[Cluster-devel] [PATCH AUTOSEL 5.12 20/49] gfs2: Prevent direct-I/O write fallback errors from getting lost

[Cluster-devel] [PATCH AUTOSEL 5.12 20/49] gfs2: Prevent direct-I/O write fallback errors from getting lost

[Sasha Levin]

From: Andreas Gruenbacher <agruenba@redhat.com>

[ Upstream commit 43a511c44e58e357a687d61a20cf5ef1dc9e5a7c ]

When a direct I/O write falls entirely and falls back to buffered I/O and the
buffered I/O fails, the write failed with return value 0 instead of the error
number reported by the buffered I/O. Fix that.

Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 fs/gfs2/file.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/fs/gfs2/file.c b/fs/gfs2/file.c
index 2d500f90cdac..a86e6810237a 100644
--- a/fs/gfs2/file.c
+++ b/fs/gfs2/file.c
@@ -935,8 +935,11 @@ static ssize_t gfs2_file_write_iter(struct kiocb *iocb, struct iov_iter *from)
 		current->backing_dev_info = inode_to_bdi(inode);
 		buffered = iomap_file_buffered_write(iocb, from, &gfs2_iomap_ops);
 		current->backing_dev_info = NULL;
-		if (unlikely(buffered <= 0))
+		if (unlikely(buffered <= 0)) {
+			if (!ret)
+				ret = buffered;
 			goto out_unlock;
+		}
 
 		/*
 		 * We need to ensure that the page cache pages are written to
-- 
2.30.2

[Cluster-devel] [PATCH AUTOSEL 5.12 21/49] gfs2: fix a deadlock on withdraw-during-mount

[Sasha Levin]

From: Bob Peterson <rpeterso@redhat.com>

[ Upstream commit 865cc3e9cc0b1d4b81c10d53174bced76decf888 ]

Before this patch, gfs2 would deadlock because of the following
sequence during mount:

mount
   gfs2_fill_super
      gfs2_make_fs_rw <--- Detects IO error with glock
         kthread_stop(sdp->sd_quotad_process);
            <--- Blocked waiting for quotad to finish

logd
   Detects IO error and the need to withdraw
   calls gfs2_withdraw
      gfs2_make_fs_ro
         kthread_stop(sdp->sd_quotad_process);
            <--- Blocked waiting for quotad to finish

gfs2_quotad
   gfs2_statfs_sync
      gfs2_glock_wait <---- Blocked waiting for statfs glock to be granted

glock_work_func
   do_xmote <---Detects IO error, can't release glock: blocked on withdraw
      glops->go_inval
      glock_blocked_by_withdraw
         requeue glock work & exit <--- work requeued, blocked by withdraw

This patch makes a special exception for the statfs system inode glock,
which allows the statfs glock UNLOCK to proceed normally. That allows the
quotad daemon to exit during the withdraw, which allows the logd daemon
to exit during the withdraw, which allows the mount to exit.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 fs/gfs2/glock.c | 24 +++++++++++++++++++++---
 1 file changed, 21 insertions(+), 3 deletions(-)

diff --git a/fs/gfs2/glock.c b/fs/gfs2/glock.c
index 9567520d79f7..142f746d7b33 100644
--- a/fs/gfs2/glock.c
+++ b/fs/gfs2/glock.c
@@ -583,6 +583,16 @@ static void finish_xmote(struct gfs2_glock *gl, unsigned int ret)
 	spin_unlock(&gl->gl_lockref.lock);
 }
 
+static bool is_system_glock(struct gfs2_glock *gl)
+{
+	struct gfs2_sbd *sdp = gl->gl_name.ln_sbd;
+	struct gfs2_inode *m_ip = GFS2_I(sdp->sd_statfs_inode);
+
+	if (gl == m_ip->i_gl)
+		return true;
+	return false;
+}
+
 /**
  * do_xmote - Calls the DLM to change the state of a lock
  * @gl: The lock state
@@ -672,17 +682,25 @@ __acquires(&gl->gl_lockref.lock)
 	 * to see sd_log_error and withdraw, and in the meantime, requeue the
 	 * work for later.
 	 *
+	 * We make a special exception for some system glocks, such as the
+	 * system statfs inode glock, which needs to be granted before the
+	 * gfs2_quotad daemon can exit, and that exit needs to finish before
+	 * we can unmount the withdrawn file system.
+	 *
 	 * However, if we're just unlocking the lock (say, for unmount, when
 	 * gfs2_gl_hash_clear calls clear_glock) and recovery is complete
 	 * then it's okay to tell dlm to unlock it.
 	 */
 	if (unlikely(sdp->sd_log_error && !gfs2_withdrawn(sdp)))
 		gfs2_withdraw_delayed(sdp);
-	if (glock_blocked_by_withdraw(gl)) {
-		if (target != LM_ST_UNLOCKED ||
-		    test_bit(SDF_WITHDRAW_RECOVERY, &sdp->sd_flags)) {
+	if (glock_blocked_by_withdraw(gl) &&
+	    (target != LM_ST_UNLOCKED ||
+	     test_bit(SDF_WITHDRAW_RECOVERY, &sdp->sd_flags))) {
+		if (!is_system_glock(gl)) {
 			gfs2_glock_queue_work(gl, GL_GLOCK_DFT_HOLD);
 			goto out;
+		} else {
+			clear_bit(GLF_INVALIDATE_IN_PROGRESS, &gl->gl_flags);
 		}
 	}
 
-- 
2.30.2

[Cluster-devel] [PATCH AUTOSEL 5.12 22/49] gfs2: Clean up revokes on normal withdraws

[Sasha Levin]

From: Bob Peterson <rpeterso@redhat.com>

[ Upstream commit f5456b5d67cf812fd31fe3e130ca216b2e0908e5 ]

Before this patch, the system ail lists were cleaned up if the logd
process withdrew, but on other withdraws, they were not cleaned up.
This included the cleaning up of the revokes as well.

This patch reorganizes things a bit so that all withdraws (not just logd)
clean up the ail lists, including any pending revokes.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 fs/gfs2/log.c  | 6 +++---
 fs/gfs2/log.h  | 1 +
 fs/gfs2/lops.c | 7 ++++++-
 fs/gfs2/lops.h | 1 +
 fs/gfs2/util.c | 1 +
 5 files changed, 12 insertions(+), 4 deletions(-)

diff --git a/fs/gfs2/log.c b/fs/gfs2/log.c
index 6410281546f9..47287a7056fe 100644
--- a/fs/gfs2/log.c
+++ b/fs/gfs2/log.c
@@ -923,10 +923,10 @@ static void log_write_header(struct gfs2_sbd *sdp, u32 flags)
 }
 
 /**
- * ail_drain - drain the ail lists after a withdraw
+ * gfs2_ail_drain - drain the ail lists after a withdraw
  * @sdp: Pointer to GFS2 superblock
  */
-static void ail_drain(struct gfs2_sbd *sdp)
+void gfs2_ail_drain(struct gfs2_sbd *sdp)
 {
 	struct gfs2_trans *tr;
 
@@ -953,6 +953,7 @@ static void ail_drain(struct gfs2_sbd *sdp)
 		list_del(&tr->tr_list);
 		gfs2_trans_free(sdp, tr);
 	}
+	gfs2_drain_revokes(sdp);
 	spin_unlock(&sdp->sd_ail_lock);
 }
 
@@ -1159,7 +1160,6 @@ void gfs2_log_flush(struct gfs2_sbd *sdp, struct gfs2_glock *gl, u32 flags)
 	if (tr && list_empty(&tr->tr_list))
 		list_add(&tr->tr_list, &sdp->sd_ail1_list);
 	spin_unlock(&sdp->sd_ail_lock);
-	ail_drain(sdp); /* frees all transactions */
 	tr = NULL;
 	goto out_end;
 }
diff --git a/fs/gfs2/log.h b/fs/gfs2/log.h
index eea58015710e..fc905c2af53c 100644
--- a/fs/gfs2/log.h
+++ b/fs/gfs2/log.h
@@ -93,5 +93,6 @@ extern int gfs2_logd(void *data);
 extern void gfs2_add_revoke(struct gfs2_sbd *sdp, struct gfs2_bufdata *bd);
 extern void gfs2_glock_remove_revoke(struct gfs2_glock *gl);
 extern void gfs2_flush_revokes(struct gfs2_sbd *sdp);
+extern void gfs2_ail_drain(struct gfs2_sbd *sdp);
 
 #endif /* __LOG_DOT_H__ */
diff --git a/fs/gfs2/lops.c b/fs/gfs2/lops.c
index a82f4747aa8d..ef44d325e518 100644
--- a/fs/gfs2/lops.c
+++ b/fs/gfs2/lops.c
@@ -882,7 +882,7 @@ static void revoke_lo_before_commit(struct gfs2_sbd *sdp, struct gfs2_trans *tr)
 	gfs2_log_write_page(sdp, page);
 }
 
-static void revoke_lo_after_commit(struct gfs2_sbd *sdp, struct gfs2_trans *tr)
+void gfs2_drain_revokes(struct gfs2_sbd *sdp)
 {
 	struct list_head *head = &sdp->sd_log_revokes;
 	struct gfs2_bufdata *bd;
@@ -897,6 +897,11 @@ static void revoke_lo_after_commit(struct gfs2_sbd *sdp, struct gfs2_trans *tr)
 	}
 }
 
+static void revoke_lo_after_commit(struct gfs2_sbd *sdp, struct gfs2_trans *tr)
+{
+	gfs2_drain_revokes(sdp);
+}
+
 static void revoke_lo_before_scan(struct gfs2_jdesc *jd,
 				  struct gfs2_log_header_host *head, int pass)
 {
diff --git a/fs/gfs2/lops.h b/fs/gfs2/lops.h
index 31b6dd0d2e5d..f707601597dc 100644
--- a/fs/gfs2/lops.h
+++ b/fs/gfs2/lops.h
@@ -20,6 +20,7 @@ extern void gfs2_log_submit_bio(struct bio **biop, int opf);
 extern void gfs2_pin(struct gfs2_sbd *sdp, struct buffer_head *bh);
 extern int gfs2_find_jhead(struct gfs2_jdesc *jd,
 			   struct gfs2_log_header_host *head, bool keep_cache);
+extern void gfs2_drain_revokes(struct gfs2_sbd *sdp);
 static inline unsigned int buf_limit(struct gfs2_sbd *sdp)
 {
 	return sdp->sd_ldptrs;
diff --git a/fs/gfs2/util.c b/fs/gfs2/util.c
index 4f034b87b427..cffb346fb9b3 100644
--- a/fs/gfs2/util.c
+++ b/fs/gfs2/util.c
@@ -130,6 +130,7 @@ static void signal_our_withdraw(struct gfs2_sbd *sdp)
 	if (test_bit(SDF_NORECOVERY, &sdp->sd_flags) || !sdp->sd_jdesc)
 		return;
 
+	gfs2_ail_drain(sdp); /* frees all transactions */
 	inode = sdp->sd_jdesc->jd_inode;
 	ip = GFS2_I(inode);
 	i_gl = ip->i_gl;
-- 
2.30.2

[Cluster-devel] [PATCH AUTOSEL 5.12 29/49] gfs2: Fix use-after-free in gfs2_glock_shrink_scan

[Sasha Levin]

From: Hillf Danton <hdanton@sina.com>

[ Upstream commit 1ab19c5de4c537ec0d9b21020395a5b5a6c059b2 ]

The GLF_LRU flag is checked under lru_lock in gfs2_glock_remove_from_lru() to
remove the glock from the lru list in __gfs2_glock_put().

On the shrink scan path, the same flag is cleared under lru_lock but because
of cond_resched_lock(&lru_lock) in gfs2_dispose_glock_lru(), progress on the
put side can be made without deleting the glock from the lru list.

Keep GLF_LRU across the race window opened by cond_resched_lock(&lru_lock) to
ensure correct behavior on both sides - clear GLF_LRU after list_del under
lru_lock.

Reported-by: syzbot <syzbot+34ba7ddbf3021981a228@syzkaller.appspotmail.com>
Signed-off-by: Hillf Danton <hdanton@sina.com>
Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 fs/gfs2/glock.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/gfs2/glock.c b/fs/gfs2/glock.c
index 142f746d7b33..a99e7295e19f 100644
--- a/fs/gfs2/glock.c
+++ b/fs/gfs2/glock.c
@@ -1790,6 +1790,7 @@ __acquires(&lru_lock)
 	while(!list_empty(list)) {
 		gl = list_first_entry(list, struct gfs2_glock, gl_lru);
 		list_del_init(&gl->gl_lru);
+		clear_bit(GLF_LRU, &gl->gl_flags);
 		if (!spin_trylock(&gl->gl_lockref.lock)) {
 add_back_to_lru:
 			list_add(&gl->gl_lru, &lru_list);
@@ -1835,7 +1836,6 @@ static long gfs2_scan_glock_lru(int nr)
 		if (!test_bit(GLF_LOCK, &gl->gl_flags)) {
 			list_move(&gl->gl_lru, &dispose);
 			atomic_dec(&lru_count);
-			clear_bit(GLF_LRU, &gl->gl_flags);
 			freed++;
 			continue;
 		}
-- 
2.30.2