Skip to contentSkip to conversation

initrd/ramdisk-Status...

parisc-linux Helge Deller 5 participants 10 messages Aug 24, 1999 – Sep 11, 1999

lore
Share & tools
lore
#0 lore raw

Hello,

After my last commits, I think it is time to inform the list of the status of initrd:

The first: initrd via bootloader should be usable now....

Internals / How to use:
I programmed it in such a way, that the bootable image in arch/parisc/boot/ holds a directory-structure for up to 63 possible ramdisks, which can be selected during boot via initrd=<number> or initrd=<name>.
The reason for adding support for multiple ramdisks is, that distributions later can add easier support for multiple/different machines in one bootfile.

Ready-made-ramdisks with a name of "ramdisk.bin", which resides in the main-directory of the Linux-Sources, will be automatically added during a "make Image". More ramdisks can be added at any time with the program arch/parisc/boot/addramdisk.

The bootloader will print at startup a list of all available ramdisks and loads the first (if any) ramdisk by default.
More, the bootloader gives you the possibility to change the commandline (and the initrd=-command) if it is started via "boot lan isl" or "boot alt isl"...

Memory-Layout:
Since I didn�t knew, where to load the ramdisk I decided to load it to the end of the available RAM. The adress will be printed during boot.

Ideas and comments are welcome,

Helge Deller.

#1 lore raw
Quoted 5 lines · Helge Deller

I programmed it in such a way, that the bootable image in arch/parisc/boot/ holds a directory-structure for up to 63 possible ramdisks, which can be selected during boot via initrd=<number> or initrd=<name>.
The reason for adding support for multiple ramdisks is, that distributions later can add easier support for multiple/different machines in one bootfile.

Hang on, this doesn't make sense. There should be no difference between the binaries, the only difference should be the kernels. I believe the plan is to put UFS-reading code into the bootloader in any case (mkp?) which makes all this a temporary hack; as it should be.

Signature · 4 lines
Matthew Wilcox <[email protected]>
"Windows and MacOS are products, contrived by engineers in the service of
specific companies. Unix, by contrast, is not so much a product as it is a
painstakingly compiled oral history of the hacker subculture." - N Stephenson
#2 lore raw

[parisc-linux] Booting 715/100

I finally managed to build a kernel from the 08/07/99 tar file. Now I can't boot it!

I might need to update the source tree again but need to figure out how CVS can operate through HP's firewall.

Here's the /etc/bootptab entry:

linuxferris:\
        hd=/linux/linux-2.2.3/arch/parisc/boot:\
        bf=Image:\
        vm=rfc1048:\
        ht=ether:\
        ha=0800097B9230:\
        sm=255.255.248.0:\
        ip=15.16.122.147

Where "/linux/linux-2.2.3" is the kernel source tree.

I also wasn't sure about the difference between CONFIG_PA7100 and CONFIG_PA7300 in the .config file. I thought the 715/100 is an older machine before 7300LC was available and thus is PA7100. Is this right?

"lifls Image" returns
lifls: Can't list Image; not a LIF volume

Whereas "file Image" says:
Image: lif file

<sigh>...can anyone explain this?

Lastly, I've attached the console output.

thanks,
grant

BOOT_ADMIN> inf

System model number: HP 9000/715/100
Processor revision 2.4 100MHz

Instruction Cache Size:  131072
Data Cache Size:         131072
Memory Size:  64 MB

Built in floating point coprocessor
Board Serial Number 40SM891817

BootRom Version 1.6

auto boot on
auto search on
fastboot on

Primary boot path:    scsi.6.0
Alternate boot path:  lan.000000-000000.0.0
Console path:         graphics_1.1
Keyboard path:        ps2

LAN Station Addresses: 080009-7B9230

                       080009-7B9231

BOOT_ADMIN> sea lan

Searching for potential boot device.
This may take several minutes.

To discontinue, press ESCAPE.

Device Path Device Type
----------- -----------
lan.15.16.121.48.3.52

BOOT_ADMIN> bo lan.15.16.121.48.3.52
0x06.C740

Attempting to boot.

Loading Initial Program Loader

0x06.8007
0x06.8007
0x06.8007
0x06.8007

ERROR: Cannot load IPL, error detected

0x06.C7F0
Boot process failed. press <ESCAPE> to Continue.

#3 lore raw

Re: [parisc-linux] Booting 715/100

Quoted 1 line · Grant Grundler

I finally managed to build a kernel from the 08/07/99 tar file.

I assume you mean 07/Aug/99 rather than 08/Jul/99? :-)

Quoted 4 lines

Now I can't boot it!

I might need to update the source tree again but need to figure out how CVS can operate through HP's firewall.

Can you ssh out? CVS_RSH=ssh should do the trick if so.

Quoted 4 lines

I also wasn't sure about the difference between CONFIG_PA7100 and CONFIG_PA7300 in the .config file. I thought the 715/100 is an older machine before 7300LC was available and thus is PA7100. Is this right?

Well, according to /usr/sam/lib/mo/sched.models (a very useful file!)

715 1.1c PA7100LC

It doesn't mention the submodels explicitly. However,

http://www.thepuffingroup.com/parisc/hp9000_models.html

says the 715/100 (both variants!) have PCXL CPUs which is the same as that on the 712 models, which we know to be the 7100LC.

Sorry I can't be of any help on the actual problem though...

Signature · 4 lines
Matthew Wilcox <[email protected]>
"Windows and MacOS are products, contrived by engineers in the service of
specific companies. Unix, by contrast, is not so much a product as it is a
painstakingly compiled oral history of the hacker subculture." - N Stephenson
#4 lore raw

Re: [parisc-linux] Booting 715/100

Quoted 4 lines · Matthew Wilcox
Quoted 1 line · Grant Grundler

I finally managed to build a kernel from the 08/07/99 tar file.

I assume you mean 07/Aug/99 rather than 08/Jul/99? :-)

Yes - sorry about the convoluted american date notation. The linux-2.2.3-19990808.tar.gz is only 291 bytes. 19990807 was the last "full" tarball.

Quoted 6 lines
Quoted 4 lines

Now I can't boot it!

I might need to update the source tree again but need to figure out how CVS can operate through HP's firewall.

Can you ssh out? CVS_RSH=ssh should do the trick if so.

Didn't have ssh...but "socksify" command seems to work rather nicely. Found it in /opt/socks/bin and cvs can now talk to the puffin.external.hp.com.

Quoted 8 lines
Quoted 4 lines

I also wasn't sure about the difference between CONFIG_PA7100 and CONFIG_PA7300 in the .config file. I thought the 715/100 is an older machine before 7300LC was available and thus is PA7100. Is this right?

Well, according to /usr/sam/lib/mo/sched.models (a very useful file!)

715 1.1c PA7100LC

Forgot about this table...good thinking.

Quoted 9 lines

It doesn't mention the submodels explicitly. However,

http://www.thepuffingroup.com/parisc/hp9000_models.html

says the 715/100 (both variants!) have PCXL CPUs which is the same as that on the 712 models, which we know to be the 7100LC.

Sorry I can't be of any help on the actual problem though...

This already helped...I'll assume CONFIG_PA7100 means PA7100LC.

BTW, I went through the linux/.config file and found conflicting options enabled (eg. 712 was enabled). So, I'm rebuilding...and we'll see what happens.

thanks,
grant

#5 lore raw

Re: [parisc-linux] Booting 715/100

Quoted 7 lines
Quoted 1 line

Sorry I can't be of any help on the actual problem though...

This already helped...I'll assume CONFIG_PA7100 means PA7100LC.

BTW, I went through the linux/.config file and found conflicting options enabled (eg. 712 was enabled). So, I'm rebuilding...and we'll see what happens.

Right now CONFIG_PA7100 and CONFIG_PA7300LC don't do anything, so you don't need to be worried about that.

#6 lore raw

Re: [parisc-linux] Booting 715/100

In reply to Grant Grundler
Quoted 3 lines · Grant Grundler

Matthew Wilcox wrote:
The linux-2.2.3-19990808.tar.gz is only 291 bytes. 19990807 was the last "full" tarball.

Actually, I've fixed this now and now you should find a splendid 13MB file named linux-2.2.1-19990824.tar.gz.

Quoted 5 lines
Quoted 1 line

Can you ssh out? CVS_RSH=ssh should do the trick if so.

Didn't have ssh...but "socksify" command seems to work rather nicely. Found it in /opt/socks/bin and cvs can now talk to the puffin.external.hp.com.

I'm heavenly ignorant of socks, but if you intend to commit code, you really have to get ssh working. It's the only way we can know that you are you when you commit code, and makes life a lot simpler.

- Alex

Signature · 3 lines
Alex deVries
Vice President of Engineering
The Puffin Group
#7 lore raw

Re: [parisc-linux] Booting 715/100

In reply to Matthew Wilcox
Quoted 8 lines · Matthew Wilcox
Quoted 4 lines

I also wasn't sure about the difference between CONFIG_PA7100 and CONFIG_PA7300 in the .config file. I thought the 715/100 is an older machine before 7300LC was available and thus is PA7100. Is this right?

Well, according to /usr/sam/lib/mo/sched.models (a very useful file!)

715 1.1c PA7100LC

On my HP-UX 10.20 machine the path is /usr/lib/sched.models

Quoted 1 line

It doesn't mention the submodels explicitly. However,

Nor the "internal" project names...most people I've worked have no clue what's in an F20 or I60 (and I don't either). One of those gaps between marketing and developement.

Quoted 4 lines

http://www.thepuffingroup.com/parisc/hp9000_models.html

says the 715/100 (both variants!) have PCXL CPUs which is the same as that on the 712 models, which we know to be the 7100LC.

Other HP folks:
1) Difference between 7100 and 7100LC the mainly integration of GSC interface? (ignore clock speed differences)
2) Is PCXL really 7100 *and* 7100LC?

I see T500 used 7100s and I'm certain no GSC bus was present...

grant

Grant Grundler
Communications Infrastructure Computer Operations
+1.408.447.7253

#8 lore raw

[parisc-linux] sched.models and HP-UX flavor

In reply to Matthew Wilcox

Matthew Wilcox wrote:
...

Quoted 1 line

Well, according to /usr/sam/lib/mo/sched.models (a very useful file!)

This is the HP-UX 11.X location.
On 10.20, it's /usr/lib/sched.models.
(Agreed - a very useful file)

I think trying to build/link parisc-linux on 10.20 has been the source of my problems. I have the impression binutils and egcs configure scripts are targeted towards hpux11.00 host since they use "elf" instead of "som" output format.
I'm moving to 11.00 host and see if that works better...

I'm not good at (mostly not interested) futzing around with build tools. It would be a good thing if someone either fixed or intentionally broke the parisc-linux configure scripts for hpux10.20 host.

cheers,
grant

ps. parisc/pci.[ch] and dino.c changes are coming along nicely.
Most of the design issues are resolved though a few are still outstanding. It looks like I should be able to drop those in by next week after they are reviewed.

Grant Grundler
Communications Infrastructure Computer Operations
+1.408.447.7253

#9 lore raw

Re: [parisc-linux] Booting 715/100

In reply to Grant Grundler
Quoted 3 lines · Grant Grundler

I finally managed to build a kernel from the 08/07/99 tar file. Now I can't boot it!

No problem. We've been through this a lot.

You should really grab the latest kernel, a lot has changed since either the July 8th or August 7th release. CVS is really the better way to handle this, but you can ftp down the latest (fixed!) source tree from ftp://puffin.external.hp.com/pub/parisc/cvs .

I know there's some problems with our LIF header not being complete, but there's not a lot we can do about this until we have docs on LIF.

- Alex

Signature · 3 lines
Alex deVries
Vice President of Engineering
The Puffin Group