From: Alexander Graf <agraf@suse.de>
To: Milton Miller <miltonm@bga.com>
Cc: linuxppc-dev <linuxppc-dev@lists.ozlabs.org>,
Avi Kivity <avi@redhat.com>,
kvm-ppc@vger.kernel.org, KVM list <kvm@vger.kernel.org>
Subject: Re: [PATCH 26/26] KVM: PPC: Add Documentation about PV interface
Date: Mon, 28 Jun 2010 09:49:33 +0200 [thread overview]
Message-ID: <92F4A3F3-A89F-418D-BD4D-66E2489F2E42@suse.de> (raw)
In-Reply-To: <1277709531_13308@mail4.comsite.net>
On 28.06.2010, at 09:18, Milton Miller wrote:
> On Sun Jun 27 around 19:33:52 EST 2010 Alexander Graf wrote:
>> Am 27.06.2010 um 10:14 schrieb Avi Kivity <avi at redhat.com>:
>>> On 06/26/2010 02:25 AM, Alexander Graf wrote:
>=20
>>>> +
>>>> +PPC hypercalls
>>>> +=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
>>>> +
>>>> +The only viable ways to reliably get from guest context to host =20=
>>>> context are:
>>>> +
>>>> + 1) Call an invalid instruction
>>>> + 2) Call the "sc" instruction with a parameter to "sc"
>>>> + 3) Call the "sc" instruction with parameters in GPRs
>>>> +
>>>> +Method 1 is always a bad idea. Invalid instructions can be =20
>>>> replaced later on
>>>> +by valid instructions, rendering the interface broken.
>>>> +
>>>> +Method 2 also has downfalls. If the parameter to "sc" is !=3D 0 =
the =20
>>>> spec is
>>>> +rather unclear if the sc is targeted directly for the hypervisor =20=
>>>> or the
>>>> +supervisor. It would also require that we read the syscall issuing =
=20
>>>> instruction
>>>> +every time a syscall is issued, slowing down guest syscalls.
>>>> +
>=20
> It goes to the hypervisor, and it would require the hypervisor to
> return to the supervisor, but I believe it just returns to the user =
with
> permission denied.
That's what I assumed, yeah :(.
>=20
>>>> +Method 3 is what KVM uses. We pass magic constants =20
>>>> (KVM_SC_MAGIC_R3 and
>>>> +KVM_SC_MAGIC_R4) in r3 and r4 respectively. If a syscall =20
>>>> instruction with these
>>>> +magic values arrives from the guest's kernel mode, we take the =20
>>>> syscall as a
>>>> +hypercall.
>>>>=20
>>>=20
>>> Is there any chance a normal syscall will have those values in r3 =20=
>>> and r4?
>>=20
>> r3 is the syscall number. So as long as the guest doesn't reuse that =20=
>> value, we're safe. Since in general syscall numbers are not randomly =20=
>> scattered throughout the number range, we should be ok here.
>>=20
>=20
> No, r0 has the system call number. Registers 3 and 4 are the first
> 2 args in c abi (or first 64 bit arg in 32 bit c abi), but the linux
> syscall abi special. (In addition, it returns success or failure in
> cr0).
Oh. Ahem :)
>=20
>>>=20
>>> If so, maybe it's better to use pc as they key for hypercalls. Let =20=
>>> the guest designate one instruction address as the hypercall call =20=
>>> point; kvm can easily check it and reflect it back to the guest if =20=
>>> it doesn't match.
>>>=20
>>=20
>> You mean the guest would tell the hv where the hypercall lies? That =20=
>> would require a hypercall, no? Defining it statically is tricky. I =20=
>> want to PV'nize osx using a kernel module later, so I don't have =20
>> control over the physical layout.
>>=20
>>> Is it valid and useful to issue sc from privileged mode anyway, =20
>>> except for calling the hypervisor?
>>=20
>> Same as a syscall on x86 really. The kernel can and does issue =20
>> syscalls within itself.
>>=20
>>=20
>=20
> I don't believe we support the kernel actually doing a syscall to =
itself
> anymore, at least on powerpc. The callers call the underlying system
> call function, or kernel_thread.
>=20
> That said, I would suggest we allocate a syscall number for this, as =
it
> would document the usage. (In additon to 0..nr_syscalls - 1 we have
> 0x1ebe in use).
That's actually a pretty good idea.
>=20
> Also, is there any desire to nest such emulation?
Nesting should just work, right? Since we only accept hypercalls from =
PR=3D0 and guests run in PR=3D1, we get the sc interrupt in the l1 guest =
by then.
The only issue I'm aware of that completely breaks when using nested KVM =
on PPC is the MSR_IR !=3D MSR_DR logic. We fetch the instruction we got =
an interrupt on for certain interrupts in the world switch handler by =
keeping MSR_IR=3D0, but setting MSR_DR=3D1. And KVM speeds up MSR_DR !=3D =
MSR_IR by mapping both of them lazily in a special address space. So if =
you access the same page as instruction and as data, you get an invalid =
result.
Alex
next prev parent reply other threads:[~2010-06-28 7:49 UTC|newest]
Thread overview: 78+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-06-25 23:24 [PATCH 00/26] KVM PPC PV framework Alexander Graf
2010-06-25 23:24 ` [PATCH 01/26] KVM: PPC: Introduce shared page Alexander Graf
2010-06-27 12:12 ` Avi Kivity
2010-06-29 9:54 ` Alexander Graf
2010-06-29 10:55 ` Avi Kivity
2010-06-25 23:24 ` [PATCH 02/26] KVM: PPC: Convert MSR to " Alexander Graf
2010-06-27 8:16 ` Avi Kivity
2010-06-27 9:38 ` Alexander Graf
2010-06-27 9:50 ` Avi Kivity
2010-06-27 10:40 ` Alexander Graf
2010-06-25 23:24 ` [PATCH 03/26] KVM: PPC: Convert DSISR " Alexander Graf
2010-06-25 23:24 ` [PATCH 04/26] KVM: PPC: Convert DAR " Alexander Graf
2010-06-25 23:24 ` [PATCH 05/26] KVM: PPC: Convert SRR0 and SRR1 " Alexander Graf
2010-06-25 23:24 ` [PATCH 06/26] KVM: PPC: Convert SPRG[0-4] " Alexander Graf
2010-06-25 23:24 ` [PATCH 07/26] KVM: PPC: Implement hypervisor interface Alexander Graf
2010-06-25 23:24 ` [PATCH 08/26] KVM: PPC: Add PV guest critical sections Alexander Graf
2010-06-27 8:21 ` Avi Kivity
2010-06-27 9:40 ` Alexander Graf
2010-06-27 9:52 ` Avi Kivity
2010-06-27 10:33 ` Alexander Graf
2010-06-27 10:59 ` Avi Kivity
2010-06-27 11:49 ` Alexander Graf
2010-06-27 11:53 ` Avi Kivity
2010-06-27 12:06 ` Alexander Graf
2010-06-27 22:03 ` Benjamin Herrenschmidt
2010-06-27 10:03 ` Avi Kivity
2010-06-27 10:35 ` Alexander Graf
2010-06-25 23:24 ` [PATCH 09/26] KVM: PPC: Add PV guest scratch registers Alexander Graf
2010-06-27 8:22 ` Avi Kivity
2010-06-27 9:41 ` Alexander Graf
2010-06-27 9:53 ` Avi Kivity
2010-06-25 23:24 ` [PATCH 10/26] KVM: PPC: Tell guest about pending interrupts Alexander Graf
2010-06-25 23:24 ` [PATCH 11/26] KVM: PPC: Make RMO a define Alexander Graf
2010-06-26 16:52 ` Segher Boessenkool
2010-06-27 9:08 ` Alexander Graf
2010-06-29 7:32 ` Segher Boessenkool
2010-06-29 7:39 ` Alexander Graf
2010-06-29 7:52 ` Segher Boessenkool
2010-06-29 8:04 ` Alexander Graf
2010-06-25 23:25 ` [PATCH 12/26] KVM: PPC: First magic page steps Alexander Graf
2010-06-27 8:24 ` Avi Kivity
2010-06-27 9:44 ` Alexander Graf
2010-06-25 23:25 ` [PATCH 13/26] KVM: PPC: Magic Page Book3s support Alexander Graf
2010-06-25 23:25 ` [PATCH 14/26] KVM: PPC: Magic Page BookE support Alexander Graf
2010-06-25 23:25 ` [PATCH 15/26] KVM: PPC: Expose magic page support to guest Alexander Graf
2010-06-25 23:25 ` [PATCH 16/26] KVM: Move kvm_guest_init out of generic code Alexander Graf
2010-06-25 23:25 ` [PATCH 17/26] KVM: PPC: Generic KVM PV guest support Alexander Graf
2010-06-25 23:25 ` [PATCH 18/26] KVM: PPC: KVM PV guest stubs Alexander Graf
2010-06-27 8:28 ` Avi Kivity
2010-06-27 9:47 ` Alexander Graf
2010-06-27 10:16 ` Avi Kivity
2010-06-27 10:38 ` Alexander Graf
2010-06-27 22:04 ` Benjamin Herrenschmidt
2010-06-28 4:39 ` Matt Evans
2010-06-28 6:33 ` Alexander Graf
2010-06-28 8:15 ` Avi Kivity
2010-06-28 8:23 ` Alexander Graf
2010-06-28 8:33 ` Avi Kivity
2010-06-25 23:25 ` [PATCH 19/26] KVM: PPC: PV instructions to loads and stores Alexander Graf
2010-06-25 23:25 ` [PATCH 20/26] KVM: PPC: PV tlbsync to nop Alexander Graf
2010-06-25 23:25 ` [PATCH 21/26] KVM: PPC: Introduce kvm_tmp framework Alexander Graf
2010-06-25 23:25 ` [PATCH 22/26] KVM: PPC: PV assembler helpers Alexander Graf
2010-06-25 23:25 ` [PATCH 23/26] KVM: PPC: PV mtmsrd L=1 Alexander Graf
2010-06-25 23:25 ` [PATCH 24/26] KVM: PPC: PV mtmsrd L=0 and mtmsr Alexander Graf
2010-06-26 17:03 ` Segher Boessenkool
2010-06-27 9:10 ` Alexander Graf
2010-06-29 7:37 ` Segher Boessenkool
2010-06-25 23:25 ` [PATCH 25/26] KVM: PPC: PV wrteei Alexander Graf
2010-06-25 23:25 ` [PATCH 26/26] KVM: PPC: Add Documentation about PV interface Alexander Graf
2010-06-27 8:14 ` Avi Kivity
2010-06-27 9:33 ` Alexander Graf
[not found] ` <4C270876.2050806%40redhat.com>
2010-06-28 7:18 ` Milton Miller
2010-06-28 7:49 ` Alexander Graf [this message]
2010-06-28 8:13 ` Avi Kivity
2010-06-28 8:21 ` Alexander Graf
2010-06-28 8:32 ` Avi Kivity
2010-06-27 8:34 ` Avi Kivity
2010-06-27 9:49 ` Alexander Graf
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=92F4A3F3-A89F-418D-BD4D-66E2489F2E42@suse.de \
--to=agraf@suse.de \
--cc=avi@redhat.com \
--cc=kvm-ppc@vger.kernel.org \
--cc=kvm@vger.kernel.org \
--cc=linuxppc-dev@lists.ozlabs.org \
--cc=miltonm@bga.com \
/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).