Solaris 11.2 ISO on UEFI systems

[Saint Atique]

It seems like Solaris hasn't got EFI supported ISO yet. Whether burning it to CD or USB it computer won't boot from the CD.

To confirm I tried some Linux CDs such as Linux Mint 17.1, Fedora and Ubuntu. All boot in this machine with gpt partition layout with EFI enabled. But Solaris won't.

Can anyone tell me what's wrong and how to fix it? I see the EFI directory in the CD missing too.

[Casper H.S. Dik]

What hardware is this? Which ISO image or USB image did you use?

Casper

[Saint Atique]

I have prepared the usbmedia using a virtual box installation of Solaris 11.2 Here's the output of the command of preparing the disk,

$ sudo usbcopy sol-11_2-live-x86.usb
Password:
Image type: dd-able x86
Found the following USB devices:
0: /dev/rdsk/c2t0d0p0 7.7 GB USB Disk 8.07
Enter the number of your choice: 0

WARNING: All data on your USB storage will be lost.
Are you sure you want to install to
USB Disk 8.07, 7700 MB at /dev/rdsk/c2t0d0p0 ? (y/n) y
Copying and verifying image to USB device
Finished 1222 MB in 1338 seconds (0.9MB/s)
0 block(s) re-written due to verification failure
Completed copy to USB

The partition table as it look like after creating usb media with solaris 11.2 ISO for x86

$ sudo parted /dev/dsk/c2t0d0p0 print
Password:
Model: Generic Ide (ide)
Disk /dev/dsk/c2t0d0p0: 8054MB
Sector size (logical/physical): 512B/512B
Partition Table: msdos

Number Start End Size Type File system Flags
1 1049kB 5243kB 4194kB primary fat16
2 5243kB 1281MB 1276MB primary solaris boot


usb file I used from the oracle site is: sol-11_2-ai-x86.usb

My machine (notebook) is 64 bit Intel Core i7 (quad core), Intel HD Graphics 4000
I keep UEFI enabled all the time (no legacy boot) and I am using GPT partition table on my harddisk.

The machine booted from the USB disk. GRUB menu appears. I choose the first option. Then it is stuck showing two lines on top of screen.(see screenshot: http://saos.azurewebsites.net/out_of_app_images/comp.unix/solaris_11.2_usb_boot_hang_cropped.jpg
Then I tried all other options. Result is same, shows two lines and nothing. I fall asleep waiting to see if anything happens. In the morning I see the same thing (two lines on top of screen) and nothing.

I was checking the grub entries. It looks like following screenshot:
http://saos.azurewebsites.net/out_of_app_images/comp.unix/Solaris_USB_media_boot_grub_entry.jpg

By the way, on the Virtual Box, to run the usbcopy command on solaris I had to copy the file from USB to solaris into home directory. It took approximately two hours or more. Why so slow file operation? Is it because I am using dynamic size parition of 20 GB on VM (given that I'm using 4 GB RAM for VM, virtual box is running on Linux Mint (64 bit) 17.1 kernel-3.18.3).

Let's get back to main point. Is it possible that someone faced similar issue trying boot using USB on a UEFI system? And did he/she succeed.

Thanks for a long reading.

[John D Groenveld]

In article <889aba98-9eb1-4129...@googlegroups.com>,

Saint Atique <uni...@gmail.com> wrote:
>I have prepared the usbmedia using a virtual box installation of Solaris
>11.2 Here's the output of the command of preparing the disk,
>
>$ sudo usbcopy sol-11_2-live-x86.usb

Try the S11.2 text installer image and also try dd'ng the image
from your host OS:
<URL:http://www.oracle.com/technetwork/server-storage/solaris11/downloads/install-2245079.html>
<URL:http://www.oracle.com/technetwork/server-storage/solaris11/documentation/solaris-11-2-faqs-2191871.pdf>

John
groe...@acm.org

[Casper H.S. Dik]

Saint Atique <uni...@gmail.com> writes:

>My machine (notebook) is 64 bit Intel Core i7 (quad core), Intel HD Graphics 4000
>I keep UEFI enabled all the time (no legacy boot) and I am using GPT partition table on my harddisk.

>The machine booted from the USB disk. GRUB menu appears.

>he same thing (two lines on top of screen) and nothing.

>I was checking the grub entries. It looks like following screenshot:

>http://saos.azurewebsites.net/out_of_app_images/comp.unix/Solaris_USB_media=
>_boot_grub_entry.jpg

>Let's get back to main point. Is it possible that someone faced similar issue
>trying boot using USB on a UEFI system? And did he/she succeed.

I have no issues doing the same (all modern systems are UEFI systems),
so I think it is something in your hardware that gives Solaris issues.

Try booting by adding "-kv" to the boot command line; the kernel
debugger will be loaded (-k) and the output will be verbose (-v).

You can then figure out where it hangs; you might want to look
at the VT-x/VT-d options; while Solaris should boot fine with
them, certain firmwares may have issues.

If it hangs, F1-A should gets you the kernel debugger but in certain
cases it the kernel hangs were the keyboard's interrupts aren't
serviced.

I don't think it is UEFI related at all, but rather an issue with
some device or lack of support for some devices.

Casper

[Saint Atique]

Tried the 11.2 text installer usb image. Exactly same result: two lines on top of screen and that's where it stays. Tried dd-ing the image from host OS: I can boot that image only by enabling legacy boot. I don't want to go the legacy way. Because I think it might trigger a legacy bios type bootloader installation( such as copying the solaris bootloader 448 bytes to active partition like we used to do when there were no EFI). This would mess up because all other OSs in the machine are using EFI boot.

Reason of thinking like that is from some paste experience. When I booted Windows with EFI I could perform EFI installation and same happened with several Linux distro. If I didn't boot with EFI it would perform a legacy bios installation provided that disk partition layout is not gpt and using legacy boot.

[Casper H.S. Dik]

Saint Atique <uni...@gmail.com> writes:

>Reason of thinking like that is from some paste experience. When I booted
>Windows with EFI I could perform EFI installation and same happened with
>several Linux distro. If I didn't boot with EFI it would perform a legacy bios
>installation provided that disk partition layout is not gpt and using legacy boot.

Please follow my advice and boot with the kernel debugger and see where it
hangs; I wouldn't think it has anything to with EFI boot.

If it was a problem , it would fail a lot earlier, I think.

Casper

[John D Groenveld]

In article <efd3c11c-c26e-4a6d...@googlegroups.com>,

Saint Atique <uni...@gmail.com> wrote:
>host OS: I can boot that image only by enabling legacy boot. I don't

You've proven that the image is good by successfully booting
in BIOS legacy boot mode.
Now you need to determine why UEFI fails on the same hardware.

BTW have you confirmed that you're running the latest BIOS?
John
groe...@acm.org

[James Lee]

On 01/02/2015 13:22, Casper H.S. Dik wrote:
> Saint Atique <uni...@gmail.com> writes:

>> Let's get back to main point. Is it possible that someone faced similar issue
>> trying boot using USB on a UEFI system? And did he/she succeed.
>
> I have no issues doing the same (all modern systems are UEFI systems),
> so I think it is something in your hardware that gives Solaris issues.

I have similar problems booting with UEFI and it is Solaris that is the
cause. My symptoms are slightly different so it might not be the same
as the OP but here is my problem and fix in the hope it helps someone.

The problem is that in UEFI mode the screen does not work. I see a snow
scene. It is obvious that the underlying process is running from the
pattern of the snow, believing this I manage to do a text install blind,
without seeing the screen. This still only gave me an OS with no
display but was fixable. An easier route is to edit the grub menu
during boot - if one knows to do it. Again this this was done blind as
the screen was not working at this point (note the OP was not blind at
this point). Here is the method I used, your system may vary:

Use sol-11_2-text-x86.usb
Enable UEFI boot and disable secure boot (requires a BIOS password)
Power on, press F12 to get the UEFI menu and select the USB
When the grub menu should appear, I see a snow scene and the screen
looks like the dots have been wiped sideways a few hundred pixels, type:
e
6 down arrows
16 right arrows
1366x768x32;
Control X
and after a short delay the boot and install proceeds visibly, "SunOS
Release 5.11 ..."

After install on first boot the same procedure was needed. Once booted
edit:
/rpool/boot/grub/grub.cfg
and add "1366x768x32;" to the line "set gfxpayload=".
Which is what the above blind typing is doing to the grub menu. I also
edited the other grub.cfg files in /boot/grub but I think it's the rpool
one that is used on boot.

The 1366x768 is my screen size, Solaris does not include this in its list.

The same procedure did not get the live USB running, it gets through the
language selection but does not start the graphics.


Some other notes and tips:

Legacy boot does not have the same screen problem but results in a
legacy installation.

I have installed triple boot of: Windows 8 because it's there, Linux
Mint Xfce 17.1 because it works and Solaris 11.2 because...

Windows must go on first. Next install Solaris because it is limited in
the number of partitions that can exit on install (WHY!?!, GPT allows
many more than 8). Using gparted I removed the Win recovery, squeezed
Win8, adding a padding partition and put a partition for Solaris at the
top. After the install of Solaris I split the padding partition and
added Linux Mint.

Finally a fix is needed as the Solaris install corrupts Windows both in
UEFI and legacy mode. Boot the Windows recovery USB, select command
line and run:
bootrec /fixmbr
bootrec /fixboot
bootrec /scanos
bootrec /rebuildbcd
then on the next attempt to reboot when it says "Preparing Automatic
Repair" it actually does the automatic repair.


Hardware:
Acer V5-171
WiFi card swapped for Intel Link 5300.
The only significant Solaris problem is wired Ethernet does not work.



James Lee.

[James Lee]

On 16/03/2015 12:28, James Lee wrote:
> On 01/02/2015 13:22, Casper H.S. Dik wrote:

>> I have no issues doing the same (all modern systems are UEFI systems),
>> so I think it is something in your hardware that gives Solaris issues.

Failure to work with UEFI is hard coded into /lib/svc/method/svc-vbiosd.

> I have installed triple boot of: Windows 8 because it's there, Linux
> Mint Xfce 17.1 because it works and Solaris 11.2 because...

What I said above is true but incomplete, hence misleading and
subsequently I have found out more. Windows 8 requires UEFI and after
working through install I can indeed boot a basic Solaris 11.2 using
UEFI, however it's not useful.

After installing in UEFI text mode Solaris will work under UEFI in text
mode. "pkg install solaris-desktop" should get graphics but Solaris can
not do this with UEFI. The failure is vbiosd on which gdm depends.

$ svcs -xL vbiosd
svc:/system/vbiosd:default (BIOS call emulation)
State: disabled since March 25, 2015 09:39:07 AM GMT
Reason: Temporarily disabled by an administrator.
See: http://support.oracle.com/msg/SMF-8000-1S
See: vbiosd(1M)
See: /var/svc/log/system-vbiosd:default.log
Impact: This service is not running.
Log:
]
[ Mar 25 09:36:06 Enabled. ]
[ Mar 25 09:36:06 Executing start method ("/lib/svc/method/svc-vbiosd
start"). ]
[ Mar 25 09:29:06 Method "start" method exited with status 101. ]
[ Mar 25 09:39:06 "start" method requested temporary disable: "vbiosd is
not supported on UEFI systems"
]
[ Mar 25 09:36:06 Enabled. ]
[ Mar 25 09:36:07 Restarting too quickly, changing to maintenance. ]
[ Mar 25 09:36:07 Leaving maintenance because disable requested. ]
[ Mar 25 09:36:07 Disabled. ]
Use: 'svcs -Lv svc:/system/vbiosd:default' to view the complete log.


$ svcs -d gdm
STATE STIME FRMI
disabled 09:39:07 svc:/system.vbiosd:default
...

The "vbiosd is not supported on UEFI" is only coming from the script
/lib/svc/method/svc-vbiosd, removing the check does get vbiosd running
but consolekit fails to run so unusually "not supported" does appear to
mean "does not work". I conclude Solaris 11.2 only partially works with
UEFI.

As it works in legacy BIOS mode I don't think I should waste any more
time looking for workarounds. I can't see an advantage in using UEFI
other than not having to press four more keys to switch OS. I have
nothing against UEFI other than it does not work.


This can be simulated on VirtualBox, select UEFI mode before install:
$ VBoxManage modifyvm "${vmname}" --firmware efi



James Lee.

[Casper H.S. Dik]

James Lee <ne...@oxdrove.co.uk> writes:

>On 16/03/2015 12:28, James Lee wrote:
>> On 01/02/2015 13:22, Casper H.S. Dik wrote:

>>> I have no issues doing the same (all modern systems are UEFI systems),
>>> so I think it is something in your hardware that gives Solaris issues.

>Failure to work with UEFI is hard coded into /lib/svc/method/svc-vbiosd.

That is strange as the system I'm typing this on is booting using UEFI
and it is using a whole-disk UEFI disk as the rpool.

>What I said above is true but incomplete, hence misleading and
>subsequently I have found out more. Windows 8 requires UEFI and after
>working through install I can indeed boot a basic Solaris 11.2 using
>UEFI, however it's not useful.

>After installing in UEFI text mode Solaris will work under UEFI in text
>mode. "pkg install solaris-desktop" should get graphics but Solaris can
>not do this with UEFI. The failure is vbiosd on which gdm depends.

>This can be simulated on VirtualBox, select UEFI mode before install:
>$ VBoxManage modifyvm "${vmname}" --firmware efi

I'll check what's up as Solaris clearly should work with modern firmware
and that is how we ship the latest Intel Solaris hardware.

I'll ask around (I'm currently in Santa Clara and can ask the engineers
responsible for x86 booting directly)

Casper

[Casper H.S. Dik]

I've asked the engineer mostly in-tune with booting.

On old systems or systems using the BIOS, you can run X without any
specific drivers for graphics by using the VESA driver.
Support for the VESA driver is in the BIOS and vbiosd is used to emulate
the real-mode BIOS code part of the option ROM.

VESA works but it is usually a lot slower than the native driver.

vbiosd is disabled when we boot using EFI as there is no BIOS code (or
if it is present then evaluating it makes little or no sense)

The EFI option ROMs are used during boot but they are 64 bit long mode
code and vbiosd does not know how to emulate those.

In virtual box, when running it with EFI, needs to have the guest additions
installed in the Solaris guest. It should then be able to run graphics
using the native driver.

>>On 16/03/2015 12:28, James Lee wrote:

>>What I said above is true but incomplete, hence misleading and
>>subsequently I have found out more. Windows 8 requires UEFI and after
>>working through install I can indeed boot a basic Solaris 11.2 using
>>UEFI, however it's not useful.

What graphics card does this system have? If it is not supported then
there will be an issue. If there is a driver but it does not work, then
clearly there is a bug. We often update both the nvidia and the i915
drivers and those seems to be the most common drivers.

Casper

[James Lee]

On 27/03/2015 15:37, Casper H.S. Dik wrote:

> I've asked the engineer mostly in-tune with booting.

Many thanks for following this up. I and many users of this group are
indebted to you.

> vbiosd is disabled when we boot using EFI as there is no BIOS code (or
> if it is present then evaluating it makes little or no sense)
>
> The EFI option ROMs are used during boot but they are 64 bit long mode
> code and vbiosd does not know how to emulate those.

This confused me. I can see vbiosd is disabled but gdm depends on
vbiosd. I've misunderstood what dependent means [in the context of
SMF]. We are saying gdm does not need vbiosd.

> In virtual box, when running it with EFI, needs to have the guest additions
> installed in the Solaris guest. It should then be able to run graphics
> using the native driver.

It does, I confirm this.

> What graphics card does this system have? If it is not supported then
> there will be an issue. If there is a driver but it does not work, then
> clearly there is a bug. We often update both the nvidia and the i915
> drivers and those seems to be the most common drivers.

ddu says
"Video Intel Corporation 3rd Gen Core processor Graphics Controller
i915"

The CPU is i3-3227U which has "Intel® HD Graphics 4000".

Running in legacy BIOS mode it has no problems drawing pictures, so
supported or not it works. The 3D screen savers run at acceptable
speeds and the compiz window manager effects work quickly enough
(although the effect is nauseating).


James Lee.




# prtdiag
System Configuration: Acer V5-171
BIOS Configuration: Acer V2.21 11/26/2013

==== Processor Sockets ====================================

Version Location Tag
-------------------------------- --------------------------
Intel(R) Core(TM) i3-3227U CPU @ 1.90GHz U3E1

==== Memory Device Sockets ================================

Type Status Set Device Locator Bank Locator
----------- ------ --- ------------------- ----------------
DDR3 in use 0 DIMM0 BANK 0
DDR3 in use 0 DIMM1 BANK 2

==== On-Board Devices =====================================
Video Graphics Controller
Lan Controller

==== Upgradeable Slots ====================================

ID Status Type Description
--- --------- ---------------- ----------------------------
1 available PCI Express x16 J5C1
2 available PCI Express x1 J6C1
3 available PCI Express x1 J6C2
4 available PCI Express x1 J6D2
5 available PCI Express x1 J7C1
6 available PCI Express x1 J7D2
7 available PCI Express x1 J8C1
8 available PCI Express x16 J8C2


# scanpci -v
...
pci bus 0x0000 cardnum 0x02 function 0x00: vendor 0x8086 device 0x0166
Intel Corporation 3rd Gen Core processor Graphics Controller
CardVendor 0x1025 card 0x0743 (Acer Incorporated [ALI], Card unknown)
STATUS 0x0090 COMMAND 0x0007
CLASS 0x03 0x00 0x00 REVISION 0x09
BIST 0x00 HEADER 0x00 LATENCY 0x00 CACHE 0x00
BASE0 0x00000000c0000000 SIZE 4194304 MEM64
BASE2 0x00000000b0000000 SIZE 268435456 MEM64 PREFETCHABLE
BASE4 0x00002000 SIZE 64 I/O
MAX_LAT 0x00 MIN_GNT 0x00 INT_PIN 0x01 INT_LINE 0x07
...

[Casper H.S. Dik]

James Lee <ne...@oxdrove.co.uk> writes:

>This confused me. I can see vbiosd is disabled but gdm depends on
>vbiosd. I've misunderstood what dependent means [in the context of
>SMF]. We are saying gdm does not need vbiosd.

There are several way in which dependencies are being used; in
some cases the first service must be running but in cases
such as vbiosd, the depenency only tells SMF that vbiosd
must be started before gdm if it is enabled

>> What graphics card does this system have? If it is not supported then
>> there will be an issue. If there is a driver but it does not work, then
>> clearly there is a bug. We often update both the nvidia and the i915
>> drivers and those seems to be the most common drivers.

>ddu says
>"Video Intel Corporation 3rd Gen Core processor Graphics Controller
> i915"

That device is supported in Solaris; it is generally updated
and is mostly written by Intel but supported directly by Oracle.

>The CPU is i3-3227U which has "Intel® HD Graphics 4000".

>Running in legacy BIOS mode it has no problems drawing pictures, so
>supported or not it works. The 3D screen savers run at acceptable
>speeds and the compiz window manager effects work quickly enough
>(although the effect is nauseating).

So you are using the i915 driver (it wouldn't work if it isn't)

What is the output of:

$ pkg list driver/graphics/drm entire

It should work as there is support for the i915; my laptop
uses the same driver. I doesn't seem we have updated the
driver in any of the SRUs.

Casper

[James Lee]

On 29/03/2015 10:57, Casper H.S. Dik wrote:

> So you are using the i915 driver (it wouldn't work if it isn't)

...in legacy BIOS mode.

> What is the output of:
>
> $ pkg list driver/graphics/drm entire

I have reinstalled in UEFI mode from the text USB, pkg update, added
solaris-desktop and drm.

$ pkg list driver/graphics/drm entire

NAME (PUBLISHER) VERSION
IFO
driver/graphics/drm
0.5.11-0.175.2.0.0.42.2 i--
entire 0.5.11-0.175.2.1.0.2.1
i--

booting gives a blank screen with a single white square the size of one
character and in the position of the first character of the text
console. Power off, using the power button, cause crash dump.

Xorg.0.log is attached. Anything else I should show?


James.

[James Lee]

On 30/03/2015 15:58, James Lee wrote:

I have gone back and done another legacy install and attached is
Xorg.0.log for comparison. The first lines that are different are:


Legacy:

[ 236.538] (WW) Falling back to old probe method for vesa
[ 236.538] (WW) VGA arbiter: cannot open kernel arbiter, no multi-card
support
[ 236.538] (II) intel(0): Creating default Display subsection in
Screen section
"Default Screen Section" for depth/fbbpp 24/32
[ 236.538] (==) intel(0): Depth 24, (--) framebuffer bpp 32
[ 236.538] (==) intel(0): RGB weight 888
[ 236.538] (==) intel(0): Default visual is TrueColor
[ 236.538] (--) intel(0): Integrated Graphics Chipset: Intel(R)
Ivybridge Mobile (GT2)
...success



UEFI:
[ 89.456] (WW) VGA arbiter: cannot open kernel arbiter, no multi-card
support
[ 89.456] (II) Loading sub module "vbe"
[ 89.456] (II) LoadModule: "vbe"
[ 89.457] (II) Loading /usr/lib/xorg/modules/libvbe.so
[ 89.457] (II) Module vbe: vendor="X.Org Foundation"
[ 89.457] compiled for 1.14.5, module version = 1.1.0
[ 89.457] ABI class: X.Org Video Driver, version 14.1
[ 89.457] (II) Loading sub module "int10"
[ 89.457] (II) LoadModule: "int10"
[ 89.457] (II) Loading /usr/lib/xorg/modules/libint10.so
[ 89.457] (II) Module int10: vendor="X.Org Foundation"
[ 89.457] compiled for 1.14.5, module version = 1.0.0
[ 89.457] ABI class: X.Org Video Driver, version 14.1
[ 89.457] (II) VESA(0): initializing int10
[ 89.458] (EE) VESA(0): V_BIOS address 0x0 out of range
[ 89.458] (II) UnloadModule: "vesa"
[ 89.458] (II) UnloadSubModule: "int10"
[ 89.458] (II) Unloading int10
[ 89.458] (II) UnloadSubModule: "vbe"
[ 89.458] (II) Unloading vbe
[ 89.458] (EE) Screen(s) found, but none have a usable configuration.
...failure




James.

[James Lee]

On 29/03/2015 10:57, Casper H.S. Dik wrote:

>> The CPU is i3-3227U which has "Intel® HD Graphics 4000".
>
>> Running in legacy BIOS mode it has no problems drawing pictures, so
>> supported or not it works. The 3D screen savers run at acceptable
>> speeds and the compiz window manager effects work quickly enough
>> (although the effect is nauseating).
>
> So you are using the i915 driver (it wouldn't work if it isn't)
>
> What is the output of:
>
> $ pkg list driver/graphics/drm entire
>
> It should work as there is support for the i915; my laptop
> uses the same driver. I doesn't seem we have updated the
> driver in any of the SRUs.

Casper, am I to accept it does not work with UEFI?
Or is there anything else to try?


James.

[Casper H.S. Dik]

James Lee <ne...@oxdrove.co.uk> writes:

>Casper, am I to accept it does not work with UEFI?
>Or is there anything else to try?

I'm assuming you've tried to update the BIOS to the latest
version (I probably asked this earlier).

We do not update that driver often; it should work with UEFI
but I can't quite see how I can further help you.


Casper

[James Lee]

On 09/04/2015 17:12, Casper H.S. Dik wrote:

> I'm assuming you've tried to update the BIOS to the latest
> version (I probably asked this earlier).

You didn't but it is. At v2.21.

A reason, possibly the only one, to keep the factory Windows on the
machine is the BIOS is only distributed as a MS exe file.

> We do not update that driver often; it should work with UEFI
> but I can't quite see how I can further help you.

Should, but for some reason doesn't. The BIOS messages from Xorg are
the only clue but I'm not sure where to look or what buttons to press to
move forward. It looks like the expedient solution is to accept legacy
boot.

It's not such a problem to swap UEFI/legacy boot, it just requires
entering the BIOS and pressing extra keys, the fact it works at all is
remarkable enough, wow, even suspend to RAM works. Screen brightness
control would be nice... but I've measured the power the screen takes
and it's not great and a wokaround is to select dark backgrounds.
Solaris does use more power than Linux - it's not all the screen - 11W
vs 7W which gives battery life of 3 hours vs 5 - but it beats my old
Solaris 10 laptop which worked its way from 40 minutes to a 40 second
battery life.



James.

[Casper H.S. Dik]

James Lee <ne...@oxdrove.co.uk> writes:

>A reason, possibly the only one, to keep the factory Windows on the
>machine is the BIOS is only distributed as a MS exe file.

Aren't those usually installable using DOS?

>> We do not update that driver often; it should work with UEFI
>> but I can't quite see how I can further help you.

>Should, but for some reason doesn't. The BIOS messages from Xorg are
>the only clue but I'm not sure where to look or what buttons to press to
>move forward. It looks like the expedient solution is to accept legacy
>boot.

I'll look into the Xorg output and may give them to person working
on these drivers.

>It's not such a problem to swap UEFI/legacy boot, it just requires
>entering the BIOS and pressing extra keys, the fact it works at all is
>remarkable enough, wow, even suspend to RAM works. Screen brightness
>control would be nice... but I've measured the power the screen takes
>and it's not great and a wokaround is to select dark backgrounds.
>Solaris does use more power than Linux - it's not all the screen - 11W
>vs 7W which gives battery life of 3 hours vs 5 - but it beats my old
>Solaris 10 laptop which worked its way from 40 minutes to a 40 second
>battery life.

Brightness control works on my laptop. We don't have specific
drivers for each and all CPUs.

I'm surprised that it doesn't work; the only issue is that it seems
to forget where it is (so when it is on low brightness, you need to
want to lower it further before it realises what the current setting
is)

Casper

[Casper H.S. Dik]

>I have gone back and done another legacy install and attached is
>Xorg.0.log for comparison. The first lines that are different are:

I'm seeing that it doesn't able to create a "screen"; would it
help to create a xorg.conf while in bios mode and
use "Xorg -configure" to create a xorg.conf file and install
it and see if it is then usable under UEFI mode.

Casper

[James Lee]

On 09/04/2015 19:42, Casper H.S. Dik wrote:

>> A reason, possibly the only one, to keep the factory Windows on the
>> machine is the BIOS is only distributed as a MS exe file.
>
> Aren't those usually installable using DOS?

Whoa there, using DOS is outside of my knowledge. The download site says
it's for Win 7 or 8 and the installer drew pictures on the screen.

> Brightness control works on my laptop. We don't have specific
> drivers for each and all CPUs.
>
> I'm surprised that it doesn't work; the only issue is that it seems
> to forget where it is (so when it is on low brightness, you need to
> want to lower it further before it realises what the current setting
> is)

Pressing the brightness buttons brings up and moves the screen slider
widget but the brightness does not change - it is trying.

Linux Mint is same as installed and needs an addition to the boot file.
With Fedora brightness worked.



James.

[James Lee]

On 10/04/2015 09:16, Casper H.S. Dik wrote:

> I'm seeing that it doesn't able to create a "screen"; would it
> help to create a xorg.conf while in bios mode and
> use "Xorg -configure" to create a xorg.conf file and install
> it and see if it is then usable under UEFI mode.

I create a xorg.conf file in BIOS mode. Reinstalled UEFI text mode,
added solaris-desktop, put the file as /etc/X11/xorg.conf and tried
xinit / Xorg again but the result is the same.

Running "Xorg -configure" under UEFI produces a similar xorg.conf. as
with BIOS. I've fiddled with some setting in the xorg.conf file but
with no success.


James.

[x86...@gmail.com]

Using: usbcopy Solaris11.3 BETA - Which was launched July 7th 2015 (Also fails)

oradba@solaris-master:~/Downloads$ sudo usbcopy sol-11_3-beta-live-x86.usb
Password:

I mage type: dd-able x86

Found the following USB devices:

0: /dev/rdsk/c3t0d0p0 3.8 GB SanDisk Cruzer 1.20
1: /dev/rdsk/c4t0d0p0 3.8 GB SanDisk Cruzer 1.20
2: /dev/rdsk/c5t0d0p0 3.8 GB SanDisk Cruzer 1.20
3: /dev/rdsk/c6t0d0p0 3.8 GB SanDisk Cruzer 1.20

Enter the number of your choice: 1

WARNING: All data on your USB storage will be lost.
Are you sure you want to install to

SanDisk Cruzer 1.20, 3800 MB at /dev/rdsk/c4t0d0p0 ? (y/n) y

Copying and verifying image to USB device

Finished 1398 MB in 510 seconds (2.7MB/s)

0 block(s) re-written due to verification failure
Completed copy to USB

No Issues.... Art.S 07/12/2015

============================================================================

While GRUB2's menu comes up perfectly fine.

When I select any Solaris 11.3 USB
Solaris 11.3 SSH
Solaris 11.3 Text

I get the following error.


Problem: LENGTH OF STORED BLOCK DOES NOT MATCH

Analyst RTC: Turned out to be GRUB2
------------------------------------
https://docs.oracle.com/cd/E26502_01/html/E28983/gkvif.html
http://www.informit.com/articles/article.aspx?p=2103810&seqNum=10
http://sourcecodebrowser.com/grub2/1.92/gzio_8c_source.html
http://www.oracle.com/technetwork/articles/servers-storage-admin/howto-autoinstall-s11dot1-1866427.html

error the length of a store block does not match (look below..)

/* go to byte boundary */
00754 DUMPBITS (k & 7);
00755
00756 /* get the length and its complement */
00757 NEEDBITS (16);
00758 gzio->block_len = ((unsigned) b & 0xffff);
00759 DUMPBITS (16);
00760 NEEDBITS (16);
00761 if (gzio->block_len != (int) ((~b) & 0xffff))
00762 grub_error (GRUB_ERR_BAD_GZIP_DATA,
00763 "the length of a stored block does not match");
00764 DUMPBITS (16);
00765
00766 /* restore global variables */
00767 gzio->bb = b;
00768 gzio->bk = k;
00769 }

[cindy.sw...@gmail.com]

Hi--

I will get someone more knowledgeable to look at this error.

Its curious to me that the error is related to GRUB/gzip. Gzip compression is not currently supported on Solaris root pools, but you would have had to specify gzip compression in an auto install manifest unless you are dropping out of the text installer and setting it manually, which I doubt is the case.

Does the install get started at all or does it fail immediately?

I wonder if you drop out of the text installer and create an pool on the disk does it work:

# zpool create rpool c4t0d0

Thanks, Cindy