From: Peter Zijlstra <peterz@infradead.org>
To: Ingo Molnar <mingo@kernel.org>
Cc: Alfredo Alvarez Fernandez <alfredoalvarezfernandez@gmail.com>,
Linus Torvalds <torvalds@linux-foundation.org>,
Sedat Dilek <sedat.dilek@gmail.com>,
"Theodore Ts'o" <tytso@mit.edu>,
linux-fsdevel <linux-fsdevel@vger.kernel.org>,
LKML <linux-kernel@vger.kernel.org>
Subject: Re: [Linux-v4.6-rc1] ext4: WARNING: CPU: 2 PID: 2692 at kernel/locking/lockdep.c:2017 __lock_acquire+0x180e/0x2260
Date: Wed, 30 Mar 2016 11:36:59 +0200 [thread overview]
Message-ID: <20160330093659.GS3408@twins.programming.kicks-ass.net> (raw)
In-Reply-To: <20160329084701.GA9393@gmail.com>
On Tue, Mar 29, 2016 at 10:47:02AM +0200, Ingo Molnar wrote:
> > You are right; this is lockdep running into a hash collision; which is a new
> > DEBUG_LOCKDEP test. See 9e4e7554e755 ("locking/lockdep: Detect chain_key
> > collisions").
>
> I've Cc:-ed Alfredo Alvarez Fernandez who added that test.
OK, so while the code in check_no_collision() seems sensible, it relies
on borken bits.
The whole chain_hlocks and /proc/lockdep_chains stuff appears to have
been buggered from the start.
The below patch should fix this.
Furthermore, our hash function has definite room for improvement.
---
include/linux/lockdep.h | 8 +++++---
kernel/locking/lockdep.c | 30 ++++++++++++++++++++++++------
kernel/locking/lockdep_proc.c | 2 ++
3 files changed, 31 insertions(+), 9 deletions(-)
diff --git a/include/linux/lockdep.h b/include/linux/lockdep.h
index d026b190c530..2568c120513b 100644
--- a/include/linux/lockdep.h
+++ b/include/linux/lockdep.h
@@ -196,9 +196,11 @@ struct lock_list {
* We record lock dependency chains, so that we can cache them:
*/
struct lock_chain {
- u8 irq_context;
- u8 depth;
- u16 base;
+ /* see BUILD_BUG_ON()s in lookup_chain_cache() */
+ unsigned int irq_context : 2,
+ depth : 6,
+ base : 24;
+ /* 4 byte hole */
struct hlist_node entry;
u64 chain_key;
};
diff --git a/kernel/locking/lockdep.c b/kernel/locking/lockdep.c
index 53ab2f85d77e..91a4b7780afb 100644
--- a/kernel/locking/lockdep.c
+++ b/kernel/locking/lockdep.c
@@ -2099,15 +2099,37 @@ static inline int lookup_chain_cache(struct task_struct *curr,
chain->irq_context = hlock->irq_context;
i = get_first_held_lock(curr, hlock);
chain->depth = curr->lockdep_depth + 1 - i;
+
+ BUILD_BUG_ON((1UL << 24) <= ARRAY_SIZE(chain_hlocks));
+ BUILD_BUG_ON((1UL << 6) <= ARRAY_SIZE(curr->held_locks));
+ BUILD_BUG_ON((1UL << 8*sizeof(chain_hlocks[0])) <= ARRAY_SIZE(lock_classes));
+
if (likely(nr_chain_hlocks + chain->depth <= MAX_LOCKDEP_CHAIN_HLOCKS)) {
chain->base = nr_chain_hlocks;
- nr_chain_hlocks += chain->depth;
for (j = 0; j < chain->depth - 1; j++, i++) {
int lock_id = curr->held_locks[i].class_idx - 1;
chain_hlocks[chain->base + j] = lock_id;
}
chain_hlocks[chain->base + j] = class - lock_classes;
}
+
+ if (nr_chain_hlocks < MAX_LOCKDEP_CHAIN_HLOCKS)
+ nr_chain_hlocks += chain->depth;
+
+#ifdef CONFIG_DEBUG_LOCKDEP
+ /*
+ * Important for check_no_collision().
+ */
+ if (unlikely(nr_chain_hlocks > MAX_LOCKDEP_CHAIN_HLOCKS)) {
+ if (debug_locks_off_graph_unlock())
+ return 0;
+
+ print_lockdep_off("BUG: MAX_LOCKDEP_CHAIN_HLOCKS too low!");
+ dump_stack();
+ return 0;
+ }
+#endif
+
hlist_add_head_rcu(&chain->entry, hash_head);
debug_atomic_inc(chain_lookup_misses);
inc_chains();
@@ -2860,11 +2882,6 @@ static int separate_irq_context(struct task_struct *curr,
{
unsigned int depth = curr->lockdep_depth;
- /*
- * Keep track of points where we cross into an interrupt context:
- */
- hlock->irq_context = 2*(curr->hardirq_context ? 1 : 0) +
- curr->softirq_context;
if (depth) {
struct held_lock *prev_hlock;
@@ -3164,6 +3181,7 @@ static int __lock_acquire(struct lockdep_map *lock, unsigned int subclass,
hlock->acquire_ip = ip;
hlock->instance = lock;
hlock->nest_lock = nest_lock;
+ hlock->irq_context = 2*(!!curr->hardirq_context) + !!curr->softirq_context;
hlock->trylock = trylock;
hlock->read = read;
hlock->check = check;
diff --git a/kernel/locking/lockdep_proc.c b/kernel/locking/lockdep_proc.c
index dbb61a302548..a0f61effad25 100644
--- a/kernel/locking/lockdep_proc.c
+++ b/kernel/locking/lockdep_proc.c
@@ -141,6 +141,8 @@ static int lc_show(struct seq_file *m, void *v)
int i;
if (v == SEQ_START_TOKEN) {
+ if (nr_chain_hlocks > MAX_LOCKDEP_CHAIN_HLOCKS)
+ seq_printf(m, "(buggered) ");
seq_printf(m, "all lock chains:\n");
return 0;
}
next prev parent reply other threads:[~2016-03-30 9:37 UTC|newest]
Thread overview: 39+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-03-27 8:15 [Linux-v4.6-rc1] ext4: WARNING: CPU: 2 PID: 2692 at kernel/locking/lockdep.c:2017 __lock_acquire+0x180e/0x2260 Sedat Dilek
2016-03-27 8:57 ` Sedat Dilek
2016-03-27 12:03 ` Linus Torvalds
2016-03-27 13:32 ` Boqun Feng
2016-03-27 18:23 ` Theodore Ts'o
2016-03-27 19:40 ` Sedat Dilek
[not found] ` <CA+55aFwoKRpgq8OCTxUaP+8gOg-mnN3nbruYgiK32a=C5U4TkQ@mail.gmail.com>
2016-03-27 20:24 ` Sedat Dilek
2016-03-27 20:48 ` Peter Zijlstra
2016-03-27 20:59 ` Sedat Dilek
2016-03-27 21:48 ` Sedat Dilek
2016-03-28 1:05 ` Boqun Feng
2016-03-28 6:33 ` Peter Zijlstra
2016-03-29 8:47 ` Ingo Molnar
2016-03-30 9:20 ` Sedat Dilek
2016-03-30 9:36 ` Sedat Dilek
2016-03-30 9:36 ` Peter Zijlstra [this message]
2016-03-30 9:49 ` Sedat Dilek
2016-03-30 10:33 ` Sedat Dilek
2016-03-30 12:43 ` Peter Zijlstra
2016-03-30 12:46 ` Sedat Dilek
2016-03-30 13:15 ` Peter Zijlstra
2016-03-30 9:50 ` Peter Zijlstra
2016-03-30 9:59 ` Boqun Feng
2016-03-30 10:36 ` Peter Zijlstra
2016-03-30 11:07 ` Sedat Dilek
2016-03-31 15:42 ` Peter Zijlstra
2016-03-31 15:52 ` Boqun Feng
2016-04-02 6:26 ` Sedat Dilek
2016-03-30 14:06 ` Peter Zijlstra
2016-03-30 15:21 ` Sedat Dilek
2016-03-30 17:03 ` [PATCH] lockdep: print chain_key collision information Alfredo Alvarez Fernandez
2016-03-30 17:19 ` Peter Zijlstra
2016-04-01 6:36 ` [tip:core/urgent] locking/lockdep: Print " tip-bot for Alfredo Alvarez Fernandez
2016-05-10 9:09 ` Peter Zijlstra
2016-04-04 15:31 ` [Linux-v4.6-rc1] ext4: WARNING: CPU: 2 PID: 2692 at kernel/locking/lockdep.c:2017 __lock_acquire+0x180e/0x2260 Sedat Dilek
2016-04-04 16:02 ` Peter Zijlstra
2016-05-09 11:37 ` Sedat Dilek
2016-06-03 15:15 ` Sedat Dilek
2016-04-23 12:54 ` [tip:locking/urgent] lockdep: Fix lock_chain::base size tip-bot for Peter Zijlstra
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=20160330093659.GS3408@twins.programming.kicks-ass.net \
--to=peterz@infradead.org \
--cc=alfredoalvarezfernandez@gmail.com \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@kernel.org \
--cc=sedat.dilek@gmail.com \
--cc=torvalds@linux-foundation.org \
--cc=tytso@mit.edu \
/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