MIME-Version: 1.0
References: <20211111191800.21281-1-david@redhat.com> <20211112070113.GA19016@MiWiFi-R3L-srv>
 <21bdcecd-127c-f70e-0c7d-cb1b97caecb0@redhat.com> <20211112090116.GC19016@MiWiFi-R3L-srv>
In-Reply-To: <20211112090116.GC19016@MiWiFi-R3L-srv>
From:   David Hildenbrand <david@redhat.com>
Date:   Fri, 12 Nov 2021 10:08:14 +0100
Message-ID: <CADFyXm7uS3GN1AnF-iLpUZMFK=MwF3=NGwSZFqXPA+kK182-cQ@mail.gmail.com>
Subject: Re: [PATCH v1] proc/vmcore: fix clearing user buffer by properly
 using clear_user()
To:     Baoquan He <bhe@redhat.com>
Cc:     Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        Dave Young <dyoung@redhat.com>,
        Vivek Goyal <vgoyal@redhat.com>,
        Andrew Morton <akpm@linux-foundation.org>,
        Philipp Rudo <prudo@redhat.com>, kexec@lists.infradead.org,
        Linux MM <linux-mm@kvack.org>,
        linux-fsdevel <linux-fsdevel@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

> > "that allows supervisor mode programs to optionally set user-space
> > memory mappings so that access to those mappings from supervisor mode
> > will cause a trap. This makes it harder for malicious programs to
> > "trick" the kernel into using instructions or data from a user-space
> > program"
>
> OK, probably. I thought it's triggered in access_ok(), and tried to
> figure out why. But seems we should do something to check this in
> access_ok(), otherwise the logic of clear_user/_clear_user is not so
> reasonable. Anyway, I have learned it, thanks a lot for digging it out.
>
> By the way, I can't open above wiki article, found below commit from
> hpa. Maybe we can add some into log to tell this, not strong opinin,
> leave it to you.

Yes, now that we know the root cause I'll add some more details to the
patch description and resend -- thanks Baoquan!