From mboxrd@z Thu Jan 1 00:00:00 1970 From: Paolo Abeni Subject: Re: linux-next: BUG: unable to handle kernel NULL pointer dereference in __sk_mem_raise_allocated() Date: Tue, 15 Nov 2016 09:57:19 +0100 Message-ID: <1479200239.4660.3.camel@redhat.com> References: <1479166501.8455.91.camel@edumazet-glaptop3.roam.corp.google.com> <1479169886.8455.111.camel@edumazet-glaptop3.roam.corp.google.com> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 7bit Cc: Andrei Vagin , Linux Kernel Network Developers To: Eric Dumazet Return-path: Received: from mx1.redhat.com ([209.132.183.28]:53922 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755456AbcKOI5V (ORCPT ); Tue, 15 Nov 2016 03:57:21 -0500 In-Reply-To: <1479169886.8455.111.camel@edumazet-glaptop3.roam.corp.google.com> Sender: netdev-owner@vger.kernel.org List-ID: On Mon, 2016-11-14 at 16:31 -0800, Eric Dumazet wrote: > On Mon, 2016-11-14 at 15:58 -0800, Andrei Vagin wrote: > > > > > Yes, you are right. It works if we set .memory_allocated and .sysctl_mem. > > Now the question would be : > > Are we okay if UDP and UDPlite share the same limits ? > > I would vote for yes, because these default limits are huge anyway > (The 50% reduction done in > https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=b66e91ccbc34ebd5a2f90f9e1bc1597e2924a500 > only impacted TCP ) Thank you for jumping on this so early! I'm sorry for the udplite left over: my fault. I agree with sharing the limits between UDP and UDPlite. I think that the current ones are so high we can apply also a similar reduction to UDP (and SCTP, too) Cheers, Paolo