Skip to contentSkip to conversation

rcu: Reduce stack usage in show_rcu_gp_kthreads()

PATCH In review Zqiang 3 participants 4 messages Jul 23 – Jul 24 RCU

lore
Share & tools
lore

Zqiang

author reviewer
#0 lore raw

PATCH 1/1 rcu: Reduce stack usage in show_rcu_gp_kthreads()

When CONFIG_KASAN=y and CONFIG_KASAN_STACK=y builds, the show_rcu_gp_kthreads() exceeds the 1024-byte frame-size limit:

make kernel/rcu/tree.o KCFLAGS="-fstack-usage"
DESCEND objtool
DESCEND bpf/resolve_btfids
INSTALL libsubcmd_headers
CC kernel/rcu/tree.o
In file included from kernel/rcu/tree.c:4998:
kernel/rcu/tree_stall.h: In function 'show_rcu_gp_kthreads':
kernel/rcu/tree_stall.h:994:1: warning: the frame size of 1656 bytes is larger than 1024 bytes [-Wframe-larger-than=]

grep show_rcu kernel/rcu/tree.su

tree_nocb.h:1622:13:show_rcu_nocb_state 896     dynamic,bounded
tree_stall.h:933:6:show_rcu_gp_kthreads 1784    dynamic,bounded
tree_stall.h:1102:13:sysrq_show_rcu     16      static

Wrap the pr_info() into two noinline_for_stack helpers function:
show_rcu_state() print rcu_state status, and show_rcu_node() print single rcu_node status.

After apply this change:

grep show_rcu kernel/rcu/tree.su
tree_stall.h:955:22:show_rcu_node 696 dynamic,bounded
tree_stall.h:930:22:show_rcu_state 872 dynamic,bounded
tree_nocb.h:1622:13:show_rcu_nocb_state 896 dynamic,bounded
tree_stall.h:972:6:show_rcu_gp_kthreads 544 static
tree_stall.h:1113:13:sysrq_show_rcu 16 static

---
 kernel/rcu/tree_stall.h | 47 +++++++++++++++++++++++++----------------
 1 file changed, 29 insertions(+), 18 deletions(-)
kernel/rcu/tree_stall.h +29 -18
index 45b9856ccd2b..a6dfd036a7e9 100644
@@ -927,20 +927,13 @@ bool rcu_check_boost_fail(unsigned long gp_state, int *cpup)
 }
 EXPORT_SYMBOL_GPL(rcu_check_boost_fail);
 
-/*
- * Show the state of the grace-period kthreads.
- */
-void show_rcu_gp_kthreads(void)
+static noinline_for_stack void show_rcu_state(void)
 {
-	unsigned long cbs = 0;
-	int cpu;
 	unsigned long j;
 	unsigned long ja;
 	unsigned long jr;
 	unsigned long js;
 	unsigned long jw;
-	struct rcu_data *rdp;
-	struct rcu_node *rnp;
 	struct task_struct *t = READ_ONCE(rcu_state.gp_kthread);
 
 	j = jiffies;
@@ -957,21 +950,39 @@ void show_rcu_gp_kthreads(void)
 		(long)data_race(READ_ONCE(rcu_get_root()->gp_seq_needed)),
 		data_race(READ_ONCE(rcu_state.gp_max)),
 		data_race(READ_ONCE(rcu_state.gp_flags)));
+}
+
+static noinline_for_stack void show_rcu_node(struct rcu_node *rnp)
+{
+	pr_info("\trcu_node %d:%d ->gp_seq %ld ->gp_seq_needed %ld ->qsmask %#lx %c%c%c%c ->n_boosts %ld\n",
+		rnp->grplo, rnp->grphi,
+		(long)data_race(READ_ONCE(rnp->gp_seq)),
+		(long)data_race(READ_ONCE(rnp->gp_seq_needed)),
+		data_race(READ_ONCE(rnp->qsmask)),
+		".b"[!!data_race(READ_ONCE(rnp->boost_kthread_task))],
+		".B"[!!data_race(READ_ONCE(rnp->boost_tasks))],
+		".E"[!!data_race(READ_ONCE(rnp->exp_tasks))],
+		".G"[!!data_race(READ_ONCE(rnp->gp_tasks))],
+		data_race(READ_ONCE(rnp->n_boosts)));
+}
+
+/*
+ * Show the state of the grace-period kthreads.
+ */
+void show_rcu_gp_kthreads(void)
+{
+	unsigned long cbs = 0;
+	int cpu;
+	struct rcu_data *rdp;
+	struct rcu_node *rnp;
+
+	show_rcu_state();
 	rcu_for_each_node_breadth_first(rnp) {
 		if (ULONG_CMP_GE(READ_ONCE(rcu_state.gp_seq), READ_ONCE(rnp->gp_seq_needed)) &&
 		    !data_race(READ_ONCE(rnp->qsmask)) && !data_race(READ_ONCE(rnp->boost_tasks)) &&
 		    !data_race(READ_ONCE(rnp->exp_tasks)) && !data_race(READ_ONCE(rnp->gp_tasks)))
 			continue;
-		pr_info("\trcu_node %d:%d ->gp_seq %ld ->gp_seq_needed %ld ->qsmask %#lx %c%c%c%c ->n_boosts %ld\n",
-			rnp->grplo, rnp->grphi,
-			(long)data_race(READ_ONCE(rnp->gp_seq)),
-			(long)data_race(READ_ONCE(rnp->gp_seq_needed)),
-			data_race(READ_ONCE(rnp->qsmask)),
-			".b"[!!data_race(READ_ONCE(rnp->boost_kthread_task))],
-			".B"[!!data_race(READ_ONCE(rnp->boost_tasks))],
-			".E"[!!data_race(READ_ONCE(rnp->exp_tasks))],
-			".G"[!!data_race(READ_ONCE(rnp->gp_tasks))],
-			data_race(READ_ONCE(rnp->n_boosts)));
+		show_rcu_node(rnp);
 		if (!rcu_is_leaf_node(rnp))
 			continue;
 		for_each_leaf_node_possible_cpu(rnp, cpu) {
Signature · 1 line
2.17.1
maintainer
#1 lore raw
Quoted 23 lines · Zqiang

When CONFIG_KASAN=y and CONFIG_KASAN_STACK=y builds, the show_rcu_gp_kthreads() exceeds the 1024-byte frame-size limit:

make kernel/rcu/tree.o KCFLAGS="-fstack-usage"
DESCEND objtool
DESCEND bpf/resolve_btfids
INSTALL libsubcmd_headers
CC kernel/rcu/tree.o
In file included from kernel/rcu/tree.c:4998:
kernel/rcu/tree_stall.h: In function 'show_rcu_gp_kthreads':
kernel/rcu/tree_stall.h:994:1: warning: the frame size of 1656 bytes is larger than 1024 bytes [-Wframe-larger-than=]

grep show_rcu kernel/rcu/tree.su

tree_nocb.h:1622:13:show_rcu_nocb_state 896     dynamic,bounded
tree_stall.h:933:6:show_rcu_gp_kthreads 1784    dynamic,bounded
tree_stall.h:1102:13:sysrq_show_rcu     16      static

Wrap the pr_info() into two noinline_for_stack helpers function:
show_rcu_state() print rcu_state status, and show_rcu_node() print single rcu_node status.

After apply this change:

Quoted 8 lines

grep show_rcu kernel/rcu/tree.su
tree_stall.h:955:22:show_rcu_node 696 dynamic,bounded
tree_stall.h:930:22:show_rcu_state 872 dynamic,bounded
tree_nocb.h:1622:13:show_rcu_nocb_state 896 dynamic,bounded
tree_stall.h:972:6:show_rcu_gp_kthreads 544 static
tree_stall.h:1113:13:sysrq_show_rcu 16 static

Queued and pushed for further review and testing, thank you!

Thanx, Paul

Quoted 83 lines
---
 kernel/rcu/tree_stall.h | 47 +++++++++++++++++++++++++----------------
 1 file changed, 29 insertions(+), 18 deletions(-)
kernel/rcu/tree_stall.h +29 -18
index 45b9856ccd2b..a6dfd036a7e9 100644
@@ -927,20 +927,13 @@ bool rcu_check_boost_fail(unsigned long gp_state, int *cpup)
 }
 EXPORT_SYMBOL_GPL(rcu_check_boost_fail);
 
-/*
- * Show the state of the grace-period kthreads.
- */
-void show_rcu_gp_kthreads(void)
+static noinline_for_stack void show_rcu_state(void)
 {
-	unsigned long cbs = 0;
-	int cpu;
 	unsigned long j;
 	unsigned long ja;
 	unsigned long jr;
 	unsigned long js;
 	unsigned long jw;
-	struct rcu_data *rdp;
-	struct rcu_node *rnp;
 	struct task_struct *t = READ_ONCE(rcu_state.gp_kthread);
 
 	j = jiffies;
@@ -957,21 +950,39 @@ void show_rcu_gp_kthreads(void)
 		(long)data_race(READ_ONCE(rcu_get_root()->gp_seq_needed)),
 		data_race(READ_ONCE(rcu_state.gp_max)),
 		data_race(READ_ONCE(rcu_state.gp_flags)));
+}
+
+static noinline_for_stack void show_rcu_node(struct rcu_node *rnp)
+{
+	pr_info("\trcu_node %d:%d ->gp_seq %ld ->gp_seq_needed %ld ->qsmask %#lx %c%c%c%c ->n_boosts %ld\n",
+		rnp->grplo, rnp->grphi,
+		(long)data_race(READ_ONCE(rnp->gp_seq)),
+		(long)data_race(READ_ONCE(rnp->gp_seq_needed)),
+		data_race(READ_ONCE(rnp->qsmask)),
+		".b"[!!data_race(READ_ONCE(rnp->boost_kthread_task))],
+		".B"[!!data_race(READ_ONCE(rnp->boost_tasks))],
+		".E"[!!data_race(READ_ONCE(rnp->exp_tasks))],
+		".G"[!!data_race(READ_ONCE(rnp->gp_tasks))],
+		data_race(READ_ONCE(rnp->n_boosts)));
+}
+
+/*
+ * Show the state of the grace-period kthreads.
+ */
+void show_rcu_gp_kthreads(void)
+{
+	unsigned long cbs = 0;
+	int cpu;
+	struct rcu_data *rdp;
+	struct rcu_node *rnp;
+
+	show_rcu_state();
 	rcu_for_each_node_breadth_first(rnp) {
 		if (ULONG_CMP_GE(READ_ONCE(rcu_state.gp_seq), READ_ONCE(rnp->gp_seq_needed)) &&
 		    !data_race(READ_ONCE(rnp->qsmask)) && !data_race(READ_ONCE(rnp->boost_tasks)) &&
 		    !data_race(READ_ONCE(rnp->exp_tasks)) && !data_race(READ_ONCE(rnp->gp_tasks)))
 			continue;
-		pr_info("\trcu_node %d:%d ->gp_seq %ld ->gp_seq_needed %ld ->qsmask %#lx %c%c%c%c ->n_boosts %ld\n",
-			rnp->grplo, rnp->grphi,
-			(long)data_race(READ_ONCE(rnp->gp_seq)),
-			(long)data_race(READ_ONCE(rnp->gp_seq_needed)),
-			data_race(READ_ONCE(rnp->qsmask)),
-			".b"[!!data_race(READ_ONCE(rnp->boost_kthread_task))],
-			".B"[!!data_race(READ_ONCE(rnp->boost_tasks))],
-			".E"[!!data_race(READ_ONCE(rnp->exp_tasks))],
-			".G"[!!data_race(READ_ONCE(rnp->gp_tasks))],
-			data_race(READ_ONCE(rnp->n_boosts)));
+		show_rcu_node(rnp);
 		if (!rcu_is_leaf_node(rnp))
 			continue;
 		for_each_leaf_node_possible_cpu(rnp, cpu) {
Signature · 1 line
2.17.1
#2 lore raw
In reply to Zqiang
Quoted 3 lines · On Thu, 23 Jul 2026 18:04:30 +0800 Zqiang <[email protected]> wrote:

When CONFIG_KASAN=y and CONFIG_KASAN_STACK=y builds, the show_rcu_gp_kthreads() exceeds the 1024-byte frame-size limit:

...

Quoted 5 lines
kernel/rcu/tree_stall.h +0 -0
 	rcu_for_each_node_breadth_first(rnp) {
 		if (ULONG_CMP_GE(READ_ONCE(rcu_state.gp_seq), READ_ONCE(rnp->gp_seq_needed)) &&
 		    !data_race(READ_ONCE(rnp->qsmask)) && !data_race(READ_ONCE(rnp->boost_tasks)) &&
 		    !data_race(READ_ONCE(rnp->exp_tasks)) && !data_race(READ_ONCE(rnp->gp_tasks)))
 			continue;

kernel/rcu/tree_stall.h void show_rcu_gp_kthreads(void) · line 974

Quoted 10 lines
kernel/rcu/tree_stall.h +0 -10
-		pr_info("\trcu_node %d:%d ->gp_seq %ld ->gp_seq_needed %ld ->qsmask %#lx %c%c%c%c ->n_boosts %ld\n",
-			rnp->grplo, rnp->grphi,
-			(long)data_race(READ_ONCE(rnp->gp_seq)),
-			(long)data_race(READ_ONCE(rnp->gp_seq_needed)),
-			data_race(READ_ONCE(rnp->qsmask)),
-			".b"[!!data_race(READ_ONCE(rnp->boost_kthread_task))],
-			".B"[!!data_race(READ_ONCE(rnp->boost_tasks))],
-			".E"[!!data_race(READ_ONCE(rnp->exp_tasks))],
-			".G"[!!data_race(READ_ONCE(rnp->gp_tasks))],
-			data_race(READ_ONCE(rnp->n_boosts)));

Isn't that code carefully reading all the values twice?

Also the "ab"[!!val] generates far worse code than the more obvious (val ? 'a' : 'b'). For the former gcc indexes a constant string, the latter is done using arithmetic.

I'm sure there is a good reason for data_race(READ_ONCE(xxx)) ...

David

maintainer
#3 lore raw
Quoted 16 lines · David Laight
Quoted 3 lines · On Thu, 23 Jul 2026 18:04:30 +0800 Zqiang <[email protected]> wrote:

When CONFIG_KASAN=y and CONFIG_KASAN_STACK=y builds, the show_rcu_gp_kthreads() exceeds the 1024-byte frame-size limit:

...

Quoted 9 lines
kernel/rcu/tree_stall.h +0 -4
 	rcu_for_each_node_breadth_first(rnp) {
 		if (ULONG_CMP_GE(READ_ONCE(rcu_state.gp_seq), READ_ONCE(rnp->gp_seq_needed)) &&
 		    !data_race(READ_ONCE(rnp->qsmask)) && !data_race(READ_ONCE(rnp->boost_tasks)) &&
 		    !data_race(READ_ONCE(rnp->exp_tasks)) && !data_race(READ_ONCE(rnp->gp_tasks)))
 			continue;
-		pr_info("\trcu_node %d:%d ->gp_seq %ld ->gp_seq_needed %ld ->qsmask %#lx %c%c%c%c ->n_boosts %ld\n",
-			rnp->grplo, rnp->grphi,
-			(long)data_race(READ_ONCE(rnp->gp_seq)),
-			(long)data_race(READ_ONCE(rnp->gp_seq_needed)),
Quoted 8 lines

kernel/rcu/tree_stall.h void show_rcu_gp_kthreads(void) · line 974

Quoted 6 lines
kernel/rcu/tree_stall.h +0 -6
-			data_race(READ_ONCE(rnp->qsmask)),
-			".b"[!!data_race(READ_ONCE(rnp->boost_kthread_task))],
-			".B"[!!data_race(READ_ONCE(rnp->boost_tasks))],
-			".E"[!!data_race(READ_ONCE(rnp->exp_tasks))],
-			".G"[!!data_race(READ_ONCE(rnp->gp_tasks))],
-			data_race(READ_ONCE(rnp->n_boosts)));

Isn't that code carefully reading all the values twice?

You are quite right, it does, and it did before Zqiang's patch. I would welcome a patch that carefully used locals to avoid the double reads, as that could avoid some debug confusion that might otherwise result due to low-probability races with updates to these same variables. However, this function is invoked quite rarely in error conditions, so I would also avoid holding anything up waiting for such a patch.

Quoted 2 lines

Also the "ab"[!!val] generates far worse code than the more obvious (val ? 'a' : 'b'). For the former gcc indexes a constant string, the latter is done using arithmetic.

This code is invoked quite rarely, so I am not concerned about the overhead and I like the compactness. In the longer term, perhaps compilers will recognize the pattern and optimize it. Or not, who knows? ;-)

Quoted 1 line

I'm sure there is a good reason for data_race(READ_ONCE(xxx)) ...

And there is!!!

The data_race() says "ignore the contents from a data-race viewpoint" and the READ_ONCE() says "carefully avoid destructive optimizations that might otherwise be applied to this load."

For an example, suppose that we had a global variable x that was accessed only under a lock. Except that we have lockless access for debugging output like that in Zqiang's patch. We want the normal lock-protected accesses to be plain C-language accesses so that any accidental (and thus likely buggy) lockless accesses are spotted by KCSAN, but we very much to *not* want to acquire the lock when doing the debug output, because if the lock was held, we might get a debug-output-free hang just when we are in the greatest need of that debugging output.

So we use normal C-language accesses for the non-debug uses, and we use data_race(READ_ONCE(x)) for the debugging output.

Does that make sense?

Thanx, Paul