On Fri, Dec 13, 2019 at 05:59:54PM -0800, Eric Dumazet wrote:
>
>
> On 12/13/19 4:47 PM, Martin KaFai Lau wrote:
> > This patch adds a helper to handle jiffies. Some of the
> > tcp_sock's timing is stored in jiffies. Although things
> > could be deduced by CONFIG_HZ, having an easy way to get
> > jiffies will make the later bpf-tcp-cc implementation easier.
> >
>
> ...
>
> > +
> > +BPF_CALL_2(bpf_jiffies, u64, in, u64, flags)
> > +{
> > + if (!flags)
> > + return get_jiffies_64();
> > +
> > + if (flags & BPF_F_NS_TO_JIFFIES) {
> > + return nsecs_to_jiffies(in);
> > + } else if (flags & BPF_F_JIFFIES_TO_NS) {
> > + if (!in)
> > + in = get_jiffies_64();
> > + return jiffies_to_nsecs(in);
> > + }
> > +
> > + return 0;
> > +}
>
> This looks a bit convoluted :)
>
> Note that we could possibly change net/ipv4/tcp_cubic.c to no longer use jiffies at all.
>
> We have in tp->tcp_mstamp an accurate timestamp (in usec) that can be converted to ms.
Thanks for the feedbacks!
I have a few questions that need some helps.
Does it mean tp->tcp_mstamp can be used as the "now" in cubic?
or tcp_clock_ns() should still be called in cubic, e.g. to replace
bictcp_clock() and tcp_jiffies32?
BPF currently has a helper calling ktime_get_mono_fast_ns() which looks
different from tcp_clock_ns().
The lsndtime is in jiffies. I think it can probably be converted to ms before
using it in cubic. There are some BICTCP_HZ logic in bictcp_update() that
is not obvious to me how to convet them to ms base also.
>
>
> Have you thought of finding a way to not duplicate the code for cubic and dctcp, maybe
> by including a template ?
>
> Maintaining two copies means that future changes need more maintenance work.
At least for bpf_dctcp.c, I did not expect it could be that close to tcp_dctcp.c
when I just started converted it. tcp_cubic/bpf_cubic still has some TBD
on jiffies/msec.
Agree that it is beneficial to have one copy. It is likely
I need to make some changes on the tcp_*.c side also. Hence, I prefer
to give it a try in a separate series, e.g. revert the kernel side
changes will be easier.
