Re: [OpenBIOS] Solaris anyone?

[Nick Couchman]

> Nick - do you think you could do a quick test on this one?

This patch for the scsi_options seems to work fine - system boots with
the patch applied and with the line removed from /etc/system.

-Nick

--------
This e-mail may contain confidential and privileged material for the sole use of the intended recipient.  If this email is not intended for you, or you are not responsible for the delivery of this message to the intended recipient, please note that this message may contain SEAKR Engineering (SEAKR) Privileged/Proprietary Information.  In such a case, you are strictly prohibited from downloading, photocopying, distributing or otherwise using this message, its contents or attachments in any way.  If you have received this message in error, please notify us immediately by replying to this e-mail and delete the message from your mailbox.  Information contained in this message that does not relate to the business of SEAKR is neither endorsed by nor attributable to SEAKR.

Solaris anyone?

[Nathan Kunkee]

> Date: Tue, 26 Apr 2011 17:02:34 -0600
> From: [e-mail]
> To: [e-mail]
> Subject: Re: [OpenBIOS] Solaris anyone?
> > Nick - do you think you could do a quick test on this one?
>
> This patch for the scsi_options seems to work fine - system boots with
> the patch applied and with the line removed from /etc/system.
>
> -Nick

Mark, thank you for your hard work on this! It's very exciting!

I've built the latest openbios and qemu, but get the following error when I try and use it:

nathan@redwall:/usr/local/build/qemu$ sparc-softmmu/qemu-system-sparc -nographic -bios openbios-builtin.elf

Configuration device id QEMU version 16777216 machine id 8192
Incompatible configuration device version, freezing
QEMU: Terminated

I built openbios on a SPARC Ultra60, and QEMU on x86. Could that be the reason the version and machine ID got swapped?

Thanks,
Nathan

Solaris anyone?

[Mark Cave-Ayland]

On 27/04/11 00:02, Nick Couchman wrote:
> This patch for the scsi_options seems to work fine - system boots with
> the patch applied and with the line removed from /etc/system.
>
> -Nick

That's great - thanks for testing!

ATB,

Mark.

Solaris anyone?

[Mark Cave-Ayland]

On 27/04/11 16:58, Nathan Kunkee wrote:
> Mark, thank you for your hard work on this! It's very exciting!
>
> I've built the latest openbios and qemu, but get the following error
> when I try and use it:
>
> nathan@redwall:/usr/local/build/qemu$ sparc-softmmu/qemu-system-sparc
> -nographic -bios openbios-builtin.elf
>
> Configuration device id QEMU version 16777216 machine id 8192
> Incompatible configuration device version, freezing
> QEMU: Terminated
>
> I built openbios on a SPARC Ultra60, and QEMU on x86. Could that be the
> reason the version and machine ID got swapped?

Gosh - that definitely looks like some kind of build error. Which 
compiler did you use? Here I cross-compile on 64-bit x86 using GCC 4.5.

I've uploaded a fresh build of SVN trunk SPARC32 OpenBIOS to 
http://www.siriusit.co.uk/tmp/openbios-sparc32 for people to experiment 
with in the window before this makes the upstream QEMU git repository.

ATB,

Mark.

Solaris anyone? Q2

[Nathan Kunkee]

> Date: Wed, 27 Apr 2011 20:28:15 +0100
> From: [e-mail]
> To: [e-mail]
> Subject: Re: [OpenBIOS] Solaris anyone?
>
> On 27/04/11 16:58, Nathan Kunkee wrote:
> > Mark, thank you for your hard work on this! It's very exciting!

[..snip..]
> > I built openbios on a SPARC Ultra60, and QEMU on x86. Could that be the
> > reason the version and machine ID got swapped?
>
> Gosh - that definitely looks like some kind of build error. Which 
> compiler did you use? Here I cross-compile on 64-bit x86 using GCC 4.5.
>
> I've uploaded a fresh build of SVN trunk SPARC32 OpenBIOS to 
> http://www.siriusit.co.uk/tmp/openbios-sparc32 for people to experiment 
> with in the window before this makes the upstream QEMU git repository.
>
> ATB,
>
> Mark.

I built that on opensolaris (so sparc64) with gcc 4.5.2. No cross-compiler needed, although the config script thinks I'm using one.

Thank you for posting the link. That image does load, which is better than I managed, but somehow isn't seeing my solaris 9 cd or iso. BTW, QEMU crashes nicely when you boot from an empty cd device.

Next I'm trying to compare what you built with my image as I'm not sure how those values are getting byte-swapped.

Thanks,
Nathan

Solaris anyone? Q2

[Mark Cave-Ayland]

On 27/04/11 23:01, Nathan Kunkee wrote:
> I built that on opensolaris (so sparc64) with gcc 4.5.2. No
> cross-compiler needed, although the config script thinks I'm using one.

Ah that will probably be the problem - config/script/switch-arch 
contains the logic which detects how 
NATIVE_BITWIDTH_EQUALS_HOST_BITWIDTH and SWAP_ENDIANNESS should be set 
and adds them to CFLAGS in Makefile.target as appropriate.
> Thank you for posting the link. That image does load, which is better
> than I managed, but somehow isn't seeing my solaris 9 cd or iso. BTW,
> QEMU crashes nicely when you boot from an empty cd device.

Strange. Can you show us the command line you are using? From your 
previous email:

nathan@redwall:/usr/local/build/qemu$ sparc-softmmu/qemu-system-sparc 
-nographic -bios openbios-builtin.elf

You are definitely missing the "-boot d" option to force boot from 
cdrom, although this is just a guess as you have shown us any error 
messages :(

ATB,

Mark.

Solaris anyone? Q2

[Nathan Kunkee]

On 04/28/11 04:00 AM, Mark Cave-Ayland wrote:
> On 27/04/11 23:01, Nathan Kunkee wrote:
> > I built that on opensolaris (so sparc64) with gcc 4.5.2. No
> > cross-compiler needed, although the config script thinks I'm using one.
>
> Ah that will probably be the problem - config/script/switch-arch 
> contains the logic which detects how 
> NATIVE_BITWIDTH_EQUALS_HOST_BITWIDTH and SWAP_ENDIANNESS should be set 
> and adds them to CFLAGS in Makefile.target as appropriate.

OK. I'll start looking to see how to change that for by build environment.
> > Thank you for posting the link. That image does load, which is better
> > than I managed, but somehow isn't seeing my solaris 9 cd or iso. BTW,
> > QEMU crashes nicely when you boot from an empty cd device.
>
> Strange. Can you show us the command line you are using? From your 
> previous email:
>
> nathan@redwall:/usr/local/build/qemu$ sparc-softmmu/qemu-system-sparc 
> -nographic -bios openbios-builtin.elf
>
> You are definitely missing the "-boot d" option to force boot from 
> cdrom, although this is just a guess as you have shown us any error 
> messages :(

Oops; it seemed obvious when I wrote it. The trick is that the cd device 
is empty, or is full of 0. Running your sparc32 image:

nathan@fangorn:/usr/local/build/qemu$ sparc-softmmu/qemu-system-sparc 
-nographic -bios  openbios-sparc32-mark
Configuration device id QEMU version 1 machine id 32
CPUs: 1 x FMI,MB86904
UUID: 00000000-0000-0000-0000-000000000000
Welcome to OpenBIOS v1.0 built on Apr 27 2011 19:13
   Type 'help' for detailed information
Trying disk...
No valid state has been set by load or init-program

0 > boot cdrom: Unhandled Exception 0x0000002a
PC = 0xffd1101c NPC = 0xffd11020
Stopping execution
QEMU 0.14.50 monitor - type 'help' for more information
(qemu) info block
scsi0-cd2: type=cdrom removable=1 locked=0 [not inserted]
floppy0: type=floppy removable=1 locked=0 [not inserted]
sd0: type=floppy removable=1 locked=0 [not inserted]
(qemu) quit

Looking at it again, OpenBIOS is bombing out rather than QEMU. I found 
this by accident trying to get the right path to my host cdrom drive. 
So, '-boot d -cdrom /dev/null' on the command line produces the same 
error at the same place. I'm not sure why a zero device would cause this 
when a missing hard disk does not.

Hard drive full of 0:
nathan@fangorn:/usr/local/build/qemu$ sparc-softmmu/qemu-system-sparc 
-nographic -bios  openbios-sparc32-mark  -hda /dev/null

Configuration device id QEMU version 1 machine id 32
CPUs: 1 x FMI,MB86904
UUID: 00000000-0000-0000-0000-000000000000
Welcome to OpenBIOS v1.0 built on Apr 27 2011 19:13
   Type 'help' for detailed information
Trying disk...
Unhandled Exception 0x0000002a
PC = 0xffd1101c NPC = 0xffd11020
Stopping execution
QEMU 0.14.50 monitor - type 'help' for more information
(qemu) info block
scsi0-hd0: type=hd removable=0 file=/dev/null ro=0 drv=raw encrypted=0
scsi0-cd2: type=cdrom removable=1 locked=0 [not inserted]
floppy0: type=floppy removable=1 locked=0 [not inserted]
sd0: type=floppy removable=1 locked=0 [not inserted]
(qemu) quit

And the happy path I usually see, just a missing hard drive:
nathan@fangorn:/usr/local/build/qemu$ sparc-softmmu/qemu-system-sparc 
-nographic -bios  openbios-sparc32-mark

Configuration device id QEMU version 1 machine id 32
CPUs: 1 x FMI,MB86904
UUID: 00000000-0000-0000-0000-000000000000
Welcome to OpenBIOS v1.0 built on Apr 27 2011 19:13
   Type 'help' for detailed information
Trying disk...
No valid state has been set by load or init-program

0 > boot Trying disk...
No valid state has been set by load or init-program
  ok
0 >

Let me know if you need more information on this.

Thanks,
Nathan

Solaris anyone? Q2

[Mark Cave-Ayland]

On 28/04/11 13:52, Nathan Kunkee wrote:
> > Ah that will probably be the problem - config/script/switch-arch
> > contains the logic which detects how
> > NATIVE_BITWIDTH_EQUALS_HOST_BITWIDTH and SWAP_ENDIANNESS should be set
> > and adds them to CFLAGS in Makefile.target as appropriate.
>
> OK. I'll start looking to see how to change that for by build environment.

Great - thanks!
> > > Thank you for posting the link. That image does load, which is better
> > > than I managed, but somehow isn't seeing my solaris 9 cd or iso. BTW,
> > > QEMU crashes nicely when you boot from an empty cd device.
> >
> > Strange. Can you show us the command line you are using? From your
> > previous email:
> >
> > nathan@redwall:/usr/local/build/qemu$ sparc-softmmu/qemu-system-sparc
> > -nographic -bios openbios-builtin.elf
> >
> > You are definitely missing the "-boot d" option to force boot from
> > cdrom, although this is just a guess as you have shown us any error
> > messages :(
>
> Oops; it seemed obvious when I wrote it. The trick is that the cd device
> is empty, or is full of 0. Running your sparc32 image:
>
> nathan@fangorn:/usr/local/build/qemu$ sparc-softmmu/qemu-system-sparc
> -nographic -bios openbios-sparc32-mark
> Configuration device id QEMU version 1 machine id 32
> CPUs: 1 x FMI,MB86904
> UUID: 00000000-0000-0000-0000-000000000000
> Welcome to OpenBIOS v1.0 built on Apr 27 2011 19:13
> Type 'help' for detailed information
> Trying disk...
> No valid state has been set by load or init-program
>
> 0 > boot cdrom: Unhandled Exception 0x0000002a
> PC = 0xffd1101c NPC = 0xffd11020
> Stopping execution

But you're still missing "-boot d" on the command line so it won't 
default to the cdrom. If you want to boot from the cdrom using the 
OpenBIOS command line manually, you also need to give the slice (d) and 
so the command you need is:

boot cdrom:d

HTH,

Mark.

Solaris anyone? Q2

[Blue Swirl]

On Thu, Apr 28, 2011 at 3:52 PM, Nathan Kunkee  wrote:
> On 04/28/11 04:00 AM, Mark Cave-Ayland wrote:
> > On 27/04/11 23:01, Nathan Kunkee wrote:
> > > I built that on opensolaris (so sparc64) with gcc 4.5.2. No
> > > cross-compiler needed, although the config script thinks I'm using one.
> >
> > Ah that will probably be the problem - config/script/switch-arch contains
> > the logic which detects how NATIVE_BITWIDTH_EQUALS_HOST_BITWIDTH and
> > SWAP_ENDIANNESS should be set and adds them to CFLAGS in Makefile.target as
> > appropriate.
>
> OK. I'll start looking to see how to change that for by build environment.

What does 'uname -m' report on your machine? We only check for
sun4u/sparc64 or sparc/sparc32.

Solaris anyone? Q2

[Nathan Kunkee]

> > > > I built that on opensolaris (so sparc64) with gcc 4.5.2. No
> > > > cross-compiler needed, although the config script thinks I'm using one.
> > >
> > > Ah that will probably be the problem - config/script/switch-arch contains
> > > the logic which detects how NATIVE_BITWIDTH_EQUALS_HOST_BITWIDTH and
> > > SWAP_ENDIANNESS should be set and adds them to CFLAGS in Makefile.target as
> > > appropriate.
> >
> > OK. I'll start looking to see how to change that for by build environment.
>
> What does 'uname -m' report on your machine? We only check for
> sun4u/sparc64 or sparc/sparc32.

nathan@valhalla:/export/home/nathan/$ uname -m
sun4u
nathan@valhalla:/export/home/nathan/$ uname -mpi
sun4u sparc sun4u
> From SVN, the switch-arch script thinks it has to use a cross-compiler:

nathan@valhalla:/export/home/nathan/openbios-devel$ env LANG=C sh -x ./config/scripts/switch-arch sparc32
+ MOLPATH=/export/home/nathan//mol-0.9.71
+ [ xsparc32 = x -o sparc32 = -help ]
+ test -f utils/dist/debian/rules
+ chmod 755 utils/dist/debian/rules
+ chmod 755 config/scripts/switch-arch
+ chmod 755 config/scripts/reldir
+ test x = x
+ archname
+ + head VERSION
VERSION=1.0
+ echo Configuring OpenBIOS on sparc64 for sparc32
Configuring OpenBIOS on sparc64 for sparc32
+ target_list=
+ target_list= builtin-sparc32 unix-sparc32
+ arch_list=
+ + echo builtin-sparc32
+ sed s/.*-//g
arch=sparc32
+ test -f config/examples/sparc32_config.xml
+ grep -q sparc32
+ echo
+ arch_list= sparc32
+ + echo unix-sparc32
+ sed s/.*-//g
arch=sparc32
+ test -f config/examples/sparc32_config.xml
+ grep -q sparc32
+ echo sparc32
+ unix=no
+ builtin=no
+ plain=no
+ mol=no
+ briq=no
+ pearpc=no
+ qemu=no
+ xbox=no
+ cross=no
+ :
+ builtin=yes
+ :
+ [ sparc32 != sparc64 ]
+ continue
+ BASEARCH=sparc32
+ select_prefix sparc
ERROR: no sparc cross-compiler found !

I got around that by adding the suffix my build of gcc is using:

--- config/scripts/switch-arch  Thu Apr 28 17:18:40 2011
+++ config/scripts/switch-arch-works    Sat Apr 23 20:21:29 2011
@@ -97,7 +101,7 @@

select_prefix()
 {
-    for TARGET in ${1}-linux-gnu- ${1}-linux- ${1}-elf- ${1}-eabi-
+    for TARGET in ${1}-linux-gnu- ${1}-linux- ${1}-elf- ${1}-eabi- ${1}-sun-solaris2.11-
     do
         if type ${TARGET}gcc > /dev/null 2>&1
         then

I'm not sure if that is the best way to get it to use the native compiler. I can't think of a reason my sparc64 build tools couldn't build the sparc32 image.

Thanks,
Nathan

Solaris anyone? Q2

[Blue Swirl]

On Fri, Apr 29, 2011 at 1:26 AM, Nathan Kunkee  wrote:
> > > > > I built that on opensolaris (so sparc64) with gcc 4.5.2. No
> > > > > cross-compiler needed, although the config script thinks I'm using
> > > > > one.
> > > >
> > > > Ah that will probably be the problem - config/script/switch-arch
> > > > contains
> > > > the logic which detects how NATIVE_BITWIDTH_EQUALS_HOST_BITWIDTH and
> > > > SWAP_ENDIANNESS should be set and adds them to CFLAGS in
> > > > Makefile.target as
> > > > appropriate.
> > >
> > > OK. I'll start looking to see how to change that for by build
> > > environment.
> >
> > What does 'uname -m' report on your machine? We only check for
> > sun4u/sparc64 or sparc/sparc32.
>
> nathan@valhalla:/export/home/nathan/$ uname -m
> sun4u
> nathan@valhalla:/export/home/nathan/$ uname -mpi
> sun4u sparc sun4u
>
> From SVN, the switch-arch script thinks it has to use a cross-compiler:
>
> nathan@valhalla:/export/home/nathan/openbios-devel$ env LANG=C sh -x
> ./config/scripts/switch-arch sparc32
> MOLPATH=/export/home/nathan//mol-0.9.71
> [ xsparc32 = x -o sparc32 = -help ]
> test -f utils/dist/debian/rules
> chmod 755 utils/dist/debian/rules
> chmod 755 config/scripts/switch-arch
> chmod 755 config/scripts/reldir
> test x = x
> archname
> head VERSION
>
> VERSION=1.0
> echo Configuring OpenBIOS on sparc64 for sparc32
>
> Configuring OpenBIOS on sparc64 for sparc32
> target_list=
> target_list= builtin-sparc32 unix-sparc32
> arch_list=
> echo builtin-sparc32
> sed s/.*-//g
>
> arch=sparc32
> test -f config/examples/sparc32_config.xml
> grep -q sparc32
> echo
> arch_list= sparc32
> echo unix-sparc32
> sed s/.*-//g
>
> arch=sparc32
> test -f config/examples/sparc32_config.xml
> grep -q sparc32
> echo sparc32
> unix=no
> builtin=no
> plain=no
> mol=no
> briq=no
> pearpc=no
> qemu=no
> xbox=no
> cross=no
> :
> builtin=yes
> :
> [ sparc32 != sparc64 ]
> continue
> BASEARCH=sparc32
> select_prefix sparc
>
> ERROR: no sparc cross-compiler found !
>
> I got around that by adding the suffix my build of gcc is using:
>
> --- config/scripts/switch-arch  Thu Apr 28 17:18:40 2011
> +++ config/scripts/switch-arch-works    Sat Apr 23 20:21:29 2011
> @@ -97,7 +101,7 @@
>
> select_prefix()
>  {
> -    for TARGET in ${1}-linux-gnu- ${1}-linux- ${1}-elf- ${1}-eabi-
> +    for TARGET in ${1}-linux-gnu- ${1}-linux- ${1}-elf- ${1}-eabi-
> ${1}-sun-solaris2.11-
>      do
>          if type ${TARGET}gcc > /dev/null 2>&1
>          then
>
> I'm not sure if that is the best way to get it to use the native compiler. I
> can't think of a reason my sparc64 build tools couldn't build the sparc32
> image.

This actually happens with OpenBSD/sparc64, the compiler can't produce
32 bit binaries at all:
$ gcc -v
Reading specs from /usr/lib/gcc-lib/sparc64-unknown-openbsd4.8/4.2.1/specs
Target: sparc64-unknown-openbsd4.8
Configured with: OpenBSD/sparc64 system compiler
Thread model: posix
gcc version 4.2.1 20070719
$ touch empty.c
$ gcc -m32 -c empty.c
empty.c:1: error: -m32 is not supported by this configuration

But I think that this is not so common setup. So we should add logic
for trying -m32 or -m64 if the host has a compatible CPU except for
word size, or a cross compiler for wrong word size is found.

Solaris anyone? Q2

[Nathan Kunkee]

On 04/28/11 11:08 AM, Mark Cave-Ayland wrote:
> On 28/04/11 13:52, Nathan Kunkee wrote:
> > > Ah that will probably be the problem - config/script/switch-arch
> > > contains the logic which detects how
> > > NATIVE_BITWIDTH_EQUALS_HOST_BITWIDTH and SWAP_ENDIANNESS should be set
> > > and adds them to CFLAGS in Makefile.target as appropriate.
> >
> > OK. I'll start looking to see how to change that for by build 
> > environment.
>
> Great - thanks!

I figured it out!

/bin/sh, the default choice if you just run switch-arch, on OpenSolaris 
is ksh93, which silently fails on the $...bigendian tests. If I run 
switch-arch with /usr/xpg4/bin/sh or bash, it works fine and my builds 
from x86 and sparc are now correct and load. It looks like testing for 
$KSH_VERSION would be a way to find out if the shell isn't going to 
work, but I don't know if this would come up enough to warrant adding 
logic to check it.

Thanks,
Nathan

Solaris anyone? Q2

[Blue Swirl]

On Sat, Apr 30, 2011 at 4:55 AM, Nathan Kunkee  wrote:
> On 04/28/11 11:08 AM, Mark Cave-Ayland wrote:
> > On 28/04/11 13:52, Nathan Kunkee wrote:
> > > > Ah that will probably be the problem - config/script/switch-arch
> > > > contains the logic which detects how
> > > > NATIVE_BITWIDTH_EQUALS_HOST_BITWIDTH and SWAP_ENDIANNESS should be set
> > > > and adds them to CFLAGS in Makefile.target as appropriate.
> > >
> > > OK. I'll start looking to see how to change that for by build
> > > environment.
> >
> > Great - thanks!
>
> I figured it out!
>
> /bin/sh, the default choice if you just run switch-arch, on OpenSolaris is
> ksh93, which silently fails on the $...bigendian tests. If I run switch-arch
> with /usr/xpg4/bin/sh or bash, it works fine and my builds from x86 and
> sparc are now correct and load. It looks like testing for $KSH_VERSION would
> be a way to find out if the shell isn't going to work, but I don't know if
> this would come up enough to warrant adding logic to check it.

Maybe we have a bashism somewhere (should be fixed), or more likely
since ksh93 is not fully Posix compatible, we use some Posix features
which are not implemented by ksh93 (maybe avoidable). Please send
again the output from sh -x for the bigendian test.

Solaris anyone? Q2

[Nathan Kunkee]

> > > > > Ah that will probably be the problem - config/script/switch-arch
> > > > > contains the logic which detects how
> > > > > NATIVE_BITWIDTH_EQUALS_HOST_BITWIDTH and SWAP_ENDIANNESS should be set
> > > > > and adds them to CFLAGS in Makefile.target as appropriate.
> > > >
> > > > OK. I'll start looking to see how to change that for by build
> > > > environment.
> > >
> > > Great - thanks!
> >
> > I figured it out!
> >
> > /bin/sh, the default choice if you just run switch-arch, on OpenSolaris is
> > ksh93, which silently fails on the $...bigendian tests. If I run switch-arch
> > with /usr/xpg4/bin/sh or bash, it works fine and my builds from x86 and
> > sparc are now correct and load. It looks like testing for $KSH_VERSION would
> > be a way to find out if the shell isn't going to work, but I don't know if
> > this would come up enough to warrant adding logic to check it.
>
> Maybe we have a bashism somewhere (should be fixed), or more likely
> since ksh93 is not fully Posix compatible, we use some Posix features
> which are not implemented by ksh93 (maybe avoidable). Please send
> again the output from sh -x for the bigendian test.

As run on my Ultra60 which should set CONFIG_BIG_ENDIAN:

nathan@valhalla:/export/home/nathan/openbios-devel$ LANG=C /bin/sh -x 
./config/scripts/switch-arch sparc32
+ MOLPATH=/export/home/nathan//mol-0.9.71
+ [ xsparc32 = x -o sparc32 = -help ]
+ test -f utils/dist/debian/rules
+ chmod 755 utils/dist/debian/rules
+ chmod 755 config/scripts/switch-arch
+ chmod 755 config/scripts/reldir
+ test x = x
+ archname
+ sed -e s/i.86/x86/ -e s/i86pc/x86/ -e s/sun4u/sparc64/ -e 
's/sparc$/sparc32/' -e 's/arm.*/arm/' -e s/sa110/arm/ -e s/x86_64/amd64/ 
-e 's/Power Macintosh/ppc/'
+ uname -m
+ HOSTARCH=sparc64
+ head VERSION
+ VERSION=1.0
+ echo 'Configuring OpenBIOS on sparc64 for sparc32'
Configuring OpenBIOS on sparc64 for sparc32
+ target_list=''
+ target_list=' builtin-sparc32 unix-sparc32'
+ arch_list=''
+ sed 's/.*-//g'
+ echo builtin-sparc32
+ arch=sparc32
+ test -f config/examples/sparc32_config.xml
+ grep -q sparc32
+ echo
+ arch_list=' sparc32'
+ sed 's/.*-//g'
+ echo unix-sparc32
+ arch=sparc32
+ test -f config/examples/sparc32_config.xml
+ grep -q sparc32
+ echo sparc32
+ unix=no
+ builtin=no
+ plain=no
+ mol=no
+ briq=no
+ pearpc=no
+ qemu=no
+ xbox=no
+ cross=no
+ :
+ builtin=yes
+ :
+ [ sparc32 != sparc64 ]
+ continue
+ BASEARCH=sparc32
+ select_prefix sparc
+ whence -v sparc-linux-gnu-gcc
+ 1> /dev/null 2>& 1
+ whence -v sparc-linux-gcc
+ 1> /dev/null 2>& 1
+ whence -v sparc-elf-gcc
+ 1> /dev/null 2>& 1
+ whence -v sparc-eabi-gcc
+ 1> /dev/null 2>& 1
+ whence -v sparc-sun-solaris2.11-gcc
+ 1> /dev/null 2>& 1
+ return
+ CFLAGS='-Wa,-xarch=v8 -Wa,-32 -m32 -mcpu=supersparc -fno-builtin'
+ AS_FLAGS='-Wa,-xarch=v8 -Wa,-32'
+ uname -s
+ [ sparc32 != sparc64 -o SunOS = Darwin ]
+ cross=yes
+ crosscflags sparc64 sparc32
+ local host=sparc64
./config/scripts/switch-arch[22]: local: not found [No such file or 
directory]
+ local target=sparc32
./config/scripts/switch-arch[23]: local: not found [No such file or 
directory]
+ test '' = powerpc -o '' = ppc -o '' = mips -o '' = s390 -o '' = 
sparc32 -o '' = sparc64 -o '' = m68k -o '' = armv4b
+ hostbigendian=no
+ test '' = sparc64 -o '' = ia64 -o '' = amd64 -o '' = x86_64 -o '' = alpha
+ hostlongbits=32
+ test unix-sparc32 = powerpc -o unix-sparc32 = ppc -o unix-sparc32 = 
powerpc64 -o unix-sparc32 = ppc64 -o unix-sparc32 = mips -o unix-sparc32 
= s390 -o unix-sparc32 = sparc32 -o unix-sparc32 = sparc64 -o 
unix-sparc32 = m68k -o unix-sparc32 = armv4b
+ targetbigendian=no
+ test unix-sparc32 = sparc64 -o unix-sparc32 = ia64 -o unix-sparc32 = 
amd64 -o unix-sparc32 = x86_64 -o unix-sparc32 = powerpc64 -o 
unix-sparc32 = ppc64 -o unix-sparc32 = alpha
+ targetlongbits=32
+ echo 'test same endian -target -host ' no no
test same endian -target -host  no no
+ test no = no
+ cflags=-USWAP_ENDIANNESS
+ test 32 = 32
+ cflags='-USWAP_ENDIANNESS -DNATIVE_BITWIDTH_EQUALS_HOST_BITWIDTH'
+ test unix-sparc32 = sparc64 -o unix-sparc32 = ia64 -o unix-sparc32 = 
amd64 -o unix-sparc32 = x86_64 -o unix-sparc32 = alpha
+ CROSSCFLAGS='-USWAP_ENDIANNESS -DNATIVE_BITWIDTH_EQUALS_HOST_BITWIDTH'
+ OBJDIR=obj-sparc32
+ printf 'Initializing build tree obj-sparc32...'
Initializing build tree obj-sparc32...+ rm -rf obj-sparc32
+ mkdir obj-sparc32
+ mkdir -p obj-sparc32/target
+ mkdir -p obj-sparc32/target/include
+ mkdir -p obj-sparc32/target/arch
+ mkdir -p obj-sparc32/target/arch/unix
+ mkdir -p obj-sparc32/target/arch/sparc32
+ mkdir -p obj-sparc32/target/libgcc
+ mkdir -p obj-sparc32/target/kernel
+ mkdir -p obj-sparc32/target/libopenbios
+ mkdir -p obj-sparc32/target/packages
+ mkdir -p obj-sparc32/target/fs
+ mkdir -p obj-sparc32/target/fs/grubfs
+ mkdir -p obj-sparc32/target/fs/hfs
+ mkdir -p obj-sparc32/target/fs/hfsplus
+ mkdir -p obj-sparc32/target/fs/iso9660
+ mkdir -p obj-sparc32/target/fs/ext2
+ mkdir -p obj-sparc32/target/drivers
+ mkdir -p obj-sparc32/target/libc
+ mkdir -p obj-sparc32/host/include
+ mkdir -p obj-sparc32/host/kernel
+ mkdir -p obj-sparc32/forth
+ ln -s ../../../include/arch/sparc32 obj-sparc32/target/include/asm
+ [ no = yes ]
+ [ no = yes ]
+ [ no = yes ]
+ [ no = yes ]
+ [ no = yes ]
+ echo ok.
ok.
+ cd obj-sparc32
+ SRCDIR=..
+ ODIR=.
+ printf 'Creating target Makefile...'
Creating target Makefile...+ echo ARCH=sparc32
+ 1> ./config.mak
+ [ yes = yes ]
+ echo TARGET=sparc-sun-solaris2.11-
+ 1>> ./config.mak
+ echo CFLAGS='-Wa,-xarch=v8 -Wa,-32 -m32 -mcpu=supersparc -fno-builtin'
+ 1>> ./config.mak
+ echo AS_FLAGS='-Wa,-xarch=v8 -Wa,-32'
+ 1>> ./config.mak
+ echo 'HOSTARCH?=sparc64'
+ 1>> ./config.mak
+ echo CROSSCFLAGS='-USWAP_ENDIANNESS 
-DNATIVE_BITWIDTH_EQUALS_HOST_BITWIDTH'
+ 1>> ./config.mak
+ echo VERSION='"1.0"'
+ 1>> ./config.mak
+ ln -s ../config/xml/rules.xml ./rules.xml
+ echo '<?xml version="1.0"?><config>'
+ 1> ./config.xml
+ config_set_boolean CONFIG_sparc32
+ 1>> ./config.xml
+ echo CONFIG_sparc32
+ tr a-z A-Z
+ option=CONFIG_SPARC32
+ echo '<option name="CONFIG_SPARC32" type="boolean" value="true" />'
+ [ sparc32 != sparc32 ]
+ [ no = yes ]
+ [ no = yes ]
+ [ no = yes ]
+ [ no = yes ]
+ [ no = yes ]
+ echo 'config test -target ' no
config test -target  no
+ [ no = yes ]
+ config_set_boolean CONFIG_LITTLE_ENDIAN
+ 1>> ./config.xml
+ echo CONFIG_LITTLE_ENDIAN
+ tr a-z A-Z
+ option=CONFIG_LITTLE_ENDIAN
+ echo '<option name="CONFIG_LITTLE_ENDIAN" type="boolean" value="true" />'
+ [ no = yes ]
+ [ yes = yes ]
+ config_set_boolean CONFIG_IMAGE_ELF_EMBEDDED
+ 1>> ./config.xml
+ echo CONFIG_IMAGE_ELF_EMBEDDED
+ tr a-z A-Z
+ option=CONFIG_IMAGE_ELF_EMBEDDED
+ echo '<option name="CONFIG_IMAGE_ELF_EMBEDDED" type="boolean" 
value="true" />'
+ [ no = yes ]
+ cat ../config/examples/sparc32_config.xml
+ 1>> ./config.xml
+ echo '</config>'
+ 1>> ./config.xml
+ ln -s ../Makefile.target ./Makefile
+ xsltproc ../config/xml/xinclude.xsl ../build.xml
+ 1> ./build-full.xml
+ xsltproc ../config/xml/makefile.xsl ./build-full.xml
+ 1> ./rules.mak
+ echo ok.
ok.
+ printf 'Creating config files...'
Creating config files...+ xsltproc ../config/xml/config-c.xsl ./config.xml
+ 1> ./host/include/autoconf.h
+ xsltproc ../config/xml/config-c.xsl ./config.xml
+ 1> ./target/include/autoconf.h
+ xsltproc ../config/xml/config-forth.xsl ./config.xml
+ 1> ./forth/config.fs
+ echo ok.
ok.
+ cd ..

Nathan

Solaris anyone? Q2

[Andreas Färber]

Am 02.05.2011 um 18:20 schrieb Nathan Kunkee:
> > > /bin/sh, the default choice if you just run switch-arch, on  
> > > OpenSolaris is
> > > ksh93, which silently fails on the $...bigendian tests. If I run  
> > > switch-arch
> > > with /usr/xpg4/bin/sh or bash, it works fine and my builds from  
> > > x86 and
> > > sparc are now correct and load. It looks like testing for  
> > > $KSH_VERSION would
> > > be a way to find out if the shell isn't going to work, but I don't  
> > > know if
> > > this would come up enough to warrant adding logic to check it.
> >
> > Maybe we have a bashism somewhere (should be fixed), or more likely
> > since ksh93 is not fully Posix compatible, we use some Posix features
> > which are not implemented by ksh93 (maybe avoidable). Please send
> > again the output from sh -x for the bigendian test.
>
> As run on my Ultra60 which should set CONFIG_BIG_ENDIAN:
>
> nathan@valhalla:/export/home/nathan/openbios-devel$ LANG=C /bin/sh - 
> x ./config/scripts/switch-arch sparc32
> MOLPATH=/export/home/nathan//mol-0.9.71
> [ xsparc32 = x -o sparc32 = -help ]
> test -f utils/dist/debian/rules
> chmod 755 utils/dist/debian/rules
> chmod 755 config/scripts/switch-arch
> chmod 755 config/scripts/reldir
> test x = x
> archname
> sed -e s/i.86/x86/ -e s/i86pc/x86/ -e s/sun4u/sparc64/ -e 's/sparc
>
> $/sparc32/' -e 's/arm.*/arm/' -e s/sa110/arm/ -e s/x86_64/amd64/ -e  
> 's/Power Macintosh/ppc/'
> uname -m
> HOSTARCH=sparc64
> head VERSION
> VERSION=1.0
> echo 'Configuring OpenBIOS on sparc64 for sparc32'
>
> Configuring OpenBIOS on sparc64 for sparc32
> target_list=''
> target_list=' builtin-sparc32 unix-sparc32'
> arch_list=''
> sed 's/.*-//g'
> echo builtin-sparc32
> arch=sparc32
> test -f config/examples/sparc32_config.xml
> grep -q sparc32
> echo
> arch_list=' sparc32'
> sed 's/.*-//g'
> echo unix-sparc32
> arch=sparc32
> test -f config/examples/sparc32_config.xml
> grep -q sparc32
> echo sparc32
> unix=no
> builtin=no
> plain=no
> mol=no
> briq=no
> pearpc=no
> qemu=no
> xbox=no
> cross=no
> :
> builtin=yes
> :
> [ sparc32 != sparc64 ]
> continue
> BASEARCH=sparc32
> select_prefix sparc
> whence -v sparc-linux-gnu-gcc
> 1> /dev/null 2>& 1
> whence -v sparc-linux-gcc
> 1> /dev/null 2>& 1
> whence -v sparc-elf-gcc
> 1> /dev/null 2>& 1
> whence -v sparc-eabi-gcc
> 1> /dev/null 2>& 1
> whence -v sparc-sun-solaris2.11-gcc
> 1> /dev/null 2>& 1
> return
> CFLAGS='-Wa,-xarch=v8 -Wa,-32 -m32 -mcpu=supersparc -fno-builtin'
> AS_FLAGS='-Wa,-xarch=v8 -Wa,-32'
> uname -s
> [ sparc32 != sparc64 -o SunOS = Darwin ]
> cross=yes
> crosscflags sparc64 sparc32
> local host=sparc64
>
> ./config/scripts/switch-arch[22]: local: not found [No such file or  
> directory]

;-)

Solaris anyone? Q2

[Kenneth Salerno]

----- Original Message -----

From: Nathan Kunkee 
To: The OpenBIOS Mailinglist 
Cc: 
Sent: Monday, May 2, 2011 12:20 PM
Subject: Re: [OpenBIOS] Solaris anyone? Q2
> > > > > Ah that will probably be the problem - config/script/switch-arch
> > > > > contains the logic which detects how
> > > > > NATIVE_BITWIDTH_EQUALS_HOST_BITWIDTH and SWAP_ENDIANNESS should be set
> > > > > and adds them to CFLAGS in Makefile.target as appropriate.
> > > >
> > > > OK. I'll start looking to see how to change that for by build
> > > > environment.
> > >
> > > Great - thanks!
> >
> > I figured it out!
> >
> > /bin/sh, the default choice if you just run switch-arch, on OpenSolaris is
> > ksh93, which silently fails on the $...bigendian tests. If I run switch-arch
> > with /usr/xpg4/bin/sh or bash, it works fine and my builds from x86 and
> > sparc are now correct and load. It looks like testing for $KSH_VERSION would
> > be a way to find out if the shell isn't going to work, but I don't know if
> > this would come up enough to warrant adding logic to check it.
>
> Maybe we have a bashism somewhere (should be fixed), or more likely
> since ksh93 is not fully Posix compatible, we use some Posix features
> which are not implemented by ksh93 (maybe avoidable). Please send
> again the output from sh -x for the bigendian test.

As run on my Ultra60 which should set CONFIG_BIG_ENDIAN:

nathan@valhalla:/export/home/nathan/openbios-devel$ LANG=C /bin/sh -x ./config/scripts/switch-arch sparc32
+ MOLPATH=/export/home/nathan//mol-0.9.71
+ [ xsparc32 = x -o sparc32 = -help ]
+ test -f utils/dist/debian/rules
+ chmod 755 utils/dist/debian/rules
+ chmod 755 config/scripts/switch-arch
+ chmod 755 config/scripts/reldir
+ test x = x
+ archname
+ sed -e s/i.86/x86/ -e s/i86pc/x86/ -e s/sun4u/sparc64/ -e 's/sparc$/sparc32/' -e 's/arm.*/arm/' -e s/sa110/arm/ -e s/x86_64/amd64/ -e 's/Power Macintosh/ppc/'
+ uname -m
+ HOSTARCH=sparc64
+ head VERSION
+ VERSION=1.0
+ echo 'Configuring OpenBIOS on sparc64 for sparc32'
Configuring OpenBIOS on sparc64 for sparc32
+ target_list=''
+ target_list=' builtin-sparc32 unix-sparc32'
+ arch_list=''
+ sed 's/.*-//g'
+ echo builtin-sparc32
+ arch=sparc32
+ test -f config/examples/sparc32_config.xml
+ grep -q sparc32
+ echo
+ arch_list=' sparc32'
+ sed 's/.*-//g'
+ echo unix-sparc32
+ arch=sparc32
+ test -f config/examples/sparc32_config.xml
+ grep -q sparc32
+ echo sparc32
+ unix=no
+ builtin=no
+ plain=no
+ mol=no
+ briq=no
+ pearpc=no
+ qemu=no
+ xbox=no
+ cross=no
+ :
+ builtin=yes
+ :
+ [ sparc32 != sparc64 ]
+ continue
+ BASEARCH=sparc32
+ select_prefix sparc
+ whence -v sparc-linux-gnu-gcc
+ 1> /dev/null 2>& 1
+ whence -v sparc-linux-gcc
+ 1> /dev/null 2>& 1
+ whence -v sparc-elf-gcc
+ 1> /dev/null 2>& 1
+ whence -v sparc-eabi-gcc
+ 1> /dev/null 2>& 1
+ whence -v sparc-sun-solaris2.11-gcc
+ 1> /dev/null 2>& 1
+ return
+ CFLAGS='-Wa,-xarch=v8 -Wa,-32 -m32 -mcpu=supersparc -fno-builtin'
+ AS_FLAGS='-Wa,-xarch=v8 -Wa,-32'
+ uname -s
+ [ sparc32 != sparc64 -o SunOS = Darwin ]
+ cross=yes
+ crosscflags sparc64 sparc32
+ local host=sparc64
./config/scripts/switch-arch[22]: local: not found [No such file or directory]
+ local target=sparc32
./config/scripts/switch-arch[23]: local: not found [No such file or directory]
+ test '' = powerpc -o '' = ppc -o '' = mips -o '' = s390 -o '' = sparc32 -o '' = sparc64 -o '' = m68k -o '' = armv4b
+ hostbigendian=no
+ test '' = sparc64 -o '' = ia64 -o '' = amd64 -o '' = x86_64 -o '' = alpha
+ hostlongbits=32
+ test unix-sparc32 = powerpc -o unix-sparc32 = ppc -o unix-sparc32 = powerpc64 -o unix-sparc32 = ppc64 -o unix-sparc32 = mips -o unix-sparc32 = s390 -o unix-sparc32 = sparc32 -o unix-sparc32 = sparc64 -o unix-sparc32 = m68k -o unix-sparc32 = armv4b
+ targetbigendian=no
+ test unix-sparc32 = sparc64 -o unix-sparc32 = ia64 -o unix-sparc32 = amd64 -o unix-sparc32 = x86_64 -o unix-sparc32 = powerpc64 -o unix-sparc32 = ppc64 -o unix-sparc32 = alpha
+ targetlongbits=32
+ echo 'test same endian -target -host ' no no
test same endian -target -host  no no
+ test no = no
+ cflags=-USWAP_ENDIANNESS
+ test 32 = 32
+ cflags='-USWAP_ENDIANNESS -DNATIVE_BITWIDTH_EQUALS_HOST_BITWIDTH'
+ test unix-sparc32 = sparc64 -o unix-sparc32 = ia64 -o unix-sparc32 = amd64 -o unix-sparc32 = x86_64 -o unix-sparc32 = alpha
+ CROSSCFLAGS='-USWAP_ENDIANNESS -DNATIVE_BITWIDTH_EQUALS_HOST_BITWIDTH'
+ OBJDIR=obj-sparc32
+ printf 'Initializing build tree obj-sparc32...'
Initializing build tree obj-sparc32...+ rm -rf obj-sparc32
+ mkdir obj-sparc32
+ mkdir -p obj-sparc32/target
+ mkdir -p obj-sparc32/target/include
+ mkdir -p obj-sparc32/target/arch
+ mkdir -p obj-sparc32/target/arch/unix
+ mkdir -p obj-sparc32/target/arch/sparc32
+ mkdir -p obj-sparc32/target/libgcc
+ mkdir -p obj-sparc32/target/kernel
+ mkdir -p obj-sparc32/target/libopenbios
+ mkdir -p obj-sparc32/target/packages
+ mkdir -p obj-sparc32/target/fs
+ mkdir -p obj-sparc32/target/fs/grubfs
+ mkdir -p obj-sparc32/target/fs/hfs
+ mkdir -p obj-sparc32/target/fs/hfsplus
+ mkdir -p obj-sparc32/target/fs/iso9660
+ mkdir -p obj-sparc32/target/fs/ext2
+ mkdir -p obj-sparc32/target/drivers
+ mkdir -p obj-sparc32/target/libc
+ mkdir -p obj-sparc32/host/include
+ mkdir -p obj-sparc32/host/kernel
+ mkdir -p obj-sparc32/forth
+ ln -s ../../../include/arch/sparc32 obj-sparc32/target/include/asm
+ [ no = yes ]
+ [ no = yes ]
+ [ no = yes ]
+ [ no = yes ]
+ [ no = yes ]
+ echo ok.
ok.
+ cd obj-sparc32
+ SRCDIR=..
+ ODIR=.
+ printf 'Creating target Makefile...'
Creating target Makefile...+ echo ARCH=sparc32
+ 1> ./config.mak
+ [ yes = yes ]
+ echo TARGET=sparc-sun-solaris2.11-
+ 1>> ./config.mak
+ echo CFLAGS='-Wa,-xarch=v8 -Wa,-32 -m32 -mcpu=supersparc -fno-builtin'
+ 1>> ./config.mak
+ echo AS_FLAGS='-Wa,-xarch=v8 -Wa,-32'
+ 1>> ./config.mak
+ echo 'HOSTARCH?=sparc64'
+ 1>> ./config.mak
+ echo CROSSCFLAGS='-USWAP_ENDIANNESS -DNATIVE_BITWIDTH_EQUALS_HOST_BITWIDTH'
+ 1>> ./config.mak
+ echo VERSION='"1.0"'
+ 1>> ./config.mak
+ ln -s ../config/xml/rules.xml ./rules.xml
+ echo '<?xml version="1.0"?><config>'
+ 1> ./config.xml
+ config_set_boolean CONFIG_sparc32
+ 1>> ./config.xml
+ echo CONFIG_sparc32
+ tr a-z A-Z
+ option=CONFIG_SPARC32
+ echo '<option name="CONFIG_SPARC32" type="boolean" value="true" />'
+ [ sparc32 != sparc32 ]
+ [ no = yes ]
+ [ no = yes ]
+ [ no = yes ]
+ [ no = yes ]
+ [ no = yes ]
+ echo 'config test -target ' no
config test -target  no
+ [ no = yes ]
+ config_set_boolean CONFIG_LITTLE_ENDIAN
+ 1>> ./config.xml
+ echo CONFIG_LITTLE_ENDIAN
+ tr a-z A-Z
+ option=CONFIG_LITTLE_ENDIAN
+ echo '<option name="CONFIG_LITTLE_ENDIAN" type="boolean" value="true" />'
+ [ no = yes ]
+ [ yes = yes ]
+ config_set_boolean CONFIG_IMAGE_ELF_EMBEDDED
+ 1>> ./config.xml
+ echo CONFIG_IMAGE_ELF_EMBEDDED
+ tr a-z A-Z
+ option=CONFIG_IMAGE_ELF_EMBEDDED
+ echo '<option name="CONFIG_IMAGE_ELF_EMBEDDED" type="boolean" value="true" />'
+ [ no = yes ]
+ cat ../config/examples/sparc32_config.xml
+ 1>> ./config.xml
+ echo '</config>'
+ 1>> ./config.xml
+ ln -s ../Makefile.target ./Makefile
+ xsltproc ../config/xml/xinclude.xsl ../build.xml
+ 1> ./build-full.xml
+ xsltproc ../config/xml/makefile.xsl ./build-full.xml
+ 1> ./rules.mak
+ echo ok.
ok.
+ printf 'Creating config files...'
Creating config files...+ xsltproc ../config/xml/config-c.xsl ./config.xml
+ 1> ./host/include/autoconf.h
+ xsltproc ../config/xml/config-c.xsl ./config.xml
+ 1> ./target/include/autoconf.h
+ xsltproc ../config/xml/config-forth.xsl ./config.xml
+ 1> ./forth/config.fs
+ echo ok.
ok.
+ cd ..

Nathan

-- OpenBIOS                 http://openbios.org/
Mailinglist:  http://lists.openbios.org/mailman/listinfo
Free your System - May the Forth be with you

This would fix the script to work on Solaris:

--- switch-arch.orig    2011-05-02 20:54:43.312500000 -0400
+++ switch-arch 2011-05-02 20:56:05.093750000 -0400
@@ -19,8 +19,8 @@
 
 crosscflags()
 {
-    local host=$1
-    local target=$2
+    host=$1
+    target=$2
 
     if test "$host" = "powerpc" -o "$host" = "ppc" \
         -o "$host" = "mips" -o "$host" = "s390" \

Ken

Solaris anyone? Q2

[Nathan Kunkee]

> This would fix the script to work on Solaris:
>
> --- switch-arch.orig    2011-05-02 20:54:43.312500000 -0400
> +++ switch-arch 2011-05-02 20:56:05.093750000 -0400
> @@ -19,8 +19,8 @@
>
> crosscflags()
>   {
> local host=$1
> local target=$2
>
> host=$1
>
> target=$2
>
> if test "$host" = "powerpc" -o "$host" = "ppc" \
>      -o "$host" = "mips" -o "$host" = "s390" \
>
> Ken

Confirmed; this produces the correct result with /bin/sh (ksh93), bash, 
and /usr/xpg4/bin/sh.

Thanks,
Nathan

Solaris anyone? Q2

[Blue Swirl]

On Tue, May 3, 2011 at 3:57 AM, Kenneth Salerno
 wrote:
> ----- Original Message -----
>
> From: Nathan Kunkee 
> To: The OpenBIOS Mailinglist 
> Cc:
> Sent: Monday, May 2, 2011 12:20 PM
> Subject: Re: [OpenBIOS] Solaris anyone? Q2
> > > > > > Ah that will probably be the problem - config/script/switch-arch
> > > > > > contains the logic which detects how
> > > > > > NATIVE_BITWIDTH_EQUALS_HOST_BITWIDTH and SWAP_ENDIANNESS should be set
> > > > > > and adds them to CFLAGS in Makefile.target as appropriate.
> > > > >
> > > > > OK. I'll start looking to see how to change that for by build
> > > > > environment.
> > > >
> > > > Great - thanks!
> > >
> > > I figured it out!
> > >
> > > /bin/sh, the default choice if you just run switch-arch, on OpenSolaris is
> > > ksh93, which silently fails on the $...bigendian tests. If I run switch-arch
> > > with /usr/xpg4/bin/sh or bash, it works fine and my builds from x86 and
> > > sparc are now correct and load. It looks like testing for $KSH_VERSION would
> > > be a way to find out if the shell isn't going to work, but I don't know if
> > > this would come up enough to warrant adding logic to check it.
> >
> > Maybe we have a bashism somewhere (should be fixed), or more likely
> > since ksh93 is not fully Posix compatible, we use some Posix features
> > which are not implemented by ksh93 (maybe avoidable). Please send
> > again the output from sh -x for the bigendian test.
>
> As run on my Ultra60 which should set CONFIG_BIG_ENDIAN:
>
> nathan@valhalla:/export/home/nathan/openbios-devel$ LANG=C /bin/sh -x ./config/scripts/switch-arch sparc32
> MOLPATH=/export/home/nathan//mol-0.9.71
> [ xsparc32 = x -o sparc32 = -help ]
> test -f utils/dist/debian/rules
> chmod 755 utils/dist/debian/rules
> chmod 755 config/scripts/switch-arch
> chmod 755 config/scripts/reldir
> test x = x
> archname
> sed -e s/i.86/x86/ -e s/i86pc/x86/ -e s/sun4u/sparc64/ -e 's/sparc$/sparc32/' -e 's/arm.*/arm/' -e s/sa110/arm/ -e s/x86_64/amd64/ -e 's/Power Macintosh/ppc/'
> uname -m
> HOSTARCH=sparc64
> head VERSION
> VERSION=1.0
> echo 'Configuring OpenBIOS on sparc64 for sparc32'
>
> Configuring OpenBIOS on sparc64 for sparc32
> target_list=''
> target_list=' builtin-sparc32 unix-sparc32'
> arch_list=''
> sed 's/.*-//g'
> echo builtin-sparc32
> arch=sparc32
> test -f config/examples/sparc32_config.xml
> grep -q sparc32
> echo
> arch_list=' sparc32'
> sed 's/.*-//g'
> echo unix-sparc32
> arch=sparc32
> test -f config/examples/sparc32_config.xml
> grep -q sparc32
> echo sparc32
> unix=no
> builtin=no
> plain=no
> mol=no
> briq=no
> pearpc=no
> qemu=no
> xbox=no
> cross=no
> :
> builtin=yes
> :
> [ sparc32 != sparc64 ]
> continue
> BASEARCH=sparc32
> select_prefix sparc
> whence -v sparc-linux-gnu-gcc
> 1> /dev/null 2>& 1
> whence -v sparc-linux-gcc
> 1> /dev/null 2>& 1
> whence -v sparc-elf-gcc
> 1> /dev/null 2>& 1
> whence -v sparc-eabi-gcc
> 1> /dev/null 2>& 1
> whence -v sparc-sun-solaris2.11-gcc
> 1> /dev/null 2>& 1
> return
> CFLAGS='-Wa,-xarch=v8 -Wa,-32 -m32 -mcpu=supersparc -fno-builtin'
> AS_FLAGS='-Wa,-xarch=v8 -Wa,-32'
> uname -s
> [ sparc32 != sparc64 -o SunOS = Darwin ]
> cross=yes
> crosscflags sparc64 sparc32
> local host=sparc64
>
> ./config/scripts/switch-arch[22]: local: not found [No such file or directory]
> local target=sparc32
>
> ./config/scripts/switch-arch[23]: local: not found [No such file or directory]
> test '' = powerpc -o '' = ppc -o '' = mips -o '' = s390 -o '' = sparc32 -o '' = sparc64 -o '' = m68k -o '' = armv4b
> hostbigendian=no
> test '' = sparc64 -o '' = ia64 -o '' = amd64 -o '' = x86_64 -o '' = alpha
> hostlongbits=32
> test unix-sparc32 = powerpc -o unix-sparc32 = ppc -o unix-sparc32 = powerpc64 -o unix-sparc32 = ppc64 -o unix-sparc32 = mips -o unix-sparc32 = s390 -o unix-sparc32 = sparc32 -o unix-sparc32 = sparc64 -o unix-sparc32 = m68k -o unix-sparc32 = armv4b
> targetbigendian=no
> test unix-sparc32 = sparc64 -o unix-sparc32 = ia64 -o unix-sparc32 = amd64 -o unix-sparc32 = x86_64 -o unix-sparc32 = powerpc64 -o unix-sparc32 = ppc64 -o unix-sparc32 = alpha
> targetlongbits=32
> echo 'test same endian -target -host ' no no
>
> test same endian -target -host  no no
> test no = no
> cflags=-USWAP_ENDIANNESS
> test 32 = 32
> cflags='-USWAP_ENDIANNESS -DNATIVE_BITWIDTH_EQUALS_HOST_BITWIDTH'
> test unix-sparc32 = sparc64 -o unix-sparc32 = ia64 -o unix-sparc32 = amd64 -o unix-sparc32 = x86_64 -o unix-sparc32 = alpha
> CROSSCFLAGS='-USWAP_ENDIANNESS -DNATIVE_BITWIDTH_EQUALS_HOST_BITWIDTH'
> OBJDIR=obj-sparc32
> printf 'Initializing build tree obj-sparc32...'
>
> Initializing build tree obj-sparc32...+ rm -rf obj-sparc32
> mkdir obj-sparc32
> mkdir -p obj-sparc32/target
> mkdir -p obj-sparc32/target/include
> mkdir -p obj-sparc32/target/arch
> mkdir -p obj-sparc32/target/arch/unix
> mkdir -p obj-sparc32/target/arch/sparc32
> mkdir -p obj-sparc32/target/libgcc
> mkdir -p obj-sparc32/target/kernel
> mkdir -p obj-sparc32/target/libopenbios
> mkdir -p obj-sparc32/target/packages
> mkdir -p obj-sparc32/target/fs
> mkdir -p obj-sparc32/target/fs/grubfs
> mkdir -p obj-sparc32/target/fs/hfs
> mkdir -p obj-sparc32/target/fs/hfsplus
> mkdir -p obj-sparc32/target/fs/iso9660
> mkdir -p obj-sparc32/target/fs/ext2
> mkdir -p obj-sparc32/target/drivers
> mkdir -p obj-sparc32/target/libc
> mkdir -p obj-sparc32/host/include
> mkdir -p obj-sparc32/host/kernel
> mkdir -p obj-sparc32/forth
> ln -s ../../../include/arch/sparc32 obj-sparc32/target/include/asm
> [ no = yes ]
> [ no = yes ]
> [ no = yes ]
> [ no = yes ]
> [ no = yes ]
> echo ok.
>
> ok.
> cd obj-sparc32
> SRCDIR=..
> ODIR=.
> printf 'Creating target Makefile...'
>
> Creating target Makefile...+ echo ARCH=sparc32
> 1> ./config.mak
> [ yes = yes ]
> echo TARGET=sparc-sun-solaris2.11-
> 1>> ./config.mak
> echo CFLAGS='-Wa,-xarch=v8 -Wa,-32 -m32 -mcpu=supersparc -fno-builtin'
> 1>> ./config.mak
> echo AS_FLAGS='-Wa,-xarch=v8 -Wa,-32'
> 1>> ./config.mak
> echo 'HOSTARCH?=sparc64'
> 1>> ./config.mak
> echo CROSSCFLAGS='-USWAP_ENDIANNESS -DNATIVE_BITWIDTH_EQUALS_HOST_BITWIDTH'
> 1>> ./config.mak
> echo VERSION='"1.0"'
> 1>> ./config.mak
> ln -s ../config/xml/rules.xml ./rules.xml
> echo '<?xml version="1.0"?><config>'
> 1> ./config.xml
> config_set_boolean CONFIG_sparc32
> 1>> ./config.xml
> echo CONFIG_sparc32
> tr a-z A-Z
> option=CONFIG_SPARC32
> echo '<option name="CONFIG_SPARC32" type="boolean" value="true" />'
> [ sparc32 != sparc32 ]
> [ no = yes ]
> [ no = yes ]
> [ no = yes ]
> [ no = yes ]
> [ no = yes ]
> echo 'config test -target ' no
>
> config test -target  no
> [ no = yes ]
> config_set_boolean CONFIG_LITTLE_ENDIAN
> 1>> ./config.xml
> echo CONFIG_LITTLE_ENDIAN
> tr a-z A-Z
> option=CONFIG_LITTLE_ENDIAN
> echo '<option name="CONFIG_LITTLE_ENDIAN" type="boolean" value="true" />'
> [ no = yes ]
> [ yes = yes ]
> config_set_boolean CONFIG_IMAGE_ELF_EMBEDDED
> 1>> ./config.xml
> echo CONFIG_IMAGE_ELF_EMBEDDED
> tr a-z A-Z
> option=CONFIG_IMAGE_ELF_EMBEDDED
> echo '<option name="CONFIG_IMAGE_ELF_EMBEDDED" type="boolean" value="true" />'
> [ no = yes ]
> cat ../config/examples/sparc32_config.xml
> 1>> ./config.xml
> echo '</config>'
> 1>> ./config.xml
> ln -s ../Makefile.target ./Makefile
> xsltproc ../config/xml/xinclude.xsl ../build.xml
> 1> ./build-full.xml
> xsltproc ../config/xml/makefile.xsl ./build-full.xml
> 1> ./rules.mak
> echo ok.
>
> ok.
> printf 'Creating config files...'
>
> Creating config files...+ xsltproc ../config/xml/config-c.xsl ./config.xml
> 1> ./host/include/autoconf.h
> xsltproc ../config/xml/config-c.xsl ./config.xml
> 1> ./target/include/autoconf.h
> xsltproc ../config/xml/config-forth.xsl ./config.xml
> 1> ./forth/config.fs
> echo ok.
>
> ok.
> cd ..
>
> Nathan
>
> -- OpenBIOS                 http://openbios.org/
> Mailinglist:  http://lists.openbios.org/mailman/listinfo
> Free your System - May the Forth be with you
>
> This would fix the script to work on Solaris:
>
> --- switch-arch.orig    2011-05-02 20:54:43.312500000 -0400
> +++ switch-arch 2011-05-02 20:56:05.093750000 -0400
> @@ -19,8 +19,8 @@
>
> crosscflags()
>  {
> -    local host=$1
> -    local target=$2
> +    host=$1
> +    target=$2

This happens to work because variable $target (used in some for loops)
is not used after the call to crosscflags. It's a bit fragile though.
Considering the various uses of $target, I'd rename $host to $host_cpu
and $target to $target_cpu in crosscflags.

Solaris anyone? Q2

[Paul Brook]

> > crosscflags()
> >   {
> > local host=$1
> > local target=$2
> > host=$1
> > target=$2
>
> Confirmed; this produces the correct result with /bin/sh (ksh93), bash,
> and /usr/xpg4/bin/sh.

Have you tried separating the declaration and the definition. i.e.:

local host
host=$1

IIRC there are sometimes issues when you combine the two.

Paul

Solaris anyone? Q2

[Kenneth Salerno]

----- Original Message -----

From: Paul Brook 
To: [e-mail]
Cc: Nathan Kunkee 
Sent: Wednesday, May 4, 2011 7:19 PM
Subject: Re: [OpenBIOS] Solaris anyone? Q2
> > crosscflags()
> >    {
> >
> > -    local host=$1
> > -    local target=$2
> > +    host=$1
> > +    target=$2
>
> Confirmed; this produces the correct result with /bin/sh (ksh93), bash,
> and /usr/xpg4/bin/sh.

Have you tried separating the declaration and the definition. i.e.:

local host
host=$1

IIRC there are sometimes issues when you combine the two.

Paul

Solaris anyone? Q2

[Blue Swirl]

On Sun, May 22, 2011 at 2:11 AM, Kenneth Salerno
 wrote:
> ----- Original Message -----
>
> From: Paul Brook 
> To: [e-mail]
> Cc: Nathan Kunkee 
> Sent: Wednesday, May 4, 2011 7:19 PM
> Subject: Re: [OpenBIOS] Solaris anyone? Q2
> > > crosscflags()
> > >    {
> > >
> > > -    local host=$1
> > > -    local target=$2
> > > +    host=$1
> > > +    target=$2
> >
> > Confirmed; this produces the correct result with /bin/sh (ksh93), bash,
> > and /usr/xpg4/bin/sh.
>
> Have you tried separating the declaration and the definition. i.e.:
>
> local host
> host=$1
>
> IIRC there are sometimes issues when you combine the two.
>
> Paul
>
> Yes, I have tried that. Korn Shell '93 (ksh93) does not include the builtin "local", that is bash/zsh-specific:
>
> kens@sunos:~$ ken() {
> > local kenny
> > kenny=1
> > }
>
> kens@sunos:~$ ken
> /bin/sh[2]: local: not found [No such file or directory]
> kens@sunos:~$ type local
> /bin/sh: whence: local: not found
> kens@sunos:~$  ^D
> 11:04 PM kens@sunos$ type local
> local is a shell builtin
>
> Also there really is no need to rename the variables from host and target because they're not used for any unrelated purposes elsewhere in switch-arch.
>
> The patch I had submitted is valid.

It looks like 'local' is a non-standard extension:
http://pubs.opengroup.org/onlinepubs/9699919799/utilities/V3_chap02.html

Please submit the patch with a description and Signed-off-by: line.

[PATCH] switch-arch: removes non-standard "local" keyword

[Kenneth Salerno]

----- Original Message -----

From: Blue Swirl 
To: The OpenBIOS Mailinglist 
Cc: 
Sent: Sunday, May 22, 2011 5:39 AM
Subject: Re: [OpenBIOS] Solaris anyone? Q2

On Sun, May 22, 2011 at 2:11 AM, Kenneth Salerno
 wrote:
> ----- Original Message -----
>
> From: Paul Brook 
> To: [e-mail]
> Cc: Nathan Kunkee 
> Sent: Wednesday, May 4, 2011 7:19 PM
> Subject: Re: [OpenBIOS] Solaris anyone? Q2
> > > crosscflags()
> > >    {
> > >
> > > -    local host=$1
> > > -    local target=$2
> > > +    host=$1
> > > +    target=$2
> >
> > Confirmed; this produces the correct result with /bin/sh (ksh93), bash,
> > and /usr/xpg4/bin/sh.
>
> Have you tried separating the declaration and the definition. i.e.:
>
> local host
> host=$1
>
> IIRC there are sometimes issues when you combine the two.
>
> Paul
>
> Yes, I have tried that. Korn Shell '93 (ksh93) does not include the builtin "local", that is bash/zsh-specific:
>
> kens@sunos:~$ ken() {
> > local kenny
> > kenny=1
> > }
>
> kens@sunos:~$ ken
> /bin/sh[2]: local: not found [No such file or directory]
> kens@sunos:~$ type local
> /bin/sh: whence: local: not found
> kens@sunos:~$  ^D
> 11:04 PM kens@sunos$ type local
> local is a shell builtin
>
> Also there really is no need to rename the variables from host and target because they're not used for any unrelated purposes elsewhere in switch-arch.
>
> The patch I had submitted is valid.

It looks like 'local' is a non-standard extension:
http://pubs.opengroup.org/onlinepubs/9699919799/utilities/V3_chap02.html

Please submit the patch with a description and Signed-off-by: line.

[PATCH] switch-arch: removes non-standard "local" keyword

[Blue Swirl]

On Sun, May 22, 2011 at 7:05 PM, Kenneth Salerno
 wrote:
> ----- Original Message -----
>
> From: Blue Swirl 
> To: The OpenBIOS Mailinglist 
> Cc:
> Sent: Sunday, May 22, 2011 5:39 AM
> Subject: Re: [OpenBIOS] Solaris anyone? Q2
>
> On Sun, May 22, 2011 at 2:11 AM, Kenneth Salerno
>  wrote:
> > ----- Original Message -----
> >
> > From: Paul Brook 
> > To: [e-mail]
> > Cc: Nathan Kunkee 
> > Sent: Wednesday, May 4, 2011 7:19 PM
> > Subject: Re: [OpenBIOS] Solaris anyone? Q2
> > > > crosscflags()
> > > >    {
> > > >
> > > > -    local host=$1
> > > > -    local target=$2
> > > > +    host=$1
> > > > +    target=$2
> > >
> > > Confirmed; this produces the correct result with /bin/sh (ksh93), bash,
> > > and /usr/xpg4/bin/sh.
> >
> > Have you tried separating the declaration and the definition. i.e.:
> >
> > local host
> > host=$1
> >
> > IIRC there are sometimes issues when you combine the two.
> >
> > Paul
> >
> > Yes, I have tried that. Korn Shell '93 (ksh93) does not include the builtin "local", that is bash/zsh-specific:
> >
> > kens@sunos:~$ ken() {
> > > local kenny
> > > kenny=1
> > > }
> >
> > kens@sunos:~$ ken
> > /bin/sh[2]: local: not found [No such file or directory]
> > kens@sunos:~$ type local
> > /bin/sh: whence: local: not found
> > kens@sunos:~$  ^D
> > 11:04 PM kens@sunos$ type local
> > local is a shell builtin
> >
> > Also there really is no need to rename the variables from host and target because they're not used for any unrelated purposes elsewhere in switch-arch.
> >
> > The patch I had submitted is valid.
>
> It looks like 'local' is a non-standard extension:
> http://pubs.opengroup.org/onlinepubs/9699919799/utilities/V3_chap02.html
>
> Please submit the patch with a description and Signed-off-by: line.
>
> Description: removes the non-standard "local" keyword from within the crosscflags() function. This was not necessary since the variables "host" and "target" were not clobbered anywhere outside of the function, but the real motivation is to support shells such as ksh93 that do not have the "local" builtin like bash.

Thanks, applied.

In the future, please submit the patches separately without the
previous discussions. The idea is that it should be possible to use
'git am' without editing the message at all and the result should look
like a proper commit with a subject line and commit description.
> Signed-off-by: Kenneth Salerno
>
> --- switch-arch.orig    2011-05-02 20:54:43.312500000 -0400
> +++ switch-arch 2011-05-02 20:56:05.093750000 -0400
> @@ -19,8 +19,8 @@
>
> crosscflags()
>  {
> -    local host=$1
> -    local target=$2
> +    host=$1
> +    target=$2
>
> if test "$host" = "powerpc" -o "$host" = "ppc" \
>          -o "$host" = "mips" -o "$host" = "s390" \
>
> Thanks,
> Ken

Solaris anyone? Q2

[Brian Vandenberg]

I've been trying to get solaris 8 to install, but I've been running
into problems.

After creating/formatting a disk in single-user mode, I reboot and
start the install.  It copies the installer to /dev/c0t0d0s1 and
reboots.  On startup, it gets to "Running in command line mode" then
seems to endlessly do nothing.

Is how long should I wait before I can safely assume it isn't going
to continue/succeed?

For reference, I'm running qemu with:

./qemu-system-sparc -cdrom solaris8_install.iso -nographic -prom-env
'auto-boot?=false' -m 256 -L . -bios openbios-sparc32_artyom.bin -hda
/scratch/qemu-images/solaris2.8.qcow2

... and booting with:

boot disk0:b

Usually I use the -v, but I forgot it for the current attempt.
Unfortunately, I can't copy logs over from the airgapped network, but
if needs be I can manually type up relevant portions.

I'm unsure whether it would help, but when I attempt to install with
the SS-5 bios image I get the following:

(...)
SunOS Release 5.8 Versio nGeneric_108528-29 32-bit
Copyright 1983-2003 Sun Microsystems, Inc.  All rights reserved.
Cannot assemble drivers for root
/iommu@0,10000000/sbus@0,10001000/espdma@5,84000000/esp@5,88000000/sd@0,0:b
Cannot mount root on
/iommu@0,10000000/sbus@0,10001000/espdma@5,84000000/esp@5,88000000/sd@0,0:b
fstype ufs

panic[cpu0]/thread=f0244020: vfs_mountroot: cannot mount root

(...)

... which is followed by 4 lines of stuff that looks somewhat like a
stack dump.

So, I suppose what I'm looking for is: how patient should I be
before I decide that the installation with openbios won't succeed?

-Brian

Solaris anyone? Q2

[Tarl Neustaedter]

On 2011-May-23 19:51 , Brian Vandenberg wrote:
> Cannot assemble drivers for root
> /iommu@0,10000000/sbus@0,10001000/espdma@5,84000000/esp@5,88000000/sd@0,0:b
> Cannot mount root on
> /iommu@0,10000000/sbus@0,10001000/espdma@5,84000000/esp@5,88000000/sd@0,0:b
> fstype ufs
>
> panic[cpu0]/thread=f0244020: vfs_mountroot: cannot mount root
>
> (...)
>
> ... which is followed by 4 lines of stuff that looks somewhat like a
> stack dump.
>
> So, I suppose what I'm looking for is: how patient should I be
> before I decide that the installation with openbios won't succeed?

Once you get "cannot assemble drivers for root", it's over. That error 
is fatal. A few lines later you see a panic "vfs_mountroot: cannot mount 
root" which is the normal endpoint for driver/hardware mismatch.

Solaris anyone? Q2

[Tarl Neustaedter]

[...]
> Once you get "cannot assemble drivers for root", it's over.

And I just failed the reading comprehension test. Sorry, that was not 
the case you were asking about:
> After creating/formatting a disk in single-user mode, I reboot and
> start the install.  It copies the installer to /dev/c0t0d0s1 and
> reboots.  On startup, it gets to "Running in command line mode" then
> seems to endlessly do nothing.
>
> Is how long should I wait before I can safely assume it isn't going
> to continue/succeed?

I can't directly answer the "how long" question, but the above seems to 
indicate that you might have your console directed to somewhere 
unexpected. Are you expecting the output to come to the video console or 
the serial port? Be aware that Solaris doesn't always make the same 
decision OBP does on this.

Solaris anyone? Q2

[Brian Vandenberg]

> Once you get "cannot assemble drivers for root", it's over. That error is
> fatal. A few lines later you see a panic "vfs_mountroot: cannot mount root"
> which is the normal endpoint for driver/hardware mismatch.

hrm.  Any pointers/tips on how to proceed?

If I'm able to get it to the "Running in command line mode" point
with openbios, and it seems to hang there, is this just silently
having related problems or do I just need to wait longer?

-Brian