MBA即Memory Bandwidth Allocation的缩写，即内存带宽分配，是针对LLC隔离的一项技术，也是RDT（Resource Director Technology）的一项指标。
关于RDT技术可参考https://github.com/intel/intel-cmt-cat/wiki/Useful-links
其中，RDT技术需要除了需要硬件支持之外，还需要linux内核支持，linux内核从4.4版本已经支持RDT技术。本文的内核版本基于5.5版本
在内核代码文件arch/x86/kernel/cpu/resctrl/core.c中
static bool __get_mem_config_intel(struct rdt_resource *r)
{
union cpuid_0x10_3_eax eax;
union cpuid_0x10_x_edx edx;
u32 ebx, ecx;
cpuid_count(0x00000010, 3, &eax.full, &ebx, &ecx, &edx.full);//通过cpuid指令获取MBA信息
r->num_closid = edx.split.cos_max + 1;
r->membw.max_delay = eax.split.max_delay + 1;
r->default_ctrl = MAX_MBA_BW;
if (ecx & MBA_IS_LINEAR) {
r->membw.delay_linear = true;
r->membw.min_bw = MAX_MBA_BW - r->membw.max_delay;
r->membw.bw_gran = MAX_MBA_BW - r->membw.max_delay;
} else {
if (!rdt_get_mb_table(r))
return false;
}
r->data_width = 3;
r->alloc_capable = true;
r->alloc_enabled = true;
return true;
}
/* Some CPUID calls want 'count' to be placed in ecx */
static inline void cpuid_count(unsigned int op, int count,
unsigned int *eax, unsigned int *ebx,
unsigned int *ecx, unsigned int *edx)
{
*eax = op;
*ecx = count;
__cpuid(eax, ebx, ecx, edx);
}
#define __cpuid native_cpuid
static inline void native_cpuid(unsigned int *eax, unsigned int *ebx,
unsigned int *ecx, unsigned int *edx)
{
/* ecx is often an input as well as an output. */
asm volatile("cpuid"
: "=a" (*eax),
"=b" (*ebx),
"=c" (*ecx),
"=d" (*edx)
: "0" (*eax), "2" (*ecx)
: "memory");
}
outputs：”=a”(a), “=b” (b), “=c” (c), “=d” (d)。表示有4个输出参数——参数0为eax（绑定到变量a）、参数1为ebx（绑定到变量b）、参数2为ecx（绑定到变量c）、参数3为edx（绑定到变量d）。
inputs：”0”(eax), “2” (ecx))。表示有2个输入参数——参数0（eax），参数2（ecx）。
调用cpuid指令 // asmstatements
将返回的eax赋给a //outputs
将返回的ebx赋给b
将返回的ecx赋给c
将返回的edx赋给d
结合intel的说明文档可以看出，cpuid指令，当eax=16，ecx=3的时候，获取到的是MBA（memory bandwidth allocation）信息
https://www.felixcloutier.com/x86/cpuid
这里我将核心代码抠出来，组成可执行程序，看下cpuid执行的结果
#include <stdio.h>
typedef unsigned int u32;
union cpuid_0x10_3_eax {
struct {
unsigned int max_delay:12;
} split;
unsigned int full;
};
union cpuid_0x10_x_edx {
struct {
unsigned int cos_max:16;
} split;
unsigned int full;
};
static inline void __cpuid(unsigned int *eax, unsigned int *ebx, unsigned int *ecx, unsigned int *edx)
{
/* ecx is often an input as well as an output. */
asm volatile("cpuid"
: "=a" (*eax),
"=b" (*ebx),
"=c" (*ecx),
"=d" (*edx)
: "0" (*eax), "2" (*ecx)
: "memory");
}
static inline void cpuid_count(unsigned int op, int count,
unsigned int *eax, unsigned int *ebx,
unsigned int *ecx, unsigned int *edx)
{
*eax = op;
*ecx = count;
__cpuid(eax, ebx, ecx, edx);
}
int main(int argc, char *argv[]) {
union cpuid_0x10_3_eax eax;
union cpuid_0x10_x_edx edx;
u32 ebx, ecx;
cpuid_count(0x00000010, 3, &eax.full, &ebx, &ecx, &edx.full);
printf("%d\n", eax.split.max_delay);
return 0;
}
因此在Intel(R) Xeon(R) Platinum 8260平台上获取到的MBA结果是89
当然上面的程序只能在intel平台上跑起来，如果是AMD平台，则需要修改op，即传给cpuid指定的操作数eax的值。所以针对AMD平台，只需要将cpuid_count(0x00000010,3, &eax.full, &ebx, &ecx, &edx.full);改为cpuid_count(0x80000020,1, &eax.full, &ebx, &ecx, &edx.full);即可。
