Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:18 as permitted sender) client-ip=2620:137:e000::1:18;
MIME-Version: 1.0
References: <20220407031525.2368067-1-yuzhao@google.com> <20220407031525.2368067-9-yuzhao@google.com>
 <20220411191621.0378467ad99ebc822d5ad005@linux-foundation.org>
 <CAOUHufYeC=Kuu59BPL_48sM67CqACxH2wWy-SYGXpadgMDmY3w@mail.gmail.com>
 <20220414185654.e7150bcbe859e0dd4b9c61af@linux-foundation.org>
 <CAOUHufYy6yQS9ARN9C5+ODkopR+ez4TH3hZNZo4HtNHBExS1mA@mail.gmail.com>
 <20220415121521.764a88dda55ae8c676ad26b0@linux-foundation.org>
 <CAOUHufYsjwMGMFCfYoh79rFZqwqS1jDihcBS9sHd-gBxEAD3Ug@mail.gmail.com>
 <20220415143220.cc37b0b0a368ed2bf2a821f8@linux-foundation.org>
 <CAHk-=whvkRTVBhAamt0kYyp925jk_+g7T0CyPke_FbCWGQ1VvA@mail.gmail.com>
 <CAOUHufZ4KrjFTYH8wtwMGd9AriZfZtO4GhbiK1SuNbY31VTT9w@mail.gmail.com> <CAHk-=whneDk3Jde3J+O-fD32VjaK+fDf9+P6jgDtr2qyo0iu2w@mail.gmail.com>
In-Reply-To: <CAHk-=whneDk3Jde3J+O-fD32VjaK+fDf9+P6jgDtr2qyo0iu2w@mail.gmail.com>
From:   Jesse Barnes <jsbarnes@google.com>
Date:   Fri, 15 Apr 2022 16:24:14 -0700
Message-ID: <CAJmaN=mJKAhRjDHbsLrqyK_uvuysx596jEcTwV2fwx4UOe4oow@mail.gmail.com>
Subject: Re: [page-reclaim] Re: [PATCH v10 08/14] mm: multi-gen LRU: support
 page table walks
To:     Linus Torvalds <torvalds@linux-foundation.org>
Cc:     Yu Zhao <yuzhao@google.com>,
        Andrew Morton <akpm@linux-foundation.org>,
        Justin Forbes <jforbes@fedoraproject.org>,
        Stephen Rothwell <sfr@rothwell.id.au>,
        Linux-MM <linux-mm@kvack.org>, Andi Kleen <ak@linux.intel.com>,
        Aneesh Kumar <aneesh.kumar@linux.ibm.com>,
        Barry Song <21cnbao@gmail.com>,
        Catalin Marinas <catalin.marinas@arm.com>,
        Dave Hansen <dave.hansen@linux.intel.com>,
        Hillf Danton <hdanton@sina.com>, Jens Axboe <axboe@kernel.dk>,
        Johannes Weiner <hannes@cmpxchg.org>,
        Jonathan Corbet <corbet@lwn.net>,
        Matthew Wilcox <willy@infradead.org>,
        Mel Gorman <mgorman@suse.de>,
        Michael Larabel <Michael@michaellarabel.com>,
        Michal Hocko <mhocko@kernel.org>,
        Mike Rapoport <rppt@kernel.org>,
        Rik van Riel <riel@surriel.com>,
        Vlastimil Babka <vbabka@suse.cz>,
        Will Deacon <will@kernel.org>,
        Ying Huang <ying.huang@intel.com>,
        Linux ARM <linux-arm-kernel@lists.infradead.org>,
        "open list:DOCUMENTATION" <linux-doc@vger.kernel.org>,
        linux-kernel <linux-kernel@vger.kernel.org>,
        Kernel Page Reclaim v2 <page-reclaim@google.com>,
        "the arch/x86 maintainers" <x86@kernel.org>,
        Brian Geffon <bgeffon@google.com>,
        Jan Alexander Steffens <heftig@archlinux.org>,
        Oleksandr Natalenko <oleksandr@natalenko.name>,
        Steven Barrett <steven@liquorix.net>,
        Suleiman Souhlal <suleiman@google.com>,
        Daniel Byrne <djbyrne@mtu.edu>,
        Donald Carr <d@chaos-reins.com>,
        =?UTF-8?Q?Holger_Hoffst=C3=A4tte?= <holger@applied-asynchrony.com>,
        Konstantin Kharlamov <Hi-Angel@yandex.ru>,
        Shuang Zhai <szhai2@cs.rochester.edu>,
        Sofia Trinh <sofia.trinh@edi.works>,
        Vaibhav Jain <vaibhav@linux.ibm.com>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Fri, Apr 15, 2022 at 4:04 PM Linus Torvalds
<torvalds@linux-foundation.org> wrote:
>
> On Fri, Apr 15, 2022 at 3:58 PM Yu Zhao <yuzhao@google.com> wrote:
> >
> > BUG_ONs are harmful but problems that trigger them would be
> > presummingly less penetrating to the user base; on the other hand,
> > from my experience working with some testers (ordinary users), they
> > ignore WARN_ON_ONCEs until the kernel crashes.
>
> I don't understand your argument.
>
> First you say that VM_BUG_ON() is only for VM developers.
>
> Then you say "some testers (ordinary users) ignore WARN_ON_ONCEs until
> the kernel crashes".
>
> So which is it?
>
> VM developers, or ordinary users?
>
> Honestly, if a VM developer is ignoring a WARN_ON_ONCE() from the VM
> subsystem, I don't even know what to say.
>
> And for ordinary users, a WARN_ON_ONCE() is about a million times
> better, becasue:
>
>  - the machine will hopefully continue working, so they can report the warning
>
>  - even when they don't notice them, distros tend to have automated
> reporting infrastructure
>
> That's why I absolutely *DETEST* those stupid BUG_ON() cases - they
> will often kill the machine with nasty locks held, resulting in a
> completely undebuggable thing that never gets reported.
>
> Yes, you can be careful and only put BUG_ON() in places where recovery
> is possible. But even then, they have no actual _advantages_ over just
> a WARN_ON_ONCE.

Generally agreed, and not to belabor this relatively small issue, but in some
environments like cloud or managed client deployments, a crash can actually
be preferable so we can get a dump, reboot the machine, and get things going
again for the application or user, then debug offline.  So having the
flexibility to
do that in those situations is helpful.  And there, a full crash dump is better
than just a log report with the WARN info, since debugging may be easier with
all the kernel memory.

Jesse