drbd: fix regression 'out of mem, failed to invoke fence-peer helper'
Subsystems (2)
A message that is not in the archive (sent off-list or not yet received): [email protected]
Philipp Reisner
maintainerPATCH 1/1 drbd: fix regression 'out of mem, failed to invoke fence-peer helper'
From: Lars Ellenberg <[email protected]>
Stable info:
This patch landed in upstream with v3.16 as commit bbc1c5e8ad6dfebf9d13b8a4ccdf66c92913eac9 it should go into v3.14+
Since linux kernel 3.13, kthread_run() internally uses wait_for_completion_killable(). We sometimes may use kthread_run() while we still have a signal pending, which we used to kick our threads out of potentially blocking network functions, causing kthread_run() to mistake that as a new fatal signal and fail.
Fix: flush_signals() before kthread_run().
--- drivers/block/drbd/drbd_nl.c | 6 ++++++ 1 file changed, 6 insertions(+)
drivers/block/drbd/drbd_nl.c +6 -0
@@ -544,6 +544,12 @@ void conn_try_outdate_peer_async(struct drbd_connection *connection)struct task_struct *opa;kref_get (&connection->kref); +/* We may just have force_sig()'ed this thread +* to get it out of some blocking network function. +* Clear signals; otherwise kthread_run(), which internally uses +* wait_on_completion_killable(), will mistake our pending signal +* for a new fatal signal and fail. */ +flush_signals (current); opa =kthread_run (_try_outdate_peer_async, connection,"drbd_async_h" );if (IS_ERR (opa)) {drbd_err (connection,"out of mem, failed to invoke fence-peer helper\n" );
Signature · 1 line
1.9.1
Quoted 29 lines · Philipp Reisner
From: Lars Ellenberg <[email protected]>
Stable info:
This patch landed in upstream with v3.16 as commitbbc1c5e8ad6dfebf9d13b8a4ccdf66c92913eac9it should go into v3.14+Since linux kernel 3.13,
kthread_run()internally useswait_for_completion_killable(). We sometimes may usekthread_run()while we still have a signal pending, which we used to kick our threads out of potentially blocking network functions, causingkthread_run()to mistake that as a new fatal signal and fail.Fix:
flush_signals()beforekthread_run().Signed-off-by Philipp Reisner <[email protected]>Signed-off-by Lars Ellenberg <[email protected]>Signed-off-by Jens Axboe <[email protected]>--- drivers/block/drbd/drbd_nl.c | 6 ++++++ 1 file changed, 6 insertions(+)drivers/block/drbd/drbd_nl.c +0
-0@@ -544,6 +544,12 @@ void conn_try_outdate_peer_async(struct drbd_connection *connection)struct task_struct *opa;
drivers/block/drbd/drbd_nl.c void conn_try_outdate_peer_async(struct drbd_connection *connection) · line 555
Quoted 10 lines
drivers/block/drbd/drbd_nl.c +6
-0kref_get (&connection->kref); +/* We may just have force_sig()'ed this thread +* to get it out of some blocking network function. +* Clear signals; otherwise kthread_run(), which internally uses +* wait_on_completion_killable(), will mistake our pending signal +* for a new fatal signal and fail. */ +flush_signals (current); opa =kthread_run (_try_outdate_peer_async, connection,"drbd_async_h" );if (IS_ERR (opa)) {drbd_err (connection,"out of mem, failed to invoke fence-peer helper\n" );
This doesn't apply to 3.16-stable or 3.14-stable, can you please provide a working backport?
thanks,
greg k-h
Lars Ellenberg
maintainerQuoted 30 lines · Greg KH
Quoted 29 lines · Philipp Reisner
From: Lars Ellenberg <[email protected]>
Stable info:
This patch landed in upstream with v3.16 as commitbbc1c5e8ad6dfebf9d13b8a4ccdf66c92913eac9it should go into v3.14+Since linux kernel 3.13,
kthread_run()internally useswait_for_completion_killable(). We sometimes may usekthread_run()while we still have a signal pending, which we used to kick our threads out of potentially blocking network functions, causingkthread_run()to mistake that as a new fatal signal and fail.Fix:
flush_signals()beforekthread_run().Signed-off-by Philipp Reisner <[email protected]>Signed-off-by Lars Ellenberg <[email protected]>Signed-off-by Jens Axboe <[email protected]>--- drivers/block/drbd/drbd_nl.c | 6 ++++++ 1 file changed, 6 insertions(+)drivers/block/drbd/drbd_nl.c +0
-0@@ -544,6 +544,12 @@ void conn_try_outdate_peer_async(struct drbd_connection *connection)struct task_struct *opa;
Quoted 13 lines
drivers/block/drbd/drbd_nl.c void conn_try_outdate_peer_async(struct drbd_connection *connection) · line 555
Quoted 10 lines
drivers/block/drbd/drbd_nl.c +6
-0kref_get (&connection->kref); +/* We may just have force_sig()'ed this thread +* to get it out of some blocking network function. +* Clear signals; otherwise kthread_run(), which internally uses +* wait_on_completion_killable(), will mistake our pending signal +* for a new fatal signal and fail. */ +flush_signals (current); opa =kthread_run (_try_outdate_peer_async, connection,"drbd_async_h" );if (IS_ERR (opa)) {drbd_err (connection,"out of mem, failed to invoke fence-peer helper\n" );This doesn't apply to 3.16-stable or 3.14-stable, can you please provide a working backport?
There was a rename of "tconn" to "connection" between 3.14 and .15. Other than that, this has not changed.
Below applies to 3.13 and 3.14 stable as of today.
Lars
8<----
From a82efa2adeb992b5ded798b01b4567bc07b6ab1b Mon Sep 17 00:00:00 2001
From: Lars Ellenberg <[email protected]>
Date: Tue, 7 Oct 2014 17:20:27 +0200
Subject: [PATCH] drbd: fix regression 'out of mem, failed to invoke fence-peer
helper'
Stable info:
This patch landed in upstream with v3.16 as commit bbc1c5e8ad6dfebf9d13b8a4ccdf66c92913eac9 it should go into v3.13+
Since linux kernel 3.13, kthread_run() internally uses wait_for_completion_killable(). We sometimes may use kthread_run() while we still have a signal pending, which we used to kick our threads out of potentially blocking network functions, causing kthread_run() to mistake that as a new fatal signal and fail.
Fix: flush_signals() before kthread_run().
--- drivers/block/drbd/drbd_nl.c | 6 ++++++ 1 file changed, 6 insertions(+)
drivers/block/drbd/drbd_nl.c +6 -0
@@ -525,6 +525,12 @@ void conn_try_outdate_peer_async(struct drbd_tconn *tconn)struct task_struct *opa;kref_get (&tconn->kref); +/* We may just have force_sig()'ed this thread +* to get it out of some blocking network function. +* Clear signals; otherwise kthread_run(), which internally uses +* wait_on_completion_killable(), will mistake our pending signal +* for a new fatal signal and fail. */ +flush_signals (current); opa =kthread_run (_try_outdate_peer_async, tconn,"drbd_async_h" );if (IS_ERR (opa)) {conn_err (tconn,"out of mem, failed to invoke fence-peer helper\n" );
Signature · 1 line
1.9.1