On Fri, Oct 26, 2018 at 07:41:54AM +0200, diego righi wrote: > So let's say I'm a fool, I use a foolish partition layout, and the intel x86 > and amd64 architectures are tricky/shitty architectures with stupid bioses > which work bad, ok? > So why openbsd 6.4 i386 and amd64 bootloaders (not biosboot, boot!) > express different behavior? Wasn't openbsd about correctness? :/ > If I'm wrong and it is documented that I can't do this fine, but so also > i386 should not work, this behavior is just strange for me, that's it.
This is simply not an interesting problem to solve. If it's interesting *for you*, I'll volunteer to take a look at your diff to fix this, once I see it appear on tech@. -ml > On Fri, Oct 26, 2018 at 4:29 AM Misc User <[email protected]> > wrote: > > > > Why are so many people lining up to die on the "One big root partition" > > hill? Partition your disks for fuck's sake, or if you are too lazy to > > do that, just let the installer do it for you. And, no, it doesn't > > "just work", it doesn't, it just hasn't broken yet. > > > > Also, both install.i386 and install.amd64 call out the fact that your > > system might break if your kernel happens to be beyond the point your > > system's BIOS can no longer read. From install.i386: > > > > `` > > The OpenBSD root partition must reside completely within the BIOS > > supported part of the hard disk -- this could typically be 504MB, 2GB, > > 8GB or 128GB, depending upon the age of the machine and its BIOS. The > > rest of the OpenBSD partitions can be anywhere that hardware supports. > > '' > > > > Beside, that limitation isn't anything any of can deal with, its an > > inherent problem in how various manufacturers implemented their BIOSes > > > > IF you are dead set on such a foolish partition layout, then go bug > > your motherboard manufacturer to completely rewrite their BIOS to > > support doing so. And since this is a restriction in the way the BIOS > > works, there is no way it would work in i386 if it doesn't in amd64. > > The only way it would works is if in your futzing about with bootloaders > > and kernels you accidentally fixed it by moving the kernel somewhere the > > BIOS can read it (in which case either boot loader would be able to read > > it). > > > > On 10/25/2018 2:14 PM, diego righi wrote: > > > So now I try to reply, I don't want to sound like a troll, because I'm > > > an openbsd > > > user and supporter since very long time and I know that with a proper bug > > > report > > > the full dmesg should be provided and possibly even more... > > > ...but to keep things short I've this ECS GF8100VM-M5 motherboard that I > > > use > > > sometimes on a home bench to test things around, disks, adapters, and so > > > on... > > > ...a lab machine let's say, with a toshiba 160Gb disk attached and 2Gb of > > > ram, > > > and one big "a" slice that was working fine with openbsd 6.3 amd64, after > > > updating to openbsd 6.4 amd64 it started to reboot as soon as "boot" took > > > over, > > > I could see nor log anything, this is why I was not yet submitting a > > > bugreport, so > > > I started to think about an hardware problem, but then I've retried with > > > openbsd > > > 6.3 amd64 and everything was working, so I've tried with openbsd 6.4 i386 > > > and > > > it was also working fine, retried with openbsd 6.4 amd64 and bam, it > > > was flipping > > > a reboot immediately no error, nothing, so since I have many spare disks > > > to > > > experiment I've installed and booted openbsd 6.4 i386 on a spare disk, > > > booted it > > > as first sata disk with the openbsd 6.4 amd64 disk as second sata disk, > > > mounted > > > it on /mnt, copied the files /usr/mdec/biosboot and /usr/mdec/biosboot in > > > /mnt/usr/mdec/ and installed the bootloader with this command: > > > installboot -r /mnt sd1 /usr/mdec/biosboot /usr/mdec/boot > > > as soon as I rebooted, openbsd amd64 booted fine with the i386 bootloader. > > > Now I'm not a programmer but the versions of the 2 "boot" differs: > > > i386 states 3.34 > > > whereas amd64 states 3.41 I don't know if it's only cosmetic, and then > > > they do the > > > same inside the code, but the version difference is confirmed in the cvs: > > > http://cvsweb.openbsd.org/src/sys/arch/i386/stand/boot/conf.c?rev=1.65&content-type=text/x-cvsweb-markup > > > http://cvsweb.openbsd.org/src/sys/arch/amd64/stand/boot/conf.c?rev=1.42&content-type=text/x-cvsweb-markup > > > (again I'm sorry for the poor report, and if I will get the famous > > > "you suck" reply I will understand =_) > > > > > > As for the FAQ and the man pages, I've actually read the man pages of > > > installboot and boot and fdisk and many > > > others since I started using openbsd from the 2.6 release, I don't > > > remember it was written that it can't boot > > > from a big "a" slice, but maybe it's my mistake and I didn't find it, > > > I totally *love* openbsd man > > > pages and they are the best of any other unix I've tried! > > > On Thu, Oct 25, 2018 at 10:09 PM Theo de Raadt <[email protected]> > > > wrote: > > >> diego righi <[email protected]> wrote: > > >> > > >>> Big "a" slice may not be advised and not secure for production, but it > > >>> always worked. > > >> Do you have evidence? > > >> > > >>> (and on i386 it still works, even on amd64 with the i386 bootloader) > > >> Evidence supplied? > > >> > > >> BTW, the i386 and amd64 bootloaders are largely identical. You better > > >> have evidence. > > >> > > >>> So I agree that it is not good practice but to quick test machines I've > > >>> did > > >>> it many times. > > >> Quick? It takes extra steps at install time. It is slower to set it up > > >> that > > >> way. > > >> > > >>> (and never found in the FAQ nor in the manpages that it should not work) > > >> Nowhere is it promised that the FAQ is incomplete, actually the FAQ > > >> recommends in strong terms to use the default setup. The manpages > > >> do not propose such decisions, but I doubt you read man pages about > > >> this and are simply making that part up... > > >
