Sorry for this really late response due to some personal stuff.
Thanks to Eduardo for clarifying, and yes, this fix is to make the bits 7:0
in Fn8000_001E_EBX be physical core_id per package/socket, instead for what
used to be physical core_id per die.
When topo for vm is like: sockets=1, dies=8, c
CCing the original author of that code (Babu Moger).
On Wed, Jun 30, 2021 at 04:25:51PM +0800, Jade Cheng wrote:
> According to AMD64 Arch Programmer's Manual Appendix D,
> bits 7:0 in Fn8000_001E_EBX should be physical core(s) per logical processor,
> not per die.
Do you mean physical cores per
ping.
On Wed, Jun 30, 2021, 16:26 wrote:
According to AMD64 Arch Programmer's Manual Appendix D, bits 7:0 in
Fn8000_001E_EBX should be physical core(s) per logical processor, not per
die. Signed-off-by: Jade Cheng ---
target/i386/cpu.c | 8 +++- 1 file changed, 7 insertions(+), 1
deletion(-) d
Hello,
I am Jade Cheng working for ByteDance, sending this email is aimed at ping
you guys to check the patch I submitted earlier, link is attached below:
https://patchew.org/QEMU/20210630082551.12956-1-chengjia...@bytedance.com/
Please do me a favor to give it a review, and let me know if you ha
According to AMD64 Arch Programmer's Manual Appendix D,
bits 7:0 in Fn8000_001E_EBX should be physical core(s) per logical processor,
not per die.
Signed-off-by: Jade Cheng
---
target/i386/cpu.c | 8 +++-
1 file changed, 7 insertions(+), 1 deletion(-)
diff --git a/target/i386/cpu.c b/targe
