From: Juergen Gross <jgross@suse.com>
To: Vitaly Kuznetsov <vkuznets@redhat.com>
Cc: Jonathan Corbet <corbet@lwn.net>,
Paolo Bonzini <pbonzini@redhat.com>,
Sean Christopherson <seanjc@google.com>,
Wanpeng Li <wanpengli@tencent.com>,
Jim Mattson <jmattson@google.com>, Joerg Roedel <joro@8bytes.org>,
Thomas Gleixner <tglx@linutronix.de>,
Ingo Molnar <mingo@redhat.com>, Borislav Petkov <bp@alien8.de>,
"H. Peter Anvin" <hpa@zytor.com>,
linux-kernel@vger.kernel.org, x86@kernel.org,
linux-doc@vger.kernel.org, kvm@vger.kernel.org
Subject: Re: [PATCH 6/6] x86/kvm: add boot parameter for setting max number of vcpus per guest
Date: Wed, 14 Jul 2021 13:24:12 +0200 [thread overview]
Message-ID: <1ddffb87-a6a2-eba3-3f34-cf606a2ecba2@suse.com> (raw)
In-Reply-To: <87h7gx2lkt.fsf@vitty.brq.redhat.com>
[-- Attachment #1.1.1: Type: text/plain, Size: 4974 bytes --]
On 14.07.21 13:15, Vitaly Kuznetsov wrote:
> Juergen Gross <jgross@suse.com> writes:
>
>> Today the maximum number of vcpus of a kvm guest is set via a #define
>> in a header file.
>>
>> In order to support higher vcpu numbers for guests without generally
>> increasing the memory consumption of guests on the host especially on
>> very large systems add a boot parameter for specifying the number of
>> allowed vcpus for guests.
>>
>> The default will still be the current setting of 288. The value 0 has
>> the special meaning to limit the number of possible vcpus to the
>> number of possible cpus of the host.
>>
>> Signed-off-by: Juergen Gross <jgross@suse.com>
>> ---
>> Documentation/admin-guide/kernel-parameters.txt | 10 ++++++++++
>> arch/x86/include/asm/kvm_host.h | 5 ++++-
>> arch/x86/kvm/x86.c | 7 +++++++
>> 3 files changed, 21 insertions(+), 1 deletion(-)
>>
>> diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
>> index 99bfa53a2bbd..8eb856396ffa 100644
>> --- a/Documentation/admin-guide/kernel-parameters.txt
>> +++ b/Documentation/admin-guide/kernel-parameters.txt
>> @@ -2373,6 +2373,16 @@
>> guest can't have more vcpus than the set value + 1.
>> Default: 1023
>>
>> + kvm.max_vcpus= [KVM,X86] Set the maximum allowed numbers of vcpus per
>> + guest. The special value 0 sets the limit to the number
>> + of physical cpus possible on the host (including not
>> + yet hotplugged cpus). Higher values will result in
>> + slightly higher memory consumption per guest. Depending
>> + on the value and the virtual topology the maximum
>> + allowed vcpu-id might need to be raised, too (see
>> + kvm.max_vcpu_id parameter).
>
> I'd suggest to at least add a sanity check: 'max_vcpu_id' should always
> be >= 'max_vcpus'. Alternatively, we can replace 'max_vcpu_id' with say
> 'vcpu_id_to_vcpus_ratio' and set it to e.g. '4' by default.
Either would be fine with me.
A default of '2' for the ratio would seem more appropriate for me,
however. A thread count per core not being a power of 2 is quite
unlikely, and the worst case scenario for cores per socket would be
2^n + 1.
>
>> + Default: 288
>> +
>> l1tf= [X86] Control mitigation of the L1TF vulnerability on
>> affected CPUs
>>
>> diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
>> index 39cbc4b6bffb..65ae82a5d444 100644
>> --- a/arch/x86/include/asm/kvm_host.h
>> +++ b/arch/x86/include/asm/kvm_host.h
>> @@ -37,7 +37,8 @@
>>
>> #define __KVM_HAVE_ARCH_VCPU_DEBUGFS
>>
>> -#define KVM_MAX_VCPUS 288
>> +#define KVM_DEFAULT_MAX_VCPUS 288
>> +#define KVM_MAX_VCPUS max_vcpus
>> #define KVM_SOFT_MAX_VCPUS 240
>> #define KVM_DEFAULT_MAX_VCPU_ID 1023
>> #define KVM_MAX_VCPU_ID max_vcpu_id
>> @@ -1509,6 +1510,8 @@ extern u64 kvm_max_tsc_scaling_ratio;
>> extern u64 kvm_default_tsc_scaling_ratio;
>> /* bus lock detection supported? */
>> extern bool kvm_has_bus_lock_exit;
>> +/* maximum number of vcpus per guest */
>> +extern unsigned int max_vcpus;
>> /* maximum vcpu-id */
>> extern unsigned int max_vcpu_id;
>> /* per cpu vcpu bitmasks (disable preemption during usage) */
>> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
>> index a9b0bb2221ea..888c4507504d 100644
>> --- a/arch/x86/kvm/x86.c
>> +++ b/arch/x86/kvm/x86.c
>> @@ -177,6 +177,10 @@ module_param(force_emulation_prefix, bool, S_IRUGO);
>> int __read_mostly pi_inject_timer = -1;
>> module_param(pi_inject_timer, bint, S_IRUGO | S_IWUSR);
>>
>> +unsigned int __read_mostly max_vcpus = KVM_DEFAULT_MAX_VCPUS;
>> +module_param(max_vcpus, uint, S_IRUGO);
>> +EXPORT_SYMBOL_GPL(max_vcpus);
>> +
>> unsigned int __read_mostly max_vcpu_id = KVM_DEFAULT_MAX_VCPU_ID;
>> module_param(max_vcpu_id, uint, S_IRUGO);
>>
>> @@ -10648,6 +10652,9 @@ int kvm_arch_hardware_setup(void *opaque)
>> if (boot_cpu_has(X86_FEATURE_XSAVES))
>> rdmsrl(MSR_IA32_XSS, host_xss);
>>
>> + if (max_vcpus == 0)
>> + max_vcpus = num_possible_cpus();
>
> Is this special case really needed? I mean 'max_vcpus' is not '0' by
> default so whoever sets it manually probably knows how big his guests
> are going to be anyway and it is not always obvious how many CPUs are
> reported by 'num_possible_cpus()' (ACPI tables can be weird for example).
The idea was to make it easy for anyone managing a large fleet of hosts
and wanting to have a common setting for all of them.
It would even be possible to use '0' as the default (probably via config
option only).
>
>> +
>> kvm_pcpu_vcpu_mask = __alloc_percpu(KVM_VCPU_MASK_SZ,
>> sizeof(unsigned long));
>> kvm_hv_vp_bitmap = __alloc_percpu(KVM_HV_VPMAP_SZ, sizeof(u64));
>
Thanks for the feedback,
Juergen
[-- Attachment #1.1.2: OpenPGP public key --]
[-- Type: application/pgp-keys, Size: 3135 bytes --]
[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 495 bytes --]
next prev parent reply other threads:[~2021-07-14 11:24 UTC|newest]
Thread overview: 21+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-07-01 15:40 [PATCH 0/6] x86/kvm: add boot parameters for max vcpu configs Juergen Gross
2021-07-01 15:41 ` [PATCH 1/6] x86/kvm: fix vcpu-id indexed array sizes Juergen Gross
2021-09-03 15:28 ` Eduardo Habkost
2021-07-01 15:41 ` [PATCH 2/6] x86/kvm: remove non-x86 stuff from arch/x86/kvm/ioapic.h Juergen Gross
2021-07-01 15:41 ` [PATCH 3/6] x86/kvm: add boot parameter for maximum vcpu-id Juergen Gross
2021-07-01 15:41 ` [PATCH 4/6] x86/kvm: introduce per cpu vcpu masks Juergen Gross
2021-07-26 13:32 ` Paolo Bonzini
2021-07-26 13:38 ` Juergen Gross
2021-07-01 15:41 ` [PATCH 5/6] kvm: allocate vcpu pointer array separately Juergen Gross
2021-07-26 13:40 ` Paolo Bonzini
2021-07-26 13:46 ` Juergen Gross
2021-07-26 13:57 ` Marc Zyngier
2021-07-01 15:41 ` [PATCH 6/6] x86/kvm: add boot parameter for setting max number of vcpus per guest Juergen Gross
2021-07-14 11:15 ` Vitaly Kuznetsov
2021-07-14 11:24 ` Juergen Gross [this message]
2021-07-14 11:45 ` Vitaly Kuznetsov
2021-07-14 13:04 ` Juergen Gross
2021-07-14 13:21 ` Vitaly Kuznetsov
2021-09-03 7:01 ` Juergen Gross
2021-09-03 7:40 ` Vitaly Kuznetsov
2021-07-26 13:41 ` [PATCH 0/6] x86/kvm: add boot parameters for max vcpu configs Paolo Bonzini
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=1ddffb87-a6a2-eba3-3f34-cf606a2ecba2@suse.com \
--to=jgross@suse.com \
--cc=bp@alien8.de \
--cc=corbet@lwn.net \
--cc=hpa@zytor.com \
--cc=jmattson@google.com \
--cc=joro@8bytes.org \
--cc=kvm@vger.kernel.org \
--cc=linux-doc@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@redhat.com \
--cc=pbonzini@redhat.com \
--cc=seanjc@google.com \
--cc=tglx@linutronix.de \
--cc=vkuznets@redhat.com \
--cc=wanpengli@tencent.com \
--cc=x86@kernel.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).