From: Jan Beulich <jbeulich@suse.com>
To: Roger Pau Monne <roger.pau@citrix.com>
Cc: xen-devel@lists.xenproject.org,
Stanislav Spassov <stanspas@amazon.de>,
Chao Gao <chao.gao@intel.com>
Subject: Re: [Xen-devel] [PATCH] pci: clear host_maskall field on assign
Date: Wed, 2 Oct 2019 15:33:43 +0200 [thread overview]
Message-ID: <c5dc7c62-ea14-7930-8f2d-62e20984da60@suse.com> (raw)
In-Reply-To: <20191002104935.60245-1-roger.pau@citrix.com>
On 02.10.2019 12:49, Roger Pau Monne wrote:
> The current implementation of host_maskall makes it sticky across
> assign and deassign calls, which means that once a guest forces Xen to
> set host_maskall the maskall bit is not going to be cleared until a
> call to PHYSDEVOP_prepare_msix is performed. Such call however
> shouldn't be part of the normal flow when doing PCI passthrough, and
> hence the flag needs to be cleared when assigning in order to prevent
> host_maskall being carried over from previous assignations.
>
> Note that other mask fields, like guest_masked or the entry maskbit
> are already reset when the msix capability is initialized.
I take it you mean a guest action here, as PHYSDEVOP_prepare_msix is
specifically about setting up the actual hardware's one? This happens
quite a bit later though, i.e. ->guest_maskall may need explicitly
setting at the same time as you clear ->host_maskall. Furthermore ...
> --- a/xen/drivers/passthrough/pci.c
> +++ b/xen/drivers/passthrough/pci.c
> @@ -1504,7 +1504,10 @@ static int assign_device(struct domain *d, u16 seg, u8 bus, u8 devfn, u32 flag)
> }
>
> if ( pdev->msix )
> + {
> msixtbl_init(d);
> + pdev->msix->host_maskall = false;
> + }
... doing just this would violate an assumed property: It ought to
be fine to assert at every entry or exit point that the physical
maskall bit of an MSI-X-enabled device is the logical OR of
->host_maskall and ->guest_maskall. I.e. I see the following
options:
1) your variant accompanied by updating of the hardware bit,
2)
pdev->msix->guest_maskall = pdev->msix->host_maskall;
pdev->msix->host_maskall = false;
leaving the hardware bit alone, as the above transformation
wouldn't change what it's supposed to be set to,
3)
pdev->msix->guest_maskall = true;
pdev->msix->host_maskall = false;
alongside setting the bit in hardware (if not already set),
4)
pdev->msix->guest_maskall = false;
pdev->msix->host_maskall = false;
alongside clearing the bit in hardware (if not already clear),
relying on all entries being individually masked (which ought
to be the state after the initial msix_capability_init()).
In all cases the operation would likely better be done by
calling a function to be put in x86/msi.c.
Jan
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel
next prev parent reply other threads:[~2019-10-02 13:33 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-10-02 10:49 [Xen-devel] [PATCH] pci: clear host_maskall field on assign Roger Pau Monne
2019-10-02 13:33 ` Jan Beulich [this message]
2019-10-08 9:23 ` Roger Pau Monné
2019-10-08 9:42 ` Jan Beulich
2019-10-08 11:09 ` Roger Pau Monné
2019-10-08 11:28 ` Jan Beulich
2019-10-08 13:14 ` Roger Pau Monné
2019-10-08 13:32 ` Jan Beulich
2019-10-08 14:16 ` Roger Pau Monné
2019-10-08 14:21 ` Jan Beulich
2019-10-04 23:58 ` Chao Gao
2019-10-07 7:38 ` Jan Beulich
2019-10-08 6:18 ` Chao Gao
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=c5dc7c62-ea14-7930-8f2d-62e20984da60@suse.com \
--to=jbeulich@suse.com \
--cc=chao.gao@intel.com \
--cc=roger.pau@citrix.com \
--cc=stanspas@amazon.de \
--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).