From: "Michał Leszczyński" <michal.leszczynski@cert.pl>
To: Julien Grall <julien@xen.org>
Cc: "Stefano Stabellini" <sstabellini@kernel.org>,
"tamas lengyel" <tamas.lengyel@intel.com>, "Wei Liu" <wl@xen.org>,
"Andrew Cooper" <andrew.cooper3@citrix.com>,
"Ian Jackson" <ian.jackson@eu.citrix.com>,
"George Dunlap" <george.dunlap@citrix.com>,
"luwei kang" <luwei.kang@intel.com>,
"Jan Beulich" <jbeulich@suse.com>,
"Anthony PERARD" <anthony.perard@citrix.com>,
xen-devel@lists.xenproject.org,
"Roger Pau Monné" <roger.pau@citrix.com>
Subject: Re: [PATCH v4 03/10] tools/libxl: add vmtrace_pt_size parameter
Date: Tue, 7 Jul 2020 13:17:19 +0200 (CEST) [thread overview]
Message-ID: <1580655090.20712847.1594120639229.JavaMail.zimbra@cert.pl> (raw)
In-Reply-To: <ab992813-4584-f8e0-b90a-7a587c396bae@xen.org>
----- 7 lip 2020 o 11:16, Julien Grall julien@xen.org napisał(a):
> On 07/07/2020 10:10, Jan Beulich wrote:
>> On 07.07.2020 10:44, Julien Grall wrote:
>>> Hi,
>>>
>>> On 06/07/2020 09:46, Jan Beulich wrote:
>>>> On 04.07.2020 19:23, Julien Grall wrote:
>>>>> Hi,
>>>>>
>>>>> On 03/07/2020 11:11, Roger Pau Monné wrote:
>>>>>> On Fri, Jul 03, 2020 at 11:56:38AM +0200, Jan Beulich wrote:
>>>>>>> On 03.07.2020 11:44, Roger Pau Monné wrote:
>>>>>>>> On Thu, Jul 02, 2020 at 06:23:28PM +0200, Michał Leszczyński wrote:
>>>>>>>>> In previous versions it was "size" but it was requested to change it
>>>>>>>>> to "order" in order to shrink the variable size from uint64_t to
>>>>>>>>> uint8_t, because there is limited space for xen_domctl_createdomain
>>>>>>>>> structure.
>>>>>>>>
>>>>>>>> It's likely I'm missing something here, but I wasn't aware
>>>>>>>> xen_domctl_createdomain had any constrains regarding it's size. It's
>>>>>>>> currently 48bytes which seems fairly small.
>>>>>>>
>>>>>>> Additionally I would guess a uint32_t could do here, if the value
>>>>>>> passed was "number of pages" rather than "number of bytes"?
>>>>> Looking at the rest of the code, the toolstack accepts a 64-bit value.
>>>>> So this would lead to truncation of the buffer if it is bigger than 2^44
>>>>> bytes.
>>>>>
>>>>> I agree such buffer is unlikely, yet I still think we want to harden the
>>>>> code whenever we can. So the solution is to either prevent check
>>>>> truncation in libxl or directly use 64-bit in the domctl.
>>>>>
>>>>> My preference is the latter.
>>>>>
>>>>>>
>>>>>> That could work, not sure if it needs to state however that those will
>>>>>> be 4K pages, since Arm can have a different minimum page size IIRC?
>>>>>> (or that's already the assumption for all number of frames fields)
>>>>>> vmtrace_nr_frames seems fine to me.
>>>>>
>>>>> The hypercalls interface is using the same page granularity as the
>>>>> hypervisor (i.e 4KB).
>>>>>
>>>>> While we already support guest using 64KB page granularity, it is
>>>>> impossible to have a 64KB Arm hypervisor in the current state. You are
>>>>> going to either break existing guest (if you switch to 64KB page
>>>>> granularity for the hypercall ABI) or render them insecure (the mimimum
>>>>> mapping in the P2M would be 64KB).
>>>>>
>>>>> DOMCTLs are not stable yet, so using a number of pages is OK. However, I
>>>>> would strongly suggest to use a number of bytes for any xl/libxl/stable
>>>>> libraries interfaces as this avoids confusion and also make more
>>>>> futureproof.
>>>>
>>>> If we can't settle on what "page size" means in the public interface
>>>> (which imo is embarrassing), then how about going with number of kb,
>>>> like other memory libxl controls do? (I guess using Mb, in line with
>>>> other config file controls, may end up being too coarse here.) This
>>>> would likely still allow for a 32-bit field to be wide enough.
>>>
>>> A 32-bit field would definitely not be able to cover a full address
>>> space. So do you mind to explain what is the upper bound you expect here?
>>
>> Do you foresee a need for buffer sizes of 4Tb and up?
>
> Not I am aware of... However, I think the question was worth it given
> that "wide enough" can mean anything.
>
> Cheers,
>
> --
> Julien Grall
So would it be OK to use uint32_t everywhere and to store the trace buffer
size as number of kB? I think this is the most straightforward option.
I would also stick with the name "processor_trace_buf_size"
everywhere, both in the hypervisor, ABI and the toolstack, with the
respective comments that the size is in kB.
Best regards,
Michał Leszczyński
CERT Polska
next prev parent reply other threads:[~2020-07-07 11:18 UTC|newest]
Thread overview: 75+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-06-30 12:33 [PATCH v4 00/10] Implement support for external IPT monitoring Michał Leszczyński
2020-06-30 12:33 ` [PATCH v4 01/10] x86/vmx: add Intel PT MSR definitions Michał Leszczyński
2020-06-30 16:23 ` Jan Beulich
2020-06-30 17:37 ` Andrew Cooper
2020-06-30 18:03 ` Tamas K Lengyel
2020-06-30 18:27 ` Michał Leszczyński
2020-07-01 17:52 ` Andrew Cooper
2020-06-30 12:33 ` [PATCH v4 02/10] x86/vmx: add IPT cpu feature Michał Leszczyński
2020-07-01 9:49 ` Roger Pau Monné
2020-07-01 15:12 ` Julien Grall
2020-07-01 16:06 ` Andrew Cooper
2020-07-01 16:17 ` Julien Grall
2020-07-01 16:18 ` Julien Grall
2020-07-01 17:26 ` Andrew Cooper
2020-07-01 18:02 ` Julien Grall
2020-07-01 18:06 ` Andrew Cooper
2020-07-01 18:09 ` Julien Grall
2020-07-02 8:29 ` Jan Beulich
2020-07-02 8:42 ` Julien Grall
2020-07-02 8:50 ` Jan Beulich
2020-07-02 8:54 ` Julien Grall
2020-07-02 9:18 ` Jan Beulich
2020-07-02 9:57 ` Julien Grall
2020-07-02 13:30 ` Jan Beulich
2020-07-02 14:14 ` Julien Grall
2020-07-02 14:17 ` Jan Beulich
2020-07-02 14:31 ` Julien Grall
2020-07-02 20:28 ` Michał Leszczyński
2020-07-03 7:58 ` Julien Grall
2020-07-04 19:16 ` Michał Leszczyński
2020-07-01 21:42 ` Andrew Cooper
2020-07-02 8:10 ` Roger Pau Monné
2020-07-02 8:34 ` Jan Beulich
2020-07-02 20:29 ` Michał Leszczyński
2020-06-30 12:33 ` [PATCH v4 03/10] tools/libxl: add vmtrace_pt_size parameter Michał Leszczyński
2020-07-01 10:05 ` Roger Pau Monné
2020-07-02 9:00 ` Roger Pau Monné
2020-07-02 16:23 ` Michał Leszczyński
2020-07-03 9:44 ` Roger Pau Monné
2020-07-03 9:56 ` Jan Beulich
2020-07-03 10:11 ` Roger Pau Monné
2020-07-04 17:23 ` Julien Grall
2020-07-06 8:46 ` Jan Beulich
2020-07-07 8:44 ` Julien Grall
2020-07-07 9:10 ` Jan Beulich
2020-07-07 9:16 ` Julien Grall
2020-07-07 11:17 ` Michał Leszczyński [this message]
2020-07-07 11:21 ` Jan Beulich
2020-07-07 11:35 ` Michał Leszczyński
2020-07-02 10:24 ` Anthony PERARD
2020-07-04 17:48 ` Julien Grall
2020-06-30 12:33 ` [PATCH v4 04/10] x86/vmx: implement processor tracing for VMX Michał Leszczyński
2020-07-01 10:30 ` Roger Pau Monné
2020-06-30 12:33 ` [PATCH v4 05/10] common/domain: allocate vmtrace_pt_buffer Michał Leszczyński
2020-07-01 10:38 ` Roger Pau Monné
2020-07-01 15:35 ` Julien Grall
2020-06-30 12:33 ` [PATCH v4 06/10] memory: batch processing in acquire_resource() Michał Leszczyński
2020-07-01 10:46 ` Roger Pau Monné
2020-07-03 10:35 ` Julien Grall
2020-07-03 10:52 ` Paul Durrant
2020-07-03 11:17 ` Julien Grall
2020-07-03 11:22 ` Jan Beulich
2020-07-03 11:36 ` Julien Grall
2020-07-03 12:50 ` Jan Beulich
2020-07-03 11:40 ` Paul Durrant
2020-06-30 12:33 ` [PATCH v4 07/10] x86/mm: add vmtrace_buf resource type Michał Leszczyński
2020-07-01 10:52 ` Roger Pau Monné
2020-06-30 12:33 ` [PATCH v4 08/10] x86/domctl: add XEN_DOMCTL_vmtrace_op Michał Leszczyński
2020-07-01 11:00 ` Roger Pau Monné
2020-06-30 12:33 ` [PATCH v4 09/10] tools/libxc: add xc_vmtrace_* functions Michał Leszczyński
2020-07-21 10:52 ` Wei Liu
2020-06-30 12:33 ` [PATCH v4 10/10] tools/proctrace: add proctrace tool Michał Leszczyński
2020-07-02 15:10 ` Andrew Cooper
2020-07-21 10:52 ` Wei Liu
2020-06-30 12:48 ` [PATCH v4 00/10] Implement support for external IPT monitoring Hubert Jasudowicz
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=1580655090.20712847.1594120639229.JavaMail.zimbra@cert.pl \
--to=michal.leszczynski@cert.pl \
--cc=andrew.cooper3@citrix.com \
--cc=anthony.perard@citrix.com \
--cc=george.dunlap@citrix.com \
--cc=ian.jackson@eu.citrix.com \
--cc=jbeulich@suse.com \
--cc=julien@xen.org \
--cc=luwei.kang@intel.com \
--cc=roger.pau@citrix.com \
--cc=sstabellini@kernel.org \
--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).