Linux CIFS filesystem development
 help / color / mirror / Atom feed
From: Sasha Levin <sashal@kernel.org>
To: linux-kernel@vger.kernel.org, stable@vger.kernel.org
Cc: Pavel Shilovsky <pshilov@microsoft.com>,
	Steve French <stfrench@microsoft.com>,
	Sasha Levin <sashal@kernel.org>,
	linux-cifs@vger.kernel.org
Subject: [PATCH AUTOSEL 4.20 40/42] CIFS: Move credit processing to mid callbacks for SMB3
Date: Sat,  9 Feb 2019 13:47:32 -0500	[thread overview]
Message-ID: <20190209184734.125935-40-sashal@kernel.org> (raw)
In-Reply-To: <20190209184734.125935-1-sashal@kernel.org>

From: Pavel Shilovsky <pshilov@microsoft.com>

[ Upstream commit ee258d79159afed52ca9372aeb9c1a51e89b32ee ]

Currently we account for credits in the thread initiating a request
and waiting for a response. The demultiplex thread receives the response,
wakes up the thread and the latter collects credits from the response
buffer and add them to the server structure on the client. This approach
is not accurate, because it may race with reconnect events in the
demultiplex thread which resets the number of credits.

Fix this by moving credit processing to new mid callbacks that collect
credits granted by the server from the response in the demultiplex thread.

Signed-off-by: Pavel Shilovsky <pshilov@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 fs/cifs/transport.c | 51 ++++++++++++++++++++++++++++++---------------
 1 file changed, 34 insertions(+), 17 deletions(-)

diff --git a/fs/cifs/transport.c b/fs/cifs/transport.c
index 4dbf62bb51b2..0dab276eced8 100644
--- a/fs/cifs/transport.c
+++ b/fs/cifs/transport.c
@@ -781,12 +781,7 @@ cifs_setup_request(struct cifs_ses *ses, struct smb_rqst *rqst)
 }
 
 static void
-cifs_noop_callback(struct mid_q_entry *mid)
-{
-}
-
-static void
-cifs_cancelled_callback(struct mid_q_entry *mid)
+cifs_compound_callback(struct mid_q_entry *mid)
 {
 	struct TCP_Server_Info *server = mid->server;
 	unsigned int optype = mid->optype;
@@ -799,10 +794,23 @@ cifs_cancelled_callback(struct mid_q_entry *mid)
 			cifs_dbg(FYI, "Bad state for cancelled MID\n");
 	}
 
-	DeleteMidQEntry(mid);
 	add_credits(server, credits_received, optype);
 }
 
+static void
+cifs_compound_last_callback(struct mid_q_entry *mid)
+{
+	cifs_compound_callback(mid);
+	cifs_wake_up_task(mid);
+}
+
+static void
+cifs_cancelled_callback(struct mid_q_entry *mid)
+{
+	cifs_compound_callback(mid);
+	DeleteMidQEntry(mid);
+}
+
 int
 compound_send_recv(const unsigned int xid, struct cifs_ses *ses,
 		   const int flags, const int num_rqst, struct smb_rqst *rqst,
@@ -880,11 +888,14 @@ compound_send_recv(const unsigned int xid, struct cifs_ses *ses,
 		midQ[i]->mid_state = MID_REQUEST_SUBMITTED;
 		midQ[i]->optype = optype;
 		/*
-		 * We don't invoke the callback compounds unless it is the last
-		 * request.
+		 * Invoke callback for every part of the compound chain
+		 * to calculate credits properly. Wake up this thread only when
+		 * the last element is received.
 		 */
 		if (i < num_rqst - 1)
-			midQ[i]->callback = cifs_noop_callback;
+			midQ[i]->callback = cifs_compound_callback;
+		else
+			midQ[i]->callback = cifs_compound_last_callback;
 	}
 	cifs_in_send_inc(ses->server);
 	rc = smb_send_rqst(ses->server, num_rqst, rqst, flags);
@@ -898,8 +909,20 @@ compound_send_recv(const unsigned int xid, struct cifs_ses *ses,
 
 	mutex_unlock(&ses->server->srv_mutex);
 
-	if (rc < 0)
+	if (rc < 0) {
+		/* Sending failed for some reason - return credits back */
+		for (i = 0; i < num_rqst; i++)
+			add_credits(ses->server, credits[i], optype);
 		goto out;
+	}
+
+	/*
+	 * At this point the request is passed to the network stack - we assume
+	 * that any credits taken from the server structure on the client have
+	 * been spent and we can't return them back. Once we receive responses
+	 * we will collect credits granted by the server in the mid callbacks
+	 * and add those credits to the server structure.
+	 */
 
 	/*
 	 * Compounding is never used during session establish.
@@ -932,11 +955,6 @@ compound_send_recv(const unsigned int xid, struct cifs_ses *ses,
 		}
 	}
 
-	for (i = 0; i < num_rqst; i++)
-		if (!cancelled_mid[i] && midQ[i]->resp_buf
-		    && (midQ[i]->mid_state == MID_RESPONSE_RECEIVED))
-			credits[i] = ses->server->ops->get_credits(midQ[i]);
-
 	for (i = 0; i < num_rqst; i++) {
 		if (rc < 0)
 			goto out;
@@ -995,7 +1013,6 @@ out:
 	for (i = 0; i < num_rqst; i++) {
 		if (!cancelled_mid[i])
 			cifs_delete_mid(midQ[i]);
-		add_credits(ses->server, credits[i], optype);
 	}
 
 	return rc;
-- 
2.19.1

  parent reply	other threads:[~2019-02-09 18:47 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
     [not found] <20190209184734.125935-1-sashal@kernel.org>
2019-02-09 18:47 ` [PATCH AUTOSEL 4.20 38/42] cifs: Limit memory used by lock request calls to a page Sasha Levin
2019-02-09 18:47 ` [PATCH AUTOSEL 4.20 39/42] CIFS: Fix credits calculation for cancelled requests Sasha Levin
2019-02-09 18:47 ` Sasha Levin [this message]
2019-02-12  1:48   ` [PATCH AUTOSEL 4.20 40/42] CIFS: Move credit processing to mid callbacks for SMB3 Pavel Shilovskiy
2019-02-09 18:47 ` [PATCH AUTOSEL 4.20 41/42] CIFS: Fix error paths in writeback code Sasha Levin

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=20190209184734.125935-40-sashal@kernel.org \
    --to=sashal@kernel.org \
    --cc=linux-cifs@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=pshilov@microsoft.com \
    --cc=stable@vger.kernel.org \
    --cc=stfrench@microsoft.com \
    /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