cluster-devel.redhat.com archive mirror
 help / color / mirror / Atom feed
From: Bob Peterson <rpeterso@redhat.com>
To: cluster-devel.redhat.com
Subject: [Cluster-devel] [PATCH 26/32] gfs2: drain the ail2 list after io errors
Date: Wed, 13 Nov 2019 15:30:24 -0600	[thread overview]
Message-ID: <20191113213030.237431-27-rpeterso@redhat.com> (raw)
In-Reply-To: <20191113213030.237431-1-rpeterso@redhat.com>

Before this patch, gfs2_logd continually tried to flush its journal
log, after the file system is withdrawn. We don't want to write anything
to the journal, lest we add corruption. Best course of action is to
drain the ail1 into the ail2 list (via gfs2_ail1_empty) then drain the
ail2 list with a new function, ail2_drain.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
---
 fs/gfs2/log.c   | 82 +++++++++++++++++++++++++++++++++++++++++++------
 fs/gfs2/trans.c |  4 +++
 2 files changed, 77 insertions(+), 9 deletions(-)

diff --git a/fs/gfs2/log.c b/fs/gfs2/log.c
index 359b85660da9..c4c832f6b6a3 100644
--- a/fs/gfs2/log.c
+++ b/fs/gfs2/log.c
@@ -302,25 +302,34 @@ static void gfs2_ail1_wait(struct gfs2_sbd *sdp)
 }
 
 /**
- * gfs2_ail2_empty_one - Check whether or not a trans in the AIL has been synced
- * @sdp: the filesystem
- * @ai: the AIL entry
- *
+ * gfs2_ail_empty_tr - empty one of the ail lists for a transaction
  */
 
-static void gfs2_ail2_empty_one(struct gfs2_sbd *sdp, struct gfs2_trans *tr)
+static void gfs2_ail_empty_tr(struct gfs2_sbd *sdp, struct gfs2_trans *tr,
+			      struct list_head *head)
 {
-	struct list_head *head = &tr->tr_ail2_list;
 	struct gfs2_bufdata *bd;
 
 	while (!list_empty(head)) {
-		bd = list_entry(head->prev, struct gfs2_bufdata,
-				bd_ail_st_list);
+		bd = list_first_entry(head, struct gfs2_bufdata,
+				      bd_ail_st_list);
 		gfs2_assert(sdp, bd->bd_tr == tr);
 		gfs2_remove_from_ail(bd);
 	}
 }
 
+/**
+ * gfs2_ail2_empty_one - Check whether or not a trans in the AIL has been synced
+ * @sdp: the filesystem
+ * @ai: the AIL entry
+ *
+ */
+
+static void gfs2_ail2_empty_one(struct gfs2_sbd *sdp, struct gfs2_trans *tr)
+{
+	gfs2_ail_empty_tr(sdp, tr, &tr->tr_ail2_list);
+}
+
 static void ail2_empty(struct gfs2_sbd *sdp, unsigned int new_tail)
 {
 	struct gfs2_trans *tr, *safe;
@@ -784,6 +793,41 @@ static void log_write_header(struct gfs2_sbd *sdp, u32 flags)
 		log_pull_tail(sdp, tail);
 }
 
+/**
+ * ail_drain - drain the ail lists after a withdraw
+ * @sdp: Pointer to GFS2 superblock
+ */
+static void ail_drain(struct gfs2_sbd *sdp)
+{
+	struct gfs2_trans *tr;
+
+	spin_lock(&sdp->sd_ail_lock);
+	/*
+	 * For transactions on the sd_ail1_list we need to drain both the
+	 * ail1 and ail2 lists. That's because function gfs2_ail1_start_one
+	 * (temporarily) moves items from its tr_ail1 list to tr_ail2 list
+	 * before revokes are sent for that block. Items on the sd_ail2_list
+	 * should have already gotten beyond that point, so no need.
+	 */
+	while (!list_empty(&sdp->sd_ail1_list)) {
+		tr = list_first_entry(&sdp->sd_ail1_list, struct gfs2_trans,
+				      tr_list);
+		gfs2_ail_empty_tr(sdp, tr, &tr->tr_ail1_list);
+		gfs2_ail_empty_tr(sdp, tr, &tr->tr_ail2_list);
+		list_del(&tr->tr_list);
+		gfs2_trans_free(sdp, tr);
+	}
+	while (!list_empty(&sdp->sd_ail2_list)) {
+		tr = list_first_entry(&sdp->sd_ail2_list, struct gfs2_trans,
+				      tr_list);
+		gfs2_ail_empty_tr(sdp, tr, &tr->tr_ail2_list);
+		/* Same as gfs2_ail2_empty_one(sdp, tr);*/
+		list_del(&tr->tr_list);
+		gfs2_trans_free(sdp, tr);
+	}
+	spin_unlock(&sdp->sd_ail_lock);
+}
+
 /**
  * gfs2_log_flush - flush incore transaction(s)
  * @sdp: the filesystem
@@ -794,11 +838,18 @@ static void log_write_header(struct gfs2_sbd *sdp, u32 flags)
 
 void gfs2_log_flush(struct gfs2_sbd *sdp, struct gfs2_glock *gl, u32 flags)
 {
-	struct gfs2_trans *tr;
+	struct gfs2_trans *tr = NULL;
 	enum gfs2_freeze_state state = atomic_read(&sdp->sd_freeze_state);
 
 	down_write(&sdp->sd_log_flush_lock);
 
+	/*
+	 * Do this check while holding the log_flush_lock to prevent new
+	 * buffers from being added to the ail via gfs2_pin()
+	 */
+	if (gfs2_withdrawn(sdp))
+		goto out;
+
 	/* Log might have been flushed while we waited for the flush lock */
 	if (gl && !test_bit(GLF_LFLUSH, &gl->gl_flags) &&
 	    atomic_read(&gl->gl_revokes) == 0) {
@@ -827,8 +878,14 @@ void gfs2_log_flush(struct gfs2_sbd *sdp, struct gfs2_glock *gl, u32 flags)
 			sdp->sd_log_num_revoke == sdp->sd_log_commited_revoke);
 
 	gfs2_ordered_write(sdp);
+	if (gfs2_withdrawn(sdp))
+		goto out;
 	lops_before_commit(sdp, tr);
+	if (gfs2_withdrawn(sdp))
+		goto out;
 	gfs2_log_submit_bio(&sdp->sd_log_bio, REQ_OP_WRITE);
+	if (gfs2_withdrawn(sdp))
+		goto out;
 
 	if (sdp->sd_log_head != sdp->sd_log_flush_head) {
 		log_flush_wait(sdp);
@@ -838,6 +895,8 @@ void gfs2_log_flush(struct gfs2_sbd *sdp, struct gfs2_glock *gl, u32 flags)
 		trace_gfs2_log_blocks(sdp, -1);
 		log_write_header(sdp, flags);
 	}
+	if (gfs2_withdrawn(sdp))
+		goto out;
 	lops_after_commit(sdp, tr);
 
 	gfs2_log_lock(sdp);
@@ -876,6 +935,11 @@ void gfs2_log_flush(struct gfs2_sbd *sdp, struct gfs2_glock *gl, u32 flags)
 	}
 
 out:
+	if (gfs2_withdrawn(sdp)) {
+		ail_drain(sdp); /* frees all transactions */
+		tr = NULL;
+	}
+
 	trace_gfs2_log_flush(sdp, 0, flags);
 	up_write(&sdp->sd_log_flush_lock);
 
diff --git a/fs/gfs2/trans.c b/fs/gfs2/trans.c
index 790fdd4e8c5e..88209a81c7bb 100644
--- a/fs/gfs2/trans.c
+++ b/fs/gfs2/trans.c
@@ -235,6 +235,10 @@ void gfs2_trans_add_meta(struct gfs2_glock *gl, struct buffer_head *bh)
 		fs_info(sdp, "GFS2:adding buf while frozen\n");
 		gfs2_assert_withdraw(sdp, 0);
 	}
+	if (unlikely(gfs2_withdrawn(sdp))) {
+		fs_info(sdp, "GFS2:adding buf while withdrawn! 0x%llx\n",
+			(unsigned long long)bd->bd_bh->b_blocknr);
+	}
 	gfs2_pin(sdp, bd->bd_bh);
 	mh->__pad0 = cpu_to_be64(0);
 	mh->mh_jid = cpu_to_be32(sdp->sd_jdesc->jd_jid);
-- 
2.23.0



  parent reply	other threads:[~2019-11-13 21:30 UTC|newest]

Thread overview: 37+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-11-13 21:29 [Cluster-devel] [PATCH 00/32] gfs2: misc recovery patch collection Bob Peterson
2019-11-13 21:29 ` [Cluster-devel] [PATCH 01/32] gfs2: Introduce concept of a pending withdraw Bob Peterson
2019-11-13 21:30 ` [Cluster-devel] [PATCH 02/32] gfs2: clear ail1 list when gfs2 withdraws Bob Peterson
2019-11-13 21:30 ` [Cluster-devel] [PATCH 03/32] gfs2: Rework how rgrp buffer_heads are managed Bob Peterson
2019-11-13 21:30 ` [Cluster-devel] [PATCH 04/32] gfs2: fix infinite loop in gfs2_ail1_flush on io error Bob Peterson
2019-11-13 21:30 ` [Cluster-devel] [PATCH 05/32] gfs2: log error reform Bob Peterson
2019-11-13 21:30 ` [Cluster-devel] [PATCH 06/32] gfs2: Only complain the first time an io error occurs in quota or log Bob Peterson
2019-11-13 21:30 ` [Cluster-devel] [PATCH 07/32] gfs2: Ignore dlm recovery requests if gfs2 is withdrawn Bob Peterson
2019-11-13 21:30 ` [Cluster-devel] [PATCH 08/32] gfs2: move check_journal_clean to util.c for future use Bob Peterson
2019-11-13 21:30 ` [Cluster-devel] [PATCH 09/32] gfs2: Allow some glocks to be used during withdraw Bob Peterson
2019-11-13 21:30 ` [Cluster-devel] [PATCH 10/32] gfs2: Don't loop forever in gfs2_freeze if withdrawn Bob Peterson
2019-11-13 21:30 ` [Cluster-devel] [PATCH 11/32] gfs2: Make secondary withdrawers wait for first withdrawer Bob Peterson
2019-11-13 21:30 ` [Cluster-devel] [PATCH 12/32] gfs2: Don't write log headers after file system withdraw Bob Peterson
2019-11-13 21:30 ` [Cluster-devel] [PATCH 13/32] gfs2: Force withdraw to replay journals and wait for it to finish Bob Peterson
2019-11-13 21:30 ` [Cluster-devel] [PATCH 14/32] gfs2: fix infinite loop when checking ail item count before go_inval Bob Peterson
2019-11-13 21:30 ` [Cluster-devel] [PATCH 15/32] gfs2: Add verbose option to check_journal_clean Bob Peterson
2019-11-13 21:30 ` [Cluster-devel] [PATCH 16/32] gfs2: Abort gfs2_freeze if io error is seen Bob Peterson
2019-11-13 21:30 ` [Cluster-devel] [PATCH 17/32] gfs2: Issue revokes more intelligently Bob Peterson
2019-11-13 21:30 ` [Cluster-devel] [PATCH 18/32] gfs2: Prepare to withdraw as soon as an IO error occurs in log write Bob Peterson
2019-11-13 21:30 ` [Cluster-devel] [PATCH 19/32] gfs2: Check for log write errors before telling dlm to unlock Bob Peterson
2019-11-13 21:30 ` [Cluster-devel] [PATCH 20/32] gfs2: new slab for transactions Bob Peterson
2019-11-13 21:30 ` [Cluster-devel] [PATCH 21/32] gfs2: Close timing window with GLF_INVALIDATE_IN_PROGRESS Bob Peterson
2019-11-13 21:30 ` [Cluster-devel] [PATCH 22/32] gfs2: Do log_flush in gfs2_ail_empty_gl even if ail list is empty Bob Peterson
2019-11-13 21:30 ` [Cluster-devel] [PATCH 23/32] gfs2: Don't skip log flush if glock still has revokes Bob Peterson
2019-11-13 21:30 ` [Cluster-devel] [PATCH 24/32] gfs2: initialize tr_ail1_list when creating transactions Bob Peterson
2019-11-13 21:30 ` [Cluster-devel] [PATCH 25/32] gfs2: Withdraw in gfs2_ail1_flush if write_cache_pages returns error Bob Peterson
2019-11-13 21:30 ` Bob Peterson [this message]
2019-11-13 21:30 ` [Cluster-devel] [PATCH 27/32] gfs2: make gfs2_log_shutdown static Bob Peterson
2019-11-13 21:30 ` [Cluster-devel] [PATCH 28/32] gfs2: Eliminate GFS2_RDF_UPTODATE flag in favor of buffer existence Bob Peterson
2019-11-14 10:42   ` Steven Whitehouse
2019-11-14 13:16     ` Bob Peterson
2019-11-13 21:30 ` [Cluster-devel] [PATCH 29/32] gfs2: if finish_open returns error, clean up iopen glock mess Bob Peterson
2019-11-13 21:30 ` [Cluster-devel] [PATCH 30/32] gfs2: Don't demote a glock until its revokes are written Bob Peterson
2019-11-14 10:45   ` Steven Whitehouse
2019-11-13 21:30 ` [Cluster-devel] [PATCH 31/32] gfs2: Do proper error checking for go_sync family of glops functions Bob Peterson
2019-11-13 21:30 ` [Cluster-devel] [PATCH 32/32] gfs2: fix glock reference problem in gfs2_trans_add_unrevoke Bob Peterson
2019-11-14 10:48 ` [Cluster-devel] [PATCH 00/32] gfs2: misc recovery patch collection Steven Whitehouse

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=20191113213030.237431-27-rpeterso@redhat.com \
    --to=rpeterso@redhat.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;
as well as URLs for NNTP newsgroup(s).