Date
1 - 3 of 3
[PATCH v3] IntelFsp2Pkg: LoadMicrocodeDefault() causing unnecessary delay.
|
Chiu, Chasel
REF: https://bugzilla.tianocore.org/show_bug.cgi?id=3D4391
FSP should support the scenario that CPU microcode already loaded before calling LoadMicrocodeDefault(), in this case it should return directly without spending more time. Also the LoadMicrocodeDefault() should only attempt to load one version of the microcode for current CPU and return directly without parsing rest of the microcode in FV. Cc: Nate DeSimone <nathaniel.l.desimone@...> Cc: Star Zeng <star.zeng@...> Cc: Ray Ni <ray.ni@...> Signed-off-by: Chasel Chiu <chasel.chiu@...> --- IntelFsp2Pkg/FspSecCore/Ia32/FspApiEntryT.nasm | 26 ++++++++++++++++++++++= ---- IntelFsp2Pkg/FspSecCore/X64/FspApiEntryT.nasm | 26 ++++++++++++++++++++++= ---- 2 files changed, 44 insertions(+), 8 deletions(-) diff --git a/IntelFsp2Pkg/FspSecCore/Ia32/FspApiEntryT.nasm b/IntelFsp2Pkg/= FspSecCore/Ia32/FspApiEntryT.nasm index 2cff8b3643..79f2a20a2c 100644 --- a/IntelFsp2Pkg/FspSecCore/Ia32/FspApiEntryT.nasm +++ b/IntelFsp2Pkg/FspSecCore/Ia32/FspApiEntryT.nasm @@ -245,6 +245,22 @@ ASM_PFX(LoadMicrocodeDefault): cmp esp, 0=0D jz ParamError=0D =0D + ;=0D + ; If microcode already loaded before this function, exit this function = with SUCCESS.=0D + ;=0D + mov ecx, MSR_IA32_BIOS_SIGN_ID=0D + xor eax, eax ; Clear EAX=0D + xor edx, edx ; Clear EDX=0D + wrmsr ; Load 0 to MSR at 8Bh=0D +=0D + mov eax, 1=0D + cpuid=0D + mov ecx, MSR_IA32_BIOS_SIGN_ID=0D + rdmsr ; Get current microcode signature=0D + xor eax, eax=0D + test edx, edx=0D + jnz Exit2=0D +=0D ; skip loading Microcode if the MicrocodeCodeSize is zero=0D ; and report error if size is less than 2k=0D ; first check UPD header revision=0D @@ -450,7 +466,7 @@ LoadCheck: =0D ; Verify this microcode update is not already loaded=0D cmp dword [esi + MicrocodeHdr.MicrocodeHdrRevision], edx=0D - je Continue=0D + je Done ; if already one version microcode loaded, go to done=0D =0D LoadMicrocode:=0D ; EAX contains the linear address of the start of the Update Data=0D @@ -465,10 +481,12 @@ LoadMicrocode: mov eax, 1=0D cpuid=0D =0D -Continue:=0D - jmp NextMicrocode=0D -=0D Done:=0D + mov ecx, MSR_IA32_BIOS_SIGN_ID=0D + xor eax, eax ; Clear EAX=0D + xor edx, edx ; Clear EDX=0D + wrmsr ; Load 0 to MSR at 8Bh=0D +=0D mov eax, 1=0D cpuid=0D mov ecx, MSR_IA32_BIOS_SIGN_ID=0D diff --git a/IntelFsp2Pkg/FspSecCore/X64/FspApiEntryT.nasm b/IntelFsp2Pkg/F= spSecCore/X64/FspApiEntryT.nasm index b32fa32a89..3e40678f47 100644 --- a/IntelFsp2Pkg/FspSecCore/X64/FspApiEntryT.nasm +++ b/IntelFsp2Pkg/FspSecCore/X64/FspApiEntryT.nasm @@ -141,6 +141,22 @@ ASM_PFX(LoadMicrocodeDefault): jz ParamError=0D mov rsp, rcx=0D =0D + ;=0D + ; If microcode already loaded before this function, exit this function = with SUCCESS.=0D + ;=0D + mov ecx, MSR_IA32_BIOS_SIGN_ID=0D + xor eax, eax ; Clear EAX=0D + xor edx, edx ; Clear EDX=0D + wrmsr ; Load 0 to MSR at 8Bh=0D +=0D + mov eax, 1=0D + cpuid=0D + mov ecx, MSR_IA32_BIOS_SIGN_ID=0D + rdmsr ; Get current microcode signature=0D + xor rax, rax=0D + test edx, edx=0D + jnz Exit2=0D +=0D ; skip loading Microcode if the MicrocodeCodeSize is zero=0D ; and report error if size is less than 2k=0D ; first check UPD header revision=0D @@ -291,7 +307,7 @@ LoadCheck: =0D ; Verify this microcode update is not already loaded=0D cmp dword [esi + MicrocodeHdr.MicrocodeHdrRevision], edx=0D - je Continue=0D + je Done ; if already one version microcode loaded, go to done=0D =0D LoadMicrocode:=0D ; EAX contains the linear address of the start of the Update Data=0D @@ -306,10 +322,12 @@ LoadMicrocode: mov eax, 1=0D cpuid=0D =0D -Continue:=0D - jmp NextMicrocode=0D -=0D Done:=0D + mov ecx, MSR_IA32_BIOS_SIGN_ID=0D + xor eax, eax ; Clear EAX=0D + xor edx, edx ; Clear EDX=0D + wrmsr ; Load 0 to MSR at 8Bh=0D +=0D mov eax, 1=0D cpuid=0D mov ecx, MSR_IA32_BIOS_SIGN_ID=0D --=20 2.35.0.windows.1 |
|
|
|
Ni, Ray
Chasel,
toggle quoted message
Show quoted text
With your changes, the flow is like: 1. check revision of loaded microcode, go to Done if it's not zero 2. find first matching microcode 3. check revision of loaded microcode, go to Done if it equals to the matching one 4. load the matching microcode Done: return fail/success depending on whether the revision of loaded microcode is zero I guess the step #3 is unnecessary because step #1 guarantees that step #3 would not go to Done. Can you please confirm? -----Original Message----- |
|
|
|
Chiu, Chasel
Hi Ray,
toggle quoted message
Show quoted text
Yes, the step 3 will be redundant after adding the check for microcode already loaded scenario in earlier point. I will remove it and sent V4 patch. Thanks, Chasel -----Original Message----- |
|
|