The get_seconds() helper returns an 'unsigned long' value, which can
overflow on 32-bit architectures. Since the interface we pass it into
already uses a 64-bit type, we can just use ktime_get_real_seconds()
instead.
While we generally prefer local timestamps in CLOCK_MONOTONIC format
(ktime_get_seconds), this keeps using the CLOCK_REALTIME version
in order to maintain compatibility with existing code.
Signed-off-by: Arnd Bergmann <[email protected]>
---
drivers/scsi/scsi_transport_fc.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/scsi/scsi_transport_fc.c b/drivers/scsi/scsi_transport_fc.c
index 1da3d71e9f61..bb6de88aa724 100644
--- a/drivers/scsi/scsi_transport_fc.c
+++ b/drivers/scsi/scsi_transport_fc.c
@@ -567,7 +567,7 @@ fc_host_post_event(struct Scsi_Host *shost, u32 event_number,
INIT_SCSI_NL_HDR(&event->snlh, SCSI_NL_TRANSPORT_FC,
FC_NL_ASYNC_EVENT, len);
- event->seconds = get_seconds();
+ event->seconds = ktime_get_real_seconds();
event->vendor_id = 0;
event->host_no = shost->host_no;
event->event_datalen = sizeof(u32); /* bytes */
@@ -635,7 +635,7 @@ fc_host_post_vendor_event(struct Scsi_Host *shost, u32 event_number,
INIT_SCSI_NL_HDR(&event->snlh, SCSI_NL_TRANSPORT_FC,
FC_NL_ASYNC_EVENT, len);
- event->seconds = get_seconds();
+ event->seconds = ktime_get_real_seconds();
event->vendor_id = vendor_id;
event->host_no = shost->host_no;
event->event_datalen = data_len; /* bytes */
--
2.9.0
Looks good,
Reviewed-by: Johannes Thumshirn <[email protected]>
--
Johannes Thumshirn Storage
[email protected] +49 911 74053 689
SUSE LINUX GmbH, Maxfeldstr. 5, 90409 N?rnberg
GF: Felix Imend?rffer, Jane Smithard, Graham Norton
HRB 21284 (AG N?rnberg)
Key fingerprint = EC38 9CAB C2C4 F25D 8600 D0D0 0393 969D 2D76 0850
On 06/18/2018 08:29 AM, Arnd Bergmann wrote:
> The get_seconds() helper returns an 'unsigned long' value, which can
> overflow on 32-bit architectures. Since the interface we pass it into
> already uses a 64-bit type, we can just use ktime_get_real_seconds()
> instead.
>
> While we generally prefer local timestamps in CLOCK_MONOTONIC format
> (ktime_get_seconds), this keeps using the CLOCK_REALTIME version
> in order to maintain compatibility with existing code.
>
> Signed-off-by: Arnd Bergmann <[email protected]>
> ---
Reviewed-by: Tyrel Datwyler <[email protected]>
On 6/18/2018 8:29 AM, Arnd Bergmann wrote:
> The get_seconds() helper returns an 'unsigned long' value, which can
> overflow on 32-bit architectures. Since the interface we pass it into
> already uses a 64-bit type, we can just use ktime_get_real_seconds()
> instead.
>
> While we generally prefer local timestamps in CLOCK_MONOTONIC format
> (ktime_get_seconds), this keeps using the CLOCK_REALTIME version
> in order to maintain compatibility with existing code.
>
> Signed-off-by: Arnd Bergmann <[email protected]>
> ---
> drivers/scsi/scsi_transport_fc.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/scsi/scsi_transport_fc.c b/drivers/scsi/scsi_transport_fc.c
> index 1da3d71e9f61..bb6de88aa724 100644
> --- a/drivers/scsi/scsi_transport_fc.c
> +++ b/drivers/scsi/scsi_transport_fc.c
> @@ -567,7 +567,7 @@ fc_host_post_event(struct Scsi_Host *shost, u32 event_number,
>
> INIT_SCSI_NL_HDR(&event->snlh, SCSI_NL_TRANSPORT_FC,
> FC_NL_ASYNC_EVENT, len);
> - event->seconds = get_seconds();
> + event->seconds = ktime_get_real_seconds();
> event->vendor_id = 0;
> event->host_no = shost->host_no;
> event->event_datalen = sizeof(u32); /* bytes */
> @@ -635,7 +635,7 @@ fc_host_post_vendor_event(struct Scsi_Host *shost, u32 event_number,
>
> INIT_SCSI_NL_HDR(&event->snlh, SCSI_NL_TRANSPORT_FC,
> FC_NL_ASYNC_EVENT, len);
> - event->seconds = get_seconds();
> + event->seconds = ktime_get_real_seconds();
> event->vendor_id = vendor_id;
> event->host_no = shost->host_no;
> event->event_datalen = data_len; /* bytes */
>
looks good
Reviewed-by: James Smart <[email protected]>
Arnd,
> The get_seconds() helper returns an 'unsigned long' value, which can
> overflow on 32-bit architectures. Since the interface we pass it into
> already uses a 64-bit type, we can just use ktime_get_real_seconds()
> instead.
>
> While we generally prefer local timestamps in CLOCK_MONOTONIC format
> (ktime_get_seconds), this keeps using the CLOCK_REALTIME version in
> order to maintain compatibility with existing code.
Applied to 4.19/scsi-queue, thanks!
--
Martin K. Petersen Oracle Linux Engineering