Re: [dpdk-dev] [PATCH] examples/ipsec-secgw: fix SPD no-match is misinterpreted

["Ananyev, Konstantin" <konstantin.ananyev@intel.com>]

Hi Akhil, 

> > > > acl_classify() returns zero value when no matching rule was found.
> > > > Currently ipsec-secgw treats it as a valid SPI value, though it has
> > > > to discard such packets.
> > > > Error could be easily observed by sending outbound unmatched packets,
> > > > user will see something like that in the log:
> > > > IPSEC: No cryptodev: core 7, cipher_algo 0, auth_algo 0, aead_algo 0
> > > >
> > > > To fix it we need to treat packets with zero result from acl_classify()
> > > > as invalid ones. Also we can change DISCARD and BYPASS values to
> > > > simplify checks and save some extra space for valid SPI values.
> > > spi value =0 is invalid but zero result may have a valid packet.
> > > consider a case:
> > > SPI = 128 or 256 or 512 and so on  => sa_idx = 0 and result will come as
> > > zero, and this would be a valid packet.
> > >
> > > I see that the sa_idx  calculation logic is not correct in first place.
> > > There will be multiple spi values for same sa_idx which is not correct.
> > > So we have 2 issues here:
> > > 1. result = 0, means sa_idx =0 which may be correct, but as you said if
> > > acl_classify fails, it also return 0.
> > > 2. SPI values which are IPSEC_SA_MAX_ENTRIES apart will have same sa_idx
> > > and will keep on overwriting the previous ones.
> > >
> >
> > Ok I see what you mean.
> > The easiest fix for that (till we'll have proper SAD) would be not to allow
> > SPIs bigger than IPSEC_SA_MAX_ENTRIES.
> > Are you ok with that?
> > Konstantin
> 
> After another thought, it seems that we can easily overcome that problem
> without introducing extra limitations - just need to store 'sa_idx + 1' in acl table.
> Will give it a try with v2.

Actually, after yet another thought - I think there is no need for that.
We can store in ACL full SPI and do SPI2IDX after acl_classify().
Just sent v2, please have a look.
Konstantin