That’s right I have also observed this, on Debug IFWI firmware it runs very slowly.

It could be due to generation of lot of prints from the firmware as it is a debug build.

So, it is expected to be very slow.

But on a release build of the firmware it works pretty quick and we don’t notice the slowness as observed with the debug firmware.

One option is to use the pre-compiled library files which are compiled to .pyc files.

So that Python interpreter boot strap process can run pretty fast.

Regards,

From: devel@edk2.groups.io <devel@edk2.groups.io> On Behalf Of Yoshinoya via groups.io
Sent: Thursday, May 15, 2025 7:39 AM
To: devel@edk2.groups.io; Jayaprakash, N <n.jayaprakash@intel.com>
Cc: Kinney, Michael D <michael.d.kinney@intel.com>; pedro.falcato@gmail.com
Subject: Re: [edk2-devel] Python-3.6.8 UEFI : fork support

Hi, JP

I found an interesting test result.

I ran shell python on qemu with debug version ovmf fd.

I executed "python" command in shell environment, it ran slowly in the first time.

then i executed "quit(0" command in python interactive environ, and launched python again in shell environment.

And it ran more slower than the first time.

If i do the same test on quem with release version ovmf fd, and it's normal.

every time, i ran python in shell, and it ran quickly.

I tried to run "python -v', and saved the output log, not find any obvious exception.

Is there any suggestion?

best wishes,

在 2025-02-18 13:07:57，"Yoshinoya" <yoshinoyatoko@163.com> 写道：

Hi, JP,

Thanks a lot!

best wishes,

At 2025-02-17 23:51:25, "Jayaprakash, N via groups.io" <n.jayaprakash=intel.com@groups.io> wrote:

Hi Yoshinoya,

The MP Services UEFI protocol has been used to run the CPUID instructions on a given CPU core and to read / write the MSRs from given CPU core.

The python functions for this purpose are supported from the edk2module built in extension module are respectively cpuid_ex, rdmsr_ex/ wrmsr_ex with the function signatures as given below.

cupid_ex has the following signature cpuid_ex(cpu, eax, ecx) -> (eax:ebx:ecx:edx)

rdmsr_ex(cpu, msr) -> (lower_32bits, higher_32bits)

wrmsr_ex(cpu, msr, lower_32bits, higher_32bits) -> None

where the cpu parameter refers to the cpu core number starting from 0, 0 being BSP.

The usage instructions and code sample are not yet updated in the wiki. I can add more details at the earliest.

Please use the edk2-libc discussions page for any further questions / discussions.

https://github.com/tianocore/edk2-libc/discussions

Regards,

JP

From: Yoshinoya <yoshinoyatoko@163.com>
Sent: Monday, February 17, 2025 4:43 PM
To: devel@edk2.groups.io; Jayaprakash, N <n.jayaprakash@intel.com>
Cc: Kinney, Michael D <michael.d.kinney@intel.com>; pedro.falcato@gmail.com
Subject: Re:Re: [edk2-devel] Python-3.6.8 UEFI : fork support

hi, JP

Got it, thank you!

I found some uefi mp service functions's call sample in edk2module.c

So, user could follow this sample and write multi process sample with python script ?

best wishes,

At 2025-02-17 13:23:58, "Jayaprakash, N via groups.io" <n.jayaprakash=intel.com@groups.io> wrote:

Hi Yoshinoya,

It’s not supported in UEFI.

You can see that macros are not defined as part of pyconfig.h

https://github.com/tianocore/edk2-libc/blob/master/AppPkg/Applications/Python/Python-3.6.8/PyMod-3.6.8/Include/pyconfig.h

So this functionality doesn’t get compiled for UEFI.

Regards,

JP

From: Yoshinoya <yoshinoyatoko@163.com>
Sent: Monday, February 17, 2025 8:19 AM
To: Kinney, Michael D <michael.d.kinney@intel.com>; pedro.falcato@gmail.com; devel@edk2.groups.io; Jayaprakash, N <n.jayaprakash@intel.com>
Subject: [edk2-devel] Python-3.6.8 UEFI : fork support

Hi, Michael

I am studying Python368 uefi code.

I have a question, ask for help.

I found some marco defines in edk2module.c

"HAVE_FORK / HAVE_FORK1 / HAVE_SPAWNV / etc"

It seems about multi-thread support functions,

so my question is:

1. Have these functions been tested on uefi shell environments?

2. Are there any code samples for study how to use them?

Thank you very much.

best wishes,