From: "Sahita, Ravi" <ravi.sahita@intel.com>
To: Jan Beulich <JBeulich@suse.com>
Cc: Tim Deegan <tim@xen.org>, "Sahita, Ravi" <ravi.sahita@intel.com>,
Wei Liu <wei.liu2@citrix.com>,
George Dunlap <george.dunlap@eu.citrix.com>,
Andrew Cooper <andrew.cooper3@citrix.com>,
Ian Jackson <ian.jackson@eu.citrix.com>,
"White, Edmund H" <edmund.h.white@intel.com>,
"xen-devel@lists.xen.org" <xen-devel@lists.xen.org>,
"tlengyel@novetta.com" <tlengyel@novetta.com>,
Daniel De Graaf <dgdegra@tycho.nsa.gov>
Subject: Re: [PATCH v5 06/15] VMX/altp2m: add code to support EPTP switching and #VE.
Date: Tue, 21 Jul 2015 05:49:46 +0000 [thread overview]
Message-ID: <DBC12B0F5509554280826E40BCDEE8BE54FDAD14@ORSMSX104.amr.corp.intel.com> (raw)
In-Reply-To: <55ACAF930200007800092D74@mail.emea.novell.com>
>From: Jan Beulich [mailto:JBeulich@suse.com]
>Sent: Sunday, July 19, 2015 11:22 PM
>
>>>> On 17.07.15 at 23:08, <ravi.sahita@intel.com> wrote:
>>> From: Jan Beulich [mailto:JBeulich@suse.com]
>>>Sent: Thursday, July 16, 2015 2:39 AM
>>>
>>>>>> On 16.07.15 at 11:20, <ravi.sahita@intel.com> wrote:
>>>>> From: Jan Beulich [mailto:JBeulich@suse.com]
>>>>>Sent: Tuesday, July 14, 2015 6:57 AM
>>>>>>>> On 14.07.15 at 02:14, <edmund.h.white@intel.com> wrote:
>>>>>> +static bool_t vmx_vcpu_emulate_ve(struct vcpu *v) {
>>>>>> + bool_t rc = 0;
>>>>>> + ve_info_t *veinfo = gfn_x(vcpu_altp2m(v).veinfo_gfn) !=
>>>>>> +INVALID_GFN
>>>>>?
>>>>>> + hvm_map_guest_frame_rw(gfn_x(vcpu_altp2m(v).veinfo_gfn),
>0) :
>>>>>> +NULL;
>>>>>> +
>>>>>> + if ( !veinfo )
>>>>>> + return 0;
>>>>>> +
>>>>>> + if ( veinfo->semaphore != 0 )
>>>>>> + goto out;
>>>>>> +
>>>>>> + rc = 1;
>>>>>> +
>>>>>> + veinfo->exit_reason = EXIT_REASON_EPT_VIOLATION;
>>>>>> + veinfo->semaphore = ~0l;
>>>>>
>>>>>Isn't semaphore a 32-bit quantity?
>>>>
>>>> Yes.
>>>
>>>I.e. the l suffix can and should be dropped.
>>>
>>
>> Ok.
>>
>>>>>> + {
>>>>>> + unsigned long idx;
>>>>>> +
>>>>>> + if ( v->arch.hvm_vmx.secondary_exec_control &
>>>>>> + SECONDARY_EXEC_ENABLE_VIRT_EXCEPTIONS )
>>>>>> + __vmread(EPTP_INDEX, &idx);
>>>>>> + else
>>>>>> + {
>>>>>> + unsigned long eptp;
>>>>>> +
>>>>>> + __vmread(EPT_POINTER, &eptp);
>>>>>> +
>>>>>> + if ( (idx = p2m_find_altp2m_by_eptp(v->domain, eptp)) ==
>>>>>> + INVALID_ALTP2M )
>>>>>> + {
>>>>>> + gdprintk(XENLOG_ERR, "EPTP not found in alternate
>>>>>> + p2m
>>>list\n");
>>>>>> + domain_crash(v->domain);
>>>>>> + }
>>>>>> + }
>>>>>> +
>>>>>> + if ( (uint16_t)idx != vcpu_altp2m(v).p2midx )
>>>>>
>>>>>Is this cast really necessary?
>>>>
>>>> Yes - The index is 16-bits, this reflects how the field is specified
>>>> in the vmcs also.
>>>
>>>While "yes" answers the question, the explanation you give suggests
>>>that the answer may be wrong: Can idx indeed have bits set beyond bit
>>>15? Because if it can't, the cast is pointless.
>>>
>>
>> We were just trying to ensure we matched the hardware behavior (I
>> think there was a message George had posted earlier for SVE that asked for
>that).
>> Since hardware considers only a 16 bit field we were doing the same.
>>
>>>>>> + {
>>>>>> + BUG_ON(idx >= MAX_ALTP2M);
>>>>>> + atomic_dec(&p2m_get_altp2m(v)->active_vcpus);
>>>>>> + vcpu_altp2m(v).p2midx = (uint16_t)idx;
>>>>>
>>>>>This one surely isn't (or else the field type is wrong).
>>>>
>>>> Again required. idx can't be uint16_t because __vmread() requires
>>>> unsigned long*, but the index is 16 bits.
>>>
>>>But it's a 16-bit VMCS field that you read it from, and hence the
>>>upper 48
>> bits
>>>are necessarily zero.
>>>
>>>Just to re-iterate: Casts are necessary in certain places, yes, but I
>>>see them used pointlessly or even wrongly more often than not.
>>
>> Same approach as above - emulating hardware exactly.
>> Should we add a comment?
>
>No, drop the casts.
>
>Jan
Ok
Ravi
next prev parent reply other threads:[~2015-07-21 5:49 UTC|newest]
Thread overview: 56+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-07-14 0:14 [PATCH v5 00/15] Alternate p2m: support multiple copies of host p2m Ed White
2015-07-14 0:14 ` [PATCH v5 01/15] common/domain: Helpers to pause a domain while in context Ed White
2015-07-14 0:14 ` [PATCH v5 02/15] VMX: VMFUNC and #VE definitions and detection Ed White
2015-07-14 0:14 ` [PATCH v5 03/15] VMX: implement suppress #VE Ed White
2015-07-14 12:46 ` Jan Beulich
2015-07-14 13:47 ` George Dunlap
2015-07-14 0:14 ` [PATCH v5 04/15] x86/HVM: Hardware alternate p2m support detection Ed White
2015-07-14 0:14 ` [PATCH v5 05/15] x86/altp2m: basic data structures and support routines Ed White
2015-07-14 13:13 ` Jan Beulich
2015-07-14 14:45 ` George Dunlap
2015-07-14 14:58 ` Jan Beulich
2015-07-16 8:57 ` Sahita, Ravi
2015-07-16 9:07 ` Jan Beulich
2015-07-17 22:36 ` Sahita, Ravi
2015-07-20 6:20 ` Jan Beulich
2015-07-21 5:18 ` Sahita, Ravi
2015-07-14 15:57 ` George Dunlap
2015-07-21 17:44 ` Sahita, Ravi
2015-07-14 0:14 ` [PATCH v5 06/15] VMX/altp2m: add code to support EPTP switching and #VE Ed White
2015-07-14 13:57 ` Jan Beulich
2015-07-16 9:20 ` Sahita, Ravi
2015-07-16 9:38 ` Jan Beulich
2015-07-17 21:08 ` Sahita, Ravi
2015-07-20 6:21 ` Jan Beulich
2015-07-21 5:49 ` Sahita, Ravi [this message]
2015-07-14 0:14 ` [PATCH v5 07/15] VMX: add VMFUNC leaf 0 (EPTP switching) to emulator Ed White
2015-07-14 14:04 ` Jan Beulich
2015-07-14 17:56 ` Sahita, Ravi
2015-07-17 22:41 ` Sahita, Ravi
2015-07-14 0:14 ` [PATCH v5 08/15] x86/altp2m: add control of suppress_ve Ed White
2015-07-14 17:03 ` George Dunlap
2015-07-14 0:14 ` [PATCH v5 09/15] x86/altp2m: alternate p2m memory events Ed White
2015-07-14 14:08 ` Jan Beulich
2015-07-16 9:22 ` Sahita, Ravi
2015-07-14 0:14 ` [PATCH v5 10/15] x86/altp2m: add remaining support routines Ed White
2015-07-14 14:31 ` Jan Beulich
2015-07-16 9:16 ` Sahita, Ravi
2015-07-16 9:34 ` Jan Beulich
2015-07-17 22:32 ` Sahita, Ravi
2015-07-20 6:53 ` Jan Beulich
2015-07-21 5:46 ` Sahita, Ravi
2015-07-21 6:38 ` Jan Beulich
2015-07-21 18:33 ` Sahita, Ravi
2015-07-22 7:33 ` Jan Beulich
2015-07-16 14:44 ` George Dunlap
2015-07-17 21:01 ` Sahita, Ravi
2015-07-14 0:14 ` [PATCH v5 11/15] x86/altp2m: define and implement alternate p2m HVMOP types Ed White
2015-07-14 14:36 ` Jan Beulich
2015-07-16 9:02 ` Sahita, Ravi
2015-07-16 9:09 ` Jan Beulich
2015-07-14 0:15 ` [PATCH v5 12/15] x86/altp2m: Add altp2mhvm HVM domain parameter Ed White
2015-07-14 0:15 ` [PATCH v5 13/15] x86/altp2m: XSM hooks for altp2m HVM ops Ed White
2015-07-14 0:15 ` [PATCH v5 14/15] tools/libxc: add support to altp2m hvmops Ed White
2015-07-14 0:15 ` [PATCH v5 15/15] tools/xen-access: altp2m testcases Ed White
2015-07-14 9:56 ` Wei Liu
2015-07-14 11:52 ` Lengyel, Tamas
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=DBC12B0F5509554280826E40BCDEE8BE54FDAD14@ORSMSX104.amr.corp.intel.com \
--to=ravi.sahita@intel.com \
--cc=JBeulich@suse.com \
--cc=andrew.cooper3@citrix.com \
--cc=dgdegra@tycho.nsa.gov \
--cc=edmund.h.white@intel.com \
--cc=george.dunlap@eu.citrix.com \
--cc=ian.jackson@eu.citrix.com \
--cc=tim@xen.org \
--cc=tlengyel@novetta.com \
--cc=wei.liu2@citrix.com \
--cc=xen-devel@lists.xen.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.