Re: [PATCH net-next v3 13/14] net: ethernet: qualcomm: Add PPE debugfs support for PPE counters

[Jie Luo <quic_luoj@quicinc.com>]

On 2/14/2025 10:02 PM, Andrew Lunn wrote:
>>>> +/* The number of packets dropped because of no buffer available, no PPE
>>>> + * buffer assigned to these packets.
>>>> + */
>>>> +static void ppe_port_rx_drop_counter_get(struct ppe_device *ppe_dev,
>>>> +					 struct seq_file *seq)
>>>> +{
>>>> +	u32 reg, drop_cnt = 0;
>>>> +	int ret, i, tag = 0;
>>>> +
>>>> +	PRINT_COUNTER_PREFIX("PRX_DROP_CNT", "SILENT_DROP:");
>>>> +	for (i = 0; i < PPE_DROP_CNT_TBL_ENTRIES; i++) {
>>>> +		reg = PPE_DROP_CNT_TBL_ADDR + i * PPE_DROP_CNT_TBL_INC;
>>>> +		ret = ppe_pkt_cnt_get(ppe_dev, reg, PPE_PKT_CNT_SIZE_1WORD,
>>>> +				      &drop_cnt, NULL);
>>>> +		if (ret) {
>>>> +			seq_printf(seq, "ERROR %d\n", ret);
>>>> +			return;
>>>> +		}
>>>
>>> This is an error getting the value from the hardware? You should not
>>> put that into the debugfs itself, you want the read() call to return
>>> it.
>>>
>>
>> Yes, this error code is returned by regmap read functions in
>> ppe_pkt_cnt_get() when the hardware counter read fails. I will
>> remove it from debugfs file and instead log the error to the
>> console (dev_info).
> 
> and return it to user space via the read() call. These functions
> normally return the number of bytes put into the buffer. But you can
> also return a negative error code which gets passed back to user space
> instead.
> 
> 	Andrew

OK, I will return the negative error code returned by the read() to the
user space, thanks.