Skip to contentSkip to conversation

Recent changes

parisc-linux Alex deVries 5 participants 12 messages Nov 3, 1999 – Nov 8, 1999 1 message not in the archive

lore
Share & tools
lore
#1 lore raw
In reply to a message not in the archive

Matt Wilcox committed some code that I know he's unable to test, so I'm mailing the bug report here.

The first is that building the kernel kernel will result in:

ld: Unsatisfied symbols:
hpux_brk (code)

so I just commented out the only reference to it, and now I end up with:

(start_code) 00001000
(end_data) 40008000
(start_stack) 20000028
(brk) 40017000

PSW : 0004f80b GR 1 : 40006928 GR 2 : 00016f5f GR 3 : 40007128 GR 4 : 40016314 GR 5 : 40006928 GR 6 : 00002001 GR 7 : 00000020 GR 8 : 40001511 GR 9 : 00000000 GR10 : 00000050 GR11 : bc0fd569 GR12 : 000043ff GR13 : 00004400 GR14 : 00004800 GR15 : 43f02aa9 GR16 : 00004000 GR17 : 00000000 GR18 : 00000c00 GR19 : 00000001 GR20 : 00000001 GR21 : 40019000 GR22 : ffffff05 GR23 : 00000000 GR24 : 00000000 GR25 : ffffffff GR26 : ffffffff GR27 : 40007128 GR28 : 4001830c GR29 : 40018ffc GR30 : 20000780 GR31 : 00033323 SR0 : 00000000 SR1 : 00000000 SR2 : 00000000 SR3 : 00000000 SR4 : 00000000 SR5 : 00000000 SR6 : 00000000 SR7 : 00000000 IAOQ : 00016f7b 00016f7f

bad address 40018ffc (code 15)

Kernel panic: bad address

In swapper task - not syncing

And that is it.

- Alex

ps. All the puffins will be out of town in Montreal on Wednesday to attend/speak at a conference and see live puffins.

Signature · 3 lines
Alex deVries <[email protected]>
Vice President Engineering
The Puffin Group
#2 lore raw
Quoted 10 lines · Alex deVries

Matt Wilcox committed some code that I know he's unable to test, so I'm mailing the bug report here.

The first is that building the kernel kernel will result in:

ld: Unsatisfied symbols:
hpux_brk (code)

so I just commented out the only reference to it, and now I end up with:

You commented it out? Aaahhh.. that was definitely the wrong thing to do. The reference is in the middle of the syscall table, so all subsequent syscalls were misnumbered.

I've changed it to call sys_brk instead; I'm not quite sure why I had thought it would need to call a different function, brk looks identical to me on both OSes.

If someone would be generous enough to test the current CVS tree for me, I'd be grateful. It may well be that sash is now able to perform `-ls'.

Quoted 2 lines

ps. All the puffins will be out of town in Montreal on Wednesday to attend/speak at a conference and see live puffins.

I'm jealous. 27 days to go.

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
In reply to Alex deVries
Quoted 2 lines · Alex deVries

ld: Unsatisfied symbols:
hpux_brk (code)

Why did the definition of hpux_brk (in kernel/sys_hpux.c) ever vanish ? Some time ago, it looked like this:

unsigned long sys_brk(unsigned long addr);

unsigned long hpux_brk(unsigned long addr)

{
        /* Sigh.  Looks like HP/UX libc relies on kernel bugs. */
        return sys_brk(addr + PAGE_SIZE);
}

Philipp Rumpf

#3 lore raw
In reply to Matthew Wilcox
Quoted 3 lines

I've changed it to call sys_brk instead; I'm not quite sure why I had thought it would need to call a different function, brk looks identical to me on both OSes.

It isn't. HP/UX brk seems to be equivalent to sys_brk(addr + PAGE_SIZE), i.e. while Linux gets passed a pointer to the first byte after the end of the brk'd segment, HP/UX brk gets passed a pointer to the first byte of the last page of it. At least, this seemed to be the case for our first binaries.

Philipp Rumpf

#5 lore raw
In reply to Philipp Rumpf
Quoted 7 lines · Philipp Rumpf
Quoted 2 lines · Alex deVries

ld: Unsatisfied symbols:
hpux_brk (code)

Why did the definition of hpux_brk (in kernel/sys_hpux.c) ever vanish ? Some time ago, it looked like this:

Quoted 7 lines

unsigned long sys_brk(unsigned long addr);

unsigned long hpux_brk(unsigned long addr)

{
        /* Sigh.  Looks like HP/UX libc relies on kernel bugs. */
        return sys_brk(addr + PAGE_SIZE);
}

It looks like I'm accidentally responsible for deleting that. Maybe this explains why grant found that `alias' wasn't working well in sash. I've just committed this function back.

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
#6 lore raw

[parisc-linux] cant use rbootd

Hi,

I am using an rbootd running on Debian 2.1 , but my HP 712/60 cant find it, I put the correct ethernet adress in the /etc/rbootd.conf

but nothing happens if i type

BOOT_ADMIN> search lan

bye
Andi

#7 lore raw

Re: cant use rbootd

Quoted 4 lines · Andi

Hi,

I am using an rbootd running on Debian 2.1 , but my HP 712/60 cant find it, I put the correct ethernet adress in the /etc/rbootd.conf

I guess you'll need the Puffins patch applied. The patch is available in the upcoming unstable aka 2.2 of Debian - I hope... In any case you'll find a non-official package at http://master.debian.org/~joey/NMU/ . On 2.1 box you'll need to recompile it.

Regards,

Joey

Signature · 2 lines
Beware of bugs in the above code; I have only proved it correct,
not tried it.  -- Donald E. Knuth
#10 lore raw

Re: [parisc-linux] cant use rbootd

In reply to Andi
Quoted 8 lines · Andi

Hi,

I am using an rbootd running on Debian 2.1 , but my HP 712/60 cant find it, I put the correct ethernet adress in the /etc/rbootd.conf

but nothing happens if i type

BOOT_ADMIN> search lan

Is rbootd definitely running? I find that if it can't change to the appropriate directory, it just exits silently.

BTW, /export/hp is exactly the same length as /usr/local, so

sed -e s:/export/hp:/usr/local:

allows you to use /usr/local/rbootd as somewhere to put kernels if you don't fancy clusteering your /.

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
#11 lore raw

Re: [parisc-linux] cant use rbootd

In reply to Andi
Quoted 8 lines · Andi

Hi,

I am using an rbootd running on Debian 2.1 , but my HP 712/60 cant find it, I put the correct ethernet adress in the /etc/rbootd.conf

but nothing happens if i type

BOOT_ADMIN> search lan

OK, forget what I wrote, I used rbootd instead of bootp... oops. Booting over the network works now. I tried the latest snapshot of the PARISC-Linux Kernel. But it doenst seem to get far...

after

"transferring control to kernel (...)"

nothing happens.

Signature · 2 lines
bye
  Andi
#8 lore raw

Re: cant use rbootd

In reply to Martin Schulze
Quoted 8 lines · Martin Schulze
Quoted 3 lines

I am using an rbootd running on Debian 2.1 , but my HP 712/60 cant find it, I put the correct ethernet adress in the /etc/rbootd.conf

I guess you'll need the Puffins patch applied. The patch is available in the upcoming unstable aka 2.2 of Debian - I hope... In any case you'll find a non-official package at http://master.debian.org/~joey/NMU/ . On 2.1 box you'll need to recompile it.

To make things clearer, I know for a fact that the rpm that's on puffin.external.hp.com will work if you create the directory that's not in the package (yes, I've been meaning to fix that package for about 6 months now). The Debian package should have the same patches applied.

Also, I'd really appreciate it if you could mail that Debian binary package to me so I can put it on puffin.external.hp.com.

- Alex "puffin" deVries.

#9 lore raw

Re: cant use rbootd

Quoted 2 lines · Alex deVries

Also, I'd really appreciate it if you could mail that Debian binary package to me so I can put it on puffin.external.hp.com.

Errr, can't you use lynx or wget?

Anyway, the files should be in your mailbox, source and binary.

I'll try to keep up updated.

Regards,

Joey

Signature · 2 lines
Beware of bugs in the above code; I have only proved it correct,
not tried it.  -- Donald E. Knuth
#12 lore raw

Re: [parisc-linux] cant use rbootd

In reply to Andi
Quoted 9 lines · Andi

OK, forget what I wrote, I used rbootd instead of bootp... oops. Booting over the network works now. I tried the latest snapshot of the PARISC-Linux Kernel. But it doenst seem to get far...

after

"transferring control to kernel (...)"

nothing happens.

See

http://puffin.external.hp.com/mailing-lists/parisc-linux/1068.html

The information contained therein is essential for anyone not using a serial console.

P.S. I've managed to find a machine to act as an rboot server now, so I've started hacking on the HIL driver to make it work on the 7xx series as well as the 300.

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