically allocated.
- Drop IORING_REGISTER_RESTRICTIONS_TASK, IORING_REGISTER_RESTRICTIONS
can just get reused with fd == -1 like other "blind" reg opcodes.
- Split IORING_OP_SOCKET filter support patch out as a separate patch
- Fix a few style issues
- Do proper cloning of registered BPF filters, if a task is forked.
These are done COW style, where initially a new task will just
inherit the existing filter with a reference to it. If new filters
are added in the new task, then the existing table is COW'ed.
- Drop ctx->bpf_restricted, just check for the presence of the filter.
- Drop various alloc helpers, not needed with other cleanups.
- Add and improve a bunch of comments.
include/linux/bpf.h | 1 +
include/linux/bpf_types.h | 4 +
include/linux/io_uring.h | 14 +-
include/linux/io_uring_types.h | 16 ++
include/linux/sched.h | 1 +
include/uapi/linux/bpf.h | 1 +
include/uapi/linux/io_uring.h | 52 +++++
io_uring/Makefile | 1 +
io_uring/bpf_filter.c | 372 +++++++++++++++++++++++++++++++++
io_uring/bpf_filter.h | 46 ++++
io_uring/io_uring.c | 27 +++
io_uring/io_uring.h | 1 +
io_uring/net.c | 9 +
io_uring/net.h | 5 +
io_uring/register.c | 73 +++++++
io_uring/tctx.c | 40 +++-
kernel/bpf/syscall.c | 9 +
kernel/fork.c | 5 +
18 files changed, 667 insertions(+), 10 deletions(-)
--
Jens Axboe
next reply other threads:[~2026-01-16 22:43 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-01-16 22:38 Jens Axboe [this message]
2026-01-16 22:38 ` [PATCH 1/5] io_uring: add support for BPF filtering for opcode restrictions Jens Axboe
2026-01-16 22:38 ` [PATCH 2/5] io_uring/net: allow filtering on IORING_OP_SOCKET data Jens Axboe
2026-01-16 22:38 ` [PATCH 3/5] io_uring/bpf_filter: add ref counts to struct io_bpf_filter Jens Axboe
2026-01-16 22:38 ` [PATCH 4/5] io_uring: add task fork hook Jens Axboe
2026-01-16 22:38 ` [PATCH 5/5] io_uring: allow registration of per-task restrictions Jens Axboe
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=20260116224356.399361-1-axboe@kernel.dk \
--to=axboe@kernel.dk \
--cc=io-uring@vger.kernel.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