[9fans] 9pccpu: can't open /dev/sdXX/nvram

[9fans] 9pccpu: can't open /dev/sdXX/nvram

[arisawa]

Hello,

I am in trouble in booting 9pccpu from usbdisk.

Boot loader detects 9fat in plan9 partition in usbdisk, and reads plan9.ini, and successfully executes 9pccpu.
However, 9pccpu says "no usb disk" in reading /dev/sdXX/nvram

Kernel messages:
usbinit...usbd...usb/hub... no usb disk...
...
auth...can't open /dev/sdXX/nvram: '/env/nvroff' file does not exist
authid: ...
...
init: starting /bin/rc

at% ls /dev/sd*
/dev/sdctl
at% 

Has anyone succeeded in booting cpu server from usbdisk?

Kenji Arisawa

Re: [9fans] 9pccpu: can't open /dev/sdXX/nvram

[David du Colombier]

> Has anyone succeeded in booting cpu server from usbdisk?

There is no problem in booting the CPU server, since
you can always specify the required values by hand
during the boot.
You problem is that you can't read/write theses
values from/to the disk.

I suppose you're using the USB image I am providing
on 9legacy.org. A nvram partition is present, but
I haven't properly configured the plan9.ini to use
an nvram on the USB disk.

You can simply mount the first FAT partition
from the USB disk, edit the plan9.ini and add
the "readparts=" parameter in the [common] section,
just before the "nvram=" parameter, like this:

readparts=
nvram=/dev/sdXX/nvram

If you are curious how this USB image was built,
you can look at the mkusbboot(8) script provided
in the Plan 9 distribution, or check my notes
available on:

http://www.9legacy.org/9legacy/doc/fossil/usb

There is nothing specific here. Everything is
already provided in Plan 9 from Bell Labs.

--
David du Colombier

Re: [9fans] 9pccpu: can't open /dev/sdXX/nvram

[arisawa]

> You can simply mount the first FAT partition
> from the USB disk, edit the plan9.ini and add
> the "readparts=" parameter in the [common] section,
> just before the "nvram=" parameter, like this:
>
> readparts=
> nvram=/dev/sdXX/nvram
>

That is exactly what I did.
The problem is in the message "no usb disk",  which is confirmed by
at% ls /dev/sd*
/dev/sdctl
at%

I believe my 9pccpu kernel is sufficiently recent.
--rwxrwxr-x M 221 sys sys 2368300 May 18 07:16 /386/9pccpu

and my script in creating bootable usb disk is as follows:
disk/partfs /dev/sdU0.0/data	# create /dev/sdXX
disk/mbr -m /386/mbr /dev/sdXX/data
disk/fdisk -baw /dev/sdXX/data
disk/prep -bw -a^(9fat nvram) /dev/sdXX/plan9
cp /dev/sdC0/nvram /dev/sdXX/nvram
cp /386/9loadusb 9load
cp /386/9pc 9pc
cp /386/9pccpu 9pccpu
disk/format -b /386/pbslba -d -r 2 /dev/sdXX/9fat 9load 9pc 9pccpu plan9.ini

If the problem is only to me, I am disliked by 9pccpu kernel.

Kenji Arisawa

Re: [9fans] 9pccpu: can't open /dev/sdXX/nvram

[David du Colombier]

> If the problem is only to me, I am disliked by 9pccpu kernel.

I don't see anything particularly wrong in your setup.
Your kernel is recent enough and should work.
Personally, I haven't encountered this problem.

I would tend to think that USB is not properly supported
on your machine. Does usb/disk see your usb disk when
you boot from another media (pxe, cd, ata...)?

Have you tried on another machine?

--
David du Colombier

Re: [9fans] 9pccpu: can't open /dev/sdXX/nvram

[arisawa]

Hello,

On 2012/07/28, at 23:33, David du Colombier wrote:

> I would tend to think that USB is not properly supported
> on your machine.

You are right. It seems the machine dislikes Plan9.
The MB is GA-H61M-USB3-B3
Thanks a lot.

Kenji Arisawa

Re: [9fans] 9pccpu: can't open /dev/sdXX/nvram

[Richard Miller]

> The MB is GA-H61M-USB3-B3
                    ^^^^

Are you sure you are plugging the usbdisk into a USB2.0 socket,
not a USB3.0 one?

Re: [9fans] 9pccpu: can't open /dev/sdXX/nvram

[arisawa]

Hello,

On 2012/07/29, at 18:43, Richard Miller wrote:

>> The MB is GA-H61M-USB3-B3
>                    ^^^^
>
> Are you sure you are plugging the usbdisk into a USB2.0 socket,
> not a USB3.0 one?
>

The usb disk is plugged into USB 2.0/1.1 socket.
Thanks Richard.

Kenji Arisawa

Re: [9fans] 9pccpu: can't open /dev/sdXX/nvram

[Richard Miller]

I'm sorry that didn't help.  My intel DH61DL motherboard has the same chipset
as yours.  If I try a usb flash disk in a USB3 port, I can boot Plan 9 from it,
but once booted the drive isn't visible.

You might try usb/probe to see if your kernel is able to see the drive at all.

Re: [9fans] 9pccpu: can't open /dev/sdXX/nvram

[arisawa]

Hello,

> You might try usb/probe to see if your kernel is able to see the drive at all.

usb/probe does not show messages such as "storage csp ...."

Thanks, Richard.

Other possibilities:
Hardware trouble? No, Xubuntu can read usb disks.

Kenji Arisawa

Re: [9fans] 9pccpu: can't open /dev/sdXX/nvram

[arisawa]

Hello,

I have another example that behaves somewhat different from GA-H61M-USB3-B3.
The MB is GA-73UM-S2H. The chipset is nVIDIA GeForce 7150/nForce 630i.
Kernel message are:

Plan 9
cpu0:  2667MHz GenuineIntel Core 2/Xeon (cpuid: AX 0x06FB DX 0xBFEBFBFF)
ELCR: 0C20
LAPIC: 0xfee00000 0xe0000000
mpinit: scanning acpi madt for extra cpus
cpu1:  2667MHz GenuineIntel Core 2/Xeon (cpuid: AX 0x06FB DX 0xBFEBFBFF)
mtrrfix1: i9: 0 404040404040404
mtrrfix1: i10: 0 404040404040404
pcirouting: ignoring south bridge PCI.0.0.0 10DE/07C0
#l0: igbepcie: 1Gbps port 0xE4080000 irq 10: 6805ca00fbd2
mpintrenable: out of choices eisa -1 isa 5 tbdf 0xc002000 irq 11
intrenable: couldn't enable irq 11, tbdf 0xC002000 for usbohci
mpintrenable: out of choices eisa -1 isa 5 tbdf 0xc002100 irq 5
intrenable: couldn't enable irq 5, tbdf 0xC002100 for usbehci
3198M memory: 149M kernel data, 3048M user, 3673M swap
usbinit...usbd...no usb disk...waiting for dhcp...
auth...can't open /dev/sdXX/nvram: '/env/nvroff' file does not exist
authid: ehci 0xe0003000: polling
usb/disk... bootes
authdom: aichi-u.ac.jp
auth password: 
secstore password: 
can't write key to nvram: '/env/nvroff' file does not exist
usbinit...partfs...version...time...
swap: /dev/swap
post...

init: starting /bin/rc
ar# ls /dev/sd*
/dev/sdU0.0/ctl
/dev/sdU0.0/data
/dev/sdU0.0/raw
/dev/sdctl
ar# usb/probe
ep1.0 roothub csp 0x000009 ports 10 ohci
ep2.0 roothub csp 0x000009 ports 10 ehci
ep3.0 storage csp 0x500608 vid 0x0411 did 0x0098 BUFFALO 'USB Flash Disk' ehci
ar#

Note that storage is detected, but "no usb disk" and then "can't open /dev/sdXX/nvram".
The message: "couldn't enable irq 5, tbdf 0xC002100 for usbehci" might be a problem.

Kenji Arisawa

Re: [9fans] 9pccpu: can't open /dev/sdXX/nvram

[erik quanstrom]

> pcirouting: ignoring south bridge PCI.0.0.0 10DE/07C0
> #l0: igbepcie: 1Gbps port 0xE4080000 irq 10: 6805ca00fbd2
> mpintrenable: out of choices eisa -1 isa 5 tbdf 0xc002000 irq 11
> intrenable: couldn't enable irq 11, tbdf 0xC002000 for usbohci
> mpintrenable: out of choices eisa -1 isa 5 tbdf 0xc002100 irq 5
> intrenable: couldn't enable irq 5, tbdf 0xC002100 for usbehci
> 3198M memory: 149M kernel data, 3048M user, 3673M swap
> usbinit...usbd...no usb disk...waiting for dhcp...

it sure looks to me like this is a usb interrupt problem.  i can't
tell from the output of your kernel if the mp table is just missing
or not.  i'm guessing that this mp table just doesn't have entries
for 1:1 mapped irqs.  my first guess (not tried myself) would be
to assume that irq 11 → GSI 11, which is usually ioapic 0, pin 11.
s/11/5 for the other usb port.

you might try adding fake irq entries to your mp table to match,
and see if that works.  :-).  usb is always level triggered.

> > authid: ehci 0xe0003000: polling

here's more confirmation of an irq problem.  i wonder if polling
isn't fully tested.

- erik

Re: [9fans] 9pccpu: can't open /dev/sdXX/nvram

[cinap_lenrek]

or use 9front with *acpi=1

--
cinap

Re: [9fans] 9pccpu: can't open /dev/sdXX/nvram

[arisawa]

Hello cinap,

There are no 9pc and 9pccpu in 9front package.
Is there any reason?

It seems 9bootpxe of 9front does better jobs than 9pxeload.
However 9bootpxe does not support menu.
Is there any reason?

On 2012/07/31, at 5:45, cinap_lenrek@gmx.de wrote:

> or use 9front with *acpi=1
>
> --
> cinap
>

Re: [9fans] 9pccpu: can't open /dev/sdXX/nvram

[cinap_lenrek]

the 9front kernels use a new rc based bootstrap system. the
kernel itself uses a compressed ram filesystem image (paqfs)
that will be mounted by /boot/boot containing the required
programs like rc, srv, ifconfig and file fileservers like
9660srv and cwfs to mount the real root filesystem.

there is no special cdboot kernel or pxe boot kernel anymore.

every kernel is capable of network booting or booting from
local disk or cdrom or usb stick.

the only thing that got specialized are the bootloaders. theres
9bootiso, 9bootfat and 9bootpxe. but once they get the kernel
loaded, stuff looks exactly the same.

this also has the advantage that you can boot to a local cwfs
just using the cdrom. or boot from the cd by using a usb stick
or network boot. ect. so thats why we removed the special netboot
kernels. it also helps rescuing a unbootable/bricked system.
you can even run rc shell from these kernels (type !rc at the
bootargs prompt) so you can inspect the system even tho it might
not be able to detect the network card or disk controller ect...

for the bootloaders, i thought that a command line/console that
lets you change specific kernel parameters before boot is more
usefull than a menu with preconfigured options. especially in the
case of testing or rescuing a system. 9boot even works if theres no
plan9.ini at all.

we also want to keep the bootloaders as simple and as stupid as
possible. the less you do at the boot stage the less can go wrong.
boot enviroment is very tricky thing. we avoid interfering with
bios at all costs and do the clever parts once we took control of the
hardware and have a running kernel.

the kernel will present you with a "menu" like list of bootable
local filesystems.

--
cinap

Re: [9fans] 9pccpu: can't open /dev/sdXX/nvram

[Skip Tavakkolian]

[-- Attachment #1: Type: text/plain, Size: 2104 bytes --]

were you able to resolve this?

i have the same problem with a diskless Supermicro 5015A-EHF-D525 server
with a 1GB ScanDisk usb.

-Skip


On Mon, Jul 30, 2012 at 2:19 AM, arisawa <arisawa@ar.aichi-u.ac.jp> wrote:

> Hello,
>
> I have another example that behaves somewhat different from
> GA-H61M-USB3-B3.
> The MB is GA-73UM-S2H. The chipset is nVIDIA GeForce 7150/nForce 630i.
> Kernel message are:
>
> Plan 9
> cpu0:  2667MHz GenuineIntel Core 2/Xeon (cpuid: AX 0x06FB DX 0xBFEBFBFF)
> ELCR: 0C20
> LAPIC: 0xfee00000 0xe0000000
> mpinit: scanning acpi madt for extra cpus
> cpu1:  2667MHz GenuineIntel Core 2/Xeon (cpuid: AX 0x06FB DX 0xBFEBFBFF)
> mtrrfix1: i9: 0 404040404040404
> mtrrfix1: i10: 0 404040404040404
> pcirouting: ignoring south bridge PCI.0.0.0 10DE/07C0
> #l0: igbepcie: 1Gbps port 0xE4080000 irq 10: 6805ca00fbd2
> mpintrenable: out of choices eisa -1 isa 5 tbdf 0xc002000 irq 11
> intrenable: couldn't enable irq 11, tbdf 0xC002000 for usbohci
> mpintrenable: out of choices eisa -1 isa 5 tbdf 0xc002100 irq 5
> intrenable: couldn't enable irq 5, tbdf 0xC002100 for usbehci
> 3198M memory: 149M kernel data, 3048M user, 3673M swap
> usbinit...usbd...no usb disk...waiting for dhcp...
> auth...can't open /dev/sdXX/nvram: '/env/nvroff' file does not exist
> authid: ehci 0xe0003000: polling
> usb/disk... bootes
> authdom: aichi-u.ac.jp
> auth password:
> secstore password:
> can't write key to nvram: '/env/nvroff' file does not exist
> usbinit...partfs...version...time...
> swap: /dev/swap
> post...
>
> init: starting /bin/rc
> ar# ls /dev/sd*
> /dev/sdU0.0/ctl
> /dev/sdU0.0/data
> /dev/sdU0.0/raw
> /dev/sdctl
> ar# usb/probe
> ep1.0 roothub csp 0x000009 ports 10 ohci
> ep2.0 roothub csp 0x000009 ports 10 ehci
> ep3.0 storage csp 0x500608 vid 0x0411 did 0x0098 BUFFALO 'USB Flash Disk'
> ehci
> ar#
>
> Note that storage is detected, but "no usb disk" and then "can't open
> /dev/sdXX/nvram".
> The message: "couldn't enable irq 5, tbdf 0xC002100 for usbehci" might be
> a problem.
>
> Kenji Arisawa
>
>
>

[-- Attachment #2: Type: text/html, Size: 2712 bytes --]

Re: [9fans] 9pccpu: can't open /dev/sdXX/nvram

[arisawa]

Hello,

not yet with usb.
my current solution is "sdcard with sata adapter" only for NVRAM.
I will be happy if someone have a solution that uses usb flash drive only.

Kenji Arisawa

On 2013/05/14, at 1:37, Skip Tavakkolian <skip.tavakkolian@gmail.com> wrote:

> were you able to resolve this?
> 
> i have the same problem with a diskless Supermicro 5015A-EHF-D525 server with a 1GB ScanDisk usb. 
> 
> -Skip
> 
> 
> On Mon, Jul 30, 2012 at 2:19 AM, arisawa <arisawa@ar.aichi-u.ac.jp> wrote:
> Hello,
> 
> I have another example that behaves somewhat different from GA-H61M-USB3-B3.
> The MB is GA-73UM-S2H. The chipset is nVIDIA GeForce 7150/nForce 630i.
> Kernel message are:
> 
> Plan 9
> cpu0:  2667MHz GenuineIntel Core 2/Xeon (cpuid: AX 0x06FB DX 0xBFEBFBFF)
> ELCR: 0C20
> LAPIC: 0xfee00000 0xe0000000
> mpinit: scanning acpi madt for extra cpus
> cpu1:  2667MHz GenuineIntel Core 2/Xeon (cpuid: AX 0x06FB DX 0xBFEBFBFF)
> mtrrfix1: i9: 0 404040404040404
> mtrrfix1: i10: 0 404040404040404
> pcirouting: ignoring south bridge PCI.0.0.0 10DE/07C0
> #l0: igbepcie: 1Gbps port 0xE4080000 irq 10: 6805ca00fbd2
> mpintrenable: out of choices eisa -1 isa 5 tbdf 0xc002000 irq 11
> intrenable: couldn't enable irq 11, tbdf 0xC002000 for usbohci
> mpintrenable: out of choices eisa -1 isa 5 tbdf 0xc002100 irq 5
> intrenable: couldn't enable irq 5, tbdf 0xC002100 for usbehci
> 3198M memory: 149M kernel data, 3048M user, 3673M swap
> usbinit...usbd...no usb disk...waiting for dhcp...
> auth...can't open /dev/sdXX/nvram: '/env/nvroff' file does not exist
> authid: ehci 0xe0003000: polling
> usb/disk... bootes
> authdom: aichi-u.ac.jp
> auth password:
> secstore password:
> can't write key to nvram: '/env/nvroff' file does not exist
> usbinit...partfs...version...time...
> swap: /dev/swap
> post...
> 
> init: starting /bin/rc
> ar# ls /dev/sd*
> /dev/sdU0.0/ctl
> /dev/sdU0.0/data
> /dev/sdU0.0/raw
> /dev/sdctl
> ar# usb/probe
> ep1.0 roothub csp 0x000009 ports 10 ohci
> ep2.0 roothub csp 0x000009 ports 10 ehci
> ep3.0 storage csp 0x500608 vid 0x0411 did 0x0098 BUFFALO 'USB Flash Disk' ehci
> ar#
> 
> Note that storage is detected, but "no usb disk" and then "can't open /dev/sdXX/nvram".
> The message: "couldn't enable irq 5, tbdf 0xC002100 for usbehci" might be a problem.
> 
> Kenji Arisawa
> 
> 
> 

Re: [9fans] 9pccpu: can't open /dev/sdXX/nvram

[Matthew Veety]

I have my nvram as a file on my fat partition. That seems to work.

On May 13, 2013, at 17:32, arisawa <arisawa@ar.aichi-u.ac.jp> wrote:

> Hello,
> 
> not yet with usb.
> my current solution is "sdcard with sata adapter" only for NVRAM.
> I will be happy if someone have a solution that uses usb flash drive only.
> 
> Kenji Arisawa
> 
> On 2013/05/14, at 1:37, Skip Tavakkolian <skip.tavakkolian@gmail.com> wrote:
> 
>> were you able to resolve this?
>> 
>> i have the same problem with a diskless Supermicro 5015A-EHF-D525 server with a 1GB ScanDisk usb. 
>> 
>> -Skip
>> 
>> 
>> On Mon, Jul 30, 2012 at 2:19 AM, arisawa <arisawa@ar.aichi-u.ac.jp> wrote:
>> Hello,
>> 
>> I have another example that behaves somewhat different from GA-H61M-USB3-B3.
>> The MB is GA-73UM-S2H. The chipset is nVIDIA GeForce 7150/nForce 630i.
>> Kernel message are:
>> 
>> Plan 9
>> cpu0:  2667MHz GenuineIntel Core 2/Xeon (cpuid: AX 0x06FB DX 0xBFEBFBFF)
>> ELCR: 0C20
>> LAPIC: 0xfee00000 0xe0000000
>> mpinit: scanning acpi madt for extra cpus
>> cpu1:  2667MHz GenuineIntel Core 2/Xeon (cpuid: AX 0x06FB DX 0xBFEBFBFF)
>> mtrrfix1: i9: 0 404040404040404
>> mtrrfix1: i10: 0 404040404040404
>> pcirouting: ignoring south bridge PCI.0.0.0 10DE/07C0
>> #l0: igbepcie: 1Gbps port 0xE4080000 irq 10: 6805ca00fbd2
>> mpintrenable: out of choices eisa -1 isa 5 tbdf 0xc002000 irq 11
>> intrenable: couldn't enable irq 11, tbdf 0xC002000 for usbohci
>> mpintrenable: out of choices eisa -1 isa 5 tbdf 0xc002100 irq 5
>> intrenable: couldn't enable irq 5, tbdf 0xC002100 for usbehci
>> 3198M memory: 149M kernel data, 3048M user, 3673M swap
>> usbinit...usbd...no usb disk...waiting for dhcp...
>> auth...can't open /dev/sdXX/nvram: '/env/nvroff' file does not exist
>> authid: ehci 0xe0003000: polling
>> usb/disk... bootes
>> authdom: aichi-u.ac.jp
>> auth password:
>> secstore password:
>> can't write key to nvram: '/env/nvroff' file does not exist
>> usbinit...partfs...version...time...
>> swap: /dev/swap
>> post...
>> 
>> init: starting /bin/rc
>> ar# ls /dev/sd*
>> /dev/sdU0.0/ctl
>> /dev/sdU0.0/data
>> /dev/sdU0.0/raw
>> /dev/sdctl
>> ar# usb/probe
>> ep1.0 roothub csp 0x000009 ports 10 ohci
>> ep2.0 roothub csp 0x000009 ports 10 ehci
>> ep3.0 storage csp 0x500608 vid 0x0411 did 0x0098 BUFFALO 'USB Flash Disk' ehci
>> ar#
>> 
>> Note that storage is detected, but "no usb disk" and then "can't open /dev/sdXX/nvram".
>> The message: "couldn't enable irq 5, tbdf 0xC002100 for usbehci" might be a problem.
>> 
>> Kenji Arisawa
> 
> 

Re: [9fans] 9pccpu: can't open /dev/sdXX/nvram

[erik quanstrom]

On Mon May 13 17:33:12 EDT 2013, arisawa@ar.aichi-u.ac.jp wrote:
> Hello,
>
> not yet with usb.
> my current solution is "sdcard with sata adapter" only for NVRAM.
> I will be happy if someone have a solution that uses usb flash drive only.

it looks to me that the problem is the fact that intrenable() doesn't
find the interrupts for usb.  polling might be slow enough to trip
up the arbitrary delays in the boot process.

that is, it looks like the mp table is incorrect.

have you tried a kernel with more support for acpi?  i believe you
may be able to try on recent 9atom kernels, (http://ftp.9atom.org/other/9paecpu)
with the following two addition to your .ini file
	*acpi=1

this is really cinap's work.  i only suggest 9atom as it may be easier
to test in your environment.

good luck.

- erik

Re: [9fans] 9pccpu: can't open /dev/sdXX/nvram

[arisawa]

Hello,

I retried cpu kernel with only usb flash drive.
this time I have finally succeeded in automatic start service.

the kernel is based on 9front.
MB is GA-G31M-S2L.
the MB is old model and very slow in polling usb even if I set
	*acpi=1
in plan9.ini.
my nvram setting is
	nvram=/dev/sdU6.0/nvram
and we must find the value 6.0 by experiment:
first, without nobootprompt and then
	nobootprompt=tcp

I have not good result for bell-labs version.

thanks all.

On 2013/05/14, at 8:06, erik quanstrom <quanstro@quanstro.net> wrote:

> On Mon May 13 17:33:12 EDT 2013, arisawa@ar.aichi-u.ac.jp wrote:
>> Hello,
>> 
>> not yet with usb.
>> my current solution is "sdcard with sata adapter" only for NVRAM.
>> I will be happy if someone have a solution that uses usb flash drive only.
> 
> it looks to me that the problem is the fact that intrenable() doesn't
> find the interrupts for usb.  polling might be slow enough to trip
> up the arbitrary delays in the boot process.
> 
> that is, it looks like the mp table is incorrect.
> 
> have you tried a kernel with more support for acpi?  i believe you
> may be able to try on recent 9atom kernels, (http://ftp.9atom.org/other/9paecpu)
> with the following two addition to your .ini file
> 	*acpi=1
> 
> this is really cinap's work.  i only suggest 9atom as it may be easier
> to test in your environment.
> 
> good luck.
> 
> - erik
> 

Re: [9fans] 9pccpu: can't open /dev/sdXX/nvram

[arisawa]

PXE + (nvram on usb flash) is much faster than usb flash only boot.

On 2013/05/15, at 18:56, arisawa <arisawa@ar.aichi-u.ac.jp> wrote:

> I retried cpu kernel with only usb flash drive.
> this time I have finally succeeded in automatic start service.
>
> the kernel is based on 9front.
> MB is GA-G31M-S2L.
> the MB is old model and very slow in polling usb even if I set
> 	*acpi=1
> in plan9.ini.
> my nvram setting is
> 	nvram=/dev/sdU6.0/nvram
> and we must find the value 6.0 by experiment:
> first, without nobootprompt and then
> 	nobootprompt=tcp

Re: [9fans] 9pccpu: can't open /dev/sdXX/nvram

[erik quanstrom]

On Wed May 15 16:57:59 EDT 2013, arisawa@ar.aichi-u.ac.jp wrote:
>
> PXE + (nvram on usb flash) is much faster than usb flash only boot.

my experience has been that usb flash boot can be faster due
to networking quirks.  for example, spanning tree or even switch uplink
connect times can be measured in many seconds.

- erik

Re: [9fans] 9pccpu: can't open /dev/sdXX/nvram

[arisawa]

Hello Erik,

the problem is not in transmission speed of network.
mine is based on 9front. 
messages that come from booting process is as follows:
(1) the content in plan9.ini.
(2) then single letter hexadecimal messages 0 1 ... a b .. f 0 1 ...
(3) and then "boot". the message comes from bootfat.
(4) finally "Plan9" message from the loaded kernel.

the message (2) is very slow and takes much time to finish.
I suspect bootfat takes too much time to find a specified kernel in fat partition.
the time is mother board dependent.
my new one is sufficiently fast, but old one (GA-G31M-S2L) is irritatingly slow.

Kenji Arisawa

P.S.
I accessed http://ftp.9atom.org/other/9paecpu to try your 9atom.
however I got "Object not found".


On 2013/05/16, at 6:18, erik quanstrom <quanstro@quanstro.net> wrote:

> On Wed May 15 16:57:59 EDT 2013, arisawa@ar.aichi-u.ac.jp wrote:
>> 
>> PXE + (nvram on usb flash) is much faster than usb flash only boot.
> 
> my experience has been that usb flash boot can be faster due
> to networking quirks.  for example, spanning tree or even switch uplink
> connect times can be measured in many seconds.
> 
> - erik
> 

Re: [9fans] 9pccpu: can't open /dev/sdXX/nvram

[erik quanstrom]

On Wed May 15 20:17:40 EDT 2013, arisawa@ar.aichi-u.ac.jp wrote:
> Hello Erik,
>
> the problem is not in transmission speed of network.
> mine is based on 9front.
> messages that come from booting process is as follows:
> (1) the content in plan9.ini.
> (2) then single letter hexadecimal messages 0 1 ... a b .. f 0 1 ...
> (3) and then "boot". the message comes from bootfat.
> (4) finally "Plan9" message from the loaded kernel.
>
> the message (2) is very slow and takes much time to finish.

i find that step (0), just getting the box to pxe the loader
can be the slowest part.

> I suspect bootfat takes too much time to find a specified kernel in fat partition.

with the same style loader, i don't have that issue.  are there a lot of files
in your fat?

> I accessed http://ftp.9atom.org/other/9paecpu to try your 9atom.
> however I got "Object not found".
>

sorry.  i didn't copy enough stuff over to the new servers.
fixed.

- erik

Re: [9fans] 9pccpu: can't open /dev/sdXX/nvram

[kernel panic]

when the hex numbers count in (2), it means its reading
the kernel file from the disk into memory.

9fronts 9boot loader uses the bios for all device access.
to read a single 512 byte block of data from a harddrive
(or usb flash drives as they are emulated as harddrives by
the bios), we switch from 32bit protected mode to realmode,
do the bios call and return to protected mode copying the
block from stack in low memory to its final destination.

doing these single block reads might be inefficient with
your bios implementation. tho it never has been a problem
with ide/sata harddrives or cdroms.

when it takes long from (1) to (2), this could be a problem
with the usleep() code. after we read plan9.ini, we wait
a bit for keyboard input so one can interrupt the automatic
boot process. maybe that wait loop is not working right?

> Gesendet: Donnerstag, 16. Mai 2013 um 02:16 Uhr
> Von: arisawa <arisawa@ar.aichi-u.ac.jp>
> An: "Fans of the OS Plan 9 from Bell Labs" <9fans@9fans.net>
> Betreff: Re: [9fans] 9pccpu: can't open /dev/sdXX/nvram
>
> Hello Erik,
>
> the problem is not in transmission speed of network.
> mine is based on 9front.
> messages that come from booting process is as follows:
> (1) the content in plan9.ini.
> (2) then single letter hexadecimal messages 0 1 ... a b .. f 0 1 ...
> (3) and then "boot". the message comes from bootfat.
> (4) finally "Plan9" message from the loaded kernel.
>
> the message (2) is very slow and takes much time to finish.
> I suspect bootfat takes too much time to find a specified kernel in fat partition.
> the time is mother board dependent.
> my new one is sufficiently fast, but old one (GA-G31M-S2L) is irritatingly slow.
>
> Kenji Arisawa
>
> P.S.
> I accessed http://ftp.9atom.org/other/9paecpu to try your 9atom.
> however I got "Object not found".
>
>
> On 2013/05/16, at 6:18, erik quanstrom <quanstro@quanstro.net> wrote:
>
> > On Wed May 15 16:57:59 EDT 2013, arisawa@ar.aichi-u.ac.jp wrote:
> >>
> >> PXE + (nvram on usb flash) is much faster than usb flash only boot.
> >
> > my experience has been that usb flash boot can be faster due
> > to networking quirks.  for example, spanning tree or even switch uplink
> > connect times can be measured in many seconds.
> >
> > - erik
> >

Re: [9fans] 9pccpu: can't open /dev/sdXX/nvram

[arisawa]

thanks panic,

it does not take long time to switch message from (1) to (2).
the problem is in (2) that read single 512 byte block of data.

Kenji Arisawa

On 2013/05/16, at 22:44, kernel panic <cinap_lenrek@gmx.de> wrote:

> when the hex numbers count in (2), it means its reading
> the kernel file from the disk into memory.
> 
> 9fronts 9boot loader uses the bios for all device access.
> to read a single 512 byte block of data from a harddrive
> (or usb flash drives as they are emulated as harddrives by
> the bios), we switch from 32bit protected mode to realmode,
> do the bios call and return to protected mode copying the
> block from stack in low memory to its final destination.
> 
> doing these single block reads might be inefficient with
> your bios implementation. tho it never has been a problem
> with ide/sata harddrives or cdroms.
> 
> when it takes long from (1) to (2), this could be a problem
> with the usleep() code. after we read plan9.ini, we wait
> a bit for keyboard input so one can interrupt the automatic
> boot process. maybe that wait loop is not working right?
> 
>> Gesendet: Donnerstag, 16. Mai 2013 um 02:16 Uhr
>> Von: arisawa <arisawa@ar.aichi-u.ac.jp>
>> An: "Fans of the OS Plan 9 from Bell Labs" <9fans@9fans.net>
>> Betreff: Re: [9fans] 9pccpu: can't open /dev/sdXX/nvram
>> 
>> Hello Erik,
>> 
>> the problem is not in transmission speed of network.
>> mine is based on 9front. 
>> messages that come from booting process is as follows:
>> (1) the content in plan9.ini.
>> (2) then single letter hexadecimal messages 0 1 ... a b .. f 0 1 ...
>> (3) and then "boot". the message comes from bootfat.
>> (4) finally "Plan9" message from the loaded kernel.
>> 
>> the message (2) is very slow and takes much time to finish.
>> I suspect bootfat takes too much time to find a specified kernel in fat partition.
>> the time is mother board dependent.
>> my new one is sufficiently fast, but old one (GA-G31M-S2L) is irritatingly slow.
>> 
>> Kenji Arisawa
>> 
>> P.S.
>> I accessed http://ftp.9atom.org/other/9paecpu to try your 9atom.
>> however I got "Object not found".
>> 
>> 
>> On 2013/05/16, at 6:18, erik quanstrom <quanstro@quanstro.net> wrote:
>> 
>>> On Wed May 15 16:57:59 EDT 2013, arisawa@ar.aichi-u.ac.jp wrote:
>>>> 
>>>> PXE + (nvram on usb flash) is much faster than usb flash only boot.
>>> 
>>> my experience has been that usb flash boot can be faster due
>>> to networking quirks.  for example, spanning tree or even switch uplink
>>> connect times can be measured in many seconds.
>>> 
>>> - erik
>>> 
>