Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Debian SPARC on SPARC M7?

15 views
Skip to first unread message

Joe

unread,
Apr 8, 2016, 2:20:03 PM4/8/16
to
    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.

Joe
   

Sam Ravnborg

unread,
Apr 8, 2016, 3:40:02 PM4/8/16
to
Hi Joe.

> 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.

Try to ask at sparclinux at vger.kernel.org.
I know some Oracle people are hanging around, and sending patches to the kernel.

Sam

Karl Volz

unread,
Apr 8, 2016, 3:50:02 PM4/8/16
to
On 04/08/2016 01:37 PM, Joe wrote:
> Ok cool. Thanks for getting back to me so fast.
>
> Joe
Send me a pointer to the latest.. I'll check it out.

Karl

Joe

unread,
Apr 8, 2016, 3:50:03 PM4/8/16
to
Ok cool. Thanks for getting back to me so fast.

Joe

On 2016-04-08 12:26 PM, Sam Ravnborg wrote:

John Paul Adrian Glaubitz

unread,
Apr 8, 2016, 4:10:03 PM4/8/16
to
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.

Adrian

Bryce

unread,
Apr 8, 2016, 8:50:03 PM4/8/16
to
Mmmm

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
https://people.debian.org/~glaubitz/debian-cd/2016-03-22/debian-9.0-sparc64-BC-1.iso
https://people.debian.org/~glaubitz/debian-cd/2016-03-22/debian-9.0-sparc64-NETINST-1.iso

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.sc...@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,...
-parp-
     ┌───────────────┤ [!!] 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
boot:
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
done.
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
/lib/udev/rules.d/10-sunv.rules:
# 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"
LABEL="vio_end"

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
lib/udev/rules.d
lib/udev/rules.d/60-persistent-storage-dm.rules
lib/udev/rules.d/55-dm.rules
lib/udev/rules.d/80-net-setup-link.rules
lib/udev/rules.d/56-lvm.rules
lib/udev/rules.d/50-firmware.rules
lib/udev/rules.d/69-lvm-metad.rules
lib/udev/rules.d/75-net-description.rules
lib/udev/rules.d/80-drivers.rules
lib/udev/rules.d/60-persistent-storage.rules
lib/udev/rules.d/50-udev-default.rules

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.

John Paul Adrian Glaubitz

unread,
Apr 9, 2016, 6:20:02 AM4/9/16
to
On 04/09/2016 01:39 AM, Bryce wrote:
> (note: I'd to manually drop to >a shell and modprobe -v sunvdc and sunvnet).

So, apparently, you're the first one to actually test the fixes that
came with the latest image because this image was supposed to fix that
very issue and I hoped that people affected by this issue would test
this:

> https://lists.debian.org/debian-sparc/2016/03/msg00017.html

But since this didn't happen, I didn't have any chance to review this
issue and make any improvements.

> 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

Just do a real minimal instllation, don't install any extra
packages as this isn't really easily possible at the moment.

> But on the plus side the kernel is alive and I still have a working shell....

That's actually the only information relevant to the M7, the rest
are generic issues.

> 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!)

Again, this is something that I need other people to test. I don't have
any modern SPARC machines available that requires these modules. All
I have is an old Sun Blade 100.

> Ok,.. so far so good,.. now installing packages,...
> -parp-
>
> ┌───────────────┤ [!!] Select and install software

Yeah, again, that doesn't really work. Unless sparc64 becomes
an official architecture, this will always be problematic.

> Close but no cigar.
> Ok,.. well we have SOMETHING on the disk so lets close out the Silo
> installation step...

Yes, that's what you actually should do. Just finish the installation by
installing SILO and install more software after rebooting into the
system.

> SILO Version 1.4.14
> boot:
> Linux LinuxOLD
>
> endless looping of
> (...)
> 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
> (...)
> yeap,.. no sunv* modules hence the initrd is probably the same
> (...)
> 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"

No, we don't. All we need is people actually testing the new images
I created on the hardware I don't own. If I had known that the two
modules are still missing, I would have probably already fixed it.

Adrian

--
.''`. John Paul Adrian Glaubitz
: :' : Debian Developer - glau...@debian.org
`. `' Freie Universitaet Berlin - glau...@physik.fu-berlin.de
`- GPG: 62FF 8A75 84E0 2956 9546 0006 7426 3B37 F5B5 F913

Anatoly Pugachev

unread,
Apr 10, 2016, 2:40:03 PM4/10/16
to
On Sat, Apr 9, 2016 at 2:39 AM, Bryce <philip....@oracle.com> wrote:
> {0} ok boot bryce-deb
> ...
> SILO Version 1.4.14
> boot:
> 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
> done.
> 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

you're right here, it's modules problem, they are not being present in
initramfs, so to fix it, boot from cdrom again into rescue mode and
add this 2 modules to:

root@deb4g:/tmp# grep -v ^# /etc/initramfs-tools/modules
sunvdc
sunvnet

and update kernel initrd from rescue shell with
# update-initramfs -u

exit rescue shell and reboot, it should boot ok now... (but there
would be no telnet console present on boot, since it needs to be
enabled, before reboot from rescue shell with :

root@debian:/# cd /etc/systemd/system/getty.target.wants#
root@debian:/etc/systemd/system/getty.target.wants# ln -sf
/lib/systemd/system/console-getty.service console-getty.service

or if you know IP address of installed debian, login via ssh and
enable with systemctl :

# systemctl enable console-getty

John Paul Adrian Glaubitz

unread,
Apr 10, 2016, 2:50:05 PM4/10/16
to
On 04/10/2016 08:30 PM, Anatoly Pugachev wrote:
> you're right here, it's modules problem, they are not being present in
> initramfs,so to fix it, boot from cdrom again into rescue mode and
> add this 2 modules to:
>
> root@deb4g:/tmp# grep -v ^# /etc/initramfs-tools/modules
> sunvdc
> sunvnet
>
> and update kernel initrd from rescue shell with
> # update-initramfs -u

See my comment to #815977 [1].

> exit rescue shell and reboot, it should boot ok now... (but there
> would be no telnet console present on boot, since it needs to be
> enabled, before reboot from rescue shell with :
>
> # systemctl enable console-getty

Just add "console=ttyS0,115200,8n1" to the command line when booting
the Debian Installer, e.g.: install console=ttys0,115200,8n1.

I'm not sure what was different with the sparc port in that regard.
Might be possible that this parameter was passed to the kernel by
default. You can verify that by checking /proc/cmdline directly
after booting the Debian Wheezy installer image.

Another possibility is that /etc/inittab contrained the necessary
entries to enable serial console by default and those settings
were lost after the systemd transition.

Adrian

> [1] https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=815977#15
0 new messages