Re: [edk2-devel] [PATCH v2 4/5] MdeModulePkg/Bus/Pci/PciBusDxe: Fix NULL_RETURNS Coverity issue

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

On 11/14/23 17:53, Ranbir Singh wrote:

> Generally speaking, there now seems to be different views coming from
> you and Laszlo.

Yes.

> We might have to wait for some sort of agreement to be
> reached.

I don't insist on CpuDeadLoop() *specifically*. Only the following two
generic points matter to me:

(1) Stop abusing ASSERT (both because it is compiled out of RELEASE
builds, and because it is conceptually unsuitable for catching data- and
environment-dependent error conditions). ASSERT must only be used for
stating (well, "asserting") algorithmic invariants.

(2) Upon detecting an algorithmic invariant failure, call *some* API
that, at the same time: (2.a) prevents execution from continuing, (2.b)
*cannot* be removed from RELEASE builds, (2.c) informs all static
analysis tools we use that execution cannot continue past that point.

For (2), Mike seems to have an additional requirement: (2.d) make the
implementation customizable by the platform, including any information
shown to, or logged for, the user (or supervisor software).

I have nothing against that additional requirement.

My concern is that "perfect" is going to get in the way of "good enough"
once again.

Laszlo



-=-=-=-=-=-=-=-=-=-=-=-
Groups.io Links: You receive all messages sent to this group.
View/Reply Online (#111248): https://edk2.groups.io/g/devel/message/111248
Mute This Topic: https://groups.io/mt/102438320/7686176
Group Owner: devel+owner@edk2.groups.io
Unsubscribe: https://edk2.groups.io/g/devel/leave/12367111/7686176/1913456212/xyzzy [rebecca@openfw.io]
-=-=-=-=-=-=-=-=-=-=-=-