From: Vitaly Kuznetsov <vkuznets@redhat.com>
To: Sean Christopherson <sean.j.christopherson@intel.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>,
Wanpeng Li <wanpengli@tencent.com>,
Jim Mattson <jmattson@google.com>, Joerg Roedel <joro@8bytes.org>,
kvm@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH 47/61] KVM: x86: Squash CPUID 0x2.0 insanity for modern CPUs
Date: Mon, 24 Feb 2020 23:35:00 +0100 [thread overview]
Message-ID: <87r1yjmx63.fsf@vitty.brq.redhat.com> (raw)
In-Reply-To: <20200201185218.24473-48-sean.j.christopherson@intel.com>
Sean Christopherson <sean.j.christopherson@intel.com> writes:
> Rework CPUID 0x2.0 to be a normal CPUID leaf if it returns "01" in AL,
> i.e. EAX & 0xff.
>
> Long ago, Intel documented CPUID 0x2.0 as being a stateful leaf, e.g. a
> version of the SDM circa 1995 states:
>
> The least-significant byte in register EAX (register AL) indicates the
> number of times the CPUID instruction must be executed with an input
> value of 2 to get a complete description of the processors's caches
> and TLBs. The Pentium Pro family of processors will return a 1.
>
> A 2000 version of the SDM only updated the paragraph to reference
> Intel's new processory family:
>
> The first member of the family of Pentium 4 processors will return a 1.
>
> Fast forward to the present, and Intel's SDM now states:
>
> The least-significant byte in register EAX (register AL) will always
> return 01H. Software should ignore this value and not interpret it as
> an information descriptor.
>
> AMD's APM simply states that CPUID 0x2 is reserved.
>
> Given that CPUID itself was introduced in the Pentium, odds are good
> that the only Intel CPU family that *maybe* implemented a stateful CPUID
> was the P5. Which obviously did not support VMX, or KVM.
>
> In other words, KVM's emulation of a stateful CPUID 0x2.0 has likely
> been dead code from the day it was introduced. This is backed up by
> commit 0fdf8e59faa5c ("KVM: Fix cpuid iteration on multiple leaves per
> eac"), whichs show that the stateful iteration code was completely
> broken when it was introduced by commit 0771671749b59 ("KVM: Enhance
> guest cpuid management"), i.e. not actually tested.
>
> Although it's _extremely_ tempting to yank KVM's stateful code, leave it
> in for now but annotate all its code paths as "unlikely". The code is
> relatively contained, and if by some miracle there is someone running KVM
> on a CPU with a stateful CPUID 0x2, more power to 'em.
>
> Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
> ---
> arch/x86/kvm/cpuid.c | 31 +++++++++++++++++++++----------
> 1 file changed, 21 insertions(+), 10 deletions(-)
>
> diff --git a/arch/x86/kvm/cpuid.c b/arch/x86/kvm/cpuid.c
> index 47f61f4497fb..ab2a34337588 100644
> --- a/arch/x86/kvm/cpuid.c
> +++ b/arch/x86/kvm/cpuid.c
> @@ -405,9 +405,6 @@ static struct kvm_cpuid_entry2 *do_host_cpuid(struct kvm_cpuid_array *array,
> &entry->eax, &entry->ebx, &entry->ecx, &entry->edx);
>
> switch (function) {
> - case 2:
> - entry->flags |= KVM_CPUID_FLAG_STATEFUL_FUNC;
> - break;
> case 4:
> case 7:
> case 0xb:
> @@ -483,17 +480,31 @@ static inline int __do_cpuid_func(struct kvm_cpuid_array *array, u32 function)
> * it since we emulate x2apic in software */
> cpuid_entry_set(entry, X86_FEATURE_X2APIC);
> break;
> - /* function 2 entries are STATEFUL. That is, repeated cpuid commands
> - * may return different values. This forces us to get_cpu() before
> - * issuing the first command, and also to emulate this annoying behavior
> - * in kvm_emulate_cpuid() using KVM_CPUID_FLAG_STATE_READ_NEXT */
> case 2:
> + /*
> + * On ancient CPUs, function 2 entries are STATEFUL. That is,
> + * CPUID(function=2, index=0) may return different results each
> + * time, with the least-significant byte in EAX enumerating the
> + * number of times software should do CPUID(2, 0).
> + *
> + * Modern CPUs (quite likely every CPU KVM has *ever* run on)
> + * are less idiotic. Intel's SDM states that EAX & 0xff "will
> + * always return 01H. Software should ignore this value and not
> + * interpret it as an informational descriptor", while AMD's
> + * APM states that CPUID(2) is reserved.
> + */
> + max_idx = entry->eax & 0xff;
> + if (likely(max_idx <= 1))
> + break;
> +
> + entry->flags |= KVM_CPUID_FLAG_STATEFUL_FUNC;
> entry->flags |= KVM_CPUID_FLAG_STATE_READ_NEXT;
>
> - for (i = 1, max_idx = entry->eax & 0xff; i < max_idx; ++i) {
> + for (i = 1; i < max_idx; ++i) {
> entry = do_host_cpuid(array, 2, 0);
> if (!entry)
> goto out;
> + entry->flags |= KVM_CPUID_FLAG_STATEFUL_FUNC;
> }
> break;
> /* functions 4 and 0x8000001d have additional index. */
> @@ -903,7 +914,7 @@ static int is_matching_cpuid_entry(struct kvm_cpuid_entry2 *e,
> return 0;
> if ((e->flags & KVM_CPUID_FLAG_SIGNIFCANT_INDEX) && e->index != index)
> return 0;
> - if ((e->flags & KVM_CPUID_FLAG_STATEFUL_FUNC) &&
> + if (unlikely(e->flags & KVM_CPUID_FLAG_STATEFUL_FUNC) &&
> !(e->flags & KVM_CPUID_FLAG_STATE_READ_NEXT))
> return 0;
> return 1;
> @@ -920,7 +931,7 @@ struct kvm_cpuid_entry2 *kvm_find_cpuid_entry(struct kvm_vcpu *vcpu,
>
> e = &vcpu->arch.cpuid_entries[i];
> if (is_matching_cpuid_entry(e, function, index)) {
> - if (e->flags & KVM_CPUID_FLAG_STATEFUL_FUNC)
> + if (unlikely(e->flags & KVM_CPUID_FLAG_STATEFUL_FUNC))
> move_to_next_stateful_cpuid_entry(vcpu, i);
> best = e;
> break;
Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>
but your history digging results make me think that killing the whole
'statefulness' thing is not a bad idea at all :-)
--
Vitaly
next prev parent reply other threads:[~2020-02-24 22:35 UTC|newest]
Thread overview: 168+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-02-01 18:51 [PATCH 00/61] KVM: x86: Introduce KVM cpu caps Sean Christopherson
2020-02-01 18:51 ` [PATCH 01/61] KVM: x86: Return -E2BIG when KVM_GET_SUPPORTED_CPUID hits max entries Sean Christopherson
2020-02-03 12:55 ` Vitaly Kuznetsov
2020-02-03 15:59 ` Sean Christopherson
2020-02-25 14:36 ` Paolo Bonzini
2020-02-01 18:51 ` [PATCH 02/61] KVM: x86: Refactor loop around do_cpuid_func() to separate helper Sean Christopherson
2020-02-06 14:59 ` Vitaly Kuznetsov
2020-02-07 19:53 ` Sean Christopherson
2020-02-25 14:37 ` Paolo Bonzini
2020-02-25 15:09 ` Vitaly Kuznetsov
2020-02-26 11:35 ` Paolo Bonzini
2020-02-01 18:51 ` [PATCH 03/61] KVM: x86: Simplify handling of Centaur CPUID leafs Sean Christopherson
2020-02-06 15:05 ` Vitaly Kuznetsov
2020-02-07 19:47 ` Sean Christopherson
2020-02-01 18:51 ` [PATCH 04/61] KVM: x86: Clean up error handling in kvm_dev_ioctl_get_cpuid() Sean Christopherson
2020-02-06 15:09 ` Vitaly Kuznetsov
2020-02-01 18:51 ` [PATCH 05/61] KVM: x86: Check userapce CPUID array size after validating sub-leaf Sean Christopherson
2020-02-06 15:24 ` Vitaly Kuznetsov
2020-02-01 18:51 ` [PATCH 06/61] KVM: x86: Move CPUID 0xD.1 handling out of the index>0 loop Sean Christopherson
2020-02-07 15:38 ` Vitaly Kuznetsov
2020-02-01 18:51 ` [PATCH 07/61] KVM: x86: Check for CPUID 0xD.N support before validating array size Sean Christopherson
2020-02-07 15:48 ` Vitaly Kuznetsov
2020-02-01 18:51 ` [PATCH 08/61] KVM: x86: Warn on zero-size save state for valid CPUID 0xD.N sub-leaf Sean Christopherson
2020-02-07 15:54 ` Vitaly Kuznetsov
2020-02-07 15:56 ` Sean Christopherson
2020-02-01 18:51 ` [PATCH 09/61] KVM: x86: Refactor CPUID 0xD.N sub-leaf entry creation Sean Christopherson
2020-02-07 15:56 ` Vitaly Kuznetsov
2020-02-01 18:51 ` [PATCH 10/61] KVM: x86: Clean up CPUID 0x7 sub-leaf loop Sean Christopherson
2020-02-21 14:20 ` Vitaly Kuznetsov
2020-02-01 18:51 ` [PATCH 11/61] KVM: x86: Drop the explicit @index from do_cpuid_7_mask() Sean Christopherson
2020-02-21 14:22 ` Vitaly Kuznetsov
2020-02-01 18:51 ` [PATCH 12/61] KVM: x86: Drop redundant boot cpu checks on SSBD feature bits Sean Christopherson
2020-02-01 18:51 ` [PATCH 13/61] KVM: x86: Consolidate CPUID array max num entries checking Sean Christopherson
2020-02-01 18:51 ` [PATCH 14/61] KVM: x86: Hoist loop counter and terminator to top of __do_cpuid_func() Sean Christopherson
2020-02-01 18:51 ` [PATCH 15/61] KVM: x86: Refactor CPUID 0x4 and 0x8000001d handling Sean Christopherson
2020-02-21 14:40 ` Vitaly Kuznetsov
2020-02-01 18:51 ` [PATCH 16/61] KVM: x86: Encapsulate CPUID entries and metadata in struct Sean Christopherson
2020-02-21 14:58 ` Vitaly Kuznetsov
2020-02-24 21:55 ` Sean Christopherson
2020-02-24 23:12 ` Vitaly Kuznetsov
2020-02-01 18:51 ` [PATCH 17/61] KVM: x86: Drop redundant array size check Sean Christopherson
2020-02-01 18:51 ` [PATCH 18/61] KVM: x86: Use common loop iterator when handling CPUID 0xD.N Sean Christopherson
2020-02-21 15:04 ` Vitaly Kuznetsov
2020-02-01 18:51 ` [PATCH 19/61] KVM: VMX: Add helpers to query Intel PT mode Sean Christopherson
[not found] ` <87pne8q8c0.fsf@vitty.brq.redhat.com>
2020-02-24 22:18 ` Sean Christopherson
2020-02-25 14:54 ` Paolo Bonzini
2020-03-03 22:41 ` Sean Christopherson
2020-02-01 18:51 ` [PATCH 20/61] KVM: x86: Calculate the supported xcr0 mask at load time Sean Christopherson
2020-02-13 14:21 ` Xiaoyao Li
2020-02-01 18:51 ` [PATCH 21/61] KVM: x86: Use supported_xcr0 to detect MPX support Sean Christopherson
2020-02-13 14:25 ` Xiaoyao Li
2020-02-21 15:32 ` Vitaly Kuznetsov
2020-02-01 18:51 ` [PATCH 22/61] KVM: x86: Make kvm_mpx_supported() an inline function Sean Christopherson
2020-02-13 14:26 ` Xiaoyao Li
2020-02-21 15:33 ` Vitaly Kuznetsov
2020-02-01 18:51 ` [PATCH 23/61] KVM: x86: Clear output regs for CPUID 0x14 if PT isn't exposed to guest Sean Christopherson
2020-02-21 15:36 ` Vitaly Kuznetsov
2020-02-01 18:51 ` [PATCH 24/61] KVM: x86: Drop explicit @func param from ->set_supported_cpuid() Sean Christopherson
2020-02-21 15:39 ` Vitaly Kuznetsov
2020-02-01 18:51 ` [PATCH 25/61] KVM: x86: Use u32 for holding CPUID register value in helpers Sean Christopherson
2020-02-21 15:43 ` Vitaly Kuznetsov
2020-02-01 18:51 ` [PATCH 26/61] KVM: x86: Introduce cpuid_entry_{get,has}() accessors Sean Christopherson
2020-02-14 9:44 ` Xiaoyao Li
2020-02-14 17:09 ` Sean Christopherson
2020-02-21 15:57 ` Vitaly Kuznetsov
2020-02-21 16:29 ` Sean Christopherson
2020-02-01 18:51 ` [PATCH 27/61] KVM: x86: Introduce cpuid_entry_{change,set,clear}() mutators Sean Christopherson
[not found] ` <87ftf0p0d0.fsf@vitty.brq.redhat.com>
2020-02-24 22:42 ` Sean Christopherson
2020-02-01 18:51 ` [PATCH 28/61] KVM: x86: Refactor cpuid_mask() to auto-retrieve the register Sean Christopherson
2020-02-24 13:49 ` Vitaly Kuznetsov
2020-02-01 18:51 ` [PATCH 29/61] KVM: x86: Add Kconfig-controlled auditing of reverse CPUID lookups Sean Christopherson
2020-02-24 13:54 ` Vitaly Kuznetsov
2020-02-24 22:46 ` Sean Christopherson
2020-02-25 15:02 ` Paolo Bonzini
2020-02-25 15:00 ` Paolo Bonzini
2020-02-01 18:51 ` [PATCH 30/61] KVM: x86: Handle MPX CPUID adjustment in VMX code Sean Christopherson
2020-02-13 13:51 ` Xiaoyao Li
2020-02-13 17:37 ` Sean Christopherson
2020-02-24 15:14 ` Vitaly Kuznetsov
2020-02-24 15:45 ` Sean Christopherson
2020-02-01 18:51 ` [PATCH 31/61] KVM: x86: Handle INVPCID " Sean Christopherson
2020-02-24 15:19 ` Vitaly Kuznetsov
2020-02-01 18:51 ` [PATCH 32/61] KVM: x86: Handle UMIP emulation " Sean Christopherson
2020-02-24 15:21 ` Vitaly Kuznetsov
2020-02-01 18:51 ` [PATCH 33/61] KVM: x86: Handle PKU " Sean Christopherson
2020-02-24 15:24 ` Vitaly Kuznetsov
2020-02-01 18:51 ` [PATCH 34/61] KVM: x86: Handle RDTSCP " Sean Christopherson
2020-02-24 15:28 ` Vitaly Kuznetsov
2020-02-01 18:51 ` [PATCH 35/61] KVM: x86: Handle Intel PT " Sean Christopherson
2020-02-24 15:30 ` Vitaly Kuznetsov
2020-02-01 18:51 ` [PATCH 36/61] KVM: x86: Handle GBPAGE CPUID adjustment for EPT " Sean Christopherson
2020-02-24 15:34 ` Vitaly Kuznetsov
2020-02-01 18:51 ` [PATCH 37/61] KVM: x86: Refactor handling of XSAVES CPUID adjustment Sean Christopherson
2020-02-24 15:39 ` Vitaly Kuznetsov
2020-02-01 18:51 ` [PATCH 38/61] KVM: x86: Introduce kvm_cpu_caps to replace runtime CPUID masking Sean Christopherson
2020-02-24 16:32 ` Vitaly Kuznetsov
2020-02-24 22:57 ` Sean Christopherson
2020-02-24 23:20 ` Vitaly Kuznetsov
2020-02-24 23:25 ` Sean Christopherson
2020-02-01 18:51 ` [PATCH 39/61] KVM: SVM: Convert feature updates from CPUID to KVM cpu caps Sean Christopherson
2020-02-24 21:33 ` Vitaly Kuznetsov
2020-02-25 15:10 ` Paolo Bonzini
2020-02-28 0:28 ` Sean Christopherson
2020-02-28 0:36 ` Sean Christopherson
2020-02-28 7:03 ` Paolo Bonzini
2020-02-28 15:09 ` Sean Christopherson
2020-02-01 18:51 ` [PATCH 40/61] KVM: VMX: " Sean Christopherson
2020-02-24 21:40 ` Vitaly Kuznetsov
2020-02-01 18:51 ` [PATCH 41/61] KVM: x86: Move XSAVES CPUID adjust to VMX's KVM cpu cap update Sean Christopherson
2020-02-24 21:43 ` Vitaly Kuznetsov
2020-02-01 18:51 ` [PATCH 42/61] KVM: x86: Add a helper to check kernel support when setting cpu cap Sean Christopherson
2020-02-24 21:47 ` Vitaly Kuznetsov
2020-02-01 18:52 ` [PATCH 43/61] KVM: x86: Use KVM cpu caps to mark CR4.LA57 as not-reserved Sean Christopherson
2020-02-24 22:08 ` Vitaly Kuznetsov
2020-02-24 23:23 ` Sean Christopherson
2020-02-25 15:12 ` Paolo Bonzini
2020-02-25 15:19 ` David Laight
2020-02-25 21:22 ` Sean Christopherson
2020-02-26 11:35 ` Paolo Bonzini
2020-02-01 18:52 ` [PATCH 44/61] KVM: x86: Use KVM cpu caps to track UMIP emulation Sean Christopherson
2020-02-24 22:13 ` Vitaly Kuznetsov
2020-02-01 18:52 ` [PATCH 45/61] KVM: x86: Fold CPUID 0x7 masking back into __do_cpuid_func() Sean Christopherson
2020-02-24 22:21 ` Vitaly Kuznetsov
2020-02-01 18:52 ` [PATCH 46/61] KVM: x86: Remove the unnecessary loop on CPUID 0x7 sub-leafs Sean Christopherson
2020-02-24 22:25 ` Vitaly Kuznetsov
2020-02-01 18:52 ` [PATCH 47/61] KVM: x86: Squash CPUID 0x2.0 insanity for modern CPUs Sean Christopherson
2020-02-24 22:35 ` Vitaly Kuznetsov [this message]
2020-02-25 15:17 ` Paolo Bonzini
2020-02-01 18:52 ` [PATCH 48/61] KVM: x86: Do host CPUID at load time to mask KVM cpu caps Sean Christopherson
[not found] ` <87o8tnmwni.fsf@vitty.brq.redhat.com>
2020-02-24 23:31 ` Sean Christopherson
2020-02-25 13:53 ` Vitaly Kuznetsov
2020-02-25 15:18 ` Paolo Bonzini
2020-02-25 21:08 ` Sean Christopherson
2020-02-29 18:38 ` Sean Christopherson
2020-02-01 18:52 ` [PATCH 49/61] KVM: x86: Override host CPUID results with kvm_cpu_caps Sean Christopherson
2020-02-24 22:57 ` Vitaly Kuznetsov
2020-02-01 18:52 ` [PATCH 50/61] KVM: x86: Set emulated/transmuted feature bits via kvm_cpu_caps Sean Christopherson
2020-02-25 13:59 ` Vitaly Kuznetsov
2020-02-01 18:52 ` [PATCH 51/61] KVM: x86: Use kvm_cpu_caps to detect Intel PT support Sean Christopherson
2020-02-25 14:06 ` Vitaly Kuznetsov
2020-02-01 18:52 ` [PATCH 52/61] KVM: x86: Use KVM cpu caps to detect MSR_TSC_AUX virt support Sean Christopherson
2020-02-25 14:08 ` Vitaly Kuznetsov
2020-02-01 18:52 ` [PATCH 53/61] KVM: VMX: Directly use VMX capabilities helper to detect RDTSCP support Sean Christopherson
2020-02-25 14:10 ` Vitaly Kuznetsov
2020-02-01 18:52 ` [PATCH 54/61] KVM: x86: Check for Intel PT MSR virtualization using KVM cpu caps Sean Christopherson
2020-02-25 14:11 ` Vitaly Kuznetsov
2020-02-01 18:52 ` [PATCH 55/61] KVM: VMX: Directly query Intel PT mode when refreshing PMUs Sean Christopherson
2020-02-25 14:16 ` Vitaly Kuznetsov
2020-02-01 18:52 ` [PATCH 56/61] KVM: SVM: Refactor logging of NPT enabled/disabled Sean Christopherson
2020-02-25 14:21 ` Vitaly Kuznetsov
2020-02-01 18:52 ` [PATCH 57/61] KVM: x86/mmu: Merge kvm_{enable,disable}_tdp() into a common function Sean Christopherson
2020-02-25 14:27 ` Vitaly Kuznetsov
2020-02-01 18:52 ` [PATCH 58/61] KVM: x86/mmu: Configure max page level during hardware setup Sean Christopherson
2020-02-25 14:43 ` Vitaly Kuznetsov
2020-02-25 21:01 ` Sean Christopherson
2020-02-26 14:55 ` Vitaly Kuznetsov
2020-02-26 15:56 ` Sean Christopherson
2020-02-01 18:52 ` [PATCH 59/61] KVM: x86: Don't propagate MMU lpage support to memslot.disallow_lpage Sean Christopherson
2020-02-25 14:55 ` Vitaly Kuznetsov
2020-02-01 18:52 ` [PATCH 60/61] KVM: Drop largepages_enabled and its accessor/mutator Sean Christopherson
2020-02-25 14:56 ` Vitaly Kuznetsov
2020-02-01 18:52 ` [PATCH 61/61] KVM: x86: Move VMX's host_efer to common x86 code Sean Christopherson
2020-02-25 15:02 ` Vitaly Kuznetsov
[not found] ` <87wo8ak84x.fsf@vitty.brq.redhat.com>
2020-02-25 15:25 ` [PATCH 00/61] KVM: x86: Introduce KVM cpu caps Paolo Bonzini
2020-02-28 1:37 ` Sean Christopherson
2020-02-28 7:04 ` Paolo Bonzini
2020-02-29 18:32 ` Sean Christopherson
2020-03-02 9:03 ` Vitaly Kuznetsov
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=87r1yjmx63.fsf@vitty.brq.redhat.com \
--to=vkuznets@redhat.com \
--cc=jmattson@google.com \
--cc=joro@8bytes.org \
--cc=kvm@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=pbonzini@redhat.com \
--cc=sean.j.christopherson@intel.com \
--cc=wanpengli@tencent.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).