Skip to contentSkip to conversation

io_uring/net: allow filtering on IORING_OP_CONNECT

PATCH Applied Shouvik Kar 2 participants 4 messages May 12 – May 14 io_uring

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 io_uring/net: allow filtering on IORING_OP_CONNECT

This adds custom filtering for IORING_OP_CONNECT, where the target family is always exposed, and (for AF_INET / AF_INET6) port and address are exposed. port and v4_addr are in network byte order so filter authors can compare against on-wire constants.

Skip population unless addr_len covers the populated fields, to avoid leaking stale io_async_msghdr data on short connects.

---
 include/uapi/linux/io_uring/bpf_filter.h | 16 +++++++++
 io_uring/net.c                           | 41 ++++++++++++++++++++++++
 io_uring/net.h                           |  7 ++++
 io_uring/opdef.c                         |  2 ++
 4 files changed, 66 insertions(+)
include/uapi/linux/io_uring/bpf_filter.h +16 -0
index 1b461d792a7b..ce7d78ab13b3 100644
@@ -27,6 +27,22 @@ struct io_uring_bpf_ctx {
 			__u64	mode;
 			__u64	resolve;
 		} open;
+		/*
+		 * For CONNECT: fields are populated only when addr_len covers
+		 * them; unpopulated fields are zero from the caller-side memset
+		 * in io_uring_populate_bpf_ctx(). port and v4_addr are network
+		 * byte order. Filters may only issue BPF_LD|BPF_W|BPF_ABS at
+		 * 4-byte aligned offsets; load + mask for sub-word fields.
+		 */
+		struct {
+			__u32	family;	/* sa_family_t zero-extended */
+			__be16	port;
+			__u8	pad[2];
+			union {
+				__be32	v4_addr;
+				__u8	v6_addr[16];
+			};
+		} connect;
 	};
 };
 
io_uring/net.c +41 -0

47 lines not on this page · raw message

io_uring/net.h +7 -0
index d4d1ddce50e3..51fda715d3c0 100644
@@ -46,6 +46,7 @@ int io_accept(struct io_kiocb *req, unsigned int issue_flags);
 int io_socket_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe);
 int io_socket(struct io_kiocb *req, unsigned int issue_flags);
 void io_socket_bpf_populate(struct io_uring_bpf_ctx *bctx, struct io_kiocb *req);
+void io_connect_bpf_populate(struct io_uring_bpf_ctx *bctx, struct io_kiocb *req);
 
 int io_connect_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe);
 int io_connect(struct io_kiocb *req, unsigned int issue_flags);
@@ -69,4 +70,10 @@ static inline void io_socket_bpf_populate(struct io_uring_bpf_ctx *bctx,
 					  struct io_kiocb *req)
 {
 }
+
+static inline void io_connect_bpf_populate(struct io_uring_bpf_ctx *bctx,
+					   struct io_kiocb *req)
+{
+}
+
 #endif
io_uring/opdef.c +2 -0
index c3ef52b70811..8ea6bd274607 100644
@@ -203,9 +203,11 @@ const struct io_issue_def io_issue_defs[] = {
 		.unbound_nonreg_file	= 1,
 		.pollout		= 1,
 #if defined(CONFIG_NET)
+		.filter_pdu_size	= sizeof_field(struct io_uring_bpf_ctx, connect),
 		.async_size		= sizeof(struct io_async_msghdr),
 		.prep			= io_connect_prep,
 		.issue			= io_connect,
+		.filter_populate	= io_connect_bpf_populate,
 #else
 		.prep			= io_eopnotsupp_prep,
 #endif
Signature · 1 line
2.53.0
maintainer
#1 lore raw
Quoted 7 lines · Shouvik Kar

This adds custom filtering for IORING_OP_CONNECT, where the target family is always exposed, and (for AF_INET / AF_INET6) port and address are exposed. port and v4_addr are in network byte order so filter authors can compare against on-wire constants.

Skip population unless addr_len covers the populated fields, to avoid leaking stale io_async_msghdr data on short connects.

Looks pretty straight forward to me. Do you have a liburing test case for this too?

Signature · 1 line
Jens Axboe
#2 lore raw

Yes -- just sent: "[PATCH liburing] tests: add cBPF filter tests for IORING_OP_CONNECT"

Regards,
Shouvik

Quoted 15 lines · Jens Axboe
Quoted 7 lines · Shouvik Kar

This adds custom filtering for IORING_OP_CONNECT, where the target family is always exposed, and (for AF_INET / AF_INET6) port and address are exposed. port and v4_addr are in network byte order so filter authors can compare against on-wire constants.

Skip population unless addr_len covers the populated fields, to avoid leaking stale io_async_msghdr data on short connects.

Looks pretty straight forward to me. Do you have a liburing test case for this too?

--
Jens Axboe

maintainer Applied
#3 lore raw
In reply to Shouvik Kar
Quoted 9 lines · Shouvik Kar

This adds custom filtering for IORING_OP_CONNECT, where the target family is always exposed, and (for AF_INET / AF_INET6) port and address are exposed. port and v4_addr are in network byte order so filter authors can compare against on-wire constants.

Skip population unless addr_len covers the populated fields, to avoid leaking stale io_async_msghdr data on short connects.

[...]

Applied, thanks!

[1/1] io_uring/net: allow filtering on IORING_OP_CONNECT

      commit: 899bea8248cee35d54760a5e7d61a76af8e64411

Best regards,

Signature · 1 line
Jens Axboe