KVM: pass through CPUID(0x80000006)
PATCH 1/1 KVM: pass through CPUID(0x80000006)
From: Eric Northup <[email protected]>
Return L2 cache and TLB information to guests.
They could have been set before, but the defaults that KVM returns will be necessary for usermode that doesn't supply their own CPUID tables.
--- arch/x86/kvm/cpuid.c | 3 +++ 1 file changed, 3 insertions(+)
arch/x86/kvm/cpuid.c +3 -0
@@ -734,6 +734,9 @@ static inline int __do_cpuid_func(struct kvm_cpuid_entry2 *entry, u32 function, entry->ecx &= kvm_cpuid_8000_0001_ecx_x86_features;cpuid_mask (&entry->ecx,CPUID_8000_0001_ECX );break ; +case 0x80000006 : +/* L2 cache and TLB: pass through host info. */ +break ;case 0x80000007 :/* Advanced power management */ /* invariant TSC is CPUID.80000007H:EDX[8] */ entry->edx &= (1 <<8 );
Signature · 1 line
2.25.1.481.gfbce0eb801-goog
Quoted 5 lines · Jon Cargille
From: Eric Northup <[email protected]>
Return L2 cache and TLB information to guests.
They could have been set before, but the defaults that KVM returns will be necessary for usermode that doesn't supply their own CPUID tables.
I don't follow the changelog. The code makes sense, but I don't understand the justification. This only affects KVM_GET_SUPPORTED_CPUID, i.e. what's advertised to userspace, it doesn't directly change CPUID emulation in any way. The "They could have been set before" blurb is especially confusing.
I assume you want to say something like:
Return the host's L2 cache and TLB information for CPUID.0x80000006 instead of zeroing out the entry as part of KVM_GET_SUPPORTED_CPUID. This allows a userspace VMM to feed KVM_GET_SUPPORTED_CPUID's output directly into KVM_SET_CPUID2 (without breaking the guest).
Quoted 4 lines
Signed-off-by Eric Northup <[email protected]>Signed-off-by Eric Northup <[email protected]>Signed-off-by Jon Cargille <[email protected]>Signed-off-by Jim Mattson <[email protected]>
Jim's tag is unnecessary, unless he was a middleman between Eric and Jon, in which case Jim's tag should also come between Eric's and Jon's.
Only one of Eric's signoffs is needed (the one that matches the From: tag, i.e. is the official author). I'm guessing Google would prefer the author to be the @google.com address.
Quoted 21 lines
--- arch/x86/kvm/cpuid.c | 3 +++ 1 file changed, 3 insertions(+)arch/x86/kvm/cpuid.c +3
-0@@ -734,6 +734,9 @@ static inline int __do_cpuid_func(struct kvm_cpuid_entry2 *entry, u32 function, entry->ecx &= kvm_cpuid_8000_0001_ecx_x86_features;cpuid_mask (&entry->ecx,CPUID_8000_0001_ECX );break ; +case 0x80000006 : +/* L2 cache and TLB: pass through host info. */ +break ;case 0x80000007 :/* Advanced power management */ /* invariant TSC is CPUID.80000007H:EDX[8] */ entry->edx &= (1 <<8 );Signature · 1 line
2.25.1.481.gfbce0eb801-goog
Quoted 24 lines · Sean Christopherson
Quoted 5 lines · Jon Cargille
From: Eric Northup <[email protected]>
Return L2 cache and TLB information to guests.
They could have been set before, but the defaults that KVM returns will be necessary for usermode that doesn't supply their own CPUID tables.I don't follow the changelog. The code makes sense, but I don't understand the justification. This only affects
KVM_GET_SUPPORTED_CPUID, i.e. what's advertised to userspace, it doesn't directly change CPUID emulation in any way. The "They could have been set before" blurb is especially confusing.I assume you want to say something like:
Return the host's L2 cache and TLB information for CPUID.0x80000006 instead of zeroing out the entry as part of
KVM_GET_SUPPORTED_CPUID. This allows a userspace VMM to feed KVM_GET_SUPPORTED_CPUID's output directly intoKVM_SET_CPUID2(without breaking the guest).Quoted 4 lines
Signed-off-by Eric Northup <[email protected]>Signed-off-by Eric Northup <[email protected]>Signed-off-by Jon Cargille <[email protected]>Signed-off-by Jim Mattson <[email protected]>
Quoted 6 lines
Jim's tag is unnecessary, unless he was a middleman between Eric and Jon, in which case Jim's tag should also come between Eric's and Jon's.
Only one of Eric's signoffs is needed (the one that matches the From: tag, i.e. is the official author). I'm guessing Google would prefer the author to be the @google.com address.
Ah, Eric's @google.com mail bounced. Maybe do:
to clarify the work was done for Google without having a double signoff and/or a dead email.
Quoted 14 lines · Sean Christopherson
Quoted 12 lines · Sean Christopherson
Quoted 5 lines · Jon Cargille
From: Eric Northup <[email protected]>
Return L2 cache and TLB information to guests.
They could have been set before, but the defaults that KVM returns will be necessary for usermode that doesn't supply their own CPUID tables.I don't follow the changelog. The code makes sense, but I don't understand the justification. This only affects
KVM_GET_SUPPORTED_CPUID, i.e. what's advertised to userspace, it doesn't directly change CPUID emulation in any way. The "They could have been set before" blurb is especially confusing.
Quoted 6 lines
Quoted 6 lines
I assume you want to say something like:
Return the host's L2 cache and TLB information for CPUID.0x80000006 instead of zeroing out the entry as part of
KVM_GET_SUPPORTED_CPUID. This allows a userspace VMM to feed KVM_GET_SUPPORTED_CPUID's output directly intoKVM_SET_CPUID2(without breaking the guest).
This is a much better commit message, thanks.
Quoted 13 lines
Quoted 6 lines
Quoted 5 lines
Signed-off-by Eric Northup <[email protected]>Signed-off-by Eric Northup <[email protected]>Signed-off-by Jon Cargille <[email protected]>Signed-off-by Jim Mattson <[email protected]>Quoted 6 lines
Jim's tag is unnecessary, unless he was a middleman between Eric and Jon, in which case Jim's tag should also come between Eric's and Jon's.
Only one of Eric's signoffs is needed (the one that matches the From: tag, i.e. is the official author). I'm guessing Google would prefer the author to be the @google.com address.
Quoted 6 lines
Ah, Eric's @google.com mail bounced. Maybe do:
Signed-off-by Eric Northup (Google) <[email protected]>to clarify the work was done for Google without having a double signoff and/or a dead email.
That works for me. Thanks for upstreaming these patches Jon!
Quoted 6 lines · Eric Northup
Quoted 4 lines
Quoted 4 lines
Return the host's L2 cache and TLB information for CPUID.0x80000006 instead of zeroing out the entry as part of
KVM_GET_SUPPORTED_CPUID. This allows a userspace VMM to feed KVM_GET_SUPPORTED_CPUID's output directly intoKVM_SET_CPUID2(without breaking the guest).This is a much better commit message, thanks.
Queued, thanks.
Paolo
Quoted 1 line
I assume you want to say something like:
That's a much better commit message--thank you, Sean!
Quoted 1 line
Jim's tag is unnecessary, unless he was a middleman between Eric and Jon,
I appreciate the feedback; I was trying to capture that Jim "was in the patch's delivery path."
(per submitting-patches.rst), but it sounds like that is intended for a more explicit middle-man
relationship than I had understood. Jim reviewed it internally before sending, which sounds
like it should be expressed as an "Acked-by" instead; is that accurate?
Quoted 4 lines
Quoted 1 line
Only one of Eric's signoffs is needed (the one that matches the From: tag,
Ah, Eric's @google.com mail bounced. Maybe do:
Signed-off-by Eric Northup (Google) <[email protected]>
Gotcha, thanks. Yes, when I conferred with Eric on submitting his commits, he had wanted to
acknowledge that the work was done while he was at Google (e.g. his old Google email addr),
and I wanted to include current contact information for him as well.
Your suggestion to combine into a single Signed-off-by is a good one, and I'll use that form
in the future.
Thanks much,
Jon
Quoted 39 lines · Sean Christopherson
Quoted 24 lines · Sean Christopherson
Quoted 5 lines · Jon Cargille
From: Eric Northup <[email protected]>
Return L2 cache and TLB information to guests.
They could have been set before, but the defaults that KVM returns will be necessary for usermode that doesn't supply their own CPUID tables.I don't follow the changelog. The code makes sense, but I don't understand the justification. This only affects
KVM_GET_SUPPORTED_CPUID, i.e. what's advertised to userspace, it doesn't directly change CPUID emulation in any way. The "They could have been set before" blurb is especially confusing.I assume you want to say something like:
Return the host's L2 cache and TLB information for CPUID.0x80000006 instead of zeroing out the entry as part of
KVM_GET_SUPPORTED_CPUID. This allows a userspace VMM to feed KVM_GET_SUPPORTED_CPUID's output directly intoKVM_SET_CPUID2(without breaking the guest).Quoted 4 lines
Signed-off-by Eric Northup <[email protected]>Signed-off-by Eric Northup <[email protected]>Signed-off-by Jon Cargille <[email protected]>Signed-off-by Jim Mattson <[email protected]>Quoted 6 lines
Jim's tag is unnecessary, unless he was a middleman between Eric and Jon, in which case Jim's tag should also come between Eric's and Jon's.
Only one of Eric's signoffs is needed (the one that matches the From: tag, i.e. is the official author). I'm guessing Google would prefer the author to be the @google.com address.
Ah, Eric's @google.com mail bounced. Maybe do:
Signed-off-by Eric Northup (Google) <[email protected]>to clarify the work was done for Google without having a double signoff and/or a dead email.
A message that is not in the archive (sent off-list or not yet received): CANxmayh4P5hhbJPxAnA2nvbzZC9EwFPeVCxDrkHzu8h6Y7JPPQ@mail.gmail.com
Quoted 10 lines · Jon Cargille
Quoted 1 line
I assume you want to say something like:
That's a much better commit message--thank you, Sean!
Quoted 1 line
Jim's tag is unnecessary, unless he was a middleman between Eric and Jon,
I appreciate the feedback; I was trying to capture that Jim "was in the patch's delivery path." (per
submitting-patches.rst), but it sounds like that is intended for a more explicit middle-man relationship than I had understood.
Yep, exactly.
Quoted 2 lines
Jim reviewed it internally before sending, which sounds like it should be expressed as an "Acked-by" instead; is that accurate?
Or Reviewed-by. The proper (and easiest) way to handle this is to use whatever tag Jim (or any other reviewer) provides, e.g. submitting-patches states, under 12) When to use Acked-by:, Cc:, and Co-developed-by:, states:
If a person has had the opportunity to comment on a patch, but has not provided such comments, you may optionally add a ``Cc:`` tag to the patch. This is the only tag which might be added without an explicit action by the person it names
I.e. all *-by tags are only supposed to be used with explicit permission from the named person. This doesn't mean the person has to literally write Reviewed-by or whatever (though that's usually the case), but it does mean you should confirm it's ok to add a tag, e.g. if someone replies "LGTM" and you want to interpret that as a Reviewed-by or Acked-by, explicitly ask if it's ok to add the tag.
Please forgive me if this is an absurd question.
Date: Tue, 14 Apr 2020 19:37:26 -0700 From: Sean Christopherson <[email protected]>
Quoted 4 lines
Return the host's L2 cache and TLB information for CPUID.0x80000006 instead of zeroing out the entry as part of
KVM_GET_SUPPORTED_CPUID. This allows a userspace VMM to feed KVM_GET_SUPPORTED_CPUID's output directly intoKVM_SET_CPUID2(without breaking the guest).
I noticed that CPUID 0x80000005 also returns cache information (L1 Cache and TLB Information) when looking at AMD APM, while it is marked reserved on Intel SDM. What do you think about passing through CPUID 0x80000005 to guests?
To be honest, I'm not sure if it is harmless from security and performance perspectives in the first place.
Regard security aspect, I'm a bit concerned that it could help malicious guests to know something to allow cache side channel attacks. However, CPUID 0x80000006 has already passed through L2 Cache and TLB and L3 Cache Information. If passing through CPUID 0x80000006 is really fine, I'm guessing it is the case with CPUID 0x80000005 as well.
In terms of performance, as far as I know, some softwares utilizes cache information to achieve better performance. To simply put, by letting guests know cache information, they may gain some benefits. Having said that, if I understand correctly, guests can be scheduled on CPUs that do not belong to the same group of CPUs that they run last time, unless guests are pinned to a specific set of host physical CPUs. In such cases, guests may not benefit from using cache information.
If I'm missing something or say something wrong, I'd appreciate it if you could correct me. If it sounds no problem, I'd like to send a patch for it.
Best regards,
Takahiro Itazuri
Sean Christopherson
maintainerTrimmed the Cc to remove folks that no longer directly work on any of this stuff.
Quoted 8 lines · Takahiro Itazuri
Please forgive me if this is an absurd question.
Date: Tue, 14 Apr 2020 19:37:26 -0700 From: Sean Christopherson <[email protected]>Quoted 4 lines
Return the host's L2 cache and TLB information for CPUID.0x80000006 instead of zeroing out the entry as part of
KVM_GET_SUPPORTED_CPUID. This allows a userspace VMM to feed KVM_GET_SUPPORTED_CPUID's output directly intoKVM_SET_CPUID2(without breaking the guest).
Ha, this confused me for a bit. While past me did technically write this changelog, I was just massaging someone else's words.
I'm honestly a bit dubious of the claim that providing a zeroed out 0x80000006 would break the guest. I'm pretty sure I chose that phrase based Eric's original wording that KVM's "defaults" would be "necessary".
Return L2 cache and TLB information to guests.
They could have been set before, but the defaults that KVM returns will be necessary for usermode that doesn't supply their own CPUID tables.
I don't think it actually matters (see below), it's just a rather odd justification.
Quoted 5 lines
I noticed that CPUID
0x80000005also returns cache information (L1 Cache and TLB Information) when looking at AMD APM, while it is marked reserved on Intel SDM. What do you think about passing through CPUID0x80000005to guests?
Quoted 8 lines
To be honest, I'm not sure if it is harmless from security and performance perspectives in the first place.
Regard security aspect, I'm a bit concerned that it could help malicious guests to know something to allow cache side channel attacks. However, CPUID
0x80000006has already passed through L2 Cache and TLB and L3 Cache Information. If passing through CPUID0x80000006is really fine, I'm guessing it is the case with CPUID0x80000005as well.
It's definitely harmless from a security perspective. Userspace already has access to this information as CPUID is NOT a priveleged instructions. And the kernel also publishes this information in sysfs, e.g. /sys/devices/system/cpu/cpuN/cache, and AFAIK that's not typically restricted.
KVM must assume that any and all information visible to userspace is also visible to the guest, e.g. even if KVM wanted to police CPUID, nothing would prevent userspace from providing a paravirtual interface to the guest to enumerate cache and TLB topology.
Quoted 7 lines
In terms of performance, as far as I know, some softwares utilizes cache information to achieve better performance. To simply put, by letting guests know cache information, they may gain some benefits. Having said that, if I understand correctly, guests can be scheduled on CPUs that do not belong to the same group of CPUs that they run last time, unless guests are pinned to a specific set of host physical CPUs. In such cases, guests may not benefit from using cache information.
I would be quite surprised if homogeneous, a.k.a. non-hybrid, CPUs ever have variable cache/TLB properties across cores. Hybrid CPUs might be a different story, but even then I gotta imagine that userspace software already has problems, e.g. userspace processes will encounter variable cache/TLB behavior unless userspace is affining all tasks.
Regardless, the decision on whether or not to report cache information via KVM_GET_SUPPORTED_CPUID was made long, long ago, as KVM has enumerated CPUID.0x4 since basically forever. So really this only affects TLB info, and since KVM already spits out 0x80000006, it's just L1 TLB info.
I'm mildly tempted to remove 0x80000006, for similar reasons as commit 45e966fcca03 ("KVM: x86: Do not return host topology information from KVM_GET_SUPPORTED_CPUID"), but I suspect that would do more harm than good, e.g. Linux falls back to 0x80000005 and 0x80000006 when running on AMD without extended topology info.
Quoted 3 lines
If I'm missing something or say something wrong, I'd appreciate it if you could correct me. If it sounds no problem, I'd like to send a patch for it.
I think it makes sense to enumerate 0x80000005. Reporting 0x80000006 but not 0x80000005 seems to be the *worst* behavior, so as I see it, the decision is really between adding 0x80000005 and removing 0x80000006. Adding 0x80000005 appears to be the least risky choice given that KVM has reported 0x80000006 for over three years.
Quoted 1 line · Sean Christopherson
Trimmed the Cc to remove folks that no longer directly work on any of this stuff.
I apologize for it and appreciate your reply on this.
Quoted 10 lines
Quoted 5 lines
Regard security aspect, I'm a bit concerned that it could help malicious guests to know something to allow cache side channel attacks. However, CPUID
0x80000006has already passed through L2 Cache and TLB and L3 Cache Information. If passing through CPUID0x80000006is really fine, I'm guessing it is the case with CPUID0x80000005as well.It's definitely harmless from a security perspective. Userspace already has access to this information as CPUID is NOT a priveleged instructions. And the kernel also publishes this information in sysfs, e.g.
/sys/devices/system/cpu/cpuN/cache, and AFAIK that's not typically restricted.
I'm releaved to hear that.
Quoted 4 lines
I'm mildly tempted to remove
0x80000006, for similar reasons as commit45e966fcca03("KVM: x86: Do not return host topology information fromKVM_GET_SUPPORTED_CPUID"), but I suspect that would do more harm than good, e.g. Linux falls back to0x80000005and0x80000006when running on AMD without extended topology info.
Actually I also saw the commit and I was a bit confused about which leaves to pass through. As you mentioned, CPUID is accessible from userspace and VMM can query it if they want.
Quoted 5 lines
I think it makes sense to enumerate
0x80000005. Reporting0x80000006but not0x80000005seems to be the *worst* behavior, so as I see it, the decision is really between adding0x80000005and removing0x80000006. Adding0x80000005appears to be the least risky choice given that KVM has reported0x80000006for over three years.
I'm on the same page that either reporting both or none of them is better. I'll create a patch for the least risky one.
Best regards,
Takahiro Itazuri