Re: page fault while in kernel mode

2008-10-21 Thread John Baldwin
On Tuesday 21 October 2008 03:22:28 pm Robert Fitzpatrick wrote:
> On Tue, 2008-10-21 at 15:09 -0400, John Baldwin wrote:
> > Alternatively, you could just remove the 'device adv' line from your
> > kernel 
> > config rather than adding lots of 'nodevice' lines at the bottom.  You
> > can 
> > usually do 'man 4 ' to see what devices it supports.  In
> > this 
> > case, adv(4) supports mostly ancient Advansys SCSI host adapters.
> > The 
> > manpage has a full list of the various model numbers, etc.
> 
> Yes, that is what I thought. Right now, I am just commenting them out,
> now I know what people mean when they say they are running a
> trimmed/clean kernel.
> 
> I did see one potential issue...
> 
> # USB support
> device  uhci# UHCI PCI->USB interface
> device  ohci# OHCI PCI->USB interface
> device  ehci# EHCI PCI->USB interface (USB 2.0)
> device  usb # USB Bus (required)
> 
> I see all of these with nodevice lines in the PAE file. Although I have
> USB ports, I don't use them, but I was concerned by the 'required' on
> the last one, is it OK to remove? Also, would I then need to disable USB
> in the BIOS to avoid errors?

Actually, USB is ok with PAE.  I recently updated the PAE configs to not 
disable PAE and at work we've run PAE kernels with USB enabled for a few 
years now on 6.x w/o any problems.

-- 
John Baldwin
___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[EMAIL PROTECTED]"


Re: page fault while in kernel mode

2008-10-21 Thread Jeremy Chadwick
On Tue, Oct 21, 2008 at 03:22:28PM -0400, Robert Fitzpatrick wrote:
> On Tue, 2008-10-21 at 15:09 -0400, John Baldwin wrote:
> > Alternatively, you could just remove the 'device adv' line from your
> > kernel 
> > config rather than adding lots of 'nodevice' lines at the bottom.  You
> > can 
> > usually do 'man 4 ' to see what devices it supports.  In
> > this 
> > case, adv(4) supports mostly ancient Advansys SCSI host adapters.
> > The 
> > manpage has a full list of the various model numbers, etc.
> 
> Yes, that is what I thought. Right now, I am just commenting them out,
> now I know what people mean when they say they are running a
> trimmed/clean kernel.
> 
> I did see one potential issue...
> 
> # USB support
> device  uhci# UHCI PCI->USB interface
> device  ohci# OHCI PCI->USB interface
> device  ehci# EHCI PCI->USB interface (USB 2.0)
> device  usb # USB Bus (required)
> 
> I see all of these with nodevice lines in the PAE file. Although I have
> USB ports, I don't use them, but I was concerned by the 'required' on
> the last one, is it OK to remove? Also, would I then need to disable USB
> in the BIOS to avoid errors?

If you remove "device usb", you will also need to remove uhci, ohci,
ehci, umass, ukbd, etc. etc. etc... from your config as well.

You do not need to disable USB support in the BIOS; the kernel will
simply state that it sees devices on the PCI bus but lacks a driver to
attach to them.  This will not harm anything.

-- 
| Jeremy Chadwickjdc at parodius.com |
| Parodius Networking   http://www.parodius.com/ |
| UNIX Systems Administrator  Mountain View, CA, USA |
| Making life hard for others since 1977.  PGP: 4BD6C0CB |

___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[EMAIL PROTECTED]"


Re: page fault while in kernel mode

2008-10-21 Thread Robert Fitzpatrick
On Tue, 2008-10-21 at 15:09 -0400, John Baldwin wrote:
> Alternatively, you could just remove the 'device adv' line from your
> kernel 
> config rather than adding lots of 'nodevice' lines at the bottom.  You
> can 
> usually do 'man 4 ' to see what devices it supports.  In
> this 
> case, adv(4) supports mostly ancient Advansys SCSI host adapters.
> The 
> manpage has a full list of the various model numbers, etc.

Yes, that is what I thought. Right now, I am just commenting them out,
now I know what people mean when they say they are running a
trimmed/clean kernel.

I did see one potential issue...

# USB support
device  uhci# UHCI PCI->USB interface
device  ohci# OHCI PCI->USB interface
device  ehci# EHCI PCI->USB interface (USB 2.0)
device  usb # USB Bus (required)

I see all of these with nodevice lines in the PAE file. Although I have
USB ports, I don't use them, but I was concerned by the 'required' on
the last one, is it OK to remove? Also, would I then need to disable USB
in the BIOS to avoid errors?

-- 
Robert

___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[EMAIL PROTECTED]"


Re: page fault while in kernel mode

2008-10-21 Thread John Baldwin
On Tuesday 21 October 2008 02:47:11 pm Jeremy Chadwick wrote:
> On Tue, Oct 21, 2008 at 02:35:22PM -0400, Robert Fitzpatrick wrote:
> > On Tue, 2008-10-21 at 12:03 -0400, John Baldwin wrote:
> > > Some drivers don't work with PAE (see all the 'nodevice' lines 
> > > in /sys/i386/conf/PAE).  You'll need to purge those drivers from your
> > > config.  
> > > If you are using the hardware those drivers support, then you can't
> > > use PAE.
> > 
> > Thanks for the help. Excuse the ignorance, I'm more a programmer than
> > system guy. How do I purge a driver, or know which driver to look for,
> > from the config and know what the driver supports? Do you mean, in this
> > case, remove 'nodevice adv' from the PAE file? If so, I don't know what
> > that supports :/
> 
> Yeah, I don't think anyone's really explaining this very well to you, so
> I'll try a different approach:
> 
> Certain FreeBSD drivers do not work in PAE mode.
> 
> The drivers which don't work are listed in the /sys/i386/conf/PAE
> file.  They're prefixed by the word "nodevice", which tells the
> kernel config reader "DO NOT build this device, because it won't
> work".
> 
> You will need to take the "nodevice" lines from /sys/i386/conf/PAE and
> put them into your kernel config file.  (There are alternative methods
> such as using "include" directives and so on, but I'm trying to keep
> this explanation simple.)
> 
> Make sense now?  :-)

Alternatively, you could just remove the 'device adv' line from your kernel 
config rather than adding lots of 'nodevice' lines at the bottom.  You can 
usually do 'man 4 ' to see what devices it supports.  In this 
case, adv(4) supports mostly ancient Advansys SCSI host adapters.  The 
manpage has a full list of the various model numbers, etc.

-- 
John Baldwin
___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[EMAIL PROTECTED]"


Re: page fault while in kernel mode

2008-10-21 Thread Robert Fitzpatrick
On Tue, 2008-10-21 at 11:47 -0700, Jeremy Chadwick wrote:
> The drivers which don't work are listed in the /sys/i386/conf/PAE
> file.  They're prefixed by the word "nodevice", which tells the
> kernel config reader "DO NOT build this device, because it won't
> work".
> 
> You will need to take the "nodevice" lines from /sys/i386/conf/PAE and
> put them into your kernel config file.  (There are alternative methods
> such as using "include" directives and so on, but I'm trying to keep
> this explanation simple.)
> 
> Make sense now?  :-)

Perfect sense now, believe it or not I was beginning to think along
these lines as I was doing some searching and found 'device adv' in my
config file and there was a description of the hardware it was for,
which I don't have. Thanks for the clarification, now let's see if I can
get this build done...thanks to all!

-- 
Robert

___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[EMAIL PROTECTED]"


Re: page fault while in kernel mode

2008-10-21 Thread Jeremy Chadwick
On Tue, Oct 21, 2008 at 02:35:22PM -0400, Robert Fitzpatrick wrote:
> On Tue, 2008-10-21 at 12:03 -0400, John Baldwin wrote:
> > Some drivers don't work with PAE (see all the 'nodevice' lines 
> > in /sys/i386/conf/PAE).  You'll need to purge those drivers from your
> > config.  
> > If you are using the hardware those drivers support, then you can't
> > use PAE.
> 
> Thanks for the help. Excuse the ignorance, I'm more a programmer than
> system guy. How do I purge a driver, or know which driver to look for,
> from the config and know what the driver supports? Do you mean, in this
> case, remove 'nodevice adv' from the PAE file? If so, I don't know what
> that supports :/

Yeah, I don't think anyone's really explaining this very well to you, so
I'll try a different approach:

Certain FreeBSD drivers do not work in PAE mode.

The drivers which don't work are listed in the /sys/i386/conf/PAE
file.  They're prefixed by the word "nodevice", which tells the
kernel config reader "DO NOT build this device, because it won't
work".

You will need to take the "nodevice" lines from /sys/i386/conf/PAE and
put them into your kernel config file.  (There are alternative methods
such as using "include" directives and so on, but I'm trying to keep
this explanation simple.)

Make sense now?  :-)

-- 
| Jeremy Chadwickjdc at parodius.com |
| Parodius Networking   http://www.parodius.com/ |
| UNIX Systems Administrator  Mountain View, CA, USA |
| Making life hard for others since 1977.  PGP: 4BD6C0CB |

___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[EMAIL PROTECTED]"


Re: page fault while in kernel mode

2008-10-21 Thread Robert Fitzpatrick
On Tue, 2008-10-21 at 12:03 -0400, John Baldwin wrote:
> Some drivers don't work with PAE (see all the 'nodevice' lines 
> in /sys/i386/conf/PAE).  You'll need to purge those drivers from your
> config.  
> If you are using the hardware those drivers support, then you can't
> use PAE.

Thanks for the help. Excuse the ignorance, I'm more a programmer than
system guy. How do I purge a driver, or know which driver to look for,
from the config and know what the driver supports? Do you mean, in this
case, remove 'nodevice adv' from the PAE file? If so, I don't know what
that supports :/

-- 
Robert

___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[EMAIL PROTECTED]"


Re: page fault while in kernel mode

2008-10-21 Thread John Baldwin
On Monday 20 October 2008 08:52:07 pm Robert Fitzpatrick wrote:
> On Mon, 2008-10-20 at 13:45 -0400, John Baldwin wrote:
> > > i386 cannot address more than 4GB unless the kernel is built with
> > PAE
> > > mode enabled.  This isn't enabled in GENERIC for many (justified)
> > > reasons.  If you have more than 4GB, you should be using amd64, so
> > you
> > > made the right decision there.
> > 
> > If you aren't using kernel modules, then PAE should work fine.  You
> > can make 
> > kernel modules work with PAE as well, but that takes more work.
> 
> Thanks for the help, I am missing AMD Features for this CPU in dmesg, so
> it looks like the CPU does not support amd64. I tried to build my own
> kernel with PAE option and getting the following error...
> 
> /usr/src/sys/dev/advansys/advansys.c: In function 'adv_action':
> /usr/src/sys/dev/advansys/advansys.c:259: warning: cast from pointer to
> integer of different size
> *** Error code 1
> 
> Any idea what I can do for this error?

Some drivers don't work with PAE (see all the 'nodevice' lines 
in /sys/i386/conf/PAE).  You'll need to purge those drivers from your config.  
If you are using the hardware those drivers support, then you can't use PAE.

-- 
John Baldwin
___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[EMAIL PROTECTED]"


Re: page fault while in kernel mode

2008-10-20 Thread Robert Fitzpatrick
On Mon, 2008-10-20 at 13:45 -0400, John Baldwin wrote:
> > i386 cannot address more than 4GB unless the kernel is built with
> PAE
> > mode enabled.  This isn't enabled in GENERIC for many (justified)
> > reasons.  If you have more than 4GB, you should be using amd64, so
> you
> > made the right decision there.
> 
> If you aren't using kernel modules, then PAE should work fine.  You
> can make 
> kernel modules work with PAE as well, but that takes more work.

Thanks for the help, I am missing AMD Features for this CPU in dmesg, so
it looks like the CPU does not support amd64. I tried to build my own
kernel with PAE option and getting the following error...

/usr/src/sys/dev/advansys/advansys.c: In function 'adv_action':
/usr/src/sys/dev/advansys/advansys.c:259: warning: cast from pointer to
integer of different size
*** Error code 1

Any idea what I can do for this error?

--
Robert

___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[EMAIL PROTECTED]"


Re: page fault while in kernel mode

2008-10-20 Thread John Baldwin
On Monday 20 October 2008 12:32:37 pm Jeremy Chadwick wrote:
> > Forgot to mention I added memory to this server as well, took it from
> > 2GB it was using under 5.4-RELEASE up to 6GB filling all slots, that is
> > why I wanted to load amd64. I reduced down to 4GB and now am able to
> > install 7.0-RELEASE i386. Does this mean that I may have a hardware
> > issue or can FreeBSD produce the page fault I was getting when using
> > over 4GB with i386?
> 
> i386 cannot address more than 4GB unless the kernel is built with PAE
> mode enabled.  This isn't enabled in GENERIC for many (justified)
> reasons.  If you have more than 4GB, you should be using amd64, so you
> made the right decision there.

If you aren't using kernel modules, then PAE should work fine.  You can make 
kernel modules work with PAE as well, but that takes more work.

> > I would love to figure out this BTX halted issue instead...any ideas
> > on that?
> 
> Boot loader problems are difficult to figure out/debug for reasons which
> should be obvious.  I'm CC'ing John Baldwin here, who has experience
> with BTX.  He might be able to shed some light on this.

You will get a BTX fault in 7.0 if your CPU does not support 64-bit "long 
mode" (i.e., amd64).  You can check to see if your CPU does support it by 
looking in the 'AMD features' line of 'dmesg' from an i386 kernel and seeing 
if you have a 'LM' feature.  If you don't, your CPU only supports i386.

-- 
John Baldwin
___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[EMAIL PROTECTED]"


Re: page fault while in kernel mode

2008-10-20 Thread Jeremy Chadwick
On Mon, Oct 20, 2008 at 12:07:17PM -0400, Robert Fitzpatrick wrote:
> On Sun, 2008-10-19 at 13:16 -0700, Jeremy Chadwick wrote:
> > On Sun, Oct 19, 2008 at 03:50:01PM -0400, Robert Fitzpatrick wrote:
> > > I took a working 5.4-i386 server and trying to convert its RAID 5 to
> > > RAID 10 and load 7.0 amd64. I kept getting BTX halted even after
> > > flashing the latest bios and firmware for the raid card, Intel SRCZCR,
> > > in this dual Xeon 2.4GHz supermicro superserver. I have another server,
> > > bit newer, but same basic hardware makeup with Xeon 3.0 procs that runs
> > > 6.1-amd64 fine. Anyway, so I have resorted to the i386 version of 7.0 to
> > > see if the server is just incapable of running amd64, which after
> > > passing the initial boot where amd64 failed, now gives me the subject
> > > error after some reference to GEOM_LABEL. I did rebuild the RAID to
> > > RAID-10, can someone tell me what this error means?
> > > 
> > > http://columbus.webtent.org/freebsd.png
> > 
> > Can you please try 7.1-BETA2 instead (ISOs are now available)?  There
> > have been fixes/improvements to BTX since 7.0-RELEASE which could fix
> > your problem.
> > 
> 
> Thanks, but that didn't work either trying 7.1-BETA2 amd64 :(
> 
> Forgot to mention I added memory to this server as well, took it from
> 2GB it was using under 5.4-RELEASE up to 6GB filling all slots, that is
> why I wanted to load amd64. I reduced down to 4GB and now am able to
> install 7.0-RELEASE i386. Does this mean that I may have a hardware
> issue or can FreeBSD produce the page fault I was getting when using
> over 4GB with i386?

i386 cannot address more than 4GB unless the kernel is built with PAE
mode enabled.  This isn't enabled in GENERIC for many (justified)
reasons.  If you have more than 4GB, you should be using amd64, so you
made the right decision there.

> I would love to figure out this BTX halted issue instead...any ideas
> on that?

Boot loader problems are difficult to figure out/debug for reasons which
should be obvious.  I'm CC'ing John Baldwin here, who has experience
with BTX.  He might be able to shed some light on this.

-- 
| Jeremy Chadwickjdc at parodius.com |
| Parodius Networking   http://www.parodius.com/ |
| UNIX Systems Administrator  Mountain View, CA, USA |
| Making life hard for others since 1977.  PGP: 4BD6C0CB |

___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[EMAIL PROTECTED]"


Re: page fault while in kernel mode

2008-10-20 Thread Robert Fitzpatrick
On Sun, 2008-10-19 at 13:16 -0700, Jeremy Chadwick wrote:
> On Sun, Oct 19, 2008 at 03:50:01PM -0400, Robert Fitzpatrick wrote:
> > I took a working 5.4-i386 server and trying to convert its RAID 5 to
> > RAID 10 and load 7.0 amd64. I kept getting BTX halted even after
> > flashing the latest bios and firmware for the raid card, Intel SRCZCR,
> > in this dual Xeon 2.4GHz supermicro superserver. I have another server,
> > bit newer, but same basic hardware makeup with Xeon 3.0 procs that runs
> > 6.1-amd64 fine. Anyway, so I have resorted to the i386 version of 7.0 to
> > see if the server is just incapable of running amd64, which after
> > passing the initial boot where amd64 failed, now gives me the subject
> > error after some reference to GEOM_LABEL. I did rebuild the RAID to
> > RAID-10, can someone tell me what this error means?
> > 
> > http://columbus.webtent.org/freebsd.png
> 
> Can you please try 7.1-BETA2 instead (ISOs are now available)?  There
> have been fixes/improvements to BTX since 7.0-RELEASE which could fix
> your problem.
> 

Thanks, but that didn't work either trying 7.1-BETA2 amd64 :(

Forgot to mention I added memory to this server as well, took it from
2GB it was using under 5.4-RELEASE up to 6GB filling all slots, that is
why I wanted to load amd64. I reduced down to 4GB and now am able to
install 7.0-RELEASE i386. Does this mean that I may have a hardware
issue or can FreeBSD produce the page fault I was getting when using
over 4GB with i386? I would love to figure out this BTX halted issue
instead...any ideas on that?

-- 
Robert

___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[EMAIL PROTECTED]"


Re: page fault while in kernel mode

2008-10-19 Thread Jeremy Chadwick
On Sun, Oct 19, 2008 at 03:50:01PM -0400, Robert Fitzpatrick wrote:
> I took a working 5.4-i386 server and trying to convert its RAID 5 to
> RAID 10 and load 7.0 amd64. I kept getting BTX halted even after
> flashing the latest bios and firmware for the raid card, Intel SRCZCR,
> in this dual Xeon 2.4GHz supermicro superserver. I have another server,
> bit newer, but same basic hardware makeup with Xeon 3.0 procs that runs
> 6.1-amd64 fine. Anyway, so I have resorted to the i386 version of 7.0 to
> see if the server is just incapable of running amd64, which after
> passing the initial boot where amd64 failed, now gives me the subject
> error after some reference to GEOM_LABEL. I did rebuild the RAID to
> RAID-10, can someone tell me what this error means?
> 
> http://columbus.webtent.org/freebsd.png

Can you please try 7.1-BETA2 instead (ISOs are now available)?  There
have been fixes/improvements to BTX since 7.0-RELEASE which could fix
your problem.

-- 
| Jeremy Chadwickjdc at parodius.com |
| Parodius Networking   http://www.parodius.com/ |
| UNIX Systems Administrator  Mountain View, CA, USA |
| Making life hard for others since 1977.  PGP: 4BD6C0CB |

___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[EMAIL PROTECTED]"


Re: page fault while in kernel mode

2005-10-25 Thread Kris Kennaway
On Mon, Oct 24, 2005 at 12:11:29PM +0100, Owen Smith wrote:

> Whats the best thing todo? debugging kernel etc or just upgrade to 5.4?

The latter.  A few hundred bugs were fixed between 5.3 and 5.4, and
the former is no longer supported anyway.

Kris


pgpT4R8dnqxnu.pgp
Description: PGP signature


Re: page fault while in kernel mode

2005-10-24 Thread stan
On Mon, Oct 24, 2005 at 12:11:29PM +0100, Owen Smith wrote:
> 
> Whats the best thing todo? debugging kernel etc or just upgrade to 5.4?
> 

Run a memory est on the machine. memtest386 works well.
-- 
U.S. Encouraged by Vietnam Vote - Officials Cite 83% Turnout Despite Vietcong 
Terror 
- New York Times 9/3/1967
___
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[EMAIL PROTECTED]"