On 29.04.2013 04:09, ron minnich wrote:
not everyone wants or needs grub. So that's part of the point.
Could we have a sane discussion about why it's not suitable for this or that scenario and what would need to be fixed? Not just quasi-fanatical "I don't want it".
Vladimir 'φ-coder/phcoder' Serbinenko wrote:
not everyone wants or needs grub. So that's part of the point.
Could we have a sane discussion about why it's not suitable
IMO it's a clusterfuck as far as usability is concerned for one. But this isn't the GRUB mailing list, so actually it's wholly inappropriate to discuss GRUB any further here. If someone wants to continue discussing with Vladimir please do so off list.
what would need to be fixed? Not just quasi-fanatical "I don't want it".
Nobody said "I don't want it" and despite what you might believe, Ron is anything but fanatical.
Remember that a grand unified solution to some given problem does not always make sense, and that GRUB is no different in that regard.
Sometimes it's simply a horrible idea.
//Peter
On 29.04.2013 04:48, Peter Stuge wrote:
Vladimir 'φ-coder/phcoder' Serbinenko wrote:
not everyone wants or needs grub. So that's part of the point.
Could we have a sane discussion about why it's not suitable
IMO it's a clusterfuck as far as usability is concerned for one. But this isn't the GRUB mailing list, so actually it's wholly inappropriate to discuss GRUB any further here. If someone wants to continue discussing with Vladimir please do so off list.
The thing is that coreboot is useless without some kind of payload. The bigggest part of effort of making coreboot-based firmware work on Loongson is porting the payload as the job of coreboot on this platform is reduced to mere 1000-2000 lines. At this the payload could simply be linked with fwstart.S and I was going to propose sharing this code with the other payload if other payload is willing to port and maintain Loongson. But for this I need the answers to questions I asked and everybody avoids answering.
On Sun, Apr 28, 2013 at 7:24 PM, Vladimir 'φ-coder/phcoder' Serbinenko phcoder@gmail.com wrote:
Could we have a sane discussion about why it's not suitable for this or that scenario and what would need to be fixed? Not just quasi-fanatical "I don't want it".
I guess I missed the part where not wanting something was considered insane.
"I don't want bubble and squeak for dinner again" "You are insane!"
"I don't want American Beer" "you are insane!"
You see where this leads :-)
Anyway ... one example where GRUB doesn't work for us. One of many. Chromebooks use a boot system called depthcharge, which is a coreboot payload. It forms a chain of verified boot from the coreboot romstage to the kernel. There's a lot of careful work that has gone into that design and implementation, involving balancing the efficiency of the hashing algorithms at various steps, how the TPM plays into it, and so on. It's quite hard to get this right. Nobody I know is interested in plugging it all into grub. You're welcome to go for it, of course, I'd love to see it happen.
As it happens, I worked with Ollie Lo, years back, on an x86 system in which we got ram working in < 200 instructions. In the end, it was still useful to put that into a coreboot (a.k.a. linuxbios) framework, because it got us huge flexibility in terms of what we did on the platform.
So, all that said, the answer still is "I am going to try to look at this code and see if a coreboot port makes sense". I understand what the U in grub stands for, and it's not "universal". And I don't see the harm in trying it out.
thanks
ron
-
Peter Stuge -
ron minnich -
Vladimir 'φ-coder/phcoder' Serbinenko