From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S937114AbeE3HmX (ORCPT ); Wed, 30 May 2018 03:42:23 -0400 Received: from mx2.suse.de ([195.135.220.15]:32810 "EHLO mx2.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S935134AbeE3HmU (ORCPT ); Wed, 30 May 2018 03:42:20 -0400 Date: Wed, 30 May 2018 09:42:16 +0200 From: Michal Hocko To: Linus Torvalds Cc: Davidlohr Bueso , Andrew Morton , Thomas Graf , Herbert Xu , Manfred Spraul , guillaume.knispel@supersonicimagine.com, Linux API , Linux Kernel Mailing List , Davidlohr Bueso Subject: Re: [PATCH 3/6] lib/bucket_locks: use kvmalloc_array() Message-ID: <20180530074216.GZ27180@dhcp22.suse.cz> References: <20180524211135.27760-1-dave@stgolabs.net> <20180524211135.27760-4-dave@stgolabs.net> <20180529144317.GA20910@dhcp22.suse.cz> <20180529145106.GV27180@dhcp22.suse.cz> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.9.5 (2018-04-13) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Tue 29-05-18 15:46:25, Linus Torvalds wrote: [...] > The whole and ONLY point of "kvmalloc()" and friends is to make it easy to > write code and _not_ have those idiotic "let's do kmalloc or kvmalloc > depending on the phase of the moon" garbage. So the warning has literally > destroyed the only value that function has! Well, I do agree but I've also seen terrible things while doing the conversion when introducing kvmalloc. So I admit that the defensive mode here is mostly inspired by existing users of vmalloc(GFP_NOFS). They are simply wrong and not really eager to be fixed from my experience. Now with kvmalloc fixing them up silently it would get even less likely to get fixed because there won't be any deadlock possible (compared to open coded kvmalloc like ext4_kvmalloc for example). My experience also tells me that most of those vmalloc NOFS users simply do not need NOFS at all because there is no risk of the reclaim recursion deadlocks. They are just used because of cargo cult which is sad and it causes some subtle problems for the direct reclaim. I would really like to eliminate those (e.g. see [1]). It is sad reality that people tend to be more sensitive to WARN splats than "look this is wrong albeit not critical in most cases). [1] http://lkml.kernel.org/r/20180424162712.GL17484@dhcp22.suse.cz That being sad, if you believe that silently fixing up a code like that is a good idea we can do the following of course: >>From c1a098e809a109800f9cfa63cb27fe9a78f3f316 Mon Sep 17 00:00:00 2001 From: Michal Hocko Date: Wed, 30 May 2018 09:34:39 +0200 Subject: [PATCH] mm: kvmalloc does not fallback to vmalloc for incompatible gfp flags kvmalloc warned about incompatible gfp_mask to catch abusers (mostly GFP_NOFS) with an intention that this will motivate authors of the code to fix those. Linus argues that this just motivates people to do even more hacks like if (gfp == GFP_KERNEL) kvmalloc else kmalloc I haven't seen this happening but it is true that we can grow those in future. Therefore Linus suggested to simply not fallback to vmalloc for incompatible gfp flags and rather stick with the kmalloc path. Requested-by: Linus Torvalds Signed-off-by: Michal Hocko --- mm/util.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/mm/util.c b/mm/util.c index 45fc3169e7b0..c6586c146995 100644 --- a/mm/util.c +++ b/mm/util.c @@ -391,7 +391,8 @@ EXPORT_SYMBOL(vm_mmap); * __GFP_RETRY_MAYFAIL is supported, and it should be used only if kmalloc is * preferable to the vmalloc fallback, due to visible performance drawbacks. * - * Any use of gfp flags outside of GFP_KERNEL should be consulted with mm people. + * Please note that any use of gfp flags outside of GFP_KERNEL is careful to not + * fall back to vmalloc. */ void *kvmalloc_node(size_t size, gfp_t flags, int node) { @@ -402,7 +403,8 @@ void *kvmalloc_node(size_t size, gfp_t flags, int node) * vmalloc uses GFP_KERNEL for some internal allocations (e.g page tables) * so the given set of flags has to be compatible. */ - WARN_ON_ONCE((flags & GFP_KERNEL) != GFP_KERNEL); + if ((flags & GFP_KERNEL) != GFP_KERNEL) + return kmalloc_node(size, flags, node); /* * We want to attempt a large physically contiguous block first because -- 2.17.0 -- Michal Hocko SUSE Labs