packages/qemu
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

i386: cache passthrough: Update AMD 8000_001D.EAX[25:14] based on vCPU topo

Browse Source 
On AMD target, when host cache passthrough is disabled we will
emulate the guest caches with default values and initialize the
shared cpu list of the caches based on vCPU topology. However
when host cache passthrough is enabled, the shared cpu list is
consistent with host regardless what the vCPU topology is.

For example, when cache passthrough is enabled, running a guest
with vThreads=1 on a host with pThreads=2, we will get that there
are every *two* logical vCPUs sharing a L1/L2 cache, which is not
consistent with the vCPU topology (vThreads=1).

So let's reinitialize BITs[25:14] of AMD CPUID 8000_001D.EAX
based on the actual vCPU topology instead of host pCPU topology.

Signed-off-by: Yanan Wang <wangyanan55@huawei.com>
This commit is contained in:
Chen Qun 2022-02-10 20:06:01 +08:00 committed by yezengruan
parent 0532b39a49
commit 4cb2eaa9a9
1 changed files with 64 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

64
i386-cache-passthrough-Update-AMD-8000_001D.EAX-25-1.patch Normal file
View File

@ -0,0 +1,64 @@
From 475988057789a1f4dcd7354c8a07fd37dcbac79f Mon Sep 17 00:00:00 2001
From: Yanan Wang <wangyanan55@huawei.com>
Date: Thu, 10 Feb 2022 20:06:01 +0800
Subject: [PATCH] i386: cache passthrough: Update AMD 8000_001D.EAX[25:14]
based on vCPU topo
On AMD target, when host cache passthrough is disabled we will
emulate the guest caches with default values and initialize the
shared cpu list of the caches based on vCPU topology. However
when host cache passthrough is enabled, the shared cpu list is
consistent with host regardless what the vCPU topology is.
For example, when cache passthrough is enabled, running a guest
with vThreads=1 on a host with pThreads=2, we will get that there
are every *two* logical vCPUs sharing a L1/L2 cache, which is not
consistent with the vCPU topology (vThreads=1).
So let's reinitialize BITs[25:14] of AMD CPUID 8000_001D.EAX
based on the actual vCPU topology instead of host pCPU topology.
Signed-off-by: Yanan Wang <wangyanan55@huawei.com>
---
target/i386/cpu.c | 22 ++++++++++++++++++++++
1 file changed, 22 insertions(+)
diff --git a/target/i386/cpu.c b/target/i386/cpu.c
index c1fe2895fd..002e32650d 100644
--- a/target/i386/cpu.c
+++ b/target/i386/cpu.c
@@ -5724,9 +5724,31 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,
}
break;
case 0x8000001D:
+ /* Populate AMD Processor Cache Information */
*eax = 0;
if (cpu->cache_info_passthrough) {
host_cpuid(index, count, eax, ebx, ecx, edx);
+
+ /*
+ * Clear BITs[25:14] and then update them based on the guest
+ * vCPU topology, like what we do in encode_cache_cpuid8000001d
+ * when cache_info_passthrough is not enabled.
+ */
+ *eax &= ~0x03FFC000;
+ switch (count) {
+ case 0: /* L1 dcache info */
+ case 1: /* L1 icache info */
+ case 2: /* L2 cache info */
+ *eax |= ((topo_info.threads_per_core - 1) << 14);
+ break;
+ case 3: /* L3 cache info */
+ *eax |= ((topo_info.cores_per_die *
+ topo_info.threads_per_core - 1) << 14);
+ break;
+ default: /* end of info */
+ *eax = *ebx = *ecx = *edx = 0;
+ break;
+ }
break;
}
switch (count) {
--
2.27.0