[PATCH RT 0/4] Linux v5.4.290-rt96-rc1

[PATCH RT 0/4] Linux v5.4.290-rt96-rc1

[zanussi]

From: Tom Zanussi <zanussi@kernel.org>

Dear RT Folks,

This is the RT stable review cycle of patch 5.4.290-rt96-rc1.

Please scream at me if I messed something up. Please test the patches
too.

The -rc release will be uploaded to kernel.org and will be deleted
when the final release is out. This is just a review release (or
release candidate).

The pre-releases will not be pushed to the git repository, only the
final release is.

If all goes well, this patch will be converted to the next main
release on 2025-03-06.

To build 5.4.290-rt96-rc1 directly, the following patches should be applied:

  https://www.kernel.org/pub/linux/kernel/v5.x/linux-5.4.tar.xz

  https://www.kernel.org/pub/linux/kernel/v5.x/patch-5.4.290.xz

  https://www.kernel.org/pub/linux/kernel/projects/rt/5.4/patch-5.4.290-rt96-rc1.patch.xz

You can also build from 5.4.290-rt95 by applying the incremental patch:

  https://www.kernel.org/pub/linux/kernel/projects/rt/5.4/incr/patch-5.4.290-rt95-rt96-rc1.patch.xz


Enjoy,

-- Tom


Ahmed S. Darwish (1):
  u64_stats: Introduce u64_stats_set()

Eric Dumazet (1):
  u64_stats: provide u64_stats_t type

Sebastian Andrzej Siewior (1):
  netfilter: nft_counter: Use u64_stats_t for statistic.

Tom Zanussi (1):
  Linux 5.4.290-rt96-rc1

 include/linux/u64_stats_sync.h | 61 ++++++++++++++++++++++++--
 localversion-rt                |  2 +-
 net/netfilter/nft_counter.c    | 78 +++++++++++++++++-----------------
 3 files changed, 98 insertions(+), 43 deletions(-)

-- 
2.43.0

[PATCH RT 1/4] u64_stats: provide u64_stats_t type

[zanussi]

From: Eric Dumazet <edumazet@google.com>

v5.4.290-rt96-rc1 stable review patch.
If anyone has any objections, please let me know.

-----------


[ Upstream commit 316580b69d0a7aeeee5063af47438b626bc47cbd ]

On 64bit arches, struct u64_stats_sync is empty and provides
no help against load/store tearing.

Using READ_ONCE()/WRITE_ONCE() would be needed.

But the update side would be slightly more expensive.

local64_t was defined so that we could use regular adds
in a manner which is atomic wrt IRQs.

However the u64_stats infra means we do not have to use
local64_t on 32bit arches since the syncp provides the needed
protection.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Tom Zanussi <zanussi@kernel.org>
---
 include/linux/u64_stats_sync.h | 51 +++++++++++++++++++++++++++++++---
 1 file changed, 47 insertions(+), 4 deletions(-)

diff --git a/include/linux/u64_stats_sync.h b/include/linux/u64_stats_sync.h
index ca40163933e33..a1f656fc317e6 100644
--- a/include/linux/u64_stats_sync.h
+++ b/include/linux/u64_stats_sync.h
@@ -40,8 +40,8 @@
  *   spin_lock_bh(...) or other synchronization to get exclusive access
  *   ...
  *   u64_stats_update_begin(&stats->syncp);
- *   stats->bytes64 += len; // non atomic operation
- *   stats->packets64++;    // non atomic operation
+ *   u64_stats_add(&stats->bytes64, len); // non atomic operation
+ *   u64_stats_inc(&stats->packets64);    // non atomic operation
  *   u64_stats_update_end(&stats->syncp);
  *
  * While a consumer (reader) should use following template to get consistent
@@ -52,8 +52,8 @@
  *
  * do {
  *         start = u64_stats_fetch_begin(&stats->syncp);
- *         tbytes = stats->bytes64; // non atomic operation
- *         tpackets = stats->packets64; // non atomic operation
+ *         tbytes = u64_stats_read(&stats->bytes64); // non atomic operation
+ *         tpackets = u64_stats_read(&stats->packets64); // non atomic operation
  * } while (u64_stats_fetch_retry(&stats->syncp, start));
  *
  *
@@ -68,6 +68,49 @@ struct u64_stats_sync {
 #endif
 };
 
+#if BITS_PER_LONG == 64
+#include <asm/local64.h>
+
+typedef struct {
+	local64_t	v;
+} u64_stats_t ;
+
+static inline u64 u64_stats_read(const u64_stats_t *p)
+{
+	return local64_read(&p->v);
+}
+
+static inline void u64_stats_add(u64_stats_t *p, unsigned long val)
+{
+	local64_add(val, &p->v);
+}
+
+static inline void u64_stats_inc(u64_stats_t *p)
+{
+	local64_inc(&p->v);
+}
+
+#else
+
+typedef struct {
+	u64		v;
+} u64_stats_t;
+
+static inline u64 u64_stats_read(const u64_stats_t *p)
+{
+	return p->v;
+}
+
+static inline void u64_stats_add(u64_stats_t *p, unsigned long val)
+{
+	p->v += val;
+}
+
+static inline void u64_stats_inc(u64_stats_t *p)
+{
+	p->v++;
+}
+#endif
 
 #if BITS_PER_LONG == 32 && defined(CONFIG_SMP)
 #define u64_stats_init(syncp)				\
-- 
2.43.0

[PATCH RT 2/4] u64_stats: Introduce u64_stats_set()

[zanussi]

From: "Ahmed S. Darwish" <a.darwish@linutronix.de>

v5.4.290-rt96-rc1 stable review patch.
If anyone has any objections, please let me know.

-----------


[ Upstream commit f2efdb17928924c9c935c136dea764a081032006 ]

Allow to directly set a u64_stats_t value which is used to provide an init
function which sets it directly to zero intead of memset() the value.

Add u64_stats_set() to the u64_stats API.

[bigeasy: commit message. ]

Signed-off-by: Ahmed S. Darwish <a.darwish@linutronix.de>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Tom Zanussi <zanussi@kernel.org>
---
 include/linux/u64_stats_sync.h | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/include/linux/u64_stats_sync.h b/include/linux/u64_stats_sync.h
index a1f656fc317e6..cae87feb82d6e 100644
--- a/include/linux/u64_stats_sync.h
+++ b/include/linux/u64_stats_sync.h
@@ -80,6 +80,11 @@ static inline u64 u64_stats_read(const u64_stats_t *p)
 	return local64_read(&p->v);
 }
 
+static inline void u64_stats_set(u64_stats_t *p, u64 val)
+{
+	local64_set(&p->v, val);
+}
+
 static inline void u64_stats_add(u64_stats_t *p, unsigned long val)
 {
 	local64_add(val, &p->v);
@@ -101,6 +106,11 @@ static inline u64 u64_stats_read(const u64_stats_t *p)
 	return p->v;
 }
 
+static inline void u64_stats_set(u64_stats_t *p, u64 val)
+{
+	p->v = val;
+}
+
 static inline void u64_stats_add(u64_stats_t *p, unsigned long val)
 {
 	p->v += val;
-- 
2.43.0

[PATCH RT 3/4] netfilter: nft_counter: Use u64_stats_t for statistic.

[zanussi]

From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

v5.4.290-rt96-rc1 stable review patch.
If anyone has any objections, please let me know.

-----------


[ Upstream commit 4a1d3acd6ea86075e77fcc1188c3fc372833ba73 ]

The nft_counter uses two s64 counters for statistics. Those two are
protected by a seqcount to ensure that the 64bit variable is always
properly seen during updates even on 32bit architectures where the store
is performed by two writes. A side effect is that the two counter (bytes
and packet) are written and read together in the same window.

This can be replaced with u64_stats_t. write_seqcount_begin()/ end() is
replaced with u64_stats_update_begin()/ end() and behaves the same way
as with seqcount_t on 32bit architectures. Additionally there is a
preempt_disable on PREEMPT_RT to ensure that a reader does not preempt a
writer.
On 64bit architectures the macros are removed and the reads happen
without any retries. This also means that the reader can observe one
counter (bytes) from before the update and the other counter (packets)
but that is okay since there is no requirement to have both counter from
the same update window.

Convert the statistic to u64_stats_t. There is one optimisation:
nft_counter_do_init() and nft_counter_clone() allocate a new per-CPU
counter and assign a value to it. During this assignment preemption is
disabled which is not needed because the counter is not yet exposed to
the system so there can not be another writer or reader. Therefore
disabling preemption is omitted and raw_cpu_ptr() is used to obtain a
pointer to a counter for the assignment.

Cc: Eric Dumazet <edumazet@google.com>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Tom Zanussi <zanussi@kernel.org>
---
 net/netfilter/nft_counter.c | 78 +++++++++++++++++++------------------
 1 file changed, 40 insertions(+), 38 deletions(-)

diff --git a/net/netfilter/nft_counter.c b/net/netfilter/nft_counter.c
index bf829fabf2785..2b3ae0fd25619 100644
--- a/net/netfilter/nft_counter.c
+++ b/net/netfilter/nft_counter.c
@@ -8,13 +8,18 @@
 #include <linux/kernel.h>
 #include <linux/init.h>
 #include <linux/module.h>
-#include <linux/seqlock.h>
+#include <linux/u64_stats_sync.h>
 #include <linux/netlink.h>
 #include <linux/netfilter.h>
 #include <linux/netfilter/nf_tables.h>
 #include <net/netfilter/nf_tables.h>
 
 struct nft_counter {
+	u64_stats_t	bytes;
+	u64_stats_t	packets;
+};
+
+struct nft_counter_tot {
 	s64		bytes;
 	s64		packets;
 };
@@ -23,25 +28,24 @@ struct nft_counter_percpu_priv {
 	struct nft_counter __percpu *counter;
 };
 
-static DEFINE_PER_CPU(seqcount_t, nft_counter_seq);
+static DEFINE_PER_CPU(struct u64_stats_sync, nft_counter_sync);
 
 static inline void nft_counter_do_eval(struct nft_counter_percpu_priv *priv,
 				       struct nft_regs *regs,
 				       const struct nft_pktinfo *pkt)
 {
+	struct u64_stats_sync *nft_sync;
 	struct nft_counter *this_cpu;
-	seqcount_t *myseq;
 
 	local_bh_disable();
 	this_cpu = this_cpu_ptr(priv->counter);
-	myseq = this_cpu_ptr(&nft_counter_seq);
-
-	write_seqcount_begin(myseq);
+	nft_sync = this_cpu_ptr(&nft_counter_sync);
 
-	this_cpu->bytes += pkt->skb->len;
-	this_cpu->packets++;
+	u64_stats_update_begin(nft_sync);
+	u64_stats_add(&this_cpu->bytes, pkt->skb->len);
+	u64_stats_inc(&this_cpu->packets);
+	u64_stats_update_end(nft_sync);
 
-	write_seqcount_end(myseq);
 	local_bh_enable();
 }
 
@@ -64,17 +68,16 @@ static int nft_counter_do_init(const struct nlattr * const tb[],
 	if (cpu_stats == NULL)
 		return -ENOMEM;
 
-	preempt_disable();
-	this_cpu = this_cpu_ptr(cpu_stats);
+	this_cpu = raw_cpu_ptr(cpu_stats);
 	if (tb[NFTA_COUNTER_PACKETS]) {
-	        this_cpu->packets =
-			be64_to_cpu(nla_get_be64(tb[NFTA_COUNTER_PACKETS]));
+		u64_stats_set(&this_cpu->packets,
+			      be64_to_cpu(nla_get_be64(tb[NFTA_COUNTER_PACKETS])));
 	}
 	if (tb[NFTA_COUNTER_BYTES]) {
-		this_cpu->bytes =
-			be64_to_cpu(nla_get_be64(tb[NFTA_COUNTER_BYTES]));
+		u64_stats_set(&this_cpu->bytes,
+			      be64_to_cpu(nla_get_be64(tb[NFTA_COUNTER_BYTES])));
 	}
-	preempt_enable();
+
 	priv->counter = cpu_stats;
 	return 0;
 }
@@ -102,40 +105,41 @@ static void nft_counter_obj_destroy(const struct nft_ctx *ctx,
 }
 
 static void nft_counter_reset(struct nft_counter_percpu_priv *priv,
-			      struct nft_counter *total)
+			      struct nft_counter_tot *total)
 {
+	struct u64_stats_sync *nft_sync;
 	struct nft_counter *this_cpu;
-	seqcount_t *myseq;
 
 	local_bh_disable();
 	this_cpu = this_cpu_ptr(priv->counter);
-	myseq = this_cpu_ptr(&nft_counter_seq);
+	nft_sync = this_cpu_ptr(&nft_counter_sync);
+
+	u64_stats_update_begin(nft_sync);
+	u64_stats_add(&this_cpu->packets, -total->packets);
+	u64_stats_add(&this_cpu->bytes, -total->bytes);
+	u64_stats_update_end(nft_sync);
 
-	write_seqcount_begin(myseq);
-	this_cpu->packets -= total->packets;
-	this_cpu->bytes -= total->bytes;
-	write_seqcount_end(myseq);
 	local_bh_enable();
 }
 
 static void nft_counter_fetch(struct nft_counter_percpu_priv *priv,
-			      struct nft_counter *total)
+			      struct nft_counter_tot *total)
 {
 	struct nft_counter *this_cpu;
-	const seqcount_t *myseq;
 	u64 bytes, packets;
 	unsigned int seq;
 	int cpu;
 
 	memset(total, 0, sizeof(*total));
 	for_each_possible_cpu(cpu) {
-		myseq = per_cpu_ptr(&nft_counter_seq, cpu);
+		struct u64_stats_sync *nft_sync = per_cpu_ptr(&nft_counter_sync, cpu);
+
 		this_cpu = per_cpu_ptr(priv->counter, cpu);
 		do {
-			seq	= read_seqcount_begin(myseq);
-			bytes	= this_cpu->bytes;
-			packets	= this_cpu->packets;
-		} while (read_seqcount_retry(myseq, seq));
+			seq	= u64_stats_fetch_begin(nft_sync);
+			bytes	= u64_stats_read(&this_cpu->bytes);
+			packets	= u64_stats_read(&this_cpu->packets);
+		} while (u64_stats_fetch_retry(nft_sync, seq));
 
 		total->bytes	+= bytes;
 		total->packets	+= packets;
@@ -146,7 +150,7 @@ static int nft_counter_do_dump(struct sk_buff *skb,
 			       struct nft_counter_percpu_priv *priv,
 			       bool reset)
 {
-	struct nft_counter total;
+	struct nft_counter_tot total;
 
 	nft_counter_fetch(priv, &total);
 
@@ -235,7 +239,7 @@ static int nft_counter_clone(struct nft_expr *dst, const struct nft_expr *src)
 	struct nft_counter_percpu_priv *priv_clone = nft_expr_priv(dst);
 	struct nft_counter __percpu *cpu_stats;
 	struct nft_counter *this_cpu;
-	struct nft_counter total;
+	struct nft_counter_tot total;
 
 	nft_counter_fetch(priv, &total);
 
@@ -243,11 +247,9 @@ static int nft_counter_clone(struct nft_expr *dst, const struct nft_expr *src)
 	if (cpu_stats == NULL)
 		return -ENOMEM;
 
-	preempt_disable();
-	this_cpu = this_cpu_ptr(cpu_stats);
-	this_cpu->packets = total.packets;
-	this_cpu->bytes = total.bytes;
-	preempt_enable();
+	this_cpu = raw_cpu_ptr(cpu_stats);
+	u64_stats_set(&this_cpu->packets, total.packets);
+	u64_stats_set(&this_cpu->bytes, total.bytes);
 
 	priv_clone->counter = cpu_stats;
 	return 0;
@@ -279,7 +281,7 @@ static int __init nft_counter_module_init(void)
 	int cpu, err;
 
 	for_each_possible_cpu(cpu)
-		seqcount_init(per_cpu_ptr(&nft_counter_seq, cpu));
+		u64_stats_init(per_cpu_ptr(&nft_counter_sync, cpu));
 
 	err = nft_register_obj(&nft_counter_obj_type);
 	if (err < 0)
-- 
2.43.0

[PATCH RT 4/4] Linux 5.4.290-rt96-rc1

[zanussi]

From: Tom Zanussi <zanussi@kernel.org>

v5.4.290-rt96-rc1 stable review patch.
If anyone has any objections, please let me know.

-----------


Signed-off-by: Tom Zanussi <zanussi@kernel.org>
---
 localversion-rt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/localversion-rt b/localversion-rt
index 5a28f0a65d160..02183db2728ee 100644
--- a/localversion-rt
+++ b/localversion-rt
@@ -1 +1 @@
--rt95
+-rt96-rc1
-- 
2.43.0