Re: Installing with new Installer-Image
On Fri, 22 Sep 2017, John Paul Adrian Glaubitz wrote:
> On 09/22/2017 03:35 AM, Finn Thain wrote:
> > The stuff in drivers/ide/ still has official maintainers, so we should
> > probably try to find the regression, if that's what it is.
>
> I have my doubts given the fact that the drivers are currently
> effectively broken and apparently no one before me noticed that.
>
> However, it should be easy to find the maintainers using
>
> scripts/get_maintainers.pl
>
I know who they are. Why else would I state that drivers/ide/ has official
maintainers?
> > Unfortunately, I don't have access to suitable hardware right now, so
> > I can't be of much help besides building kernel binaries for
> > bisection.
>
> You don't need access to hardware. You can test the kernel just fine on
> Aranym.
I suppose that even if Aranym's IDE emulation is not 100% accurate, it
should be good enough to expose various kinds of kernel bugs. So I agree
that it is worth trying to bisect this on Aranym.
> All you need is a simple root filesystem and an initrd which I can
> provide. In fact, you can just use the kernel and initrd from the
> debian-installer distribution:
>
> > https://people.debian.org/~glaubitz/debian-cd/debian-installer-images_20170828_m68k.tar.gz
>
> Use the common kernel and the initrd for the CD-ROM installation.
>
Not useful for bisection.
> >> For Falcon IDE, you can try the new driver under drivers/ata/, just
> >> enable CONFIG_ATA and CONFIG_PATA_FALCON.
> >>
> >
> > Last I heard, the IDE driver/libata conversion plan had been NAK'd.
> > Has the IDE subsystem maintainer changed his mind?
>
> Why it should it be NAK'd?
See link below.
> The IDE/ATA subsystem maintainers want to get all the drivers ported
> over to libata so they can eventually get rid of the old IDE stack.
>
Needless to say. And I know about the efforts of b.zolnierkie@samsung.com
towards that end. https://www.spinics.net/lists/linux-ide/msg53897.html
Perhaps it would help if I said that I don't have any objection to libata.
What interests me here is avoiding regressions in a useful body of code
like the IDE core.
If Debian Installer testers are also interested in that goal, all the
better.
--
> Adrian
>
>
Reply to: