Hi everyone,
My name is Nick Bowes and I'm looking to create a small (got to start somewhere) Foss-only computer build and supply business. To make that viable, I need a free bios implementation for one or two new mobos, and coreboot seems to hold the most promise. I'm interested in doing this in the hope that I can help the foss community get away from the emerging and likely pervasive UEFI stupidness thats coming. As you probably know better then I do, creating alternatives to the things that cause us all problems is the only way to truely solve them.
Having heard recently that Google released coreboot updates for both Sandy Bridge and Ivy Bridge chipsets, I thought now would be a good time to ask this mailing list a few questions about the viability of funding some development.
As a future small distributor of foss-only computers and likely a one man band to start with, I would need to keep things very simple. That means creating a product line (if you can call it that) of perhaps only 3-5 different desktops/laptops in total. Idealy I'd like to offer a range of 3 itx desktops, Ivy Bridge i3 i5 i7, or, Trinity A6 A8 A10, and one or two laptops, one budget, one highend.
So, if you don't mind, and this is the right place(if it isn't, please point me in the right direction) I'd like to ask you all...
1. What do you all think the requirements will be for replacing UEFI on future mobos. Will it likely be a case of switching out the bios chip or just flashing it?
2. What % of flashes resulted in bricked mobos? Do most new boards come with a backup chip which can restore life after a failure?
3. Can anyone reccomend a possible target ivy bridge or Trinity itx mobo for coreboot development? Perhaps one that's already being worked on?
4. If we can identify a good itx mobo for the desktop line-up, would someone here be able to asses how much time a fairly full featured and reliable implementation of coreboot would take to develop?
note: to me "fairly full featured would be to have all the fundamentals up and avaliable for the OS to pick up, such as pci-e usb 3.0, hdmi, sata 3, wifi, and working reliably.
5. Is there any special requirement for getting amd/nvidia gfx cards working with coreboot?
6. Which hardware have you found to be the simplist to fully implement coreboot on, and which hardware (if any) should be considered a no go?
7. What type of computer would most on this list be interested to develop coreboot for. As in, if you were going to buy a desktop or laptop today, or in the near future, what combination of hardware would you go for?
8. What do you think about the viability of a kickstarter campaign to raise development funds. Has anyone tried this yet?
I would truely love to distribute computers which run entirely on free software. Unfortunately, I'm no developer so If funding some development is the only way I can achieve this, I will find a way to do it.
I look forward to any insights you provide.
Thanks Nick
Am 04.06.2012 14:42, schrieb mailinglists:
What do you all think the requirements will be for replacing UEFI onfuture mobos. Will it likely be a case of switching out the bios chip or just flashing it?
We generally reuse the chips. If you're doing business, you should be able to afford an external flasher, so getting rid of the original BIOS is no problem, even if vendor BIOS locks down the chip on boot.
Some chip types are easier to flash externally than others, but that's from a hobbyist perspective. As a business, investing some money in the right adapter for a large number of boards isn't too bad, it's just hard to justify spending $50 (to pick some upper bound) on a single-use item when it's just a hobby.
What % of flashes resulted in bricked mobos? Do most new boards comewith a backup chip which can restore life after a failure?
With external flashing you can recover by just writing again. "bricked" means wrong content and thus unable to boot, not destroyed chip (usually - I've seen the latter case, but it's very rare).
Can anyone reccomend a possible target ivy bridge or Trinity itx mobofor coreboot development? Perhaps one that's already being worked on?
As yet, Trinity is only released for notebooks, at least that's what I gathered from the tech media. Desktop and server Trinity are scheduled for summer and fall releases (AFAIK).
As for Ivybridge, please note that it requires a couple of binary-only components (beyond those we usually need, see below): RAM init is done with the Intel reference code, and to turn on the system in the first place, the Management Engine (some embedded controller in the chipset) requires a binary-only component as well.
It's remotely feasible to replace the RAM init with source (multimonth effort, after obtaining access to the documentation), but the ME code will remain a requirement (if only because it's said to be signed by Intel, so replacing it requires cracking their signature scheme - good luck).
If we can identify a good itx mobo for the desktop line-up, wouldsomeone here be able to asses how much time a fairly full featured and reliable implementation of coreboot would take to develop?
note: to me "fairly full featured would be to have all thefundamentals up and avaliable for the OS to pick up, such as pci-e usb 3.0, hdmi, sata 3, wifi, and working reliably.
This depends (among other things) on the kind of OS support you desire. OSS systems are generally more forgiving for incomplete configuration than Windows - but with "FOSS-only computers" Windows might not be a priority.
Is there any special requirement for getting amd/nvidia gfx cardsworking with coreboot?
Minor hacks might be necessary to get the IGD/GFX switch to work right. On systems with integrated graphics we generally expect IGD to manage the primary display. Details on that vary by chipset and/or vendor.
Which hardware have you found to be the simplist to fully implementcoreboot on, and which hardware (if any) should be considered a no go?
nVidia chipsets are no-go (there was a lucky strike once that gave us the nVidia support we have - I don't expect that to happen again). Intel chipsets are complicated. AMD is the best choice for coreboot support these days.
We don't have to care about many other things (eg. the Wifi card - we don't handle Wifi, we just have to make sure the card is found on the bus)
- What do you think about the viability of a kickstarter campaign to raise
development funds. Has anyone tried this yet?
Last I heard Kickstarter has a backlog of several thousand projects.
Kickstarter seems to be first and foremost a social media popularity contest at the moment, for which coreboot is probably not a fancy enough topic. I'd assume the same for "FOSS Hardware", but feel free to try, if you feel like it.
Make clear that you're not collecting funds for the coreboot project - our situation is that we can't accept donations of any kind (nor take money with any other designation).
I would truely love to distribute computers which run entirely on free software. Unfortunately, I'm no developer so If funding some development is the only way I can achieve this, I will find a way to do it.
"run entirely on free software": We usually reuse the VGABIOS image delivered by the hardware vendor, which comes in binary-only form. There are a couple of ways around it, but they're only appropriate for systems where you control the OS that's used (eg. not doing VGA init at all and defer things to Linux KMS), or they're still experimental (see i915tool - it's an attempt to create a coreboot-level driver from KMS sources).
This means that for a "generic" box you're still bound to the VGABIOS to some degree (you can't rely on KMS being available in the OS when using a bootloader menu).
The other aspect is that you generally have an embedded controller on notebooks (in addition to the management engine on newer Intel chipsets). It usually runs its own firmware, which is also not availabe in source form. If you're lucky, it's part of the EC chip as read-only memory - in this case, you can safely consider it part of the hardware (even under FSF's strict definition, as far as I understand it).
Regards, Patrick
Hi Patrick,
Sorry, I was hopping to have the time to reply sooner.
On Monday 04 Jun 2012 16:12:50 Patrick Georgi wrote:
Am 04.06.2012 14:42, schrieb mailinglists:
What do you all think the requirements will be for replacing UEFIon future mobos. Will it likely be a case of switching out the bios chip or just flashing it?
We generally reuse the chips. If you're doing business, you should be able to afford an external flasher, so getting rid of the original BIOS is no problem, even if vendor BIOS locks down the chip on boot.
Some chip types are easier to flash externally than others, but that's from a hobbyist perspective. As a business, investing some money in the right adapter for a large number of boards isn't too bad, it's just hard to justify spending $50 (to pick some upper bound) on a single-use item when it's just a hobby.
Unfortunately, as you might have guessed, I’m fairly technical but my experience is at the OS level. I found this
http://www.ebay.com/itm/SP8-A-Universal-EEPROM-Flash-SPI-BIOS-24-25-BR90-93- USB-Programmer-4000- CHIPS-/180852019132?pt=LH_DefaultDomain_0&hash=item2a1b9ecfbc#ht_31613wt_1163
I'm guessing that's the type of equipment you're referring to? no problem to invest in one or a few of those.
What % of flashes resulted in bricked mobos? Do most new boardscome with a backup chip which can restore life after a failure?
With external flashing you can recover by just writing again. "bricked" means wrong content and thus unable to boot, not destroyed chip (usually
- I've seen the latter case, but it's very rare).
Excellent.
Can anyone recommend a possible target ivy bridge or Trinity itxmobo for coreboot development? Perhaps one that's already being worked on?
As yet, Trinity is only released for notebooks, at least that's what I gathered from the tech media. Desktop and server Trinity are scheduled for summer and fall releases (AFAIK).
Unfortunately it seems you're right, I must not have paid close enough attention. Though, I did find the link below. The name of the site doesn't inspire confidence however;)
At the bottom of the article "Updated: June 4th, 2012 9am. AMD has responded with the following. So we’re confirming that we’ll have OEM Trinity desktops on shelves this week."
http://semiaccurate.com/2012/06/01/amd-delays-desktop-trinity-one-quarter/
As for Ivybridge, please note that it requires a couple of binary-only components (beyond those we usually need, see below): RAM init is done with the Intel reference code, and to turn on the system in the first place, the Management Engine (some embedded controller in the chipset) requires a binary-only component as well.
It's remotely feasible to replace the RAM init with source (multimonth effort, after obtaining access to the documentation), but the ME code will remain a requirement (if only because it's said to be signed by Intel, so replacing it requires cracking their signature scheme - good luck).
If there is no hope of replacing these bits then a FOSS-almost computer will suffice. Thankfully, the only real concern at the moment is to sidestep Microsoft's influence on OEMs. You guys have already done a fantastic job of that, we just need to get that work onto some hardware that people will be buying in future. If we can do that, it will inspire others to follow suit.
If we can identify a good itx mobo for the desktop line-up, wouldsomeone here be able to asses how much time a fairly full featured and reliable implementation of coreboot would take to develop?
note: to me "fairly full featured would be to have all thefundamentals up and avaliable for the OS to pick up, such as pci-e usb 3.0, hdmi, sata 3, wifi, and working reliably.
This depends (among other things) on the kind of OS support you desire. OSS systems are generally more forgiving for incomplete configuration than Windows - but with "FOSS-only computers" Windows might not be a priority.
Windows is definitely not a priority. In an ideal world, I’d be looking to offer Linux distributions without binary blobs, but that is probably out of the question for now. Never mind though, getting close is a very good first step.
Is there any special requirement for getting AMD/nvidia gfx cardsworking with coreboot?
Minor hacks might be necessary to get the IGD/GFX switch to work right. On systems with integrated graphics we generally expect IGD to manage the primary display. Details on that vary by chipset and/or vendor.
Sounds promising.
Which hardware have you found to be the simplist to fullyimplement coreboot on, and which hardware (if any) should be considered a no go?
nVidia chipsets are no-go (there was a lucky strike once that gave us the nVidia support we have - I don't expect that to happen again). Intel chipsets are complicated. AMD is the best choice for coreboot support these days.
Ok, AMD it is then. I only hope the semiaccurate.com article above has some merit. FM1 chips don't really compete with the new intel offerings. I'll contact someone at AMD to get a status on the trinity chips.
Are Intel chipsets complicated due to lack of documentation? If so, can I contact/badger the right people on your behalf?
We don't have to care about many other things (eg. the Wifi card - we don't handle Wifi, we just have to make sure the card is found on the bus)
Good to know, thank you.
- What do you think about the viability of a kickstarter campaign to
raise development funds. Has anyone tried this yet?
Last I heard Kickstarter has a backlog of several thousand projects.
Kickstarter seems to be first and foremost a social media popularity contest at the moment, for which coreboot is probably not a fancy enough topic. I'd assume the same for "FOSS Hardware", but feel free to try, if you feel like it.
Yes it does seem to be like that. Though If I can make the case for UEFI-less computers compelling enough, and market it through foss type news and review sites, a crowd sourced project, what ever site it is posted on, might just capture enough of the right audience to fund the initial requirements. I can but try...
Make clear that you're not collecting funds for the coreboot project - our situation is that we can't accept donations of any kind (nor take money with any other designation).
No problem, it would of course be deceptive of me to use a name I'm not affiliated with.
I would imagine the best way to fund development (initially) would be to put the chosen mobo into interested developers hands (Acting on their own behalf of course).
I would truely love to distribute computers which run entirely on free software. Unfortunately, I'm no developer so If funding some development is the only way I can achieve this, I will find a way to do it.
"run entirely on free software": We usually reuse the VGABIOS image delivered by the hardware vendor, which comes in binary-only form. There are a couple of ways around it, but they're only appropriate for systems where you control the OS that's used (eg. not doing VGA init at all and defer things to Linux KMS), or they're still experimental (see i915tool - it's an attempt to create a coreboot-level driver from KMS sources).
This means that for a "generic" box you're still bound to the VGABIOS to some degree (you can't rely on KMS being available in the OS when using a bootloader menu).
The other aspect is that you generally have an embedded controller on notebooks (in addition to the management engine on newer Intel chipsets). It usually runs its own firmware, which is also not availabe in source form. If you're lucky, it's part of the EC chip as read-only memory - in this case, you can safely consider it part of the hardware (even under FSF's strict definition, as far as I understand it).
As long as implementing coreboot removes anything to do with secure boot, I and most others concerned about it will remain fairly satisfied.
My ultimate aim will remain a FOSS-only computer business, but the timeframe requires some sacrifice for now.
Regards, Patrick
Thank you very much for your thoughtful reply. You have given me many things to research.
-
mailinglists -
Patrick Georgi