Re: inpcb sip hash mutex contention

2023-06-26 Thread Alexander Bluhm
On Sat, Jun 24, 2023 at 11:20:50AM +1000, David Gwynne wrote: > maybe it's time to re-evaluate siphash? In one of our products we have replaced SipHash with xxHash. https://xxhash.com/ https://github.com/Cyan4973/xxHash bluhm

Re: inpcb sip hash mutex contention

2023-06-24 Thread Vitaliy Makkoveev
> On 22 Jun 2023, at 22:50, Alexander Bluhm wrote: > > Hi, > > I am working on a diff to run UDP input in parallel. Btrace kstack > analysis shows that SIP hash for PCB lookup is quite expensive. > When running in parallel we get lock contention on the PCB table > mutex. > > So it results in b

Re: inpcb sip hash mutex contention

2023-06-23 Thread David Gwynne
makes sense to me, ok. maybe it's time to re-evaluate siphash? > On 23 Jun 2023, at 05:50, Alexander Bluhm wrote: > > Hi, > > I am working on a diff to run UDP input in parallel. Btrace kstack > analysis shows that SIP hash for PCB lookup is quite expensive. > When running in parallel we get

inpcb sip hash mutex contention

2023-06-22 Thread Alexander Bluhm
Hi, I am working on a diff to run UDP input in parallel. Btrace kstack analysis shows that SIP hash for PCB lookup is quite expensive. When running in parallel we get lock contention on the PCB table mutex. So it results in better performance to calculate the hash value before taking the mutex.