As I work my way to unlocked and zero-copy TLS Rx the obvious bugs in the splice_read implementation get harder and harder to ignore. This is to say the fixes here are discovered by code inspection, I'm not aware of anyone actually using splice_read.
Jakub Kicinski (9): selftests: tls: add helper for creating sock pairs selftests: tls: factor out cmsg send/receive selftests: tls: add tests for handling of bad records tls: splice_read: fix record type check selftests: tls: test splicing cmsgs tls: splice_read: fix accessing pre-processed records selftests: tls: test splicing decrypted records tls: fix replacing proto_ops selftests: tls: test for correct proto_ops
net/tls/tls_main.c | 47 ++- net/tls/tls_sw.c | 40 ++- tools/testing/selftests/net/tls.c | 521 ++++++++++++++++++++++-------- 3 files changed, 456 insertions(+), 152 deletions(-)
We have the same code 3 times, about to add a fourth copy.
Signed-off-by: Jakub Kicinski kuba@kernel.org --- tools/testing/selftests/net/tls.c | 128 +++++++----------------------- 1 file changed, 29 insertions(+), 99 deletions(-)
diff --git a/tools/testing/selftests/net/tls.c b/tools/testing/selftests/net/tls.c index e61fc4c32ba2..8fb7cf8c4bfb 100644 --- a/tools/testing/selftests/net/tls.c +++ b/tools/testing/selftests/net/tls.c @@ -78,26 +78,21 @@ static void memrnd(void *s, size_t n) *byte++ = rand(); }
-FIXTURE(tls_basic) -{ - int fd, cfd; - bool notls; -}; - -FIXTURE_SETUP(tls_basic) +static void ulp_sock_pair(struct __test_metadata *_metadata, + int *fd, int *cfd, bool *notls) { struct sockaddr_in addr; socklen_t len; int sfd, ret;
- self->notls = false; + *notls = false; len = sizeof(addr);
addr.sin_family = AF_INET; addr.sin_addr.s_addr = htonl(INADDR_ANY); addr.sin_port = 0;
- self->fd = socket(AF_INET, SOCK_STREAM, 0); + *fd = socket(AF_INET, SOCK_STREAM, 0); sfd = socket(AF_INET, SOCK_STREAM, 0);
ret = bind(sfd, &addr, sizeof(addr)); @@ -108,26 +103,37 @@ FIXTURE_SETUP(tls_basic) ret = getsockname(sfd, &addr, &len); ASSERT_EQ(ret, 0);
- ret = connect(self->fd, &addr, sizeof(addr)); + ret = connect(*fd, &addr, sizeof(addr)); ASSERT_EQ(ret, 0);
- self->cfd = accept(sfd, &addr, &len); - ASSERT_GE(self->cfd, 0); + *cfd = accept(sfd, &addr, &len); + ASSERT_GE(*cfd, 0);
close(sfd);
- ret = setsockopt(self->fd, IPPROTO_TCP, TCP_ULP, "tls", sizeof("tls")); + ret = setsockopt(*fd, IPPROTO_TCP, TCP_ULP, "tls", sizeof("tls")); if (ret != 0) { ASSERT_EQ(errno, ENOENT); - self->notls = true; + *notls = true; printf("Failure setting TCP_ULP, testing without tls\n"); return; }
- ret = setsockopt(self->cfd, IPPROTO_TCP, TCP_ULP, "tls", sizeof("tls")); + ret = setsockopt(*cfd, IPPROTO_TCP, TCP_ULP, "tls", sizeof("tls")); ASSERT_EQ(ret, 0); }
+FIXTURE(tls_basic) +{ + int fd, cfd; + bool notls; +}; + +FIXTURE_SETUP(tls_basic) +{ + ulp_sock_pair(_metadata, &self->fd, &self->cfd, &self->notls); +} + FIXTURE_TEARDOWN(tls_basic) { close(self->fd); @@ -199,60 +205,21 @@ FIXTURE_VARIANT_ADD(tls, 13_sm4_ccm) FIXTURE_SETUP(tls) { struct tls_crypto_info_keys tls12; - struct sockaddr_in addr; - socklen_t len; - int sfd, ret; - - self->notls = false; - len = sizeof(addr); + int ret;
tls_crypto_info_init(variant->tls_version, variant->cipher_type, &tls12);
- addr.sin_family = AF_INET; - addr.sin_addr.s_addr = htonl(INADDR_ANY); - addr.sin_port = 0; + ulp_sock_pair(_metadata, &self->fd, &self->cfd, &self->notls);
- self->fd = socket(AF_INET, SOCK_STREAM, 0); - sfd = socket(AF_INET, SOCK_STREAM, 0); - - ret = bind(sfd, &addr, sizeof(addr)); - ASSERT_EQ(ret, 0); - ret = listen(sfd, 10); - ASSERT_EQ(ret, 0); + if (self->notls) + return;
- ret = getsockname(sfd, &addr, &len); + ret = setsockopt(self->fd, SOL_TLS, TLS_TX, &tls12, tls12.len); ASSERT_EQ(ret, 0);
- ret = connect(self->fd, &addr, sizeof(addr)); + ret = setsockopt(self->cfd, SOL_TLS, TLS_RX, &tls12, tls12.len); ASSERT_EQ(ret, 0); - - ret = setsockopt(self->fd, IPPROTO_TCP, TCP_ULP, "tls", sizeof("tls")); - if (ret != 0) { - self->notls = true; - printf("Failure setting TCP_ULP, testing without tls\n"); - } - - if (!self->notls) { - ret = setsockopt(self->fd, SOL_TLS, TLS_TX, &tls12, - tls12.len); - ASSERT_EQ(ret, 0); - } - - self->cfd = accept(sfd, &addr, &len); - ASSERT_GE(self->cfd, 0); - - if (!self->notls) { - ret = setsockopt(self->cfd, IPPROTO_TCP, TCP_ULP, "tls", - sizeof("tls")); - ASSERT_EQ(ret, 0); - - ret = setsockopt(self->cfd, SOL_TLS, TLS_RX, &tls12, - tls12.len); - ASSERT_EQ(ret, 0); - } - - close(sfd); }
FIXTURE_TEARDOWN(tls) @@ -1355,62 +1322,25 @@ TEST(non_established) {
TEST(keysizes) { struct tls12_crypto_info_aes_gcm_256 tls12; - struct sockaddr_in addr; - int sfd, ret, fd, cfd; - socklen_t len; + int ret, fd, cfd; bool notls;
- notls = false; - len = sizeof(addr); - memset(&tls12, 0, sizeof(tls12)); tls12.info.version = TLS_1_2_VERSION; tls12.info.cipher_type = TLS_CIPHER_AES_GCM_256;
- addr.sin_family = AF_INET; - addr.sin_addr.s_addr = htonl(INADDR_ANY); - addr.sin_port = 0; - - fd = socket(AF_INET, SOCK_STREAM, 0); - sfd = socket(AF_INET, SOCK_STREAM, 0); - - ret = bind(sfd, &addr, sizeof(addr)); - ASSERT_EQ(ret, 0); - ret = listen(sfd, 10); - ASSERT_EQ(ret, 0); - - ret = getsockname(sfd, &addr, &len); - ASSERT_EQ(ret, 0); - - ret = connect(fd, &addr, sizeof(addr)); - ASSERT_EQ(ret, 0); - - ret = setsockopt(fd, IPPROTO_TCP, TCP_ULP, "tls", sizeof("tls")); - if (ret != 0) { - notls = true; - printf("Failure setting TCP_ULP, testing without tls\n"); - } + ulp_sock_pair(_metadata, &fd, &cfd, ¬ls);
if (!notls) { ret = setsockopt(fd, SOL_TLS, TLS_TX, &tls12, sizeof(tls12)); EXPECT_EQ(ret, 0); - } - - cfd = accept(sfd, &addr, &len); - ASSERT_GE(cfd, 0); - - if (!notls) { - ret = setsockopt(cfd, IPPROTO_TCP, TCP_ULP, "tls", - sizeof("tls")); - EXPECT_EQ(ret, 0);
ret = setsockopt(cfd, SOL_TLS, TLS_RX, &tls12, sizeof(tls12)); EXPECT_EQ(ret, 0); }
- close(sfd); close(fd); close(cfd); }
Add helpers for sending and receiving special record types.
Signed-off-by: Jakub Kicinski kuba@kernel.org --- tools/testing/selftests/net/tls.c | 111 +++++++++++++++++++----------- 1 file changed, 70 insertions(+), 41 deletions(-)
diff --git a/tools/testing/selftests/net/tls.c b/tools/testing/selftests/net/tls.c index 8fb7cf8c4bfb..642d1d629b28 100644 --- a/tools/testing/selftests/net/tls.c +++ b/tools/testing/selftests/net/tls.c @@ -123,6 +123,65 @@ static void ulp_sock_pair(struct __test_metadata *_metadata, ASSERT_EQ(ret, 0); }
+/* Produce a basic cmsg */ +static int tls_send_cmsg(int fd, unsigned char record_type, + void *data, size_t len, int flags) +{ + char cbuf[CMSG_SPACE(sizeof(char))]; + int cmsg_len = sizeof(char); + struct cmsghdr *cmsg; + struct msghdr msg; + struct iovec vec; + + vec.iov_base = data; + vec.iov_len = len; + memset(&msg, 0, sizeof(struct msghdr)); + msg.msg_iov = &vec; + msg.msg_iovlen = 1; + msg.msg_control = cbuf; + msg.msg_controllen = sizeof(cbuf); + cmsg = CMSG_FIRSTHDR(&msg); + cmsg->cmsg_level = SOL_TLS; + /* test sending non-record types. */ + cmsg->cmsg_type = TLS_SET_RECORD_TYPE; + cmsg->cmsg_len = CMSG_LEN(cmsg_len); + *CMSG_DATA(cmsg) = record_type; + msg.msg_controllen = cmsg->cmsg_len; + + return sendmsg(fd, &msg, flags); +} + +static int tls_recv_cmsg(struct __test_metadata *_metadata, + int fd, unsigned char record_type, + void *data, size_t len, int flags) +{ + char cbuf[CMSG_SPACE(sizeof(char))]; + struct cmsghdr *cmsg; + unsigned char ctype; + struct msghdr msg; + struct iovec vec; + int n; + + vec.iov_base = data; + vec.iov_len = len; + memset(&msg, 0, sizeof(struct msghdr)); + msg.msg_iov = &vec; + msg.msg_iovlen = 1; + msg.msg_control = cbuf; + msg.msg_controllen = sizeof(cbuf); + + n = recvmsg(fd, &msg, flags); + + cmsg = CMSG_FIRSTHDR(&msg); + EXPECT_NE(cmsg, NULL); + EXPECT_EQ(cmsg->cmsg_level, SOL_TLS); + EXPECT_EQ(cmsg->cmsg_type, TLS_GET_RECORD_TYPE); + ctype = *((unsigned char *)CMSG_DATA(cmsg)); + EXPECT_EQ(ctype, record_type); + + return n; +} + FIXTURE(tls_basic) { int fd, cfd; @@ -1160,60 +1219,30 @@ TEST_F(tls, mutliproc_sendpage_writers)
TEST_F(tls, control_msg) { - if (self->notls) - return; - - char cbuf[CMSG_SPACE(sizeof(char))]; - char const *test_str = "test_read"; - int cmsg_len = sizeof(char); + char *test_str = "test_read"; char record_type = 100; - struct cmsghdr *cmsg; - struct msghdr msg; int send_len = 10; - struct iovec vec; char buf[10];
- vec.iov_base = (char *)test_str; - vec.iov_len = 10; - memset(&msg, 0, sizeof(struct msghdr)); - msg.msg_iov = &vec; - msg.msg_iovlen = 1; - msg.msg_control = cbuf; - msg.msg_controllen = sizeof(cbuf); - cmsg = CMSG_FIRSTHDR(&msg); - cmsg->cmsg_level = SOL_TLS; - /* test sending non-record types. */ - cmsg->cmsg_type = TLS_SET_RECORD_TYPE; - cmsg->cmsg_len = CMSG_LEN(cmsg_len); - *CMSG_DATA(cmsg) = record_type; - msg.msg_controllen = cmsg->cmsg_len; + if (self->notls) + SKIP(return, "no TLS support");
- EXPECT_EQ(sendmsg(self->fd, &msg, 0), send_len); + EXPECT_EQ(tls_send_cmsg(self->fd, record_type, test_str, send_len, 0), + send_len); /* Should fail because we didn't provide a control message */ EXPECT_EQ(recv(self->cfd, buf, send_len, 0), -1);
- vec.iov_base = buf; - EXPECT_EQ(recvmsg(self->cfd, &msg, MSG_WAITALL | MSG_PEEK), send_len); - - cmsg = CMSG_FIRSTHDR(&msg); - EXPECT_NE(cmsg, NULL); - EXPECT_EQ(cmsg->cmsg_level, SOL_TLS); - EXPECT_EQ(cmsg->cmsg_type, TLS_GET_RECORD_TYPE); - record_type = *((unsigned char *)CMSG_DATA(cmsg)); - EXPECT_EQ(record_type, 100); + EXPECT_EQ(tls_recv_cmsg(_metadata, self->cfd, record_type, + buf, sizeof(buf), MSG_WAITALL | MSG_PEEK), + send_len); EXPECT_EQ(memcmp(buf, test_str, send_len), 0);
/* Recv the message again without MSG_PEEK */ - record_type = 0; memset(buf, 0, sizeof(buf));
- EXPECT_EQ(recvmsg(self->cfd, &msg, MSG_WAITALL), send_len); - cmsg = CMSG_FIRSTHDR(&msg); - EXPECT_NE(cmsg, NULL); - EXPECT_EQ(cmsg->cmsg_level, SOL_TLS); - EXPECT_EQ(cmsg->cmsg_type, TLS_GET_RECORD_TYPE); - record_type = *((unsigned char *)CMSG_DATA(cmsg)); - EXPECT_EQ(record_type, 100); + EXPECT_EQ(tls_recv_cmsg(_metadata, self->cfd, record_type, + buf, sizeof(buf), MSG_WAITALL), + send_len); EXPECT_EQ(memcmp(buf, test_str, send_len), 0); }
Test broken records.
Signed-off-by: Jakub Kicinski kuba@kernel.org --- tools/testing/selftests/net/tls.c | 154 ++++++++++++++++++++++++++++++ 1 file changed, 154 insertions(+)
diff --git a/tools/testing/selftests/net/tls.c b/tools/testing/selftests/net/tls.c index 642d1d629b28..2108b197d3f6 100644 --- a/tools/testing/selftests/net/tls.c +++ b/tools/testing/selftests/net/tls.c @@ -1297,6 +1297,160 @@ TEST_F(tls, shutdown_reuse) EXPECT_EQ(errno, EISCONN); }
+FIXTURE(tls_err) +{ + int fd, cfd; + int fd2, cfd2; + bool notls; +}; + +FIXTURE_VARIANT(tls_err) +{ + uint16_t tls_version; +}; + +FIXTURE_VARIANT_ADD(tls_err, 12_aes_gcm) +{ + .tls_version = TLS_1_2_VERSION, +}; + +FIXTURE_VARIANT_ADD(tls_err, 13_aes_gcm) +{ + .tls_version = TLS_1_3_VERSION, +}; + +FIXTURE_SETUP(tls_err) +{ + struct tls_crypto_info_keys tls12; + int ret; + + tls_crypto_info_init(variant->tls_version, TLS_CIPHER_AES_GCM_128, + &tls12); + + ulp_sock_pair(_metadata, &self->fd, &self->cfd, &self->notls); + ulp_sock_pair(_metadata, &self->fd2, &self->cfd2, &self->notls); + if (self->notls) + return; + + ret = setsockopt(self->fd, SOL_TLS, TLS_TX, &tls12, tls12.len); + ASSERT_EQ(ret, 0); + + ret = setsockopt(self->cfd2, SOL_TLS, TLS_RX, &tls12, tls12.len); + ASSERT_EQ(ret, 0); +} + +FIXTURE_TEARDOWN(tls_err) +{ + close(self->fd); + close(self->cfd); + close(self->fd2); + close(self->cfd2); +} + +TEST_F(tls_err, bad_rec) +{ + char buf[64]; + + if (self->notls) + SKIP(return, "no TLS support"); + + memset(buf, 0x55, sizeof(buf)); + EXPECT_EQ(send(self->fd2, buf, sizeof(buf), 0), sizeof(buf)); + EXPECT_EQ(recv(self->cfd2, buf, sizeof(buf), 0), -1); + EXPECT_EQ(errno, EMSGSIZE); + EXPECT_EQ(recv(self->cfd2, buf, sizeof(buf), MSG_DONTWAIT), -1); + EXPECT_EQ(errno, EAGAIN); +} + +TEST_F(tls_err, bad_auth) +{ + char buf[128]; + int n; + + if (self->notls) + SKIP(return, "no TLS support"); + + memrnd(buf, sizeof(buf) / 2); + EXPECT_EQ(send(self->fd, buf, sizeof(buf) / 2, 0), sizeof(buf) / 2); + n = recv(self->cfd, buf, sizeof(buf), 0); + EXPECT_GT(n, sizeof(buf) / 2); + + buf[n - 1]++; + + EXPECT_EQ(send(self->fd2, buf, n, 0), n); + EXPECT_EQ(recv(self->cfd2, buf, sizeof(buf), 0), -1); + EXPECT_EQ(errno, EBADMSG); + EXPECT_EQ(recv(self->cfd2, buf, sizeof(buf), 0), -1); + EXPECT_EQ(errno, EBADMSG); +} + +TEST_F(tls_err, bad_in_large_read) +{ + char txt[3][64]; + char cip[3][128]; + char buf[3 * 128]; + int i, n; + + if (self->notls) + SKIP(return, "no TLS support"); + + /* Put 3 records in the sockets */ + for (i = 0; i < 3; i++) { + memrnd(txt[i], sizeof(txt[i])); + EXPECT_EQ(send(self->fd, txt[i], sizeof(txt[i]), 0), + sizeof(txt[i])); + n = recv(self->cfd, cip[i], sizeof(cip[i]), 0); + EXPECT_GT(n, sizeof(txt[i])); + /* Break the third message */ + if (i == 2) + cip[2][n - 1]++; + EXPECT_EQ(send(self->fd2, cip[i], n, 0), n); + } + + /* We should be able to receive the first two messages */ + EXPECT_EQ(recv(self->cfd2, buf, sizeof(buf), 0), sizeof(txt[0]) * 2); + EXPECT_EQ(memcmp(buf, txt[0], sizeof(txt[0])), 0); + EXPECT_EQ(memcmp(buf + sizeof(txt[0]), txt[1], sizeof(txt[1])), 0); + /* Third mesasge is bad */ + EXPECT_EQ(recv(self->cfd2, buf, sizeof(buf), 0), -1); + EXPECT_EQ(errno, EBADMSG); + EXPECT_EQ(recv(self->cfd2, buf, sizeof(buf), 0), -1); + EXPECT_EQ(errno, EBADMSG); +} + +TEST_F(tls_err, bad_cmsg) +{ + char *test_str = "test_read"; + int send_len = 10; + char cip[128]; + char buf[128]; + char txt[64]; + int n; + + if (self->notls) + SKIP(return, "no TLS support"); + + /* Queue up one data record */ + memrnd(txt, sizeof(txt)); + EXPECT_EQ(send(self->fd, txt, sizeof(txt), 0), sizeof(txt)); + n = recv(self->cfd, cip, sizeof(cip), 0); + EXPECT_GT(n, sizeof(txt)); + EXPECT_EQ(send(self->fd2, cip, n, 0), n); + + EXPECT_EQ(tls_send_cmsg(self->fd, 100, test_str, send_len, 0), 10); + n = recv(self->cfd, cip, sizeof(cip), 0); + cip[n - 1]++; /* Break it */ + EXPECT_GT(n, send_len); + EXPECT_EQ(send(self->fd2, cip, n, 0), n); + + EXPECT_EQ(recv(self->cfd2, buf, sizeof(buf), 0), sizeof(txt)); + EXPECT_EQ(memcmp(buf, txt, sizeof(txt)), 0); + EXPECT_EQ(recv(self->cfd2, buf, sizeof(buf), 0), -1); + EXPECT_EQ(errno, EBADMSG); + EXPECT_EQ(recv(self->cfd2, buf, sizeof(buf), 0), -1); + EXPECT_EQ(errno, EBADMSG); +} + TEST(non_established) { struct tls12_crypto_info_aes_gcm_256 tls12; struct sockaddr_in addr;
We don't support splicing control records. TLS 1.3 changes moved the record type check into the decrypt if(). The skb may already be decrypted and still be an alert.
Note that decrypt_skb_update() is idempotent and updates ctx->decrypted so the if() is pointless.
Reorder the check for decryption errors with the content type check while touching them. This part is not really a bug, because if decryption failed in TLS 1.3 content type will be DATA, and for TLS 1.2 it will be correct. Nevertheless its strange to touch output before checking if the function has failed.
Fixes: fedf201e1296 ("net: tls: Refactor control message handling on recv") Signed-off-by: Jakub Kicinski kuba@kernel.org --- net/tls/tls_sw.c | 23 ++++++++++------------- 1 file changed, 10 insertions(+), 13 deletions(-)
diff --git a/net/tls/tls_sw.c b/net/tls/tls_sw.c index d81564078557..2f11f1db917a 100644 --- a/net/tls/tls_sw.c +++ b/net/tls/tls_sw.c @@ -2018,21 +2018,18 @@ ssize_t tls_sw_splice_read(struct socket *sock, loff_t *ppos, if (!skb) goto splice_read_end;
- if (!ctx->decrypted) { - err = decrypt_skb_update(sk, skb, NULL, &chunk, &zc, false); - - /* splice does not support reading control messages */ - if (ctx->control != TLS_RECORD_TYPE_DATA) { - err = -EINVAL; - goto splice_read_end; - } + err = decrypt_skb_update(sk, skb, NULL, &chunk, &zc, false); + if (err < 0) { + tls_err_abort(sk, -EBADMSG); + goto splice_read_end; + }
- if (err < 0) { - tls_err_abort(sk, -EBADMSG); - goto splice_read_end; - } - ctx->decrypted = 1; + /* splice does not support reading control messages */ + if (ctx->control != TLS_RECORD_TYPE_DATA) { + err = -EINVAL; + goto splice_read_end; } + rxm = strp_msg(skb);
chunk = min_t(unsigned int, rxm->full_len, len);
Make sure we correctly reject splicing non-data records.
Signed-off-by: Jakub Kicinski kuba@kernel.org --- tools/testing/selftests/net/tls.c | 40 +++++++++++++++++++++++++++++++ 1 file changed, 40 insertions(+)
diff --git a/tools/testing/selftests/net/tls.c b/tools/testing/selftests/net/tls.c index 2108b197d3f6..3dfa9d7dd4cc 100644 --- a/tools/testing/selftests/net/tls.c +++ b/tools/testing/selftests/net/tls.c @@ -639,6 +639,46 @@ TEST_F(tls, splice_to_pipe) EXPECT_EQ(memcmp(mem_send, mem_recv, send_len), 0); }
+TEST_F(tls, splice_cmsg_to_pipe) +{ + char *test_str = "test_read"; + char record_type = 100; + int send_len = 10; + char buf[10]; + int p[2]; + + ASSERT_GE(pipe(p), 0); + EXPECT_EQ(tls_send_cmsg(self->fd, 100, test_str, send_len, 0), 10); + EXPECT_EQ(splice(self->cfd, NULL, p[1], NULL, send_len, 0), -1); + EXPECT_EQ(errno, EINVAL); + EXPECT_EQ(recv(self->cfd, buf, send_len, 0), -1); + EXPECT_EQ(errno, EIO); + EXPECT_EQ(tls_recv_cmsg(_metadata, self->cfd, record_type, + buf, sizeof(buf), MSG_WAITALL), + send_len); + EXPECT_EQ(memcmp(test_str, buf, send_len), 0); +} + +TEST_F(tls, splice_dec_cmsg_to_pipe) +{ + char *test_str = "test_read"; + char record_type = 100; + int send_len = 10; + char buf[10]; + int p[2]; + + ASSERT_GE(pipe(p), 0); + EXPECT_EQ(tls_send_cmsg(self->fd, 100, test_str, send_len, 0), 10); + EXPECT_EQ(recv(self->cfd, buf, send_len, 0), -1); + EXPECT_EQ(errno, EIO); + EXPECT_EQ(splice(self->cfd, NULL, p[1], NULL, send_len, 0), -1); + EXPECT_EQ(errno, EINVAL); + EXPECT_EQ(tls_recv_cmsg(_metadata, self->cfd, record_type, + buf, sizeof(buf), MSG_WAITALL), + send_len); + EXPECT_EQ(memcmp(test_str, buf, send_len), 0); +} + TEST_F(tls, recvmsg_single) { char const *test_str = "test_recvmsg_single";
recvmsg() will put peek()ed and partially read records onto the rx_list. splice_read() needs to consult that list otherwise it may miss data. Align with recvmsg() and also put partially-read records onto rx_list. tls_sw_advance_skb() is pretty pointless now and will be removed in net-next.
Fixes: 692d7b5d1f91 ("tls: Fix recvmsg() to be able to peek across multiple records") Signed-off-by: Jakub Kicinski kuba@kernel.org --- net/tls/tls_sw.c | 33 +++++++++++++++++++++++++-------- 1 file changed, 25 insertions(+), 8 deletions(-)
diff --git a/net/tls/tls_sw.c b/net/tls/tls_sw.c index 2f11f1db917a..d3e7ff90889e 100644 --- a/net/tls/tls_sw.c +++ b/net/tls/tls_sw.c @@ -2005,6 +2005,7 @@ ssize_t tls_sw_splice_read(struct socket *sock, loff_t *ppos, struct sock *sk = sock->sk; struct sk_buff *skb; ssize_t copied = 0; + bool from_queue; int err = 0; long timeo; int chunk; @@ -2014,14 +2015,20 @@ ssize_t tls_sw_splice_read(struct socket *sock, loff_t *ppos,
timeo = sock_rcvtimeo(sk, flags & SPLICE_F_NONBLOCK);
- skb = tls_wait_data(sk, NULL, flags & SPLICE_F_NONBLOCK, timeo, &err); - if (!skb) - goto splice_read_end; + from_queue = !skb_queue_empty(&ctx->rx_list); + if (from_queue) { + skb = __skb_dequeue(&ctx->rx_list); + } else { + skb = tls_wait_data(sk, NULL, flags & SPLICE_F_NONBLOCK, timeo, + &err); + if (!skb) + goto splice_read_end;
- err = decrypt_skb_update(sk, skb, NULL, &chunk, &zc, false); - if (err < 0) { - tls_err_abort(sk, -EBADMSG); - goto splice_read_end; + err = decrypt_skb_update(sk, skb, NULL, &chunk, &zc, false); + if (err < 0) { + tls_err_abort(sk, -EBADMSG); + goto splice_read_end; + } }
/* splice does not support reading control messages */ @@ -2037,7 +2044,17 @@ ssize_t tls_sw_splice_read(struct socket *sock, loff_t *ppos, if (copied < 0) goto splice_read_end;
- tls_sw_advance_skb(sk, skb, copied); + if (!from_queue) { + ctx->recv_pkt = NULL; + __strp_unpause(&ctx->strp); + } + if (chunk < rxm->full_len) { + __skb_queue_head(&ctx->rx_list, skb); + rxm->offset += len; + rxm->full_len -= len; + } else { + consume_skb(skb); + }
splice_read_end: release_sock(sk);
Add tests for half-received and peeked records.
Signed-off-by: Jakub Kicinski kuba@kernel.org --- tools/testing/selftests/net/tls.c | 49 +++++++++++++++++++++++++++++++ 1 file changed, 49 insertions(+)
diff --git a/tools/testing/selftests/net/tls.c b/tools/testing/selftests/net/tls.c index 3dfa9d7dd4cc..6e78d7207cc1 100644 --- a/tools/testing/selftests/net/tls.c +++ b/tools/testing/selftests/net/tls.c @@ -679,6 +679,55 @@ TEST_F(tls, splice_dec_cmsg_to_pipe) EXPECT_EQ(memcmp(test_str, buf, send_len), 0); }
+TEST_F(tls, recv_and_splice) +{ + int send_len = TLS_PAYLOAD_MAX_LEN; + char mem_send[TLS_PAYLOAD_MAX_LEN]; + char mem_recv[TLS_PAYLOAD_MAX_LEN]; + int half = send_len / 2; + int p[2]; + + ASSERT_GE(pipe(p), 0); + EXPECT_EQ(send(self->fd, mem_send, send_len, 0), send_len); + /* Recv hald of the record, splice the other half */ + EXPECT_EQ(recv(self->cfd, mem_recv, half, MSG_WAITALL), half); + EXPECT_EQ(splice(self->cfd, NULL, p[1], NULL, half, SPLICE_F_NONBLOCK), + half); + EXPECT_EQ(read(p[0], &mem_recv[half], half), half); + EXPECT_EQ(memcmp(mem_send, mem_recv, send_len), 0); +} + +TEST_F(tls, peek_and_splice) +{ + int send_len = TLS_PAYLOAD_MAX_LEN; + char mem_send[TLS_PAYLOAD_MAX_LEN]; + char mem_recv[TLS_PAYLOAD_MAX_LEN]; + int chunk = TLS_PAYLOAD_MAX_LEN / 4; + int n, i, p[2]; + + memrnd(mem_send, sizeof(mem_send)); + + ASSERT_GE(pipe(p), 0); + for (i = 0; i < 4; i++) + EXPECT_EQ(send(self->fd, &mem_send[chunk * i], chunk, 0), + chunk); + + EXPECT_EQ(recv(self->cfd, mem_recv, chunk * 5 / 2, + MSG_WAITALL | MSG_PEEK), + chunk * 5 / 2); + EXPECT_EQ(memcmp(mem_send, mem_recv, chunk * 5 / 2), 0); + + n = 0; + while (n < send_len) { + i = splice(self->cfd, NULL, p[1], NULL, send_len - n, 0); + EXPECT_GT(i, 0); + n += i; + } + EXPECT_EQ(n, send_len); + EXPECT_EQ(read(p[0], mem_recv, send_len), send_len); + EXPECT_EQ(memcmp(mem_send, mem_recv, send_len), 0); +} + TEST_F(tls, recvmsg_single) { char const *test_str = "test_recvmsg_single";
We replace proto_ops whenever TLS is configured for RX. But our replacement also overrides sendpage_locked, which will crash unless TX is also configured. Similarly we plug both of those in for TLS_HW (NIC crypto offload) even tho TLS_HW has a completely different implementation for TX.
Last but not least we always plug in something based on inet_stream_ops even though a few of the callbacks differ for IPv6 (getname, release, bind).
Use a callback building method similar to what we do for struct proto.
Fixes: c46234ebb4d1 ("tls: RX path for ktls") Fixes: d4ffb02dee2f ("net/tls: enable sk_msg redirect to tls socket egress") Signed-off-by: Jakub Kicinski kuba@kernel.org --- net/tls/tls_main.c | 47 +++++++++++++++++++++++++++++++++++++++------- 1 file changed, 40 insertions(+), 7 deletions(-)
diff --git a/net/tls/tls_main.c b/net/tls/tls_main.c index acfba9f1ba72..6bc2879ba637 100644 --- a/net/tls/tls_main.c +++ b/net/tls/tls_main.c @@ -61,7 +61,7 @@ static DEFINE_MUTEX(tcpv6_prot_mutex); static const struct proto *saved_tcpv4_prot; static DEFINE_MUTEX(tcpv4_prot_mutex); static struct proto tls_prots[TLS_NUM_PROTS][TLS_NUM_CONFIG][TLS_NUM_CONFIG]; -static struct proto_ops tls_sw_proto_ops; +static struct proto_ops tls_proto_ops[TLS_NUM_PROTS][TLS_NUM_CONFIG][TLS_NUM_CONFIG]; static void build_protos(struct proto prot[TLS_NUM_CONFIG][TLS_NUM_CONFIG], const struct proto *base);
@@ -71,6 +71,8 @@ void update_sk_prot(struct sock *sk, struct tls_context *ctx)
WRITE_ONCE(sk->sk_prot, &tls_prots[ip_ver][ctx->tx_conf][ctx->rx_conf]); + WRITE_ONCE(sk->sk_socket->ops, + &tls_proto_ops[ip_ver][ctx->tx_conf][ctx->rx_conf]); }
int wait_on_pending_writer(struct sock *sk, long *timeo) @@ -669,8 +671,6 @@ static int do_tls_setsockopt_conf(struct sock *sk, sockptr_t optval, if (tx) { ctx->sk_write_space = sk->sk_write_space; sk->sk_write_space = tls_write_space; - } else { - sk->sk_socket->ops = &tls_sw_proto_ops; } goto out;
@@ -728,6 +728,39 @@ struct tls_context *tls_ctx_create(struct sock *sk) return ctx; }
+static void build_proto_ops(struct proto_ops ops[TLS_NUM_CONFIG][TLS_NUM_CONFIG], + const struct proto_ops *base) +{ + ops[TLS_BASE][TLS_BASE] = *base; + + ops[TLS_SW ][TLS_BASE] = ops[TLS_BASE][TLS_BASE]; + ops[TLS_SW ][TLS_BASE].sendpage_locked = tls_sw_sendpage_locked; + + ops[TLS_BASE][TLS_SW ] = ops[TLS_BASE][TLS_BASE]; + ops[TLS_BASE][TLS_SW ].splice_read = tls_sw_splice_read; + + ops[TLS_SW ][TLS_SW ] = ops[TLS_SW ][TLS_BASE]; + ops[TLS_SW ][TLS_SW ].splice_read = tls_sw_splice_read; + +#ifdef CONFIG_TLS_DEVICE + ops[TLS_HW ][TLS_BASE] = ops[TLS_BASE][TLS_BASE]; + ops[TLS_HW ][TLS_BASE].sendpage_locked = NULL; + + ops[TLS_HW ][TLS_SW ] = ops[TLS_BASE][TLS_SW ]; + ops[TLS_HW ][TLS_SW ].sendpage_locked = NULL; + + ops[TLS_BASE][TLS_HW ] = ops[TLS_BASE][TLS_SW ]; + + ops[TLS_SW ][TLS_HW ] = ops[TLS_SW ][TLS_SW ]; + + ops[TLS_HW ][TLS_HW ] = ops[TLS_HW ][TLS_SW ]; + ops[TLS_HW ][TLS_HW ].sendpage_locked = NULL; +#endif +#ifdef CONFIG_TLS_TOE + ops[TLS_HW_RECORD][TLS_HW_RECORD] = *base; +#endif +} + static void tls_build_proto(struct sock *sk) { int ip_ver = sk->sk_family == AF_INET6 ? TLSV6 : TLSV4; @@ -739,6 +772,8 @@ static void tls_build_proto(struct sock *sk) mutex_lock(&tcpv6_prot_mutex); if (likely(prot != saved_tcpv6_prot)) { build_protos(tls_prots[TLSV6], prot); + build_proto_ops(tls_proto_ops[TLSV6], + sk->sk_socket->ops); smp_store_release(&saved_tcpv6_prot, prot); } mutex_unlock(&tcpv6_prot_mutex); @@ -749,6 +784,8 @@ static void tls_build_proto(struct sock *sk) mutex_lock(&tcpv4_prot_mutex); if (likely(prot != saved_tcpv4_prot)) { build_protos(tls_prots[TLSV4], prot); + build_proto_ops(tls_proto_ops[TLSV4], + sk->sk_socket->ops); smp_store_release(&saved_tcpv4_prot, prot); } mutex_unlock(&tcpv4_prot_mutex); @@ -959,10 +996,6 @@ static int __init tls_register(void) if (err) return err;
- tls_sw_proto_ops = inet_stream_ops; - tls_sw_proto_ops.splice_read = tls_sw_splice_read; - tls_sw_proto_ops.sendpage_locked = tls_sw_sendpage_locked; - tls_device_init(); tcp_register_ulp(&tcp_tls_ulp_ops);
Previous patch fixes overriding callbacks incorrectly. Triggering the crash in sendpage_locked would be more spectacular but it's hard to get to, so take the easier path of proving this is broken and call getname. We're currently getting IPv4 socket info on an IPv6 socket.
Signed-off-by: Jakub Kicinski kuba@kernel.org --- tools/testing/selftests/net/tls.c | 55 +++++++++++++++++++++++++++++++ 1 file changed, 55 insertions(+)
diff --git a/tools/testing/selftests/net/tls.c b/tools/testing/selftests/net/tls.c index 6e78d7207cc1..8a22db0cca49 100644 --- a/tools/testing/selftests/net/tls.c +++ b/tools/testing/selftests/net/tls.c @@ -1617,4 +1617,59 @@ TEST(keysizes) { close(cfd); }
+TEST(tls_v6ops) { + struct tls_crypto_info_keys tls12; + struct sockaddr_in6 addr, addr2; + int sfd, ret, fd; + socklen_t len, len2; + + tls_crypto_info_init(TLS_1_2_VERSION, TLS_CIPHER_AES_GCM_128, &tls12); + + addr.sin6_family = AF_INET6; + addr.sin6_addr = in6addr_any; + addr.sin6_port = 0; + + fd = socket(AF_INET6, SOCK_STREAM, 0); + sfd = socket(AF_INET6, SOCK_STREAM, 0); + + ret = bind(sfd, &addr, sizeof(addr)); + ASSERT_EQ(ret, 0); + ret = listen(sfd, 10); + ASSERT_EQ(ret, 0); + + len = sizeof(addr); + ret = getsockname(sfd, &addr, &len); + ASSERT_EQ(ret, 0); + + ret = connect(fd, &addr, sizeof(addr)); + ASSERT_EQ(ret, 0); + + len = sizeof(addr); + ret = getsockname(fd, &addr, &len); + ASSERT_EQ(ret, 0); + + ret = setsockopt(fd, IPPROTO_TCP, TCP_ULP, "tls", sizeof("tls")); + if (ret) { + ASSERT_EQ(errno, ENOENT); + SKIP(return, "no TLS support"); + } + ASSERT_EQ(ret, 0); + + ret = setsockopt(fd, SOL_TLS, TLS_TX, &tls12, tls12.len); + ASSERT_EQ(ret, 0); + + ret = setsockopt(fd, SOL_TLS, TLS_RX, &tls12, tls12.len); + ASSERT_EQ(ret, 0); + + len2 = sizeof(addr2); + ret = getsockname(fd, &addr2, &len2); + ASSERT_EQ(ret, 0); + + EXPECT_EQ(len2, len); + EXPECT_EQ(memcmp(&addr, &addr2, len), 0); + + close(fd); + close(sfd); +} + TEST_HARNESS_MAIN
Hello:
This series was applied to netdev/net.git (master) by Jakub Kicinski kuba@kernel.org:
On Wed, 24 Nov 2021 15:25:48 -0800 you wrote:
As I work my way to unlocked and zero-copy TLS Rx the obvious bugs in the splice_read implementation get harder and harder to ignore. This is to say the fixes here are discovered by code inspection, I'm not aware of anyone actually using splice_read.
Jakub Kicinski (9): selftests: tls: add helper for creating sock pairs selftests: tls: factor out cmsg send/receive selftests: tls: add tests for handling of bad records tls: splice_read: fix record type check selftests: tls: test splicing cmsgs tls: splice_read: fix accessing pre-processed records selftests: tls: test splicing decrypted records tls: fix replacing proto_ops selftests: tls: test for correct proto_ops
[...]
Here is the summary with links: - [net,1/9] selftests: tls: add helper for creating sock pairs https://git.kernel.org/netdev/net/c/a125f91fe783 - [net,2/9] selftests: tls: factor out cmsg send/receive https://git.kernel.org/netdev/net/c/31180adb0bed - [net,3/9] selftests: tls: add tests for handling of bad records https://git.kernel.org/netdev/net/c/ef0fc0b3cc2b - [net,4/9] tls: splice_read: fix record type check https://git.kernel.org/netdev/net/c/520493f66f68 - [net,5/9] selftests: tls: test splicing cmsgs https://git.kernel.org/netdev/net/c/d87d67fd61ef - [net,6/9] tls: splice_read: fix accessing pre-processed records https://git.kernel.org/netdev/net/c/e062fe99cccd - [net,7/9] selftests: tls: test splicing decrypted records https://git.kernel.org/netdev/net/c/274af0f9e279 - [net,8/9] tls: fix replacing proto_ops https://git.kernel.org/netdev/net/c/f3911f73f51d - [net,9/9] selftests: tls: test for correct proto_ops https://git.kernel.org/netdev/net/c/f884a3426291
You are awesome, thank you!
linux-kselftest-mirror@lists.linaro.org
-
Jakub Kicinski -
patchwork-bot+netdevbpf＠kernel.org