Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Wed, 22 Mar 2023 11:31:47 +0000
From:   Lorenzo Stoakes <lstoakes@gmail.com>
To:     David Hildenbrand <david@redhat.com>
Cc:     linux-mm@kvack.org, linux-kernel@vger.kernel.org,
        linux-fsdevel@vger.kernel.org,
        Andrew Morton <akpm@linux-foundation.org>,
        Baoquan He <bhe@redhat.com>,
        Uladzislau Rezki <urezki@gmail.com>,
        Matthew Wilcox <willy@infradead.org>,
        Liu Shixin <liushixin2@huawei.com>,
        Jiri Olsa <jolsa@kernel.org>, Jens Axboe <axboe@kernel.dk>,
        Alexander Viro <viro@zeniv.linux.org.uk>
Subject: Re: [PATCH v4 4/4] mm: vmalloc: convert vread() to vread_iter()
Message-ID: <fe8f9909-332f-41c3-b672-a352cc6218d7@lucifer.local>
References: <cover.1679431886.git.lstoakes@gmail.com>
 <6b3899bbbf1f4bd6b7133c8b6f27b3a8791607b0.1679431886.git.lstoakes@gmail.com>
 <8ba0360e-57eb-93b0-3ae6-612f6b371bff@redhat.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <8ba0360e-57eb-93b0-3ae6-612f6b371bff@redhat.com>
Precedence: bulk

On Wed, Mar 22, 2023 at 12:18:08PM +0100, David Hildenbrand wrote:
> On 21.03.23 21:54, Lorenzo Stoakes wrote:
> > Having previously laid the foundation for converting vread() to an iterator
> > function, pull the trigger and do so.
> >
> > This patch attempts to provide minimal refactoring and to reflect the
> > existing logic as best we can, for example we continue to zero portions of
> > memory not read, as before.
> >
> > Overall, there should be no functional difference other than a performance
> > improvement in /proc/kcore access to vmalloc regions.
> >
> > Now we have eliminated the need for a bounce buffer in read_kcore_iter(),
> > we dispense with it. We need to ensure userland pages are faulted in before
> > proceeding, as we take spin locks.
> >
> > Additionally, we must account for the fact that at any point a copy may
> > fail if this happens, we exit indicating fewer bytes retrieved than
> > expected.
> >
> > Signed-off-by: Lorenzo Stoakes <lstoakes@gmail.com>
> > ---
> >   fs/proc/kcore.c         |  26 ++---
> >   include/linux/vmalloc.h |   3 +-
> >   mm/nommu.c              |  10 +-
> >   mm/vmalloc.c            | 234 +++++++++++++++++++++++++---------------
> >   4 files changed, 160 insertions(+), 113 deletions(-)
> >
> > diff --git a/fs/proc/kcore.c b/fs/proc/kcore.c
> > index 25e0eeb8d498..221e16f75ba5 100644
> > --- a/fs/proc/kcore.c
> > +++ b/fs/proc/kcore.c
> > @@ -307,13 +307,9 @@ static void append_kcore_note(char *notes, size_t *i, const char *name,
> >   	*i = ALIGN(*i + descsz, 4);
> >   }
> > -static ssize_t
> > -read_kcore_iter(struct kiocb *iocb, struct iov_iter *iter)
> > +static ssize_t read_kcore_iter(struct kiocb *iocb, struct iov_iter *iter)
> >   {
> > -	struct file *file = iocb->ki_filp;
> > -	char *buf = file->private_data;
> >   	loff_t *ppos = &iocb->ki_pos;
> > -
> >   	size_t phdrs_offset, notes_offset, data_offset;
> >   	size_t page_offline_frozen = 1;
> >   	size_t phdrs_len, notes_len;
> > @@ -507,9 +503,12 @@ read_kcore_iter(struct kiocb *iocb, struct iov_iter *iter)
> >   		switch (m->type) {
> >   		case KCORE_VMALLOC:
> > -			vread(buf, (char *)start, tsz);
> > -			/* we have to zero-fill user buffer even if no read */
> > -			if (copy_to_iter(buf, tsz, iter) != tsz) {
> > +			/*
> > +			 * Make sure user pages are faulted in as we acquire
> > +			 * spinlocks in vread_iter().
> > +			 */
> > +			if (fault_in_iov_iter_writeable(iter, tsz) ||
> > +			    vread_iter(iter, (char *)start, tsz) != tsz) {
> >   				ret = -EFAULT;
> >   				goto out;
> >   			}
>
> What if we race with swapout after faulting the pages in? Or some other
> mechanism to write-protect the user space pages?
>
> Also, "This is primarily useful when we already know that some or all of the
> pages in @i aren't in memory". This order of events might slow down things
> quite a bit if I am not wrong.
>
>
> Wouldn't you want to have something like:
>
> while (vread_iter(iter, (char *)start, tsz) != tsz) {
> 	if (fault_in_iov_iter_writeable(iter, tsz)) {
> 		ret = -EFAULT;
> 		goto out;
> 	}
> }
>
> Or am I missing something?
>

Indeed, I was thinking of this as:-

- prefault
- try (possibly fail if race) copy operation

However it does make more sense, and makes it explicit that it's an attempt that
might fail requiring a fault-in in the while form.

I think the upcoming change to explicitly make the iter function
copy_folio_to_iter_nofault() makes it clear from end-to-end what is being done -
we mustn't fault (spinlocks held) and if a fault would occur we error out, if we
error out try faulting in, if this fails then abort.

I will fixup in respin.

> --
> Thanks,
>
> David / dhildenb
>