[PATCH 3.12 101/116] libceph: don't bail early from try_read() when skipping a message

From: Jiri Slaby
Date: Fri Mar 04 2016 - 04:16:27 EST


From: Ilya Dryomov <idryomov@xxxxxxxxx>

3.12-stable review patch. If anyone has any objections, please let me know.

===============

commit e7a88e82fe380459b864e05b372638aeacb0f52d upstream.

The contract between try_read() and try_write() is that when called
each processes as much data as possible. When instructed by osd_client
to skip a message, try_read() is violating this contract by returning
after receiving and discarding a single message instead of checking for
more. try_write() then gets a chance to write out more requests,
generating more replies/skips for try_read() to handle, forcing the
messenger into a starvation loop.

Reported-by: Varada Kari <Varada.Kari@xxxxxxxxxxx>
Signed-off-by: Ilya Dryomov <idryomov@xxxxxxxxx>
Tested-by: Varada Kari <Varada.Kari@xxxxxxxxxxx>
Reviewed-by: Alex Elder <elder@xxxxxxxxxx>
Signed-off-by: Jiri Slaby <jslaby@xxxxxxx>
---
net/ceph/messenger.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/net/ceph/messenger.c b/net/ceph/messenger.c
index 057017bd3b42..469f3138d0f6 100644
--- a/net/ceph/messenger.c
+++ b/net/ceph/messenger.c
@@ -2280,7 +2280,7 @@ static int read_partial_message(struct ceph_connection *con)
con->in_base_pos = -front_len - middle_len - data_len -
sizeof(m->footer);
con->in_tag = CEPH_MSGR_TAG_READY;
- return 0;
+ return 1;
} else if ((s64)seq - (s64)con->in_seq > 1) {
pr_err("read_partial_message bad seq %lld expected %lld\n",
seq, con->in_seq + 1);
@@ -2313,7 +2313,7 @@ static int read_partial_message(struct ceph_connection *con)
sizeof(m->footer);
con->in_tag = CEPH_MSGR_TAG_READY;
con->in_seq++;
- return 0;
+ return 1;
}

BUG_ON(!con->in_msg);
--
2.7.2