Re: [net-next PATCH v3] octeontx2-pf: Use PTP HW timestamp counter atomic update feature

[Leon Romanovsky]

On Thu, Aug 17, 2023 at 11:13:51PM +0530, Sai Krishna wrote:
> Some of the newer silicon versions in CN10K series supports a feature
> where in the current PTP timestamp in HW can be updated atomically
> without losing any cpu cycles unlike read/modify/write register.
> This patch uses this feature so that PTP accuracy can be improved
> while adjusting the master offset in HW. There is no need for SW
> timecounter when using this feature. So removed references to SW
> timecounter wherever appropriate.
> 
> Signed-off-by: Sai Krishna <saikrishnag@xxxxxxxxxxx>
> Signed-off-by: Naveen Mamindlapalli <naveenm@xxxxxxxxxxx>
> Signed-off-by: Sunil Kovvuri Goutham <sgoutham@xxxxxxxxxxx>
> Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@xxxxxxxxxxxx>
> ---
> v3:
>     - Addressed review comments given by Jakub Kicinski
>         1. Fixed re-ordering of headers in alphabetical order
>         2. Refactored SoC revision identification logic
>         3. CN10K errata revisions can be different from atomic update
>            supported revision devices.
>         4. Removed ptp device check.
> v2:
>     - Addressed review comments given by Simon Horman, Kalesh Anakkur Purayil
> 	1. Removed inline keyword for function in .c file
>         2. Modified/optimized conditions related boolean
> 
>  .../net/ethernet/marvell/octeontx2/af/mbox.h  |  12 ++
>  .../net/ethernet/marvell/octeontx2/af/ptp.c   | 155 ++++++++++++++--
>  .../net/ethernet/marvell/octeontx2/af/ptp.h   |   3 +-
>  .../net/ethernet/marvell/octeontx2/af/rvu.c   |   2 +-
>  .../net/ethernet/marvell/octeontx2/af/rvu.h   |  12 ++
>  .../marvell/octeontx2/nic/otx2_common.h       |   1 +
>  .../ethernet/marvell/octeontx2/nic/otx2_ptp.c | 174 ++++++++++++++----
>  7 files changed, 304 insertions(+), 55 deletions(-)

<...>

> +static bool is_tstmp_atomic_update_supported(struct otx2_ptp *ptp)
> +{
> +	struct ptp_get_cap_rsp *rsp;
> +	struct msg_req *req;
> +	int err;
> +
> +	if (!ptp->nic)
> +		return false;
> +
> +	mutex_lock(&ptp->nic->mbox.lock);
> +	req = otx2_mbox_alloc_msg_ptp_get_cap(&ptp->nic->mbox);
> +	if (!req)
> +		return false;
> +
> +	err = otx2_sync_mbox_msg(&ptp->nic->mbox);
> +	if (err)
> +		return false;

Shouldn't you call to mutex_unlock() in two returns above?

Thanks

> +
> +	rsp = (struct ptp_get_cap_rsp *)otx2_mbox_get_rsp(&ptp->nic->mbox.mbox, 0,
> +							  &req->hdr);
> +	mutex_unlock(&ptp->nic->mbox.lock);
> +
> +	if (IS_ERR(rsp))
> +		return false;
> +
> +	if (rsp->cap & PTP_CAP_HW_ATOMIC_UPDATE)
> +		return true;
> +
> +	return false;
> +}