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;
Subject: Re: [PATCH v1 1/6] fs/readdir: Fix filldir() and filldir64() use of
 user_access_begin()
To:     Al Viro <viro@zeniv.linux.org.uk>,
        Linus Torvalds <torvalds@linux-foundation.org>
Cc:     Benjamin Herrenschmidt <benh@kernel.crashing.org>,
        Paul Mackerras <paulus@samba.org>,
        Michael Ellerman <mpe@ellerman.id.au>,
        Andrew Morton <akpm@linux-foundation.org>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        linuxppc-dev <linuxppc-dev@lists.ozlabs.org>,
        linux-fsdevel <linux-fsdevel@vger.kernel.org>,
        Linux-MM <linux-mm@kvack.org>
References: <a02d3426f93f7eb04960a4d9140902d278cab0bb.1579697910.git.christophe.leroy@c-s.fr>
 <CAHk-=whTzEu5=sMEVLzuf7uOnoCyUs8wbfw87njes9FyE=mj1w@mail.gmail.com>
 <20200122174129.GH23230@ZenIV.linux.org.uk>
From:   Christophe Leroy <christophe.leroy@c-s.fr>
Message-ID: <e6423f62-c29a-1a67-fb75-1330f5ef1348@c-s.fr>
Date:   Wed, 22 Jan 2020 18:54:45 +0100
User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:60.0) Gecko/20100101
 Thunderbird/60.9.1
MIME-Version: 1.0
In-Reply-To: <20200122174129.GH23230@ZenIV.linux.org.uk>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Language: fr
Content-Transfer-Encoding: 8bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk


Le 22/01/2020 à 18:41, Al Viro a écrit :
> On Wed, Jan 22, 2020 at 08:13:12AM -0800, Linus Torvalds wrote:
>> On Wed, Jan 22, 2020 at 5:00 AM Christophe Leroy
>> <christophe.leroy@c-s.fr> wrote:
>>>
>>> Modify filldir() and filldir64() to request the real area they need
>>> to get access to.
>>
>> Not like this.
>>
>> This makes the situation for architectures like x86 much worse, since
>> you now use "put_user()" for the previous dirent filling. Which does
>> that expensive user access setup/teardown twice again.
>>
>> So either you need to cover both the dirent's with one call, or you
>> just need to cover the whole (original) user buffer passed in. But not
>> this unholy mixing of both unsafe_put_user() and regular put_user().
> 
> I would suggest simply covering the range from dirent->d_off to
> buf->current_dir->d_name[namelen]; they are going to be close to
> each other and we need those addresses anyway...
> 

In v2, I'm covering from the beginning of parent dirent to the end of 
current dirent.

Christophe