Hi Babu, On 9/27/2022 1:25 PM, Babu Moger wrote: > Add the new AMD feature X86_FEATURE_SMBA. With this feature, the QOS > enforcement policies can be applied to external slow memory connected > to the host. QOS enforcement is accomplished by assigning a Class Of > Service (COS) to a processor and specifying allocations or limits for > that COS for each resource to be allocated. > > This feature is identified by the CPUID Function 8000_0020_EBX_x0. > > CPUID Fn8000_0020_EBX_x0 AMD Bandwidth Enforcement Feature Identifiers > (ECX=0) > > Bits Field Name Description > 2 L3SBE L3 external slow memory bandwidth enforcement > > > Currently, CXL.memory is the only supported "slow" memory device. With > the support of SMBA feature, the hardware enables bandwidth allocation > on the slow memory devices. If there are multiple slow memory devices > in the system, then the throttling logic groups all the slow sources > together and applies the limit on them as a whole. > > The presence of the SMBA feature(with CXL.memory) is independent of > whether slow memory device is actually present in the system. If there > is no slow memory in the system, then setting a SMBA limit will have no > impact on the performance of the system. > > Presence of CXL memory can be identified by numactl command. > > $numactl -H > available: 2 nodes (0-1) > node 0 cpus: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 > node 0 size: 63678 MB node 0 free: 59542 MB > node 1 cpus: > node 1 size: 16122 MB > node 1 free: 15627 MB > node distances: > node 0 1 > 0: 10 50 > 1: 50 10 > > CPU list for CXL memory will be empty. The cpu-cxl node distance is > greater than cpu-to-cpu distances. Node 1 has the CXL memory in this > case. CXL memory can also be identified using ACPI SRAT table and > memory maps. > > Feature description is available in the specification, "AMD64 > Technology Platform Quality of Service Extensions, Revision: 1.03 > Publication # 56375 Revision: 1.03 Issue Date: February 2022". > > Link: https://www.amd.com/en/support/tech-docs/amd64-technology-platform-quality-service-extensions > Link: https://bugzilla.kernel.org/show_bug.cgi?id=206537 > Signed-off-by: Babu Moger <babu.moger@xxxxxxx> > --- > arch/x86/include/asm/cpufeatures.h | 1 + > arch/x86/kernel/cpu/scattered.c | 1 + > 2 files changed, 2 insertions(+) > > diff --git a/arch/x86/include/asm/cpufeatures.h b/arch/x86/include/asm/cpufeatures.h > index ef4775c6db01..349852b9daa4 100644 > --- a/arch/x86/include/asm/cpufeatures.h > +++ b/arch/x86/include/asm/cpufeatures.h > @@ -304,6 +304,7 @@ > #define X86_FEATURE_UNRET (11*32+15) /* "" AMD BTB untrain return */ > #define X86_FEATURE_USE_IBPB_FW (11*32+16) /* "" Use IBPB during runtime firmware calls */ > #define X86_FEATURE_RSB_VMEXIT_LITE (11*32+17) /* "" Fill RSB on VM exit when EIBRS is enabled */ > +#define X86_FEATURE_SMBA (11*32+18) /* Slow Memory Bandwidth Allocation */ > > /* Intel-defined CPU features, CPUID level 0x00000007:1 (EAX), word 12 */ > #define X86_FEATURE_AVX_VNNI (12*32+ 4) /* AVX VNNI instructions */ > diff --git a/arch/x86/kernel/cpu/scattered.c b/arch/x86/kernel/cpu/scattered.c > index fd44b54c90d5..885ecf46abb2 100644 > --- a/arch/x86/kernel/cpu/scattered.c > +++ b/arch/x86/kernel/cpu/scattered.c > @@ -44,6 +44,7 @@ static const struct cpuid_bit cpuid_bits[] = { > { X86_FEATURE_CPB, CPUID_EDX, 9, 0x80000007, 0 }, > { X86_FEATURE_PROC_FEEDBACK, CPUID_EDX, 11, 0x80000007, 0 }, > { X86_FEATURE_MBA, CPUID_EBX, 6, 0x80000008, 0 }, > + { X86_FEATURE_SMBA, CPUID_EBX, 2, 0x80000020, 0 }, > { X86_FEATURE_PERFMON_V2, CPUID_EAX, 0, 0x80000022, 0 }, > { 0, 0, 0, 0, 0 } > }; > > Please respect the coding style of the area you are modifying. This is the same feedback as provided in v4 in https://lore.kernel.org/lkml/ba36c68c-0b13-e8a2-fb45-8b84ea9f7259@xxxxxxxxx/ Looking ahead the same issue also remains in patch 3 as previously mentioned in v4 feedback. https://lore.kernel.org/lkml/c4a9ea23-4280-d54c-263b-354ea321f746@xxxxxxxxx/ Also missing is highlighting that configuration has changed from per-domain to per-CPU and why. It does not seem as though this series is ready. I will wait for next version to have existing review comments addressed before trying to look at new changes. Reinette
