On Wed, Feb 21, 2024 at 11:32 PM Ferruh Yigit <ferruh.yigit@amd.com> wrote:
On 2/16/2024 1:56 PM, Ferruh Yigit wrote:
> On 2/16/2024 3:47 AM, Kumara Parameshwaran wrote:
>> In heavy-weight mode GRO which is based on timer, the GRO packets
>> will not be flushed in spite of timer expiry if there is no packet
>> in the current poll. If timer mode GRO is enabled the
>> rte_gro_timeout_flush API should be invoked.
>>
>> Fixes: b7091f1dcfbc ("app/testpmd: enable the heavyweight mode TCP/IPv4 GRO")
>> Cc: hujiayu.hu@foxmail.com
>>
>> Signed-off-by: Kumara Parameshwaran <kumaraparamesh92@gmail.com>
>> ---
>> v1:
>>     Changes to make sure that the GRO flush API is invoked if there are no packets in
>>     current poll and timer expiry.
>>
>> v2:
>>     Fix code organisation issue
>>
>> v3:
>>     Fix warnings
>>
>> v4:
>>     Fix error and warnings
>>
>> v5:
>>     Fix compilation issue when GRO is not defined
>>
>> v6:
>>     Address review comments
>>
>> v7:
>>     Address review comments
>>
>> v8:
>>     Fix spell check warnings
>>
>>  app/test-pmd/csumonly.c | 22 ++++++++++++++++++----
>>  1 file changed, 18 insertions(+), 4 deletions(-)
>>
>> diff --git a/app/test-pmd/csumonly.c b/app/test-pmd/csumonly.c
>> index c103e54111..a922160f6d 100644
>> --- a/app/test-pmd/csumonly.c
>> +++ b/app/test-pmd/csumonly.c
>> @@ -863,16 +863,29 @@ pkt_burst_checksum_forward(struct fwd_stream *fs)
>> 
>>      /* receive a burst of packet */
>>      nb_rx = common_fwd_stream_receive(fs, pkts_burst, nb_pkt_per_burst);
>> -    if (unlikely(nb_rx == 0))
>> +    if (unlikely(nb_rx == 0)) {
>> +#ifndef RTE_LIB_GRO
>>              return false;
>> +#else
>> +            gro_enable = gro_ports[fs->rx_port].enable;
>> +            /*
>> +             * Make sure that in case of Heavyweight mode GRO the packets in
>> +             * GRO cache should be flushed as the timer could have expired.
>> +             *
>> +             * The order of conditions should be the same as gro_ctx is valid
>> +             * only when gro_flush_cycles is not the GRO_DEFAULT_FLUSH_CYCLES which
>> +             * indicates light weight mode GRO
>> +             */
>>
>
> Updated comment as below to make it terse, what do you think:
>  /*
>  * Check if packets need to be flushed in the GRO context
>  * due to a timeout.
>  *
>  * Continue only in GRO heavyweight mode and if there are
>  * packets in the GRO context.
>  */
>
>
>> +            if (!gro_enable || (gro_flush_cycles == GRO_DEFAULT_FLUSH_CYCLES) ||
>> +                    (rte_gro_get_pkt_count(current_fwd_lcore()->gro_ctx) == 0))
>> +                    return false;
>> +#endif
>> +    }
>>
>
> Another issue but also related to your patch, if there is no packet to
> Tx after GRO block, should we add another zero packet check:
> if (unlikely(nb_rx == 0))
>       return false;
>
> To prevent executing GSO and Tx path code with zero packet, do you think
> does this make sense?
>
>

Patch looks good to me, with above comment update, but I am worried
about side impacts of this patch that we might be missing, that is why I
would like it to be in -rc1, so that it can be tested better. Hence,


Reviewed-by: Ferruh Yigit <ferruh.yigit@amd.com>

Applied to dpdk-next-net/main, thanks.
(Updated comment as suggested above while merging.)


Lets continue to discuss return on "nb_rx == 0" case after GRO block,
incremental to this patch.

I was not able to get to this. I will also take a look at the code to see if this can cause any issues.
Thanks.