Skip to contentSkip to conversation

net: alacritech: Use u64_stats_t with u64_stats_sync properly

PATCH Applied David Yang 5 participants 5 messages Jan 22 – Jan 27

Subsystems (3)
lore
Share & tools
lore

David Yang

author
#0 lore raw

PATCH 1/1 net: alacritech: Use u64_stats_t with u64_stats_sync properly

On 64bit arches, struct u64_stats_sync is empty and provides no help against load/store tearing. Convert to u64_stats_t to ensure atomic operations.

---
 drivers/net/ethernet/alacritech/slic.h    | 50 +++++++++++------------
 drivers/net/ethernet/alacritech/slicoss.c |  8 ++--
 2 files changed, 29 insertions(+), 29 deletions(-)
drivers/net/ethernet/alacritech/slic.h +25 -25
index 82071d0e5f7f..f5bb2d9a61be 100644
@@ -284,7 +284,7 @@
 #define SLIC_INC_STATS_COUNTER(st, counter)	\
 do {						\
 	u64_stats_update_begin(&(st)->syncp);	\
-	(st)->counter++;			\
+	u64_stats_inc(&(st)->counter);		\
 	u64_stats_update_end(&(st)->syncp);	\
 } while (0)
 
@@ -293,7 +293,7 @@ do {						\
 	unsigned int start;					\
 	do {							\
 		start = u64_stats_fetch_begin(&(st)->syncp);	\
-		newst = (st)->counter;				\
+		newst = u64_stats_read(&(st)->counter);		\
 	} while (u64_stats_fetch_retry(&(st)->syncp, start));	\
 }
 
@@ -407,34 +407,34 @@ struct slic_oasis_eeprom {
 };
 
 struct slic_stats {
-	u64 rx_packets;
-	u64 rx_bytes;
-	u64 rx_mcasts;
-	u64 rx_errors;
-	u64 tx_packets;
-	u64 tx_bytes;
+	u64_stats_t rx_packets;
+	u64_stats_t rx_bytes;
+	u64_stats_t rx_mcasts;
+	u64_stats_t rx_errors;
+	u64_stats_t tx_packets;
+	u64_stats_t tx_bytes;
 	/* HW STATS */
-	u64 rx_buff_miss;
-	u64 tx_dropped;
-	u64 irq_errs;
+	u64_stats_t rx_buff_miss;
+	u64_stats_t tx_dropped;
+	u64_stats_t irq_errs;
 	/* transport layer */
-	u64 rx_tpcsum;
-	u64 rx_tpoflow;
-	u64 rx_tphlen;
+	u64_stats_t rx_tpcsum;
+	u64_stats_t rx_tpoflow;
+	u64_stats_t rx_tphlen;
 	/* ip layer */
-	u64 rx_ipcsum;
-	u64 rx_iplen;
-	u64 rx_iphlen;
+	u64_stats_t rx_ipcsum;
+	u64_stats_t rx_iplen;
+	u64_stats_t rx_iphlen;
 	/* link layer */
-	u64 rx_early;
-	u64 rx_buffoflow;
-	u64 rx_lcode;
-	u64 rx_drbl;
-	u64 rx_crc;
-	u64 rx_oflow802;
-	u64 rx_uflow802;
+	u64_stats_t rx_early;
+	u64_stats_t rx_buffoflow;
+	u64_stats_t rx_lcode;
+	u64_stats_t rx_drbl;
+	u64_stats_t rx_crc;
+	u64_stats_t rx_oflow802;
+	u64_stats_t rx_uflow802;
 	/* oasis only */
-	u64 tx_carrier;
+	u64_stats_t tx_carrier;
 	struct u64_stats_sync syncp;
 };
 
drivers/net/ethernet/alacritech/slicoss.c +4 -4
index f62851708d4f..7488fb6ace0b 100644
@@ -378,8 +378,8 @@ static void slic_xmit_complete(struct slic_device *sdev)
 	smp_wmb();
 
 	u64_stats_update_begin(&sdev->stats.syncp);
-	sdev->stats.tx_bytes += bytes;
-	sdev->stats.tx_packets += frames;
+	u64_stats_add(&sdev->stats.tx_bytes, bytes);
+	u64_stats_add(&sdev->stats.tx_packets, frames);
 	u64_stats_update_end(&sdev->stats.syncp);
 
 	netif_tx_lock(dev);
@@ -615,8 +615,8 @@ static void slic_handle_receive(struct slic_device *sdev, unsigned int todo,
 	}
 
 	u64_stats_update_begin(&sdev->stats.syncp);
-	sdev->stats.rx_bytes += bytes;
-	sdev->stats.rx_packets += frames;
+	u64_stats_add(&sdev->stats.rx_bytes, bytes);
+	u64_stats_add(&sdev->stats.rx_packets, frames);
 	u64_stats_update_end(&sdev->stats.syncp);
 
 	slic_refill_rx_queue(sdev, GFP_ATOMIC);
Signature · 1 line
2.51.0
maintainer
#1 lore raw
Quoted 3 lines · On Fri, 23 Jan 2026 02:51:07 +0800 David Yang wrote:

On 64bit arches, struct u64_stats_sync is empty and provides no help against load/store tearing. Convert to u64_stats_t to ensure atomic operations.

Eric, could you comment if this matters? I expect David may send a non-trivial number of these, I want to make sure we're making good use of everyone's time here. 64b accesses are not torn on 64b arches, and never used for control flow.

#2 lore raw
Quoted 10 lines · Jakub Kicinski
Quoted 3 lines · On Fri, 23 Jan 2026 02:51:07 +0800 David Yang wrote:

On 64bit arches, struct u64_stats_sync is empty and provides no help against load/store tearing. Convert to u64_stats_t to ensure atomic operations.

Eric, could you comment if this matters? I expect David may send a non-trivial number of these, I want to make sure we're making good use of everyone's time here. 64b accesses are not torn on 64b arches, and never used for control flow.

I added u64_stats_add()/u64_stats_inc() back in 2019, and never bothered sending dozens of patches.

Presumably compilers could be really dumb back then, I am not sure this is still the case today.

#3 lore raw
Quoted 12 lines · On Mon, 26 Jan 2026 11:11:09 +0100 Eric Dumazet <[email protected]> wrote:
Quoted 10 lines · Jakub Kicinski
Quoted 3 lines · On Fri, 23 Jan 2026 02:51:07 +0800 David Yang wrote:

On 64bit arches, struct u64_stats_sync is empty and provides no help against load/store tearing. Convert to u64_stats_t to ensure atomic operations.

Eric, could you comment if this matters? I expect David may send a non-trivial number of these, I want to make sure we're making good use of everyone's time here. 64b accesses are not torn on 64b arches, and never used for control flow.

Quoted 6 lines

I added u64_stats_add()/u64_stats_inc() back in 2019, and never bothered sending dozens of patches.

Presumably compilers could be really dumb back then, I am not sure this is still the case today.

gcc still generates 'crap code' for var_64 += var_32 on x86-32. Basically it zero-extends then does a 64bit add rather than using 'adc $0, %reg_hi', the extra register is likely to force a spill to stack.
But that is 'fixed' by using add_u64_u32().

Oh and you really don't want to support adding a signed 32bit value for one place that wants to do a subtract.

David

#4 lore raw
In reply to David Yang
This patch was applied to netdev/net-next.git (main) by Jakub Kicinski <[email protected]>:

Hello:

This patch was applied to netdev/net-next.git (main) by Jakub Kicinski <[email protected]>:

Quoted 9 lines · On Fri, 23 Jan 2026 02:51:07 +0800 you wrote:

On 64bit arches, struct u64_stats_sync is empty and provides no help against load/store tearing. Convert to u64_stats_t to ensure atomic operations.

---
 drivers/net/ethernet/alacritech/slic.h    | 50 +++++++++++------------
 drivers/net/ethernet/alacritech/slicoss.c |  8 ++--
 2 files changed, 29 insertions(+), 29 deletions(-)

Here is the summary with links:
- [net-next] net: alacritech: Use u64_stats_t with u64_stats_sync properly

    https://git.kernel.org/netdev/net-next/c/6e9674a980f5

You are awesome, thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html