[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Debian SPARC on SPARC M7?

On 08/04/16 21:00, John Paul Adrian Glaubitz wrote:
Since I assume that current 4.4.x kernels already support M7, you should also be able to run Debian's sparc64 port since it comes with that kernel.

However, since I have never even seen an M7 in real life, I can't really tell.

You might just try to install it in an Linux LDOM.


On Apr 8, 2016, at 8:09 PM, Joe <confirmed101@yahoo.com> wrote:

    Hi there everyone, I was wondering, will the Debian SPARC port, as it is now, run on the new SPARC M7 processor that just came out from Oracle? I've heard a lot about this processor, and it's supposed to be quite groundbreaking, and I am intrigued. I know this might not be the best place to ask this, but I've scoured the net and can't find an answer.

    Thanks so much, looking forward to a reply. I've been following the Debian SPARC mailing lists for several months now and haven't heard anything about the M7 here, so I would love to hear whats up with it.



well since it's a Friday and it's been a crap week, since karl is suffering badly from working in the salt mines 24/7, I   went off and pulled down what I assume to be the current ISO images

I didn't have a baremetal T7 box to hand since those are in use for other devel tasks but I did create a random spare LDOM (think of it conceptually as a xen instance but for solaris) and starts installing the debian-9.0-sparc64-BC-1.iso image. (note: I'd to manually drop to a shell and modprobe -v sunvdc and sunvnet). I got most of the way through the installer to the "Install the base system' task but it blows out there because it gets upset about signatures and signing keys

  ┌─────┌───────────────┤ [!!] Install the base system ├───────────────┐────┐
  │     │                                                              │    │
  │     │                      Debootstrap Error                       │    │
  │     │ Release file signed by unknown key (key id B4C86482705A2CE1) │    │
  │ Chec│                                                              │    │
  │     │     <Go Back>                                 <Continue>     │    │
  └─────│                                                              │ ───┘

I believe Rod <r.schnell@mythos.freeddns.org> hit the same  back at the end of January but your indicated solution has a caveat,.. no gpg in the install root
Apr  8 22:49:40 main-menu[172]: INFO: Menu item 'bootstrap-base' selected
Apr  8 22:49:40 debootstrap: gpgv: Signature made Fri Apr  8 20:02:55 2016 UTC using RSA key ID 705A2CE1
Apr  8 22:49:40 debootstrap:
Apr  8 22:49:40 debootstrap: gpgv: Can't check signature: No public key
Apr  8 22:49:40 debootstrap:
Apr  8 22:51:50 base-installer: error: exiting on error base-installer/debootstrap-failed
Apr  8 22:52:01 main-menu[172]: WARNING **: Configuring 'bootstrap-base' failed with error code 1
Apr  8 22:52:01 main-menu[172]: WARNING **: Menu item 'bootstrap-base' failed.
/var/log # gpg --keyserver pgp.mit.edu --recv-keys 705A2CE1 ; gpg --armor --export 705A2CE1 |apt-key add - && apt-get update
/bin/sh: gpg: not found
/bin/sh: apt-key: not found
/bin/sh: gpg: not found

But on the plus side the kernel is alive and I still have a working shell....
/var/log # cat /proc/cpuinfo 
cpu             : SPARC-M7
fpu             : SPARC-M7 integrated FPU
pmu             : sparc-m7
prom            : OBP 4.37.0.build_27 2015/01/07 14:08
type            : sun4v
ncpus probed    : 8
ncpus active    : 1
D$ parity tl1   : 0
I$ parity tl1   : 0
Cpu0ClkTck      : 00000000f65c15b0
cpucaps         : flush,stbar,swap,muldiv,v9,blkinit,n2,mul32,div32,v8plus,popc,vis,vis2,ASIBlkInit,fmaf,vis3,hpc,ima,pause,cbcond,adp,aes,des,camellia,md5,sha1,sha256,sha512,mpmul,montmul,montsqr,crc32c
MMU Type        : Hypervisor (sun4v)
MMU PGSZs       : 8K,64K,4MB,256MB,2GB,16GB

/var/log # uname -a
Linux bryce-deb 4.4.0-1-sparc64 #1 Debian 4.4.6-1 (2016-03-17) sparc64 GNU/Linux
Ok thats not gonna fly anytime soon so lets try the debian-9.0-sparc64-NETINST-1.iso image. Again need to load sunvdc and sunvnet by hand in the shell before installing but thats just minor problem when using LDM to create a linux instance (as long as you know about it!)
Ok,.. so far so good,.. now installing packages,...
     ┌───────────────┤ [!!] Select and install software ├────────────────┐
     │                                                                   │
     │                     Installation step failed                      │
     │ An installation step failed. You can try to run the failing item  │
     │ again from the menu, or skip it and choose something else. The    │
     │ failing step is: Select and install software                      │
     │                                                                   │
     │                            <Continue>                             │
     │                                                                   │

Apr  8 23:18:13 in-target: WARNING: The following packages cannot be authenticated!
Apr  8 23:18:13 in-target:   dash ncurses-bin ncurses-base gcc-5-base libgcc1 libstdc++6 libapt-pkg5.0
Apr  8 23:18:13 in-target:   libapt-inst2.0 apt apt-utils gpgv gnupg libpcre3 console-setup-linux
Apr  8 23:18:13 in-target:   console-setup keyboard-configuration libtinfo5 libncurses5 libncursesw5
Apr  8 23:18:13 in-target:   tzdata kmod libkmod2 libdebconfclient0 bsdmainutils isc-dhcp-common
Apr  8 23:18:13 in-target:   libslang2 libxapian22v5 vim-tiny vim-common
Apr  8 23:18:13 in-target: E
Apr  8 23:18:13 in-target: :
Apr  8 23:18:13 in-target: There were unauthenticated packages and -y was used without --allow-unauthenticated
Apr  8 23:18:13 in-target:
Apr  8 23:18:14 kernel: [5207573.026870] ISO 9660 Extensions: RRIP_1991A
Apr  8 23:18:17 main-menu[172]: WARNING **: Configuring 'pkgsel' failed with error code 100
Apr  8 23:18:17 main-menu[172]: WARNING **: Menu item 'pkgsel' failed.
Apr  8 23:18:45 main-menu[172]: INFO: Modifying debconf priority limit from 'high' to 'medium'
Apr  8 23:18:45 debconf: Setting debconf/priority to medium
~ # gpg --keyserver pgp.mit.edu --recv-keys 705A2CE1 ; gpg --armor --export 705A2CE1 |apt-key add - && apt-get update
/bin/sh: gpg: not found
/bin/sh: apt-key: not found
/bin/sh: gpg: not found
Close but no cigar.
Ok,.. well we have SOMETHING on the disk so lets close out the Silo installation step...
   ┌──────────┤ [?] Install the SILO boot loader on a hard disk ├──────────┐
  ┌│                                                                       │
  ││                     SILO installation successful                      │
  ││ The SILO boot loader has been successfully installed on your new boot │
  ││ partition, and your system should now be able to boot itself.         │
  ││                                                                       │
  ││     <Go Back>                                          <Continue>     │
  └│                                                                       │
Installation complete (but not really)
boot: halt
Program terminated
{0} ok boot bryce-deb
SILO Version 1.4.14
Linux                    LinuxOLD     

endless looping of
Begin: Running /scripts/local-block ...   lvmetad is not active yet, using direct activation during sysinit
  Volume group "bryce-deb-vg" not found
  Cannot process volume group bryce-deb-vg
ALERT!  /dev/mapper/bryce--deb--vg-root does not exist.  Dropping to a shell!
Gave up waiting for root device.  Common problems:
 - Boot args (cat /proc/cmdline)
   - Check rootdelay= (did the system wait long enough?)
   - Check root= (did the system wait for the right device?)
 - Missing modules (cat /proc/modules; ls /dev)

at a guess I thik this is because the system is running in an ldm and hasn;t caught on that it needs to load the sunvdc sunvnet modules
that'll be a udev item

I can't check what's actually been installed
but you'd want something like this
# Theory
# Linux under solaris's ldm exposes a pile of /devices/channel-devices/v*
# devices. The drivers should not be reloaded as that would likely
# crash the system.
# If the vio subsystem exists, we check for an environment var (sunv_ran)
# if it doesn't exist or does not have the value '1' then we look for
# a glob match for each driver,.. should we find one we set sunv_ran to
# '1' permanently using ':=' and load the associated module

SUBSYSTEM!="vio", ENV{sunv_ran}!="1", GOTO="vio_end"
DEVPATH=="/devices/channel-devices/vnet-*", ENV{sunv_ran}:="1", RUN+="/sbin/modprobe -b sunvnet"
DEVPATH=="/devices/channel-devices/vdc-*", ENV{sunv_ran}:="1", RUN+="/sbin/modprobe -b sunvdc"

Lets start up in rescue mode and see if that's whats happening.
             │  Execute a shell in /dev/bryce-deb-vg/root         │
# cd /lib/udev/rules.d
# grep sunv *
yeap,.. no sunv* modules hence the initrd is probably the same
root@bryce-deb:/boot# cat initrd.img-4.4.0-1-sparc64-smp | zcat | cpio --list | grep udev/\rules.d

BUT keep in mind this is a T7-1 and that the kernel is runnniing, I have a shell and I can almost do stuff 8)
So,.. the answer to your original question is ... "'kinda' we need to make a tweak for next ISO for LDM abusers"

Karl may followup with a baremetal attempt.

Reply to: