From: Peter Zijlstra <peterz@infradead.org>
To: Byungchul Park <byungchul.park@lge.com>
Cc: mingo@kernel.org, tglx@linutronix.de, walken@google.com,
boqun.feng@gmail.com, kirill@shutemov.name,
linux-kernel@vger.kernel.org, linux-mm@kvack.org,
akpm@linux-foundation.org, willy@infradead.org,
npiggin@gmail.com, kernel-team@lge.com
Subject: Re: [PATCH v7 06/16] lockdep: Detect and handle hist_lock ring buffer overwrite
Date: Thu, 13 Jul 2017 10:14:42 +0200 [thread overview]
Message-ID: <20170713081442.GA439@worktop> (raw)
In-Reply-To: <20170713020745.GG20323@X58A-UD3R>
On Thu, Jul 13, 2017 at 11:07:45AM +0900, Byungchul Park wrote:
> Does my approach have problems, rewinding to 'original idx' on exit and
> deciding whether overwrite or not? I think, this way, no need to do the
> drastic work. Or.. does my one get more overhead in usual case?
So I think that invalidating just the one entry doesn't work; the moment
you fill that up the iteration in commit_xhlocks() will again use the
next one etc.. even though you wanted it not to.
So we need to wipe the _entire_ history.
So I _think_ the below should work, but its not been near a compiler.
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -822,6 +822,7 @@ struct task_struct {
unsigned int xhlock_idx_soft; /* For restoring at softirq exit */
unsigned int xhlock_idx_hard; /* For restoring at hardirq exit */
unsigned int xhlock_idx_hist; /* For restoring at history boundaries */
+ unsigned int xhlock_idX_max;
#endif
#ifdef CONFIG_UBSAN
unsigned int in_ubsan;
--- a/kernel/locking/lockdep.c
+++ b/kernel/locking/lockdep.c
@@ -4746,6 +4746,14 @@ EXPORT_SYMBOL_GPL(lockdep_rcu_suspicious
static atomic_t cross_gen_id; /* Can be wrapped */
/*
+ * make xhlock_valid() false.
+ */
+static inline void invalidate_xhlock(struct hist_lock *xhlock)
+{
+ xhlock->hlock.instance = NULL;
+}
+
+/*
* Lock history stacks; we have 3 nested lock history stacks:
*
* Hard IRQ
@@ -4764,28 +4772,58 @@ static atomic_t cross_gen_id; /* Can be
* MAX_XHLOCKS_NR ? Possibly re-instroduce hist_gen_id ?
*/
-void crossrelease_hardirq_start(void)
+static inline void __crossrelease_start(unsigned int *stamp)
{
if (current->xhlocks)
- current->xhlock_idx_hard = current->xhlock_idx;
+ *stamp = current->xhlock_idx;
+}
+
+static void __crossrelease_end(unsigned int *stamp)
+{
+ int i;
+
+ if (!current->xhlocks)
+ return;
+
+ current->xhlock_idx = *stamp;
+
+ /*
+ * If we rewind past the tail; all of history is lost.
+ */
+ if ((current->xhlock_idx_max - *stamp) < MAX_XHLOCKS_NR)
+ return;
+
+ /*
+ * Invalidate the entire history..
+ */
+ for (i = 0; i < MAX_XHLOCKS_NR; i++)
+ invalidate_xhlock(&xhlock(i));
+
+ current->xhlock_idx = 0;
+ current->xhlock_idx_hard = 0;
+ current->xhlock_idx_soft = 0;
+ current->xhlock_idx_hist = 0;
+ current->xhlock_idx_max = 0;
+}
+
+void crossrelease_hardirq_start(void)
+{
+ __crossrelease_start(¤t->xhlock_idx_hard);
}
void crossrelease_hardirq_end(void)
{
- if (current->xhlocks)
- current->xhlock_idx = current->xhlock_idx_hard;
+ __crossrelease_end(¤t->xhlock_idx_hard);
}
void crossrelease_softirq_start(void)
{
- if (current->xhlocks)
- current->xhlock_idx_soft = current->xhlock_idx;
+ __crossrelease_start(¤t->xhlock_idx_soft);
}
void crossrelease_softirq_end(void)
{
- if (current->xhlocks)
- current->xhlock_idx = current->xhlock_idx_soft;
+ __crossrelease_end(¤t->xhlock_idx_soft);
}
/*
@@ -4806,14 +4844,12 @@ void crossrelease_softirq_end(void)
*/
void crossrelease_hist_start(void)
{
- if (current->xhlocks)
- current->xhlock_idx_hist = current->xhlock_idx;
+ __crossrelease_start(¤t->xhlock_idx_hist);
}
void crossrelease_hist_end(void)
{
- if (current->xhlocks)
- current->xhlock_idx = current->xhlock_idx_hist;
+ __crossrelease_end(¤t->xhlock_idx_hist);
}
static int cross_lock(struct lockdep_map *lock)
@@ -4880,6 +4916,9 @@ static void add_xhlock(struct held_lock
unsigned int idx = ++current->xhlock_idx;
struct hist_lock *xhlock = &xhlock(idx);
+ if ((int)(current->xhlock_idx_max - idx) < 0)
+ current->xhlock_idx_max = idx;
+
#ifdef CONFIG_DEBUG_LOCKDEP
/*
* This can be done locklessly because they are all task-local
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
next prev parent reply other threads:[~2017-07-13 8:14 UTC|newest]
Thread overview: 41+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-05-24 8:59 [PATCH v7 00/16] lockdep: Implement crossrelease feature Byungchul Park
2017-05-24 8:59 ` [PATCH v7 01/16] lockdep: Refactor lookup_chain_cache() Byungchul Park
2017-05-24 8:59 ` [PATCH v7 02/16] lockdep: Add a function building a chain between two classes Byungchul Park
2017-05-24 8:59 ` [PATCH v7 03/16] lockdep: Change the meaning of check_prev_add()'s return value Byungchul Park
2017-05-24 8:59 ` [PATCH v7 04/16] lockdep: Make check_prev_add() able to handle external stack_trace Byungchul Park
2017-05-24 8:59 ` [PATCH v7 05/16] lockdep: Implement crossrelease feature Byungchul Park
2017-06-13 0:33 ` Byungchul Park
2017-06-22 23:27 ` Byungchul Park
2017-07-11 16:04 ` Peter Zijlstra
2017-07-12 2:24 ` Byungchul Park
2017-05-24 8:59 ` [PATCH v7 06/16] lockdep: Detect and handle hist_lock ring buffer overwrite Byungchul Park
2017-07-11 16:12 ` Peter Zijlstra
2017-07-12 2:00 ` Byungchul Park
2017-07-12 7:56 ` Peter Zijlstra
2017-07-13 2:07 ` Byungchul Park
2017-07-13 8:14 ` Peter Zijlstra [this message]
2017-07-13 8:57 ` Byungchul Park
2017-07-13 9:50 ` Peter Zijlstra
2017-07-13 10:09 ` Byungchul Park
2017-07-13 10:29 ` Peter Zijlstra
2017-07-13 11:12 ` Peter Zijlstra
2017-07-13 11:23 ` Byungchul Park
2017-07-14 1:41 ` Byungchul Park
2017-07-14 6:42 ` Byungchul Park
2017-07-21 13:54 ` Peter Zijlstra
2017-07-25 6:29 ` Byungchul Park
2017-07-25 8:45 ` Peter Zijlstra
2017-07-13 11:19 ` Byungchul Park
2017-07-18 1:25 ` Byungchul Park
2017-05-24 8:59 ` [PATCH v7 07/16] lockdep: Handle non(or multi)-acquisition of a crosslock Byungchul Park
2017-05-24 8:59 ` [PATCH v7 08/16] lockdep: Avoid adding redundant direct links of crosslocks Byungchul Park
2017-07-25 15:41 ` Peter Zijlstra
2017-07-26 7:16 ` Byungchul Park
2017-05-24 8:59 ` [PATCH v7 09/16] lockdep: Fix incorrect condition to print bug msgs for MAX_LOCKDEP_CHAIN_HLOCKS Byungchul Park
2017-05-24 8:59 ` [PATCH v7 10/16] lockdep: Make print_circular_bug() aware of crossrelease Byungchul Park
2017-05-24 8:59 ` [PATCH v7 11/16] lockdep: Apply crossrelease to completions Byungchul Park
2017-05-24 8:59 ` [PATCH v7 12/16] pagemap.h: Remove trailing white space Byungchul Park
2017-05-24 8:59 ` [PATCH v7 13/16] lockdep: Apply crossrelease to PG_locked locks Byungchul Park
2017-05-24 8:59 ` [PATCH v7 14/16] lockdep: Apply lock_acquire(release) on __Set(__Clear)PageLocked Byungchul Park
2017-05-24 8:59 ` [PATCH v7 15/16] lockdep: Move data of CONFIG_LOCKDEP_PAGELOCK from page to page_ext Byungchul Park
2017-05-24 8:59 ` [PATCH v7 16/16] lockdep: Crossrelease feature documentation Byungchul Park
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=20170713081442.GA439@worktop \
--to=peterz@infradead.org \
--cc=akpm@linux-foundation.org \
--cc=boqun.feng@gmail.com \
--cc=byungchul.park@lge.com \
--cc=kernel-team@lge.com \
--cc=kirill@shutemov.name \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=mingo@kernel.org \
--cc=npiggin@gmail.com \
--cc=tglx@linutronix.de \
--cc=walken@google.com \
--cc=willy@infradead.org \
/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).