[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20171011004512.7949-11-alexander.levin@verizon.com>
Date: Wed, 11 Oct 2017 00:45:21 +0000
From: "Levin, Alexander (Sasha Levin)" <alexander.levin@...izon.com>
To: "gregkh@...uxfoundation.org" <gregkh@...uxfoundation.org>
CC: "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
"stable@...r.kernel.org" <stable@...r.kernel.org>
Subject: [GIT PULL for-4.9 10/48] net/mlx4_en: fix overflow in
mlx4_en_init_timestamp()
From: Eric Dumazet <edumazet@...gle.com>
[ Upstream commit 47d3a07528ecbbccf53bc4390d70b4e3d1c04fcf ]
The cited commit makes a great job of finding optimal shift/multiplier
values assuming a 10 seconds wrap around, but forgot to change the
overflow_period computation.
It overflows in cyclecounter_cyc2ns(), and the final result is 804 ms,
which is silly.
Lets simply use 5 seconds, no need to recompute this, given how it is
supposed to work.
Later, we will use a timer instead of a work queue, since the new RX
allocation schem will no longer need mlx4_en_recover_from_oom() and the
service_task firing every 250 ms.
Fixes: 31c128b66e5b ("net/mlx4_en: Choose time-stamping shift value according to HW frequency")
Signed-off-by: Eric Dumazet <edumazet@...gle.com>
Cc: Tariq Toukan <tariqt@...lanox.com>
Cc: Eugenia Emantayev <eugenia@...lanox.com>
Reviewed-by: Tariq Toukan <tariqt@...lanox.com>
Signed-off-by: David S. Miller <davem@...emloft.net>
Signed-off-by: Sasha Levin <alexander.levin@...izon.com>
---
drivers/net/ethernet/mellanox/mlx4/en_clock.c | 18 ++++++++----------
drivers/net/ethernet/mellanox/mlx4/mlx4_en.h | 1 -
2 files changed, 8 insertions(+), 11 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx4/en_clock.c b/drivers/net/ethernet/mellanox/mlx4/en_clock.c
index a5fc46bbcbe2..d4d97ca12e83 100644
--- a/drivers/net/ethernet/mellanox/mlx4/en_clock.c
+++ b/drivers/net/ethernet/mellanox/mlx4/en_clock.c
@@ -88,10 +88,17 @@ void mlx4_en_remove_timestamp(struct mlx4_en_dev *mdev)
}
}
+#define MLX4_EN_WRAP_AROUND_SEC 10UL
+/* By scheduling the overflow check every 5 seconds, we have a reasonably
+ * good chance we wont miss a wrap around.
+ * TOTO: Use a timer instead of a work queue to increase the guarantee.
+ */
+#define MLX4_EN_OVERFLOW_PERIOD (MLX4_EN_WRAP_AROUND_SEC * HZ / 2)
+
void mlx4_en_ptp_overflow_check(struct mlx4_en_dev *mdev)
{
bool timeout = time_is_before_jiffies(mdev->last_overflow_check +
- mdev->overflow_period);
+ MLX4_EN_OVERFLOW_PERIOD);
unsigned long flags;
if (timeout) {
@@ -236,7 +243,6 @@ static const struct ptp_clock_info mlx4_en_ptp_clock_info = {
.enable = mlx4_en_phc_enable,
};
-#define MLX4_EN_WRAP_AROUND_SEC 10ULL
/* This function calculates the max shift that enables the user range
* of MLX4_EN_WRAP_AROUND_SEC values in the cycles register.
@@ -261,7 +267,6 @@ void mlx4_en_init_timestamp(struct mlx4_en_dev *mdev)
{
struct mlx4_dev *dev = mdev->dev;
unsigned long flags;
- u64 ns, zero = 0;
/* mlx4_en_init_timestamp is called for each netdev.
* mdev->ptp_clock is common for all ports, skip initialization if
@@ -285,13 +290,6 @@ void mlx4_en_init_timestamp(struct mlx4_en_dev *mdev)
ktime_to_ns(ktime_get_real()));
write_unlock_irqrestore(&mdev->clock_lock, flags);
- /* Calculate period in seconds to call the overflow watchdog - to make
- * sure counter is checked at least once every wrap around.
- */
- ns = cyclecounter_cyc2ns(&mdev->cycles, mdev->cycles.mask, zero, &zero);
- do_div(ns, NSEC_PER_SEC / 2 / HZ);
- mdev->overflow_period = ns;
-
/* Configure the PHC */
mdev->ptp_clock_info = mlx4_en_ptp_clock_info;
snprintf(mdev->ptp_clock_info.name, 16, "mlx4 ptp");
diff --git a/drivers/net/ethernet/mellanox/mlx4/mlx4_en.h b/drivers/net/ethernet/mellanox/mlx4/mlx4_en.h
index a3528dd1e72e..df0f39611c5e 100644
--- a/drivers/net/ethernet/mellanox/mlx4/mlx4_en.h
+++ b/drivers/net/ethernet/mellanox/mlx4/mlx4_en.h
@@ -419,7 +419,6 @@ struct mlx4_en_dev {
struct cyclecounter cycles;
struct timecounter clock;
unsigned long last_overflow_check;
- unsigned long overflow_period;
struct ptp_clock *ptp_clock;
struct ptp_clock_info ptp_clock_info;
struct notifier_block nb;
--
2.11.0
Powered by blists - more mailing lists