[External] QEMU models for new BMC chips

[Sumeet Kochar]

Community came together and drove vendors to deliver drivers to device tree bindings. Most of us who are part of this community purchase some BMC solutions in one form or the other and a large group in here has vested interest in deploying a stack that is vetted out through CI as best it can be. Solving it as a general requirement is more powerful in my mind than some of us solving on the side with a vendor. Vendors are part of this community so would love their feedback. I see mutual benefit. In long term as OpenBMC improves the CI model to me it would make more sense that code/drivers submitted for review have atleast gone through the CI flow without issues vs. the community running the CI and/or during manual reviews finding  issues late which could be discovered earlier through automation ? We certainly will contribute what we can individually and upstream what is ok to share publically respecting vendor IP on a new chip but to me that is catching the problem late. Doing what we can to avoid finding things late by making initial checklist more automated and effective can only help save limited cycles we all have. I'll leave it at that and let the community decide. Thanks again for response. 

Date: Wed, 01 May 2019 00:32:18 -0400
From: "Andrew Jeffery" <andrew at aj.id.au>
To: "Sumeet Kochar" <skochar at lenovo.com>, "openbmc at lists.ozlabs.org"
	<openbmc at lists.ozlabs.org>
Subject: Re: [External]  openbmc Digest, Vol 44, Issue 120
Message-ID: <52d67ef7-a7a4-4a5f-8af0-3a5cda3e4130 at www.fastmail.com>
Content-Type: text/plain

Hi Sumeet,

On Wed, 1 May 2019, at 10:56, Sumeet Kochar wrote:
> I want to seek input from the community on QEMU support from BMC chip 
> vendors for their new BMC chips.

I agree it would be great to have the chip vendors on board. I think some have seen the utility of qemu, but ultimately it's up to them to either act on it for their own purposes or for their customers to start pressuring for this as part of their purchase arrangement.

We can start advocating for it as a community, but given the community isn't the one purchasing the chips we don't have a lot of sway. As such from a community perspective it's largely still a scratch-your-own-itch problem.

As such if you want to see improved qemu support please don't hesitate to start contributing! Just like the kernel, we want the patches upstream, so send them there first and we can backport as required.

>> 

Cheers,

Andrew