Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
MIME-Version: 1.0
References: <cover.thread-4f3684.your-ad-here.call-01526696764-ext-9375@work.hours>
 <patch-1.thread-4f3684.git-4f3684ca2aad.your-ad-here.call-01526696764-ext-9375@work.hours>
In-Reply-To: <patch-1.thread-4f3684.git-4f3684ca2aad.your-ad-here.call-01526696764-ext-9375@work.hours>
From:   Linus Torvalds <torvalds@linux-foundation.org>
Date:   Fri, 18 May 2018 20:20:26 -0700
Message-ID: <CA+55aFxJb9P2TMCc2Pao-uXcVACK-CTBHmzObAQu=rvs87itvw@mail.gmail.com>
Subject: Re: [PATCH] procfs: fix mmap() for /proc/vmcore
To:     gor@linux.ibm.com
Cc:     Andrew Morton <akpm@linux-foundation.org>,
        Al Viro <viro@zeniv.linux.org.uk>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        Heiko Carstens <heiko.carstens@de.ibm.com>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Fri, May 18, 2018 at 8:15 PM Vasily Gorbik <gor@linux.ibm.com> wrote:

> Commit be83bbf80682 ("mmap: introduce sane default mmap limits")
> introduced "pgoff" limits checks for mmap, considering fs max file size
> as upper limit, which made it impossible to mmap /proc/vmcore file
> contents above 2Gb (/proc/vmcore appears to be the only procfs file
> supporting mmap).

> Reuse MAX_LFS_FILESIZE as procfs s_maxbytes value.

Ugh. /proc is where a lot of problems *have* been.

Admittedly not as much as random drivers, but still. If proc doesn't set
s_maxbytes, then we should not raise it here magically, there might be
various /proc files that get offset accounting wrong.

I'd *much* rather just set FMODE_UNSIGNED_OFFSET for /proc/vmcore _only_,
rather than open up all proc files to issues with 4G+ offsets.

                Linus