Hi, Everyone,
This patch series remove redundant steps and fix one bug of the
ntb_hw_switchtec module.
When a re-initialization is caused by a link event, the driver will
re-setup the shared memory windows. But at that time, the shared memory
is still valid, and it's unnecessary to free, reallocate and then
initialize it again. Remove these redundant steps.
In case of NTB crosslink topology, the setting of shared memory window
in the virtual partition doesn't reset on peer's reboot. So skip the
re-setup of shared memory window for that case.
Switchtec does not support setting multiple MWs simultaneously. However,
there's a race condition when a re-initialization is caused by a link
event, the driver will re-setup the shared memory window asynchronously
and this races with the client setting up its memory windows on the
link up event. Fix this by ensure do the entire initialization in a work
queue and signal the client once it's done.
Regard,
Kelvin
--
Changed since v1:
- It's a second resend of v1
--
Joey Zhang (2):
ntb_hw_switchtec: Remove redundant steps of
switchtec_ntb_reinit_peer() function
ntb_hw_switchtec: Fix setup MW with failure bug
Wesley Sheng (1):
ntb_hw_switchtec: Skip unnecessary re-setup of shared memory window
for crosslink case
drivers/ntb/hw/mscc/ntb_hw_switchtec.c | 80 +++++++++++++++++++++-------------
1 file changed, 49 insertions(+), 31 deletions(-)
--
2.7.4
From: Joey Zhang <[email protected]>
When a re-initialization is caused by a link event, the driver will
re-setup the shared memory window. But at that time, the shared memory
is still valid, and it's unnecessary to free, reallocate and then
initialize it again. We only need to reconfigure the hardware
registers. Remove the redundant steps from
switchtec_ntb_reinit_peer() function.
Signed-off-by: Joey Zhang <[email protected]>
Signed-off-by: Wesley Sheng <[email protected]>
Reviewed-by: Logan Gunthorpe <[email protected]>
---
drivers/ntb/hw/mscc/ntb_hw_switchtec.c | 11 +++++++----
1 file changed, 7 insertions(+), 4 deletions(-)
diff --git a/drivers/ntb/hw/mscc/ntb_hw_switchtec.c b/drivers/ntb/hw/mscc/ntb_hw_switchtec.c
index d905d36..947ed0b 100644
--- a/drivers/ntb/hw/mscc/ntb_hw_switchtec.c
+++ b/drivers/ntb/hw/mscc/ntb_hw_switchtec.c
@@ -1457,10 +1457,13 @@ static void switchtec_ntb_deinit_db_msg_irq(struct switchtec_ntb *sndev)
static int switchtec_ntb_reinit_peer(struct switchtec_ntb *sndev)
{
- dev_info(&sndev->stdev->dev, "peer reinitialized\n");
- switchtec_ntb_deinit_shared_mw(sndev);
- switchtec_ntb_init_mw(sndev);
- return switchtec_ntb_init_shared_mw(sndev);
+ int rc;
+
+ dev_info(&sndev->stdev->dev, "reinitialize shared memory window\n");
+ rc = config_rsvd_lut_win(sndev, sndev->mmio_peer_ctrl, 0,
+ sndev->self_partition,
+ sndev->self_shared_dma);
+ return rc;
}
static int switchtec_ntb_add(struct device *dev,
--
2.7.4
From: Wesley Sheng <[email protected]>
In case of NTB crosslink topology, the setting of shared memory window in
the virtual partition doesn't reset on peer's reboot. So skip the
unnecessary re-setup of shared memory window for that case.
Signed-off-by: Wesley Sheng <[email protected]>
---
drivers/ntb/hw/mscc/ntb_hw_switchtec.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/drivers/ntb/hw/mscc/ntb_hw_switchtec.c b/drivers/ntb/hw/mscc/ntb_hw_switchtec.c
index 947ed0b..6cf15c18 100644
--- a/drivers/ntb/hw/mscc/ntb_hw_switchtec.c
+++ b/drivers/ntb/hw/mscc/ntb_hw_switchtec.c
@@ -1459,6 +1459,9 @@ static int switchtec_ntb_reinit_peer(struct switchtec_ntb *sndev)
{
int rc;
+ if (crosslink_is_enabled(sndev))
+ return 0;
+
dev_info(&sndev->stdev->dev, "reinitialize shared memory window\n");
rc = config_rsvd_lut_win(sndev, sndev->mmio_peer_ctrl, 0,
sndev->self_partition,
--
2.7.4
From: Joey Zhang <[email protected]>
Switchtec does not support setting multiple MWs simultaneously. The
driver takes a hardware lock to ensure that two peers are not doing this
simultaneously and it fails if someone else takes the lock. In most
cases, this is fine as clients only setup the MWs once on one side of
the link.
However, there's a race condition when a re-initialization is caused by
a link event. The driver will re-setup the shared memory window
asynchronously and this races with the client setting up it's memory
windows on the link up event.
To fix this we ensure do the entire initialization in a work queue and
signal the client once it's done.
Signed-off-by: Joey Zhang <[email protected]>
Signed-off-by: Wesley Sheng <[email protected]>
---
drivers/ntb/hw/mscc/ntb_hw_switchtec.c | 66 ++++++++++++++++++++--------------
1 file changed, 39 insertions(+), 27 deletions(-)
diff --git a/drivers/ntb/hw/mscc/ntb_hw_switchtec.c b/drivers/ntb/hw/mscc/ntb_hw_switchtec.c
index 6cf15c18..fffff9a 100644
--- a/drivers/ntb/hw/mscc/ntb_hw_switchtec.c
+++ b/drivers/ntb/hw/mscc/ntb_hw_switchtec.c
@@ -95,7 +95,8 @@ struct switchtec_ntb {
bool link_is_up;
enum ntb_speed link_speed;
enum ntb_width link_width;
- struct work_struct link_reinit_work;
+ struct work_struct check_link_status_work;
+ bool link_force_down;
};
static struct switchtec_ntb *ntb_sndev(struct ntb_dev *ntb)
@@ -494,33 +495,11 @@ enum switchtec_msg {
static int switchtec_ntb_reinit_peer(struct switchtec_ntb *sndev);
-static void link_reinit_work(struct work_struct *work)
-{
- struct switchtec_ntb *sndev;
-
- sndev = container_of(work, struct switchtec_ntb, link_reinit_work);
-
- switchtec_ntb_reinit_peer(sndev);
-}
-
-static void switchtec_ntb_check_link(struct switchtec_ntb *sndev,
- enum switchtec_msg msg)
+static void switchtec_ntb_link_status_update(struct switchtec_ntb *sndev)
{
int link_sta;
int old = sndev->link_is_up;
- if (msg == MSG_LINK_FORCE_DOWN) {
- schedule_work(&sndev->link_reinit_work);
-
- if (sndev->link_is_up) {
- sndev->link_is_up = 0;
- ntb_link_event(&sndev->ntb);
- dev_info(&sndev->stdev->dev, "ntb link forced down\n");
- }
-
- return;
- }
-
link_sta = sndev->self_shared->link_sta;
if (link_sta) {
u64 peer = ioread64(&sndev->peer_shared->magic);
@@ -545,6 +524,38 @@ static void switchtec_ntb_check_link(struct switchtec_ntb *sndev,
}
}
+static void check_link_status_work(struct work_struct *work)
+{
+ struct switchtec_ntb *sndev;
+
+ sndev = container_of(work, struct switchtec_ntb,
+ check_link_status_work);
+
+ if (sndev->link_force_down) {
+ sndev->link_force_down = false;
+ switchtec_ntb_reinit_peer(sndev);
+
+ if (sndev->link_is_up) {
+ sndev->link_is_up = 0;
+ ntb_link_event(&sndev->ntb);
+ dev_info(&sndev->stdev->dev, "ntb link forced down\n");
+ }
+
+ return;
+ }
+
+ switchtec_ntb_link_status_update(sndev);
+}
+
+static void switchtec_ntb_check_link(struct switchtec_ntb *sndev,
+ enum switchtec_msg msg)
+{
+ if (msg == MSG_LINK_FORCE_DOWN)
+ sndev->link_force_down = true;
+
+ schedule_work(&sndev->check_link_status_work);
+}
+
static void switchtec_ntb_link_notification(struct switchtec_dev *stdev)
{
struct switchtec_ntb *sndev = stdev->sndev;
@@ -577,7 +588,7 @@ static int switchtec_ntb_link_enable(struct ntb_dev *ntb,
sndev->self_shared->link_sta = 1;
switchtec_ntb_send_msg(sndev, LINK_MESSAGE, MSG_LINK_UP);
- switchtec_ntb_check_link(sndev, MSG_CHECK_LINK);
+ switchtec_ntb_link_status_update(sndev);
return 0;
}
@@ -591,7 +602,7 @@ static int switchtec_ntb_link_disable(struct ntb_dev *ntb)
sndev->self_shared->link_sta = 0;
switchtec_ntb_send_msg(sndev, LINK_MESSAGE, MSG_LINK_DOWN);
- switchtec_ntb_check_link(sndev, MSG_CHECK_LINK);
+ switchtec_ntb_link_status_update(sndev);
return 0;
}
@@ -844,7 +855,8 @@ static int switchtec_ntb_init_sndev(struct switchtec_ntb *sndev)
sndev->ntb.topo = NTB_TOPO_SWITCH;
sndev->ntb.ops = &switchtec_ntb_ops;
- INIT_WORK(&sndev->link_reinit_work, link_reinit_work);
+ INIT_WORK(&sndev->check_link_status_work, check_link_status_work);
+ sndev->link_force_down = false;
sndev->self_partition = sndev->stdev->partition;
--
2.7.4
On Thu, Jun 06, 2019 at 03:09:41PM +0800, Kelvin Cao wrote:
> Hi, Everyone,
>
> This patch series remove redundant steps and fix one bug of the
> ntb_hw_switchtec module.
>
> When a re-initialization is caused by a link event, the driver will
> re-setup the shared memory windows. But at that time, the shared memory
> is still valid, and it's unnecessary to free, reallocate and then
> initialize it again. Remove these redundant steps.
>
> In case of NTB crosslink topology, the setting of shared memory window
> in the virtual partition doesn't reset on peer's reboot. So skip the
> re-setup of shared memory window for that case.
>
> Switchtec does not support setting multiple MWs simultaneously. However,
> there's a race condition when a re-initialization is caused by a link
> event, the driver will re-setup the shared memory window asynchronously
> and this races with the client setting up its memory windows on the
> link up event. Fix this by ensure do the entire initialization in a work
> queue and signal the client once it's done.
>
> Regard,
> Kelvin
>
> --
>
> Changed since v1:
> - It's a second resend of v1
Sorry for the delay. The series is now in the ntb branch. We've
missed window for 5.2, but it will be in the 5.3 pull request.
Thanks,
Jon
> --
>
> Joey Zhang (2):
> ntb_hw_switchtec: Remove redundant steps of
> switchtec_ntb_reinit_peer() function
> ntb_hw_switchtec: Fix setup MW with failure bug
>
> Wesley Sheng (1):
> ntb_hw_switchtec: Skip unnecessary re-setup of shared memory window
> for crosslink case
>
> drivers/ntb/hw/mscc/ntb_hw_switchtec.c | 80 +++++++++++++++++++++-------------
> 1 file changed, 49 insertions(+), 31 deletions(-)
>
> --
> 2.7.4
>