The new adjustment should be based on the base frequency, not the I40E_PTP_40GB_INCVAL in i40e_ptp_adjfine().
This issue was introduced in commit 3626a690b717 ("i40e: use mul_u64_u64_div_u64 for PTP frequency calculation"), frequency is left just as base I40E_PTP_40GB_INCVAL before the commit. After the commit, frequency is the I40E_PTP_40GB_INCVAL times the ptp_adj_mult value. But then the diff is applied on the wrong value, and no multiplication is done afterwards.
It was accidentally fixed in commit 1060707e3809 ("ptp: introduce helpers to adjust by scaled parts per million"). It uses adjust_by_scaled_ppm correctly performs the calculation and uses the base adjustment, so there's no error here. But it is a new feature and doesn't need to backported to the stable releases.
This issue affects both v6.0 and v6.1, and the v6.1 version is an LTS release. Therefore, the patch only needs to be applied to v6.1 stable.
Fixes: 3626a690b717 ("i40e: use mul_u64_u64_div_u64 for PTP frequency calculation") Cc: stable@vger.kernel.org # 6.1 Cc: Tony Nguyen anthony.l.nguyen@intel.com Signed-off-by: Yajun Deng yajun.deng@linux.dev Reviewed-by: Jacob Keller jacob.e.keller@intel.com --- drivers/net/ethernet/intel/i40e/i40e_ptp.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/net/ethernet/intel/i40e/i40e_ptp.c b/drivers/net/ethernet/intel/i40e/i40e_ptp.c index ffea0c9c82f1..97a9efe7b713 100644 --- a/drivers/net/ethernet/intel/i40e/i40e_ptp.c +++ b/drivers/net/ethernet/intel/i40e/i40e_ptp.c @@ -361,9 +361,9 @@ static int i40e_ptp_adjfine(struct ptp_clock_info *ptp, long scaled_ppm) 1000000ULL << 16);
if (neg_adj) - adj = I40E_PTP_40GB_INCVAL - diff; + adj = freq - diff; else - adj = I40E_PTP_40GB_INCVAL + diff; + adj = freq + diff;
wr32(hw, I40E_PRTTSYN_INC_L, adj & 0xFFFFFFFF); wr32(hw, I40E_PRTTSYN_INC_H, adj >> 32);