Skip to contentSkip to conversation

libceph: Partially revert changes to support MSG_SPLICE_PAGES

PATCH Superseded David Howells 3 participants 8 messages Jun 26, 2023

Subsystems (3)

v1 of 3 Superseded Latest is v3 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 libceph: Partially revert changes to support MSG_SPLICE_PAGES

Fix the mishandling of MSG_DONTWAIT and also reinstates the per-page checking of the source pages (which might have come from a DIO write by userspace) by partially reverting the changes to support MSG_SPLICE_PAGES and doing things a little differently. In messenger_v1:

(1) The ceph_tcp_sendpage() is resurrected and the callers reverted to use that.

(2) The callers now pass MSG_MORE unconditionally. Previously, they were passing in MSG_MORE|MSG_SENDPAGE_NOTLAST and then degrading that to just MSG_MORE on the last call to ->sendpage().

(3) Make ceph_tcp_sendpage() a wrapper around sendmsg() rather than sendpage(), setting MSG_SPLICE_PAGES if sendpage_ok() returns true on the page.

In messenger_v2:

(4) Bring back do_try_sendpage() and make the callers use that.

(5) Make do_try_sendpage() use sendmsg() for both cases and set MSG_SPLICE_PAGES if sendpage_ok() is set.

Note that libceph is using sendpage_ok() for this, but it should not really be considered exhaustive in this regard - this is, however, what libceph is doing upstream.

Fixes 40a8c17aa770 ("ceph: Use sendmsg(MSG_SPLICE_PAGES) rather than sendpage")
Fixes fa094ccae1e7 ("ceph: Use sendmsg(MSG_SPLICE_PAGES) rather than sendpage()")
cc Ilya Dryomov <[email protected]>
cc Xiubo Li <[email protected]>
cc Jeff Layton <[email protected]>
cc "David S. Miller" <[email protected]>
cc Eric Dumazet <[email protected]>
cc Jakub Kicinski <[email protected]>
cc Paolo Abeni <[email protected]>
cc Jens Axboe <[email protected]>
cc Matthew Wilcox <[email protected]>
---
 net/ceph/messenger_v1.c |   57 +++++++++++++++++++------------
 net/ceph/messenger_v2.c |   87 ++++++++++++++++++++++++++++++++++++++----------
 2 files changed, 105 insertions(+), 39 deletions(-)
net/ceph/messenger_v1.c +36 -21
index 814579f27f04..1ee58cbaf151 100644
@@ -74,6 +74,38 @@ static int ceph_tcp_sendmsg(struct socket *sock, struct kvec *iov,
 	return r;
 }
 
+/*
+ * @more: either or both of MSG_MORE and MSG_SENDPAGE_NOTLAST
+ */
+static int ceph_tcp_sendpage(struct socket *sock, struct page *page,
+			     int offset, size_t size, int more)
+{
+	struct msghdr msg = {
+		.msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL | more,
+	};
+	struct bio_vec bvec;
+	int ret;
+
+	/*
+	 * sendpage cannot properly handle pages with page_count == 0,
+	 * we need to fall back to sendmsg if that's the case.
+	 *
+	 * Same goes for slab pages: skb_can_coalesce() allows
+	 * coalescing neighboring slab objects into a single frag which
+	 * triggers one of hardened usercopy checks.
+	 */
+	if (sendpage_ok(page))
+		msg.msg_flags |= MSG_SPLICE_PAGES;
+
+	bvec_set_page(&bvec, page, size, offset);
+
+	ret = sock_sendmsg(sock, &msg);
+	if (ret == -EAGAIN)
+		ret = 0;
+
+	return ret;
+}
+
 static void con_out_kvec_reset(struct ceph_connection *con)
 {
 	BUG_ON(con->v1.out_skip);
@@ -450,10 +482,6 @@ static int write_partial_message_data(struct ceph_connection *con)
 	 */
 	crc = do_datacrc ? le32_to_cpu(msg->footer.data_crc) : 0;
 	while (cursor->total_resid) {
-		struct bio_vec bvec;
-		struct msghdr msghdr = {
-			.msg_flags = MSG_SPLICE_PAGES,
-		};
 		struct page *page;
 		size_t page_offset;
 		size_t length;
@@ -465,13 +493,8 @@ static int write_partial_message_data(struct ceph_connection *con)
 		}
 
 		page = ceph_msg_data_next(cursor, &page_offset, &length);
-		if (length != cursor->total_resid)
-			msghdr.msg_flags |= MSG_MORE;
-
-		bvec_set_page(&bvec, page, length, page_offset);
-		iov_iter_bvec(&msghdr.msg_iter, ITER_SOURCE, &bvec, 1, length);
-
-		ret = sock_sendmsg(con->sock, &msghdr);
+		ret = ceph_tcp_sendpage(con->sock, page, page_offset, length,
+					MSG_MORE);
 		if (ret <= 0) {
 			if (do_datacrc)
 				msg->footer.data_crc = cpu_to_le32(crc);
@@ -501,22 +524,14 @@ static int write_partial_message_data(struct ceph_connection *con)
  */
 static int write_partial_skip(struct ceph_connection *con)
 {
-	struct bio_vec bvec;
-	struct msghdr msghdr = {
-		.msg_flags = MSG_SPLICE_PAGES | MSG_MORE,
-	};
 	int ret;
 
 	dout("%s %p %d left\n", __func__, con, con->v1.out_skip);
 	while (con->v1.out_skip > 0) {
 		size_t size = min(con->v1.out_skip, (int)PAGE_SIZE);
 
-		if (size == con->v1.out_skip)
-			msghdr.msg_flags &= ~MSG_MORE;
-		bvec_set_page(&bvec, ZERO_PAGE(0), size, 0);
-		iov_iter_bvec(&msghdr.msg_iter, ITER_SOURCE, &bvec, 1, size);
-
-		ret = sock_sendmsg(con->sock, &msghdr);
+		ret = ceph_tcp_sendpage(con->sock, ceph_zero_page, 0, size,
+					MSG_MORE);
 		if (ret <= 0)
 			goto out;
 		con->v1.out_skip -= ret;
net/ceph/messenger_v2.c +69 -18

107 lines not on this page · raw message

maintainer
#1 lore raw
Quoted 249 lines · David Howells

Fix the mishandling of MSG_DONTWAIT and also reinstates the per-page checking of the source pages (which might have come from a DIO write by userspace) by partially reverting the changes to support MSG_SPLICE_PAGES and doing things a little differently. In messenger_v1:

(1) The ceph_tcp_sendpage() is resurrected and the callers reverted to use that.

(2) The callers now pass MSG_MORE unconditionally. Previously, they were passing in MSG_MORE|MSG_SENDPAGE_NOTLAST and then degrading that to just MSG_MORE on the last call to ->sendpage().

(3) Make ceph_tcp_sendpage() a wrapper around sendmsg() rather than sendpage(), setting MSG_SPLICE_PAGES if sendpage_ok() returns true on the page.

In messenger_v2:

(4) Bring back do_try_sendpage() and make the callers use that.

(5) Make do_try_sendpage() use sendmsg() for both cases and set MSG_SPLICE_PAGES if sendpage_ok() is set.

Note that libceph is using sendpage_ok() for this, but it should not really be considered exhaustive in this regard - this is, however, what libceph is doing upstream.

Fixes 40a8c17aa770 ("ceph: Use sendmsg(MSG_SPLICE_PAGES) rather than sendpage")
Fixes fa094ccae1e7 ("ceph: Use sendmsg(MSG_SPLICE_PAGES) rather than sendpage()")
cc Ilya Dryomov <[email protected]>
cc Xiubo Li <[email protected]>
cc Jeff Layton <[email protected]>
cc "David S. Miller" <[email protected]>
cc Eric Dumazet <[email protected]>
cc Jakub Kicinski <[email protected]>
cc Paolo Abeni <[email protected]>
cc Jens Axboe <[email protected]>
cc Matthew Wilcox <[email protected]>
---
 net/ceph/messenger_v1.c |   57 +++++++++++++++++++------------
 net/ceph/messenger_v2.c |   87 ++++++++++++++++++++++++++++++++++++++----------
 2 files changed, 105 insertions(+), 39 deletions(-)
net/ceph/messenger_v1.c +36 -21
index 814579f27f04..1ee58cbaf151 100644
@@ -74,6 +74,38 @@ static int ceph_tcp_sendmsg(struct socket *sock, struct kvec *iov,
        return r;
 }
 
+/*
+ * @more: either or both of MSG_MORE and MSG_SENDPAGE_NOTLAST
+ */
+static int ceph_tcp_sendpage(struct socket *sock, struct page *page,
+                            int offset, size_t size, int more)
+{
+       struct msghdr msg = {
+               .msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL | more,
+       };
+       struct bio_vec bvec;
+       int ret;
+
+       /*
+        * sendpage cannot properly handle pages with page_count == 0,
+        * we need to fall back to sendmsg if that's the case.
+        *
+        * Same goes for slab pages: skb_can_coalesce() allows
+        * coalescing neighboring slab objects into a single frag which
+        * triggers one of hardened usercopy checks.
+        */
+       if (sendpage_ok(page))
+               msg.msg_flags |= MSG_SPLICE_PAGES;
+
+       bvec_set_page(&bvec, page, size, offset);
+
+       ret = sock_sendmsg(sock, &msg);
+       if (ret == -EAGAIN)
+               ret = 0;
+
+       return ret;
+}
+
 static void con_out_kvec_reset(struct ceph_connection *con)
 {
        BUG_ON(con->v1.out_skip);
@@ -450,10 +482,6 @@ static int write_partial_message_data(struct ceph_connection *con)
         */
        crc = do_datacrc ? le32_to_cpu(msg->footer.data_crc) : 0;
        while (cursor->total_resid) {
-               struct bio_vec bvec;
-               struct msghdr msghdr = {
-                       .msg_flags = MSG_SPLICE_PAGES,
-               };
                struct page *page;
                size_t page_offset;
                size_t length;
@@ -465,13 +493,8 @@ static int write_partial_message_data(struct ceph_connection *con)
                }
 
                page = ceph_msg_data_next(cursor, &page_offset, &length);
-               if (length != cursor->total_resid)
-                       msghdr.msg_flags |= MSG_MORE;
-
-               bvec_set_page(&bvec, page, length, page_offset);
-               iov_iter_bvec(&msghdr.msg_iter, ITER_SOURCE, &bvec, 1, length);
-
-               ret = sock_sendmsg(con->sock, &msghdr);
+               ret = ceph_tcp_sendpage(con->sock, page, page_offset, length,
+                                       MSG_MORE);
                if (ret <= 0) {
                        if (do_datacrc)
                                msg->footer.data_crc = cpu_to_le32(crc);
@@ -501,22 +524,14 @@ static int write_partial_message_data(struct ceph_connection *con)
  */
 static int write_partial_skip(struct ceph_connection *con)
 {
-       struct bio_vec bvec;
-       struct msghdr msghdr = {
-               .msg_flags = MSG_SPLICE_PAGES | MSG_MORE,
-       };
        int ret;
 
        dout("%s %p %d left\n", __func__, con, con->v1.out_skip);
        while (con->v1.out_skip > 0) {
                size_t size = min(con->v1.out_skip, (int)PAGE_SIZE);
 
-               if (size == con->v1.out_skip)
-                       msghdr.msg_flags &= ~MSG_MORE;
-               bvec_set_page(&bvec, ZERO_PAGE(0), size, 0);
-               iov_iter_bvec(&msghdr.msg_iter, ITER_SOURCE, &bvec, 1, size);
-
-               ret = sock_sendmsg(con->sock, &msghdr);
+               ret = ceph_tcp_sendpage(con->sock, ceph_zero_page, 0, size,
+                                       MSG_MORE);
                if (ret <= 0)
                        goto out;
                con->v1.out_skip -= ret;
net/ceph/messenger_v2.c +65 -15
index 87ac97073e75..139b61598f4f 100644
@@ -117,38 +117,89 @@ static int ceph_tcp_recv(struct ceph_connection *con)
        return ret;
 }
 
+static int do_sendmsg(struct socket *sock, struct iov_iter *it)
+{
+       struct msghdr msg = { .msg_flags = CEPH_MSG_FLAGS };
+       int ret;
+
+       msg.msg_iter = *it;
+       while (iov_iter_count(it)) {
+               ret = sock_sendmsg(sock, &msg);
+               if (ret <= 0) {
+                       if (ret == -EAGAIN)
+                               ret = 0;
+                       return ret;
+               }
+
+               iov_iter_advance(it, ret);
+       }
+
+       WARN_ON(msg_data_left(&msg));
+       return 1;
+}
+
+static int do_try_sendpage(struct socket *sock, struct iov_iter *it)
+{
+       struct msghdr msg = {};
+       struct bio_vec bv;
+       int ret;
+
+       if (WARN_ON(!iov_iter_is_bvec(it)))
+               return -EINVAL;
+
+       while (iov_iter_count(it)) {
+               msg.msg_flags = CEPH_MSG_FLAGS;
+               if (sendpage_ok(bv.bv_page))
+                       msg.msg_flags |= MSG_SPLICE_PAGES;
+
+               /* iov_iter_iovec() for ITER_BVEC */
+               bvec_set_page(&bv, it->bvec->bv_page,
+                             min(iov_iter_count(it),
+                                 it->bvec->bv_len - it->iov_offset),
+                             it->bvec->bv_offset + it->iov_offset);
+
+               /*
+                * MSG_SPLICE_PAGES cannot properly handle pages with
+                * page_count == 0, we need to fall back to sendmsg if
+                * that's the case.
+                *
+                * Same goes for slab pages: skb_can_coalesce() allows
+                * coalescing neighboring slab objects into a single frag
+                * which triggers one of hardened usercopy checks.
+                */
+               iov_iter_bvec(&msg.msg_iter, ITER_SOURCE, &bv, 1, bv.bv_len);
+               ret = sock_sendmsg(sock, &msg);
+               if (ret <= 0) {
+                       if (ret == -EAGAIN)
+                               ret = 0;
+                       return ret;
+               }
+
+               iov_iter_advance(it, ret);
+       }
+
+       return 1;
+}
+
 /*
  * Write as much as possible.  The socket is expected to be corked,
  * so we don't bother with MSG_MORE here.
  *
  * Return:
- *  >0 - done, nothing (else) to write
+ *   1 - done, nothing (else) to write
  *   0 - socket is full, need to wait
  *  <0 - error
  */
 static int ceph_tcp_send(struct ceph_connection *con)
 {
-       struct msghdr msg = {
-               .msg_iter       = con->v2.out_iter,
-               .msg_flags      = CEPH_MSG_FLAGS,
-       };
        int ret;
 
-       if (WARN_ON(!iov_iter_is_bvec(&con->v2.out_iter)))
-               return -EINVAL;
-
-       if (con->v2.out_iter_sendpage)
-               msg.msg_flags |= MSG_SPLICE_PAGES;
-
        dout("%s con %p have %zu try_sendpage %d\n", __func__, con,
             iov_iter_count(&con->v2.out_iter), con->v2.out_iter_sendpage);
-
-       ret = sock_sendmsg(con->sock, &msg);
-       if (ret > 0)
-               iov_iter_advance(&con->v2.out_iter, ret);

net/ceph/messenger_v2.c static int ceph_tcp_recv(struct ceph_connection *con) · line 205

Quoted 10 lines
net/ceph/messenger_v2.c +4 -3
-       else if (ret == -EAGAIN)
-               ret = 0;
-
+       if (con->v2.out_iter_sendpage)
+               ret = do_try_sendpage(con->sock, &con->v2.out_iter);
+       else
+               ret = do_sendmsg(con->sock, &con->v2.out_iter);
        dout("%s con %p ret %d left %zu\n", __func__, con, ret,
             iov_iter_count(&con->v2.out_iter));
        return ret;

Hi David,

This patch appears to be mangled.

Thanks,

Ilya

#2 lore raw
Quoted 1 line · Ilya Dryomov

This patch appears to be mangled.

Mangled? Did you see tab -> space conversion? I see that in your reply, but in the copy I got it's fine.

David

maintainer
#3 lore raw
Quoted 7 lines · David Howells
Quoted 1 line · Ilya Dryomov

This patch appears to be mangled.

Mangled? Did you see tab -> space conversion? I see that in your reply, but in the copy I got it's fine.

It turns out that it applies fine with "git am" but "git apply --check" complains as follows:

error: patch fragment without header at line 203: @@ -450,10 +482,6 @@ static int write_partial_message_data(struct ceph_con=

I think it's due to the quoted-printable encoding. I didn't have any issues with your previous patches using the same automation so I just assumed the patch to be mangled in the usual way.

Sorry for the noise, now that I can see the cumulative diff, I will review shortly.

Thanks,

Ilya

maintainer
#5 lore raw
In reply to David Howells
Quoted 20 lines · David Howells

Fix the mishandling of MSG_DONTWAIT and also reinstates the per-page checking of the source pages (which might have come from a DIO write by userspace) by partially reverting the changes to support MSG_SPLICE_PAGES and doing things a little differently. In messenger_v1:

(1) The ceph_tcp_sendpage() is resurrected and the callers reverted to use that.

(2) The callers now pass MSG_MORE unconditionally. Previously, they were passing in MSG_MORE|MSG_SENDPAGE_NOTLAST and then degrading that to just MSG_MORE on the last call to ->sendpage().

(3) Make ceph_tcp_sendpage() a wrapper around sendmsg() rather than sendpage(), setting MSG_SPLICE_PAGES if sendpage_ok() returns true on the page.

In messenger_v2:

Quoted 8 lines

(4) Bring back do_try_sendpage() and make the callers use that.

(5) Make do_try_sendpage() use sendmsg() for both cases and set MSG_SPLICE_PAGES if sendpage_ok() is set.

Note that libceph is using sendpage_ok() for this, but it should not really be considered exhaustive in this regard - this is, however, what libceph is doing upstream.

Such use spans far beyond libceph: drbd, iSCSI, NVMe-over-TCP, etc. sendpage_ok() the only tool we have got...

Quoted 28 lines
Fixes 40a8c17aa770 ("ceph: Use sendmsg(MSG_SPLICE_PAGES) rather than sendpage")
Fixes fa094ccae1e7 ("ceph: Use sendmsg(MSG_SPLICE_PAGES) rather than sendpage()")
cc Ilya Dryomov <[email protected]>
cc Xiubo Li <[email protected]>
cc Jeff Layton <[email protected]>
cc "David S. Miller" <[email protected]>
cc Eric Dumazet <[email protected]>
cc Jakub Kicinski <[email protected]>
cc Paolo Abeni <[email protected]>
cc Jens Axboe <[email protected]>
cc Matthew Wilcox <[email protected]>
---
 net/ceph/messenger_v1.c |   57 +++++++++++++++++++------------
 net/ceph/messenger_v2.c |   87 ++++++++++++++++++++++++++++++++++++++----------
 2 files changed, 105 insertions(+), 39 deletions(-)
net/ceph/messenger_v1.c +0 -0
index 814579f27f04..1ee58cbaf151 100644

net/ceph/messenger_v1.c static int ceph_tcp_sendmsg(struct socket *sock, struct kvec *iov, · line 78

Quoted 5 lines
net/ceph/messenger_v1.c +2 -0
@@ -74,6 +74,38 @@ static int ceph_tcp_sendmsg(struct socket *sock, struct kvec *iov,
        return r;
 }
 
+/*
+ * @more: either or both of MSG_MORE and MSG_SENDPAGE_NOTLAST

MSG_SENDPAGE_NOTLAST flag doesn't exist anymore.

net/ceph/messenger_v1.c static int ceph_tcp_sendmsg(struct socket *sock, struct kvec *iov, · line 90

Quoted 12 lines
net/ceph/messenger_v1.c +12 -0
+ */
+static int ceph_tcp_sendpage(struct socket *sock, struct page *page,
+                            int offset, size_t size, int more)
+{
+       struct msghdr msg = {
+               .msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL | more,
+       };
+       struct bio_vec bvec;
+       int ret;
+
+       /*
+        * sendpage cannot properly handle pages with page_count == 0,

I would suggest sendpage -> MSG_SPLICE_PAGES here like was done in the other file.

net/ceph/messenger_v1.c static int ceph_tcp_sendmsg(struct socket *sock, struct kvec *iov, · line 100

Quoted 10 lines
net/ceph/messenger_v1.c +10 -0
+        * we need to fall back to sendmsg if that's the case.
+        *
+        * Same goes for slab pages: skb_can_coalesce() allows
+        * coalescing neighboring slab objects into a single frag which
+        * triggers one of hardened usercopy checks.
+        */
+       if (sendpage_ok(page))
+               msg.msg_flags |= MSG_SPLICE_PAGES;
+
+       bvec_set_page(&bvec, page, size, offset);

The bvec is not added to the message, so I don't think this actually sends anything...

Quoted 111 lines
net/ceph/messenger_v1.c +12 -21
+
+       ret = sock_sendmsg(sock, &msg);
+       if (ret == -EAGAIN)
+               ret = 0;
+
+       return ret;
+}
+
 static void con_out_kvec_reset(struct ceph_connection *con)
 {
        BUG_ON(con->v1.out_skip);
@@ -450,10 +482,6 @@ static int write_partial_message_data(struct ceph_connection *con)
         */
        crc = do_datacrc ? le32_to_cpu(msg->footer.data_crc) : 0;
        while (cursor->total_resid) {
-               struct bio_vec bvec;
-               struct msghdr msghdr = {
-                       .msg_flags = MSG_SPLICE_PAGES,
-               };
                struct page *page;
                size_t page_offset;
                size_t length;
@@ -465,13 +493,8 @@ static int write_partial_message_data(struct ceph_connection *con)
                }
 
                page = ceph_msg_data_next(cursor, &page_offset, &length);
-               if (length != cursor->total_resid)
-                       msghdr.msg_flags |= MSG_MORE;
-
-               bvec_set_page(&bvec, page, length, page_offset);
-               iov_iter_bvec(&msghdr.msg_iter, ITER_SOURCE, &bvec, 1, length);
-
-               ret = sock_sendmsg(con->sock, &msghdr);
+               ret = ceph_tcp_sendpage(con->sock, page, page_offset, length,
+                                       MSG_MORE);
                if (ret <= 0) {
                        if (do_datacrc)
                                msg->footer.data_crc = cpu_to_le32(crc);
@@ -501,22 +524,14 @@ static int write_partial_message_data(struct ceph_connection *con)
  */
 static int write_partial_skip(struct ceph_connection *con)
 {
-       struct bio_vec bvec;
-       struct msghdr msghdr = {
-               .msg_flags = MSG_SPLICE_PAGES | MSG_MORE,
-       };
        int ret;
 
        dout("%s %p %d left\n", __func__, con, con->v1.out_skip);
        while (con->v1.out_skip > 0) {
                size_t size = min(con->v1.out_skip, (int)PAGE_SIZE);
 
-               if (size == con->v1.out_skip)
-                       msghdr.msg_flags &= ~MSG_MORE;
-               bvec_set_page(&bvec, ZERO_PAGE(0), size, 0);
-               iov_iter_bvec(&msghdr.msg_iter, ITER_SOURCE, &bvec, 1, size);
-
-               ret = sock_sendmsg(con->sock, &msghdr);
+               ret = ceph_tcp_sendpage(con->sock, ceph_zero_page, 0, size,
+                                       MSG_MORE);
                if (ret <= 0)
                        goto out;
                con->v1.out_skip -= ret;
net/ceph/messenger_v2.c +40 -0
index 87ac97073e75..139b61598f4f 100644
@@ -117,38 +117,89 @@ static int ceph_tcp_recv(struct ceph_connection *con)
        return ret;
 }
 
+static int do_sendmsg(struct socket *sock, struct iov_iter *it)
+{
+       struct msghdr msg = { .msg_flags = CEPH_MSG_FLAGS };
+       int ret;
+
+       msg.msg_iter = *it;
+       while (iov_iter_count(it)) {
+               ret = sock_sendmsg(sock, &msg);
+               if (ret <= 0) {
+                       if (ret == -EAGAIN)
+                               ret = 0;
+                       return ret;
+               }
+
+               iov_iter_advance(it, ret);
+       }
+
+       WARN_ON(msg_data_left(&msg));
+       return 1;
+}
+
+static int do_try_sendpage(struct socket *sock, struct iov_iter *it)
+{
+       struct msghdr msg = {};
+       struct bio_vec bv;
+       int ret;
+
+       if (WARN_ON(!iov_iter_is_bvec(it)))
+               return -EINVAL;
+
+       while (iov_iter_count(it)) {
+               msg.msg_flags = CEPH_MSG_FLAGS;
+               if (sendpage_ok(bv.bv_page))
+                       msg.msg_flags |= MSG_SPLICE_PAGES;
+
+               /* iov_iter_iovec() for ITER_BVEC */
+               bvec_set_page(&bv, it->bvec->bv_page,
+                             min(iov_iter_count(it),
+                                 it->bvec->bv_len - it->iov_offset),
+                             it->bvec->bv_offset + it->iov_offset);

net/ceph/messenger_v2.c static int ceph_tcp_recv(struct ceph_connection *con) · line 169

Quoted 10 lines
net/ceph/messenger_v2.c +10 -0
+
+               /*
+                * MSG_SPLICE_PAGES cannot properly handle pages with
+                * page_count == 0, we need to fall back to sendmsg if
+                * that's the case.
+                *
+                * Same goes for slab pages: skb_can_coalesce() allows
+                * coalescing neighboring slab objects into a single frag
+                * which triggers one of hardened usercopy checks.
+                */

This comment is now misplaced. I would suggest

  if (sendpage_ok(bv.bv_page))
          msg.msg_flags |= MSG_SPLICE_PAGES;
  else
          msg.msg_flags &= ~MSG_SPLICE_PAGES;

here and keeping the initialization

  struct msghdr msg = { .msg_flags = CEPH_MSG_FLAGS };

above thus staying closer to the original.

Thanks,

Ilya

#6 lore raw
Quoted 4 lines · Ilya Dryomov
  if (sendpage_ok(bv.bv_page))
          msg.msg_flags |= MSG_SPLICE_PAGES;
  else
          msg.msg_flags &= ~MSG_SPLICE_PAGES;

Hmmm... I'm not sure there's any guarantee that msg, including msg_flags, won't get altered by ->sendmsg().

David

maintainer
#7 lore raw
Quoted 10 lines · David Howells
Quoted 4 lines · Ilya Dryomov
  if (sendpage_ok(bv.bv_page))
          msg.msg_flags |= MSG_SPLICE_PAGES;
  else
          msg.msg_flags &= ~MSG_SPLICE_PAGES;

Hmmm... I'm not sure there's any guarantee that msg, including msg_flags, won't get altered by ->sendmsg().

If this is indeed an issue, do_sendmsg() should be fixed too. I would like to avoid having do_try_sendpage() do one thing and do_sendmsg() do something subtly different.

But then, even with the current patch, only msg_flags is reinitialized on the next loop iteration, not the entire message. Should the entire message be reinitialized?

Thanks,

Ilya