From: Andrew Cooper <andrew.cooper3@citrix.com>
To: Jan Beulich <jbeulich@suse.com>,
"xen-devel@lists.xenproject.org" <xen-devel@lists.xenproject.org>
Cc: "Wei Liu" <wl@xen.org>, "Roger Pau Monné" <roger.pau@citrix.com>
Subject: Re: [PATCH] x86: refine guest_mode()
Date: Mon, 27 Apr 2020 15:35:02 +0100 [thread overview]
Message-ID: <1704f4f6-7e77-971c-2c94-4f6a6719c34a@citrix.com> (raw)
In-Reply-To: <7b62d06c-1369-2857-81c0-45e2434357f4@suse.com>
On 27/04/2020 09:03, Jan Beulich wrote:
> The 2nd of the assertions as well as the macro's return value have been
> assuming we're on the primary stack. While for most IST exceptions we
> eventually switch back to the main one,
"we switch to the main one when interrupting user mode".
"eventually" isn't accurate as it is before we enter C.
> for #DF we intentionally never
> do, and hence a #DF actually triggering on a user mode insn (which then
> is still a Xen bug) would in turn trigger this assertion, rather than
> cleanly logging state.
>
> Reported-by: Andrew Cooper <andrew.cooper3@citrix.com>
> Signed-off-by: Jan Beulich <jbeulich@suse.com>
> ---
> While we could go further and also assert we're on the correct IST
> stack in an "else" ti the "if()" added, I'm not fully convinced this
> would be generally helpful. I'll be happy to adjust accordingly if
> others think differently; at such a point though I think this should
> then no longer be a macro.
>
> --- a/xen/include/asm-x86/regs.h
> +++ b/xen/include/asm-x86/regs.h
> @@ -10,9 +10,10 @@
> /* Frame pointer must point into current CPU stack. */ \
> ASSERT(diff < STACK_SIZE); \
> /* If not a guest frame, it must be a hypervisor frame. */ \
> - ASSERT((diff == 0) || (r->cs == __HYPERVISOR_CS)); \
> + if ( diff < PRIMARY_STACK_SIZE ) \
> + ASSERT(!diff || ((r)->cs == __HYPERVISOR_CS)); \
> /* Return TRUE if it's a guest frame. */ \
> - (diff == 0); \
> + !diff || ((r)->cs != __HYPERVISOR_CS); \
The (diff == 0) already worried me before because it doesn't fail safe,
but this makes things more problematic. Consider the case back when we
had __HYPERVISOR_CS32.
Guest mode is strictly "(r)->cs & 3".
Everything else is expectations about how things ought to be laid out,
but for safety in release builds, the final judgement should not depend
on the expectations evaluating true.
~Andrew
next prev parent reply other threads:[~2020-04-27 14:35 UTC|newest]
Thread overview: 17+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-04-27 8:03 [PATCH] x86: refine guest_mode() Jan Beulich
2020-04-27 9:59 ` Roger Pau Monné
2020-04-27 14:08 ` Jan Beulich
2020-04-27 16:00 ` Roger Pau Monné
2020-04-27 14:35 ` Andrew Cooper [this message]
2020-04-27 15:15 ` Jan Beulich
2020-04-27 20:11 ` Andrew Cooper
2020-04-28 6:30 ` Jan Beulich
2020-05-18 14:51 ` Roger Pau Monné
2020-05-20 8:56 ` Jan Beulich
2020-05-20 15:13 ` Roger Pau Monné
2020-05-22 9:52 ` Jan Beulich
2020-05-22 10:48 ` Roger Pau Monné
2020-05-22 12:00 ` Jan Beulich
2020-05-26 10:56 ` Roger Pau Monné
2020-05-26 13:55 ` Jan Beulich
2020-05-27 15:17 ` Roger Pau Monné
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=1704f4f6-7e77-971c-2c94-4f6a6719c34a@citrix.com \
--to=andrew.cooper3@citrix.com \
--cc=jbeulich@suse.com \
--cc=roger.pau@citrix.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).