From: Tamas K Lengyel <tamas.k.lengyel@gmail.com>
To: "Roger Pau Monné" <roger.pau@citrix.com>
Cc: Kevin Tian <kevin.tian@intel.com>,
Tamas K Lengyel <tamas.lengyel@intel.com>,
Jun Nakajima <jun.nakajima@intel.com>, Wei Liu <wl@xen.org>,
Paul Durrant <paul@xen.org>,
Andrew Cooper <andrew.cooper3@citrix.com>,
Jan Beulich <jbeulich@suse.com>,
Xen-devel <xen-devel@lists.xenproject.org>
Subject: Re: [PATCH v2 for-4.14] x86/vmx: use P2M_ALLOC in vmx_load_pdptrs instead of P2M_UNSHARE
Date: Thu, 18 Jun 2020 06:21:42 -0600 [thread overview]
Message-ID: <CABfawhmsBijg=EA2=J3DOTZHByrcOYk5NOQGUrwv24o8nEptTw@mail.gmail.com> (raw)
In-Reply-To: <20200618094014.GH735@Air-de-Roger>
On Thu, Jun 18, 2020 at 3:42 AM Roger Pau Monné <roger.pau@citrix.com> wrote:
>
> On Thu, Jun 18, 2020 at 08:30:08AM +0200, Jan Beulich wrote:
> > On 17.06.2020 18:19, Tamas K Lengyel wrote:
> > > While forking VMs running a small RTOS system (Zephyr) a Xen crash has been
> > > observed due to a mm-lock order violation while copying the HVM CPU context
> > > from the parent. This issue has been identified to be due to
> > > hap_update_paging_modes first getting a lock on the gfn using get_gfn. This
> > > call also creates a shared entry in the fork's memory map for the cr3 gfn. The
> > > function later calls hap_update_cr3 while holding the paging_lock, which
> > > results in the lock-order violation in vmx_load_pdptrs when it tries to unshare
> > > the above entry when it grabs the page with the P2M_UNSHARE flag set.
> > >
> > > Since vmx_load_pdptrs only reads from the page its usage of P2M_UNSHARE was
> > > unnecessary to start with. Using P2M_ALLOC is the appropriate flag to ensure
> > > the p2m is properly populated and to avoid the lock-order violation we
> > > observed.
> >
> > Using P2M_ALLOC is not going to address the original problem though
> > afaict: You may hit the mem_sharing_fork_page() path that way, and
> > via nominate_page() => __grab_shared_page() => mem_sharing_page_lock()
> > you'd run into a lock order violation again.
>
> Well, I guess Tamas avoids this because of the get_gfn call in
> hap_update_paging_modes will have already populated the entry, so it's
> never going to hit the p2m_is_hole check in __get_gfn_type_access.
>
> > The change is an improvement, so I'd be fine with it going in this
> > way, but only as long as the description mentions that there's still
> > an open issue here (which may be non-trivial to address). Or perhaps
> > combining with your v1 change is the way to go (for now or even
> > permanently)?
>
> If vmx_load_pdptrs only requires P2M_ALLOC then this is already
> covered by the call to get_gfn performed in hap_update_paging_modes,
> so I don't think there's much point in merging with v1, as forcing
> hap_update_paging_modes to unshare the entry won't affect
> vmx_load_pdptrs anymore.
>
> I'm however worried about other code paths that can call into
> vmx_load_pdptrs with mm locks taken, and I agree it would be safer to
> assert that all the higher layers make sure the cr3 loaded is
> correctly populated for a query without P2M_ALLOC to succeed.
Using P2M_ALLOC is always safe if 1) the entry is already populated
like in this case but also in 2) in case the gfn is a hole and gets
forked. In mem_sharing the paging lock order is only applicable when
an already present entry is getting converted to a shared type or a
shared typed is getting unshared. It does not apply when a hole is
being plugged. So this is safe in other paths as well where the entry
is not yet present.
Tamas
next prev parent reply other threads:[~2020-06-18 12:22 UTC|newest]
Thread overview: 12+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-06-17 16:19 [PATCH v2 for-4.14] x86/vmx: use P2M_ALLOC in vmx_load_pdptrs instead of P2M_UNSHARE Tamas K Lengyel
2020-06-18 6:30 ` Jan Beulich
2020-06-18 9:40 ` Roger Pau Monné
2020-06-18 11:32 ` Jan Beulich
2020-06-18 12:21 ` Tamas K Lengyel [this message]
2020-06-18 12:49 ` Roger Pau Monné
2020-06-18 12:39 ` Tamas K Lengyel
2020-06-18 12:46 ` Jan Beulich
2020-06-18 12:52 ` Roger Pau Monné
2020-06-18 13:00 ` Tamas K Lengyel
2020-06-18 13:26 ` Jan Beulich
2020-06-18 13:34 ` Tamas K Lengyel
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to='CABfawhmsBijg=EA2=J3DOTZHByrcOYk5NOQGUrwv24o8nEptTw@mail.gmail.com' \
--to=tamas.k.lengyel@gmail.com \
--cc=andrew.cooper3@citrix.com \
--cc=jbeulich@suse.com \
--cc=jun.nakajima@intel.com \
--cc=kevin.tian@intel.com \
--cc=paul@xen.org \
--cc=roger.pau@citrix.com \
--cc=tamas.lengyel@intel.com \
--cc=wl@xen.org \
--cc=xen-devel@lists.xenproject.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).