On 07/06/2018 06:56 AM, Paolo Abeni wrote: > With: > > schedule_work_on(smp_processor_id(), #... ) > > We can be sure to run exclusively on the cpu handling the RX queue even with > the worker. > Make sure to test your patch with 16 RX queues (16 cpus) feeding the defrag unit. In this (common) mode, then having one cpu trying to purge frags wont be enough. Really the GC can not cope with DDOS, we tried that in the past (IPv4 route cache) and failed miserably.
- Re: [RFC PATCH] ip: re-introduce fragments cache worker Paolo Abeni
- Re: [RFC PATCH] ip: re-introduce fragments cache worker Eric Dumazet
- Re: [RFC PATCH] ip: re-introduce fragments cache wor... Paolo Abeni
- Re: [RFC PATCH] ip: re-introduce fragments cache... Eric Dumazet
- Re: [RFC PATCH] ip: re-introduce fragments cache... Eric Dumazet
- Re: [RFC PATCH] ip: re-introduce fragments cache... Eric Dumazet
- Re: [RFC PATCH] ip: re-introduce fragments cache... Paolo Abeni
- Re: [RFC PATCH] ip: re-introduce fragments cache... Eric Dumazet
- Re: [RFC PATCH] ip: re-introduce fragments cache... Paolo Abeni
- Re: [RFC PATCH] ip: re-introduce fragments cache... Eric Dumazet
- Re: [RFC PATCH] ip: re-introduce fragments cache worker Eric Dumazet