Message-ID: <f62085ff-ab39-4452-8862-7352901f1d86@uwaterloo.ca>
Date: Fri, 29 Aug 2025 19:37:15 -0400
From: Martin Karsten <mkarsten@...terloo.ca>
To: Samiullah Khawaja <skhawaja@...gle.com>
Cc: Jakub Kicinski <kuba@...nel.org>, "David S . Miller"
 <davem@...emloft.net>, Eric Dumazet <edumazet@...gle.com>,
 Paolo Abeni <pabeni@...hat.com>, almasrymina@...gle.com, willemb@...gle.com,
 Joe Damato <joe@...a.to>, netdev@...r.kernel.org
Subject: Re: [PATCH net-next v8 0/2] Add support to do threaded napi busy poll

On 2025-08-29 19:31, Samiullah Khawaja wrote:
> On Fri, Aug 29, 2025 at 3:56 PM Martin Karsten <mkarsten@...terloo.ca> wrote:
>>
>> On 2025-08-29 18:25, Samiullah Khawaja wrote:
>>> On Fri, Aug 29, 2025 at 3:19 PM Martin Karsten <mkarsten@...terloo.ca> wrote:
>>>>
>>>> On 2025-08-29 14:08, Martin Karsten wrote:
>>>>> On 2025-08-29 13:50, Samiullah Khawaja wrote:
>>>>>> On Thu, Aug 28, 2025 at 8:15 PM Martin Karsten <mkarsten@...terloo.ca>
>>>>>> wrote:
>>>>>>>
>>>>>>> On 2025-08-28 21:16, Samiullah Khawaja wrote:
>>>>>>>> Extend the already existing support of threaded napi poll to do
>>>>>>>> continuous
>>>>>>>> busy polling.
>>>>>>>>
>>>>>>>> This is used for doing continuous polling of napi to fetch descriptors
>>>>>>>> from backing RX/TX queues for low latency applications. Allow enabling
>>>>>>>> of threaded busypoll using netlink so this can be enabled on a set of
>>>>>>>> dedicated napis for low latency applications.
>>>>>>>>
>>>>>>>> Once enabled user can fetch the PID of the kthread doing NAPI polling
>>>>>>>> and set affinity, priority and scheduler for it depending on the
>>>>>>>> low-latency requirements.
>>>>>>>>
>>>>>>>> Extend the netlink interface to allow enabling/disabling threaded
>>>>>>>> busypolling at individual napi level.
>>>>>>>>
>>>>>>>> We use this for our AF_XDP based hard low-latency usecase with usecs
>>>>>>>> level latency requirement. For our usecase we want low jitter and
>>>>>>>> stable
>>>>>>>> latency at P99.
>>>>>>>>
>>>>>>>> Following is an analysis and comparison of available (and compatible)
>>>>>>>> busy poll interfaces for a low latency usecase with stable P99. This
>>>>>>>> can
>>>>>>>> be suitable for applications that want very low latency at the expense
>>>>>>>> of cpu usage and efficiency.
>>>>>>>>
>>>>>>>> Already existing APIs (SO_BUSYPOLL and epoll) allow busy polling a NAPI
>>>>>>>> backing a socket, but the missing piece is a mechanism to busy poll a
>>>>>>>> NAPI instance in a dedicated thread while ignoring available events or
>>>>>>>> packets, regardless of the userspace API. Most existing mechanisms are
>>>>>>>> designed to work in a pattern where you poll until new packets or
>>>>>>>> events
>>>>>>>> are received, after which userspace is expected to handle them.
>>>>>>>>
>>>>>>>> As a result, one has to hack together a solution using a mechanism
>>>>>>>> intended to receive packets or events, not to simply NAPI poll. NAPI
>>>>>>>> threaded busy polling, on the other hand, provides this capability
>>>>>>>> natively, independent of any userspace API. This makes it really
>>>>>>>> easy to
>>>>>>>> setup and manage.
>>>>>>>>
>>>>>>>> For analysis we use an AF_XDP based benchmarking tool `xsk_rr`. The
>>>>>>>> description of the tool and how it tries to simulate the real workload
>>>>>>>> is following,
>>>>>>>>
>>>>>>>> - It sends UDP packets between 2 machines.
>>>>>>>> - The client machine sends packets at a fixed frequency. To maintain
>>>>>>>> the
>>>>>>>>       frequency of the packet being sent, we use open-loop sampling.
>>>>>>>> That is
>>>>>>>>       the packets are sent in a separate thread.
>>>>>>>> - The server replies to the packet inline by reading the pkt from the
>>>>>>>>       recv ring and replies using the tx ring.
>>>>>>>> - To simulate the application processing time, we use a configurable
>>>>>>>>       delay in usecs on the client side after a reply is received from
>>>>>>>> the
>>>>>>>>       server.
>>>>>>>>
>>>>>>>> The xsk_rr tool is posted separately as an RFC for tools/testing/
>>>>>>>> selftest.
>>>>>>>>
>>>>>>>> We use this tool with following napi polling configurations,
>>>>>>>>
>>>>>>>> - Interrupts only
>>>>>>>> - SO_BUSYPOLL (inline in the same thread where the client receives the
>>>>>>>>       packet).
>>>>>>>> - SO_BUSYPOLL (separate thread and separate core)
>>>>>>>> - Threaded NAPI busypoll
>>>>>>>>
>>>>>>>> System is configured using following script in all 4 cases,
>>>>>>>>
>>>>>>>> ```
>>>>>>>> echo 0 | sudo tee /sys/class/net/eth0/threaded
>>>>>>>> echo 0 | sudo tee /proc/sys/kernel/timer_migration
>>>>>>>> echo off | sudo tee  /sys/devices/system/cpu/smt/control
>>>>>>>>
>>>>>>>> sudo ethtool -L eth0 rx 1 tx 1
>>>>>>>> sudo ethtool -G eth0 rx 1024
>>>>>>>>
>>>>>>>> echo 0 | sudo tee /proc/sys/net/core/rps_sock_flow_entries
>>>>>>>> echo 0 | sudo tee /sys/class/net/eth0/queues/rx-0/rps_cpus
>>>>>>>>
>>>>>>>>      # pin IRQs on CPU 2
>>>>>>>> IRQS="$(gawk '/eth0-(TxRx-)?1/ {match($1, /([0-9]+)/, arr); \
>>>>>>>>                                  print arr[0]}' < /proc/interrupts)"
>>>>>>>> for irq in "${IRQS}"; \
>>>>>>>>          do echo 2 | sudo tee /proc/irq/$irq/smp_affinity_list; done
>>>>>>>>
>>>>>>>> echo -1 | sudo tee /proc/sys/kernel/sched_rt_runtime_us
>>>>>>>>
>>>>>>>> for i in /sys/devices/virtual/workqueue/*/cpumask; \
>>>>>>>>                          do echo $i; echo 1,2,3,4,5,6 > $i; done
>>>>>>>>
>>>>>>>> if [[ -z "$1" ]]; then
>>>>>>>>       echo 400 | sudo tee /proc/sys/net/core/busy_read
>>>>>>>>       echo 100 | sudo tee /sys/class/net/eth0/napi_defer_hard_irqs
>>>>>>>>       echo 15000   | sudo tee /sys/class/net/eth0/gro_flush_timeout
>>>>>>>> fi
>>>>>>>>
>>>>>>>> sudo ethtool -C eth0 adaptive-rx off adaptive-tx off rx-usecs 0 tx-
>>>>>>>> usecs 0
>>>>>>>>
>>>>>>>> if [[ "$1" == "enable_threaded" ]]; then
>>>>>>>>       echo 0 | sudo tee /proc/sys/net/core/busy_poll
>>>>>>>>       echo 0 | sudo tee /proc/sys/net/core/busy_read
>>>>>>>>       echo 100 | sudo tee /sys/class/net/eth0/napi_defer_hard_irqs
>>>>>>>>       echo 15000 | sudo tee /sys/class/net/eth0/gro_flush_timeout
>>>>>>>>       echo 2 | sudo tee /sys/class/net/eth0/threaded
>>>>>>>>       NAPI_T=$(ps -ef | grep napi | grep -v grep | awk '{ print $2 }')
>>>>>>>>       sudo chrt -f  -p 50 $NAPI_T
>>>>>>>>
>>>>>>>>       # pin threaded poll thread to CPU 2
>>>>>>>>       sudo taskset -pc 2 $NAPI_T
>>>>>>>> fi
>>>>>>>>
>>>>>>>> if [[ "$1" == "enable_interrupt" ]]; then
>>>>>>>>       echo 0 | sudo tee /proc/sys/net/core/busy_read
>>>>>>>>       echo 0 | sudo tee /sys/class/net/eth0/napi_defer_hard_irqs
>>>>>>>>       echo 15000 | sudo tee /sys/class/net/eth0/gro_flush_timeout
>>>>>>>> fi
>>>>>>>> ```
>>>>>>>
>>>>>>> The experiment script above does not work, because the sysfs parameter
>>>>>>> does not exist anymore in this version.
>>>>>>>
>>>>>>>> To enable various configurations, script can be run as following,
>>>>>>>>
>>>>>>>> - Interrupt Only
>>>>>>>>       ```
>>>>>>>>       <script> enable_interrupt
>>>>>>>>       ```
>>>>>>>>
>>>>>>>> - SO_BUSYPOLL (no arguments to script)
>>>>>>>>       ```
>>>>>>>>       <script>
>>>>>>>>       ```
>>>>>>>>
>>>>>>>> - NAPI threaded busypoll
>>>>>>>>       ```
>>>>>>>>       <script> enable_threaded
>>>>>>>>       ```
>>>>>>>>
>>>>>>>> If using idpf, the script needs to be run again after launching the
>>>>>>>> workload just to make sure that the configurations are not reverted. As
>>>>>>>> idpf reverts some configurations on software reset when AF_XDP program
>>>>>>>> is attached.
>>>>>>>>
>>>>>>>> Once configured, the workload is run with various configurations using
>>>>>>>> following commands. Set period (1/frequency) and delay in usecs to
>>>>>>>> produce results for packet frequency and application processing delay.
>>>>>>>>
>>>>>>>>      ## Interrupt Only and SO_BUSYPOLL (inline)
>>>>>>>>
>>>>>>>> - Server
>>>>>>>> ```
>>>>>>>> sudo chrt -f 50 taskset -c 3-5 ./xsk_rr -o 0 -B 400 -i eth0 -4 \
>>>>>>>>          -D <IP-dest> -S <IP-src> -M <MAC-dst> -m <MAC-src> -p 54321 -
>>>>>>>> h -v
>>>>>>>> ```
>>>>>>>>
>>>>>>>> - Client
>>>>>>>> ```
>>>>>>>> sudo chrt -f 50 taskset -c 3-5 ./xsk_rr -o 0 -B 400 -i eth0 -4 \
>>>>>>>>          -S <IP-src> -D <IP-dest> -m <MAC-src> -M <MAC-dst> -p 54321 \
>>>>>>>>          -P <Period-usecs> -d <Delay-usecs>  -T -l 1 -v
>>>>>>>> ```
>>>>>>>>
>>>>>>>>      ## SO_BUSYPOLL(done in separate core using recvfrom)
> Defines this test case clearly here.
>>>>>>>>
>>>>>>>> Argument -t spawns a seprate thread and continuously calls recvfrom.
> This defines the -t argument and clearly states that it spawns the
> separate thread.
>>>>>>>>
>>>>>>>> - Server
>>>>>>>> ```
>>>>>>>> sudo chrt -f 50 taskset -c 3-5 ./xsk_rr -o 0 -B 400 -i eth0 -4 \
>>>>>>>>          -D <IP-dest> -S <IP-src> -M <MAC-dst> -m <MAC-src> -p 54321 \
>>>>>>>>          -h -v -t
>>>>>>>> ```
>>>>>>>>
>>>>>>>> - Client
>>>>>>>> ```
>>>>>>>> sudo chrt -f 50 taskset -c 3-5 ./xsk_rr -o 0 -B 400 -i eth0 -4 \
>>>>>>>>          -S <IP-src> -D <IP-dest> -m <MAC-src> -M <MAC-dst> -p 54321 \
>>>>>>>>          -P <Period-usecs> -d <Delay-usecs>  -T -l 1 -v -t
>>>>>>>> ```
>>
>> see below
>>>>>>>>      ## NAPI Threaded Busy Poll
> Section for NAPI Threaded Busy Poll scenario
>>>>>>>>
>>>>>>>> Argument -n skips the recvfrom call as there is no recv kick needed.
> States -n argument and defines it.
>>>>>>>>
>>>>>>>> - Server
>>>>>>>> ```
>>>>>>>> sudo chrt -f 50 taskset -c 3-5 ./xsk_rr -o 0 -B 400 -i eth0 -4 \
>>>>>>>>          -D <IP-dest> -S <IP-src> -M <MAC-dst> -m <MAC-src> -p 54321 \
>>>>>>>>          -h -v -n
>>>>>>>> ```
>>>>>>>>
>>>>>>>> - Client
>>>>>>>> ```
>>>>>>>> sudo chrt -f 50 taskset -c 3-5 ./xsk_rr -o 0 -B 400 -i eth0 -4 \
>>>>>>>>          -S <IP-src> -D <IP-dest> -m <MAC-src> -M <MAC-dst> -p 54321 \
>>>>>>>>          -P <Period-usecs> -d <Delay-usecs>  -T -l 1 -v -n
>>>>>>>> ```
>>
>> see below
>>>>>>> I believe there's a bug when disabling busy-polled napi threading after
>>>>>>> an experiment. My system hangs and needs a hard reset.
>>>>>>>
>>>>>>>> | Experiment | interrupts | SO_BUSYPOLL | SO_BUSYPOLL(separate) |
>>>>>>>> NAPI threaded |
>>>>>>>> |---|---|---|---|---|
>>>>>>>> | 12 Kpkt/s + 0us delay | | | | |
>>>>>>>> |  | p5: 12700 | p5: 12900 | p5: 13300 | p5: 12800 |
>>>>>>>> |  | p50: 13100 | p50: 13600 | p50: 14100 | p50: 13000 |
>>>>>>>> |  | p95: 13200 | p95: 13800 | p95: 14400 | p95: 13000 |
>>>>>>>> |  | p99: 13200 | p99: 13800 | p99: 14400 | p99: 13000 |
>>>>>>>> | 32 Kpkt/s + 30us delay | | | | |
>>>>>>>> |  | p5: 19900 | p5: 16600 | p5: 13100 | p5: 12800 |
>>>>>>>> |  | p50: 21100 | p50: 17000 | p50: 13700 | p50: 13000 |
>>>>>>>> |  | p95: 21200 | p95: 17100 | p95: 14000 | p95: 13000 |
>>>>>>>> |  | p99: 21200 | p99: 17100 | p99: 14000 | p99: 13000 |
>>>>>>>> | 125 Kpkt/s + 6us delay | | | | |
>>>>>>>> |  | p5: 14600 | p5: 17100 | p5: 13300 | p5: 12900 |
>>>>>>>> |  | p50: 15400 | p50: 17400 | p50: 13800 | p50: 13100 |
>>>>>>>> |  | p95: 15600 | p95: 17600 | p95: 14000 | p95: 13100 |
>>>>>>>> |  | p99: 15600 | p99: 17600 | p99: 14000 | p99: 13100 |
>>>>>>>> | 12 Kpkt/s + 78us delay | | | | |
>>>>>>>> |  | p5: 14100 | p5: 16700 | p5: 13200 | p5: 12600 |
>>>>>>>> |  | p50: 14300 | p50: 17100 | p50: 13900 | p50: 12800 |
>>>>>>>> |  | p95: 14300 | p95: 17200 | p95: 14200 | p95: 12800 |
>>>>>>>> |  | p99: 14300 | p99: 17200 | p99: 14200 | p99: 12800 |
>>>>>>>> | 25 Kpkt/s + 38us delay | | | | |
>>>>>>>> |  | p5: 19900 | p5: 16600 | p5: 13000 | p5: 12700 |
>>>>>>>> |  | p50: 21000 | p50: 17100 | p50: 13800 | p50: 12900 |
>>>>>>>> |  | p95: 21100 | p95: 17100 | p95: 14100 | p95: 12900 |
>>>>>>>> |  | p99: 21100 | p99: 17100 | p99: 14100 | p99: 12900 |
>>>>>>>
>>>>>>> On my system, routing the irq to same core where xsk_rr runs results in
>>>>>>> lower latency than routing the irq to a different core. To me that makes
>>>>>>> sense in a low-rate latency-sensitive scenario where interrupts are not
>>>>>>> causing much trouble, but the resulting locality might be beneficial. I
>>>>>>> think you should test this as well.
>>>>>>>
>>>>>>> The experiments reported above (except for the first one) are
>>>>>>> cherry-picking parameter combinations that result in a near-100% load
>>>>>>> and ignore anything else. Near-100% load is a highly unlikely scenario
>>>>>>> for a latency-sensitive workload.
>>>>>>>
>>>>>>> When combining the above two paragraphs, I believe other interesting
>>>>>>> setups are missing from the experiments, such as comparing to two pairs
>>>>>>> of xsk_rr under high load (as mentioned in my previous emails).
>>>>>> This is to support an existing real workload. We cannot easily modify
>>>>>> its threading model. The two xsk_rr model would be a different
>>>>>> workload.
>>>>>
>>>>> That's fine, but:
>>>>>
>>>>> - In principle I don't think it's a good justification for a kernel
>>>>> change that an application cannot be rewritten.
>>>>>
>>>>> - I believe it is your responsibility to more comprehensively document
>>>>> the impact of your proposed changes beyond your one particular workload.>
>>>> A few more observations from my tests for the "SO_BUSYPOLL(separate)" case:
>>>>
>>>> - Using -t for the client reduces latency compared to -T.
>>> That is understandable and also it is part of the data I presented. -t
>>> means running the SO_BUSY_POLL in a separate thread. Removing -T would
>>> invalidate the workload by making the rate unpredictable.
>>
>> That's another problem with your cover letter then. The experiment as
>> described should match the data presented. See above.
> The experiments are described clearly. I have pointed out the areas in
> the cover letter where these are documented. Where is the mismatch?

Ah, I missed the -t at the end, sorry, my bad.

>>>> - Using poll instead of recvfrom in xsk_rr in rx_polling_run() also
>>>> reduces latency.
>>
>> Any thoughts on this one?
> I think we discussed this already in the previous iteration, with
> Stanislav, and how it will suffer the same way SO_BUSYPOLL suffers. As
> I have already stated, for my workload every microsecond matters and
> the CPU efficiency is not an issue.

Discussing is one thing. Testing is another. In my setup I observe a 
noticeable difference between using recvfrom and poll.

Thanks,
Martin

Powered by blists - more mailing lists