You can use bool here and later on true and false.

Would it make sense to put the return (0); before the #endif and add an #else with return(1)?

That's fancy. Done.

I don't know. I chose the way with less code lines.

I don't know. I chose the way with less code lines.

If neither INET nor INET6 is defined, doesn't the function now returns 0? I would expect 1 for errors.

This function returns 1 before the patch in that caste. To keep the functional changes small, I changed the code as suggested.

Since this is a general fix, I took it out in D52046.

Thanks! Done by rebase.

The number of times a packet with a request for receive or transmit checksum offload was received and this request failed.
The different reasons for the failure are counted by
.Va dev.vtnet. Ns Ar X Ns Va .rx_csum_inaccessible_ipproto ,
.Va dev.vtnet. Ns Ar X Ns Va .rx_csum_bad_ipproto ,
.Va dev.vtnet. Ns Ar X Ns Va .rx_csum_bad_ethtype ,
or
.Va dev.vtnet. Ns Ar X Ns Va .rx_csum_bad_offset .

The number of times a packet with a request for receive or transmit checksum offload was received where the IP protocol was not accessible.

The number of times fixing the checksum required by
.Va hw.vtnet.fixup_needs_csum
or
.Va hw.vtnet. Ns Ar X Ns Va .fixup_needs_csum
was attempted for a packet where the csum is not located in the first mbuf.

The number of times a packet with a request for receive or transmit checksum offload was received where the IP protocol was neither TCP nor UDP.

The number of times a packet with a request for receive or transmit checksum offload was received where the EtherType was neither IPv4 nor IPv6.

Shouldn't we also increment some error counter here? Why not use rx_csum_bad_offset?

Use %x for flags.

Only declare variables at the beginning of a block.

Comment should be a sentence. So start with a capital letter and end with a full stop.

Use checksum offloading instead of checksum offload, also below.
Finally use and instead of or.