qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Markus Armbruster <armbru@redhat.com>
To: Akihiko Odaki <akihiko.odaki@daynix.com>
Cc: qemu-devel@nongnu.org,  qemu-block@nongnu.org,
	 qemu-arm@nongnu.org,
	Alex Williamson <alex.williamson@redhat.com>,
	 "Michael S. Tsirkin" <mst@redhat.com>,
	 Marcel Apfelbaum <marcel.apfelbaum@gmail.com>,
	 Gerd Hoffmann <kraxel@redhat.com>,
	 Paolo Bonzini <pbonzini@redhat.com>,
	Richard Henderson <richard.henderson@linaro.org>,
	 Eduardo Habkost <eduardo@habkost.net>,
	 John Snow <jsnow@redhat.com>,
	 Dmitry Fleytman <dmitry.fleytman@gmail.com>,
	 Jason Wang <jasowang@redhat.com>,  Stefan Weil <sw@weilnetz.de>,
	 Keith Busch <kbusch@kernel.org>,
	 Klaus Jensen <its@irrelevant.dk>,
	 Peter Maydell <peter.maydell@linaro.org>,
	 Andrey Smirnov <andrew.smirnov@gmail.com>,
	 Paul Burton <paulburton@kernel.org>,
	Aleksandar Rikalo <aleksandar.rikalo@syrmia.com>
Subject: Re: [PATCH v2] pci: Assert that capabilities never overlap
Date: Mon, 05 Sep 2022 11:26:34 +0200	[thread overview]
Message-ID: <875yi2mat1.fsf@pond.sub.org> (raw)
In-Reply-To: <CAE=JJXdg=Miisek8WeqQ12NqL8obzmuyzD0mbv1SfiJTyVBLuw@mail.gmail.com> (Akihiko Odaki's message of "Sun, 4 Sep 2022 16:06:55 +0900")

Akihiko Odaki <akihiko.odaki@daynix.com> writes:

> On Fri, Sep 2, 2022 at 7:23 PM Markus Armbruster <armbru@redhat.com> wrote:
>>
>> Akihiko Odaki <akihiko.odaki@daynix.com> writes:
>>
>> > pci_add_capability appears most PCI devices. Its error handling required
>> > lots of code, and led to inconsistent behaviors such as:
>> > - passing error_abort
>> > - passing error_fatal
>> > - asserting the returned value
>> > - propagating the error to the caller
>> > - skipping the rest of the function
>> > - just ignoring
>> >
>> > The code generating errors in pci_add_capability had a comment which
>> > says:
>> >> Verify that capabilities don't overlap.  Note: device assignment
>> >> depends on this check to verify that the device is not broken.
>> >> Should never trigger for emulated devices, but it's helpful for
>> >> debugging these.
>> >
>> > Indeed vfio has some code that passes capability offsets and sizes from
>> > a physical device, but it explicitly pays attention so that the
>> > capabilities never overlap.
>>
>> I can't see that at a glance.  Can you give me a clue?
>>
>> >                             Therefore, we can always assert that
>> > capabilities never overlap when pci_add_capability is called, resolving
>> > these inconsistencies.
>> >
>> > Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
>>
>
> Looking at vfio_add_std_cap(), and vfio_add_ext_cap() it seems that
> they are clipping the size of capabilities so that they do not
> overlap, if I read it correctly.

If we want to deal gracefully with buggy physical devices, we need to
treat pdev->config[] as untrusted input.

As far as I can tell:

* vfio_add_capabilities() replicates the physical device's capabilities
  (starting at pdev->config[PCI_CAPABILITY_LIST]) in the virtual device.

* vfio_add_std_cap() is a helper to add the tail starting at
  pdev->config[pos].

Could the physical device's capabilities overlap?  If yes, what would
happen before and after your series?



  reply	other threads:[~2022-09-05  9:29 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-08-31  1:32 [PATCH v2] pci: Assert that capabilities never overlap Akihiko Odaki
2022-09-02 10:23 ` Markus Armbruster
2022-09-04  7:06   ` Akihiko Odaki
2022-09-05  9:26     ` Markus Armbruster [this message]
2022-09-05 10:11       ` Akihiko Odaki
2022-09-29  9:25         ` Akihiko Odaki
2022-09-29 10:55         ` Markus Armbruster
2022-10-13  5:39           ` Akihiko Odaki
2022-10-13 14:45             ` Michael S. Tsirkin

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=875yi2mat1.fsf@pond.sub.org \
    --to=armbru@redhat.com \
    --cc=akihiko.odaki@daynix.com \
    --cc=aleksandar.rikalo@syrmia.com \
    --cc=alex.williamson@redhat.com \
    --cc=andrew.smirnov@gmail.com \
    --cc=dmitry.fleytman@gmail.com \
    --cc=eduardo@habkost.net \
    --cc=its@irrelevant.dk \
    --cc=jasowang@redhat.com \
    --cc=jsnow@redhat.com \
    --cc=kbusch@kernel.org \
    --cc=kraxel@redhat.com \
    --cc=marcel.apfelbaum@gmail.com \
    --cc=mst@redhat.com \
    --cc=paulburton@kernel.org \
    --cc=pbonzini@redhat.com \
    --cc=peter.maydell@linaro.org \
    --cc=qemu-arm@nongnu.org \
    --cc=qemu-block@nongnu.org \
    --cc=qemu-devel@nongnu.org \
    --cc=richard.henderson@linaro.org \
    --cc=sw@weilnetz.de \
    /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).