On 9/26/21 9:34 AM, Biju Das wrote:
>> -----Original Message-----
>> Subject: Re: [RFC/PATCH 12/18] ravb: Add timestamp to struct ravb_hw_info
>>
>> On 9/23/21 5:08 PM, Biju Das wrote:
>>
>>> R-Car AVB-DMAC supports timestamp feature.
>>> Add a timestamp hw feature bit to struct ravb_hw_info to add this
>>> feature only for R-Car.
>>>
>>> Signed-off-by: Biju Das <biju.das.jz@xxxxxxxxxxxxxx>
>>> ---
>>> drivers/net/ethernet/renesas/ravb.h | 2 +
>>> drivers/net/ethernet/renesas/ravb_main.c | 68
>>> +++++++++++++++---------
>>> 2 files changed, 45 insertions(+), 25 deletions(-)
>>>
>>> diff --git a/drivers/net/ethernet/renesas/ravb.h
>>> b/drivers/net/ethernet/renesas/ravb.h
>>> index ab4909244276..2505de5d4a28 100644
>>> --- a/drivers/net/ethernet/renesas/ravb.h
>>> +++ b/drivers/net/ethernet/renesas/ravb.h
>>> @@ -1034,6 +1034,7 @@ struct ravb_hw_info {
>>> unsigned mii_rgmii_selection:1; /* E-MAC supports mii/rgmii
>> selection */
>>> unsigned half_duplex:1; /* E-MAC supports half duplex mode */
>>> unsigned rx_2k_buffers:1; /* AVB-DMAC has Max 2K buf size on RX
>> */
>>> + unsigned timestamp:1; /* AVB-DMAC has timestamp */
>>
>> Isn't this a matter of the gPTP support as well, i.e. no separate flag
>> needed?
>
> Agreed. Previously it is suggested to use timestamp. I will change it to as part of gPTP support cases.
TIA. :-)
>>
>> [...]
>>> @@ -1089,6 +1090,7 @@ struct ravb_private {
>>> unsigned int num_tx_desc; /* TX descriptors per packet */
>>>
>>> int duplex;
>>> + struct ravb_rx_desc *rgeth_rx_ring[NUM_RX_QUEUE];
>>
>> Strange place to declare this...
>
> Agreed. This has to be on later patch. Will move it.
I only meant that these fields should go together with rx_ring[]. Apparently
we have a case of wrong patch ordering here (as this patch needs this field declared)...
>>> const struct ravb_hw_info *info;
>>> struct reset_control *rstc;
>>> diff --git a/drivers/net/ethernet/renesas/ravb_main.c
>>> b/drivers/net/ethernet/renesas/ravb_main.c
>>> index 9c0d35f4b221..2c375002ebcb 100644
>>> --- a/drivers/net/ethernet/renesas/ravb_main.c
>>> +++ b/drivers/net/ethernet/renesas/ravb_main.c
>>> @@ -949,11 +949,14 @@ static bool ravb_queue_interrupt(struct
>>> net_device *ndev, int q)
>>>
>>> static bool ravb_timestamp_interrupt(struct net_device *ndev) {
>>> + struct ravb_private *priv = netdev_priv(ndev);
>>> + const struct ravb_hw_info *info = priv->info;
>>> u32 tis = ravb_read(ndev, TIS);
>>>
>>> if (tis & TIS_TFUF) {
>>> ravb_write(ndev, ~(TIS_TFUF | TIS_RESERVED), TIS);
>>> - ravb_get_tx_tstamp(ndev);
>>> + if (info->timestamp)
>>> + ravb_get_tx_tstamp(ndev);
>>
>> Shouldn't we just disable TIS.TFUF permanently instead for the non-gPTP
>> case?
>
> Good catch. As ravb_dmac_init_rgeth(will be renamed to "ravb_dmac_init_gbeth") is not enabling this interrupt as it is not documented in RZ/G2L hardware manual.
> So this function never gets called for non-gPTP case.
>
> I will remove this check.
TIA!
> Regards,
> Biju
MBR, Sergey
