Re: [PATCH v6 net-next 1/1] tcp: Replace min_tso_segs() with tso_segs() CC callback

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] tcp_tso_autosize is not exported as a BPF kfunc, making it impossible for custom BPF CC programs to utilize standard TSO autosizing.
--

--- Patch [1]: [PATCH v6 net-next 1/1] tcp: Replace min_tso_segs() with tso_segs() CC callback ---
commit e9e6269217c20aef163ee8ebfb7cab71f918807f
Author: Chia-Yu Chang <[email protected]>

tcp: Replace min_tso_segs() with tso_segs() CC callback

This patch replaces the existing min_tso_segs() CC callback with a new
tso_segs() callback. The new callback provides the explicit TSO segment
number for each data burst, completely bypassing tcp_tso_autosize().

> diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
> index d7c1444b5e304..d9bda3a84f261 100644
> --- a/net/ipv4/tcp_output.c
> +++ b/net/ipv4/tcp_output.c
[ ... ]
> @@ -2269,6 +2269,7 @@ static u32 tcp_tso_autosize(const struct sock *sk, unsigned int mss_now,
>  
>  	return max_t(u32, bytes / mss_now, min_tso_segs);
>  }
> +EXPORT_SYMBOL(tcp_tso_autosize);

[Severity: High]
Since BPF congestion control programs now need to compute the final TSO
segments value, do they need access to tcp_tso_autosize() to retain standard
autosizing?

The commit exports tcp_tso_autosize for native kernel modules, but it appears
missing from the BTF kfuncs set (bpf_tcp_ca_check_kfunc_ids) in
net/ipv4/bpf_tcp_ca.c:

BTF_KFUNCS_START(bpf_tcp_ca_check_kfunc_ids)
BTF_ID_FLAGS(func, tcp_reno_ssthresh)
BTF_ID_FLAGS(func, tcp_reno_cong_avoid)
...
BTF_KFUNCS_END(bpf_tcp_ca_check_kfunc_ids)

Without being registered here, custom BPF CC programs cannot call
tcp_tso_autosize(). This might prevent them from properly migrating to the
new tso_segs callback if they want to retain standard TSO autosizing.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1