Re: [PATCH net-next 2/2] net:add socket option for low latency polling

On Tue, 2013-06-11 at 18:37 +0300, Eliezer Tamir wrote:

On 11/06/2013 17:45, Eric Dumazet wrote:

quoted

On Tue, 2013-06-11 at 17:24 +0300, Eliezer Tamir wrote:

quoted

adds a socket option for low latency polling.
This allows overriding the global sysctl value with a per-socket one.

Signed-off-by: Eliezer Tamir <redacted>
---


-static inline cycles_t ll_end_time(void)
+static inline cycles_t ll_end_time(struct sock *sk)
  {
-	return TSC_MHZ * ACCESS_ONCE(sysctl_net_ll_poll) + get_cycles();
+	return TSC_MHZ * ACCESS_ONCE(sk->sk_ll_usec) + get_cycles();
  }

Hmm, sk_ll_usec is an unsigned int.

We changed it to an unsigned long in v7, I guess that was gratuitous.
Re-reading your comments on v6 2/5 I realize a cast would have sufficed.
Should I send a patch to revert it to an unsigned int?

One sysctl as unsigned long was not a big deal so I was ok with your
change ;)

unsigned int for sk_ll_usec is enough, but using a 32x32->64bit multiply
is probably safer.





------------------------------------------------------------------------------
This SF.net email is sponsored by Windows:

Build for Windows Store.

http://p.sf.net/sfu/windows-dev2dev
_______________________________________________
E1000-devel mailing list
E1000-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/e1000-devel
To learn more about Intel® Ethernet, visit http://communities.intel.com/community/wired