* [PATCH 1/3] i386/cpu: rename feature name avx10-vnni-int to avx10-v1-aux
2026-09-30 4:52 [PATCH 0/3] i386/cpu: miscellaneous fixes for DiamondRapids Zhao Liu
@ 2026-09-30 4:52 ` Zhao Liu
2026-09-30 4:52 ` [PATCH 2/3] i386/cpu: rename CPUID_24_1_ECX_AVX10_VNNI_INT macro to CPUID_24_1_ECX_AVX10_V1_AUX Zhao Liu
` (2 subsequent siblings)
3 siblings, 0 replies; 5+ messages in thread
From: Zhao Liu @ 2026-09-30 4:52 UTC (permalink / raw)
To: Paolo Bonzini, Philippe Mathieu-Daudé; +Cc: qemu-devel, Zhao Liu
Since the AVX10.2 Spec rev. 6.0, CPUID feature bit AVX10_VNNI_INT is
renamed to AVX10_V1_AUX to cover additional more convert instructions:
VCVT[,2]PH2[B,H]F8[,S], VCVT2PS2PHX,VCVTBIASPH2[B,H]F8[,S], VCVTHF82PH.
Therefore, rename the feature name to avx10-v1-aux and use the old name
avx10-vnni-int as an alias for compatibility.
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
---
target/i386/cpu.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/target/i386/cpu.c b/target/i386/cpu.c
index f3bc6911c10f..e1c012cf2dfa 100644
--- a/target/i386/cpu.c
+++ b/target/i386/cpu.c
@@ -1370,7 +1370,7 @@ FeatureWordInfo feature_word_info[FEATURE_WORDS] = {
[FEAT_24_1_ECX] = {
.type = CPUID_FEATURE_WORD,
.feat_names = {
- [2] = "avx10-vnni-int",
+ [2] = "avx10-v1-aux",
},
.cpuid = {
.eax = 0x24,
@@ -10556,6 +10556,7 @@ static void x86_cpu_initfn(Object *obj)
object_property_add_alias(obj, "hv-apicv", obj, "hv-avic");
cpu->lbr_fmt = ~PERF_CAP_LBR_FMT;
object_property_add_alias(obj, "lbr_fmt", obj, "lbr-fmt");
+ object_property_add_alias(obj, "avx10-vnni-int", obj, "avx10-v1-aux");
if (xcc->model) {
x86_cpu_load_model(cpu, xcc->model);
--
2.34.1
^ permalink raw reply related [flat|nested] 5+ messages in thread* [PATCH 2/3] i386/cpu: rename CPUID_24_1_ECX_AVX10_VNNI_INT macro to CPUID_24_1_ECX_AVX10_V1_AUX
2026-09-30 4:52 [PATCH 0/3] i386/cpu: miscellaneous fixes for DiamondRapids Zhao Liu
2026-09-30 4:52 ` [PATCH 1/3] i386/cpu: rename feature name avx10-vnni-int to avx10-v1-aux Zhao Liu
@ 2026-09-30 4:52 ` Zhao Liu
2026-09-30 4:52 ` [PATCH 3/3] hw/core/machine-smp: allow thread level cache Zhao Liu
2026-11-20 10:01 ` [PATCH 0/3] i386/cpu: miscellaneous fixes for DiamondRapids Zhao Liu
3 siblings, 0 replies; 5+ messages in thread
From: Zhao Liu @ 2026-09-30 4:52 UTC (permalink / raw)
To: Paolo Bonzini, Philippe Mathieu-Daudé; +Cc: qemu-devel, Zhao Liu
Use CPUID_24_1_ECX_AVX10_V1_AUX macro to align with the official feature
name in AVX10.2 (since AVX10.2 spec rev 6.0).
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
---
target/i386/cpu.c | 4 ++--
target/i386/cpu.h | 4 ++--
2 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/target/i386/cpu.c b/target/i386/cpu.c
index e1c012cf2dfa..6f27c2e7b00b 100644
--- a/target/i386/cpu.c
+++ b/target/i386/cpu.c
@@ -2395,7 +2395,7 @@ static const AVX10VersionDefinition builtin_avx10_defs[] = {
.version = 2,
.max_subleaf = 1,
.features = (FeatureMask[]) {
- { FEAT_24_1_ECX, CPUID_24_1_ECX_AVX10_VNNI_INT },
+ { FEAT_24_1_ECX, CPUID_24_1_ECX_AVX10_V1_AUX },
{ /* end of list */ }
}
},
@@ -5733,7 +5733,7 @@ static const X86CPUDefinition builtin_x86_defs[] = {
* Though this bit will be set by avx_version=2, it's better to
* explicitly enumerate this feature here.
*/
- .features[FEAT_24_1_ECX] = CPUID_24_1_ECX_AVX10_VNNI_INT,
+ .features[FEAT_24_1_ECX] = CPUID_24_1_ECX_AVX10_V1_AUX,
.features[FEAT_8000_0001_ECX] =
CPUID_EXT3_LAHF_LM | CPUID_EXT3_ABM | CPUID_EXT3_3DNOWPREFETCH,
.features[FEAT_8000_0001_EDX] =
diff --git a/target/i386/cpu.h b/target/i386/cpu.h
index 786957fadd08..771673d0902a 100644
--- a/target/i386/cpu.h
+++ b/target/i386/cpu.h
@@ -1136,8 +1136,8 @@ uint64_t x86_cpu_get_supported_feature_word(X86CPU *cpu, FeatureWord w);
CPUID_24_0_EBX_AVX10_256 | \
CPUID_24_0_EBX_AVX10_512)
-/* AVX10_VNNI_INT instruction */
-#define CPUID_24_1_ECX_AVX10_VNNI_INT (1U << 2)
+/* AVX10_V1_AUX instruction */
+#define CPUID_24_1_ECX_AVX10_V1_AUX (1U << 2)
/*
* New Conditional Instructions (NCIs), explicit New Data Destination (NDD)
--
2.34.1
^ permalink raw reply related [flat|nested] 5+ messages in thread* [PATCH 3/3] hw/core/machine-smp: allow thread level cache
2026-09-30 4:52 [PATCH 0/3] i386/cpu: miscellaneous fixes for DiamondRapids Zhao Liu
2026-09-30 4:52 ` [PATCH 1/3] i386/cpu: rename feature name avx10-vnni-int to avx10-v1-aux Zhao Liu
2026-09-30 4:52 ` [PATCH 2/3] i386/cpu: rename CPUID_24_1_ECX_AVX10_VNNI_INT macro to CPUID_24_1_ECX_AVX10_V1_AUX Zhao Liu
@ 2026-09-30 4:52 ` Zhao Liu
2026-11-20 10:01 ` [PATCH 0/3] i386/cpu: miscellaneous fixes for DiamondRapids Zhao Liu
3 siblings, 0 replies; 5+ messages in thread
From: Zhao Liu @ 2026-09-30 4:52 UTC (permalink / raw)
To: Paolo Bonzini, Philippe Mathieu-Daudé
Cc: qemu-devel, Zhao Liu, Badari Pulavarty
Commit 87ae7ab84fff allowed i386 CPU to configure thread level cache,
However, without support for the cache model, users still need to
configure it through -machine smp-cache.
Remove the thread-level restriction in the smp-cache option.
Reported-by: Badari Pulavarty <badari.pulavarty@intel.com>
Fixes: commit 87ae7ab84fff ("i386/cpu: Allow cache to be shared at thread level")
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
---
hw/core/machine-smp.c | 7 -------
1 file changed, 7 deletions(-)
diff --git a/hw/core/machine-smp.c b/hw/core/machine-smp.c
index bef04aa2d713..baf06bca2aeb 100644
--- a/hw/core/machine-smp.c
+++ b/hw/core/machine-smp.c
@@ -321,13 +321,6 @@ bool machine_parse_smp_cache(MachineState *ms,
return false;
}
- if (props->topology == CPU_TOPOLOGY_LEVEL_THREAD) {
- error_setg(errp,
- "%s level cache not supported by this machine",
- CpuTopologyLevel_str(props->topology));
- return false;
- }
-
if (!machine_check_topo_support(ms, props->topology, errp)) {
return false;
}
--
2.34.1
^ permalink raw reply related [flat|nested] 5+ messages in thread* Re: [PATCH 0/3] i386/cpu: miscellaneous fixes for DiamondRapids
2026-09-30 4:52 [PATCH 0/3] i386/cpu: miscellaneous fixes for DiamondRapids Zhao Liu
` (2 preceding siblings ...)
2026-09-30 4:52 ` [PATCH 3/3] hw/core/machine-smp: allow thread level cache Zhao Liu
@ 2026-11-20 10:01 ` Zhao Liu
3 siblings, 0 replies; 5+ messages in thread
From: Zhao Liu @ 2026-11-20 10:01 UTC (permalink / raw)
To: Paolo Bonzini, Philippe Mathieu-Daudé, Zhao Liu; +Cc: qemu-devel
Hi maintainers,
Just a gentle ping.
Regards,
Zhao
On Wed, Sep 30, 2026 at 12:52:34PM +0800, Zhao Liu wrote:
> Date: Wed, 30 Sep 2026 12:52:34 +0800
> From: Zhao Liu <zhao1.liu@intel.com>
> Subject: [PATCH 0/3] i386/cpu: miscellaneous fixes for DiamondRapids
> X-Mailer: git-send-email 2.34.1
>
> Hi,
>
> This series includes:
> * AVX10.2 feature renaming to keep up with the latest spec changes.
> * A fix for DMR cache topology emulation.
>
> In addition, there is another TF32 fix remaining in the mail list,
> which is a change in hardware implementation:
>
> https://lore.kernel.org/qemu-devel/20260720205227.519647-1-chang.seok.bae@intel.com/
>
> Thanks and Best Regards,
> Zhao
> ---
> Zhao Liu (3):
> i386/cpu: rename feature name avx10-vnni-int to avx10-v1-aux
> i386/cpu: rename CPUID_24_1_ECX_AVX10_VNNI_INT macro to
> CPUID_24_1_ECX_AVX10_V1_AUX
> hw/core/machine-smp: allow thread level cache
>
> hw/core/machine-smp.c | 7 -------
> target/i386/cpu.c | 7 ++++---
> target/i386/cpu.h | 4 ++--
> 3 files changed, 6 insertions(+), 12 deletions(-)
>
> --
> 2.34.1
>
^ permalink raw reply [flat|nested] 5+ messages in thread