Re: [edk2-devel] [PATCH] UefiCpuPkg: Adding a new TSC library by using CPUID(0x15) TSC leaf

["Laszlo Ersek" <lersek@redhat.com>]

On 08/15/19 06:02, Gao, Liming wrote:
> Donald: This change is a new feature. Now, it is not in edk2 feature
> planning list. If you want to catch it into 201908 stable tag, please
> get approve from Stewards first. I have cc this mail to all Stewards.
- I don't mind adding a new feature, as long as it gets properly
reviewed by package owners before we enter the soft feature freeze.

- Looking at the BZ
<https://bugzilla.tianocore.org/show_bug.cgi?id=1909>, a bit more
documentation would be nice.

- On the negative side, I'm very much *not* a fan of adding features to
the open source edk2 tree without actually *consuming* the feature in an
open source tree. Are the new library instances going to be put to use
in edk2-platforms, perhaps?

We discussed this topic earlier on some of the stewards' calls. On one
hand, it's not uncommon to see library instances from Intel enter core
edk2 packages without any dependent platform code, or even a detailed
problem statement / purpose description (see e.g. commit 5c9bb86f171c
and its surrounding commits). On the other hand, attempts in the past,
to add libraries with well demonstrated and direct in-tree use cases, to
edk2 core, have been rejected, from other submitters. (Here's one
example: <https://bugzilla.tianocore.org/show_bug.cgi?id=957>.) I'm not
prying at proprietary platform information, but a new library added to
edk2 core *should* be well-justified.

The commit message on this patch is empty. It only references
<https://bugzilla.tianocore.org/show_bug.cgi?id=1909>. And if I open the
BZ, this is all I get:

    Need a new TSC library to check the CPUID leaf (EAX=0x15) for TSC.
    For new platform (start from SKL) can use CPUID and retire/remove
    the current override from AcpiTimerLib.

Does this read like an actual feature request? (TimerLib is an MdePkg
library class, so not exactly "niche".)

Plus, the BZ isn't even marked as a feature request, to begin with (the
Product field is wrong, it says "EDK2", which is for bugs, not
features). Also, the patch is on the list, and the status is still
CONFIRMED, and not IN_PROGRESS.

Why are we making a joke of bug tracking? The purpose of a public
bugzilla instance on the web is not just more red tape, not just to have
one more administrative tool that we can abuse, ignore, and write off
with make-believe actions.

It's not like another issue tracker system (JIRA, Launchpad, GitHub,
GitLab, you name it) would properly work with this level of neglect, either.

In summary: I don't mind the feature, but the documentation around it
should be *much* better.

Thanks
Laszlo