Frank van der Linden | 1 Mar 20:25 2005
Picon

Re: amd64 and PCMCIA

On Mon, Feb 28, 2005 at 02:05:00PM -0600, Samuel Kesterson wrote:
> 	I've purchased a new laptop that's based on the amd64. It's an
> HP zv5410us. I put the dmesg output from both the i386 (GENERIC_LAPTOP)
> and amd64 (GENERIC) kernels up at
> http://www.k-labs.com/~samuelk/dmesg.i386 and
> http://www.k-labs.com/~samuelk/dmesg.amd64 
> 
> 	Here's my problem. The amd64 kernel doesn't seem to support
> PCMCIA. I tried to add it by copying the GENERIC config to
> GENERIC_LAPTOP_64 and uncommenting acpi0, cbb, cardbus, and pcmcia but
> config just tells me:
> 
> GENERIC_LAPTOP_64:285: cardslot*: unknown device 'cardslot'
> GENERIC_LAPTOP_64:288: cardbus*: unknown device 'cardbus'
> GENERIC_LAPTOP_64:289: pcmcia*: unknown device 'pcmcia'
> *** Stop.
> 	
> 	I'd really like to run the amd64 port if I could get this PCMCIA
> stuff working. Does anyone have any suggestions?

That's not hard to do.. So far it hasn't been added because noone has
tested it yet, but it's likely to "just work".

What you need to do is to take the cardbus+pcmcia stuff out of
sys/arch/i386/conf/files.i386, and put it in sys/arch/amd64/conf/files.amd64,
at the same spot. Then reconfigure the kernel.

If this works for you, let me know, and I'll enable it by default.

- Frank
(Continue reading)

Sami Kantoluoto | 1 Mar 22:01 2005
Picon

Re: amd64 and PCMCIA

On Tue, Mar 01, 2005 at 08:25:20PM +0100, Frank van der Linden wrote:
> On Mon, Feb 28, 2005 at 02:05:00PM -0600, Samuel Kesterson wrote:
> > 	I've purchased a new laptop that's based on the amd64. It's an

[snip]

> > 	I'd really like to run the amd64 port if I could get this PCMCIA
> > stuff working. Does anyone have any suggestions?
> 
> That's not hard to do.. So far it hasn't been added because noone has
> tested it yet, but it's likely to "just work".
>
> What you need to do is to take the cardbus+pcmcia stuff out of
> sys/arch/i386/conf/files.i386, and put it in sys/arch/amd64/conf/files.amd64,
> at the same spot. Then reconfigure the kernel.
> 
> If this works for you, let me know, and I'll enable it by default.

This alone was not just enough but I copied
sys/arch/i386/i386/rbus_machdep.c and sys/arch/i386/include/rbus_machdep.h
to appropriate places and with these I got it compiled (NOTE: I don't have
any understanding about those files). But:

cbb0 at pci0 dev 9 function 0: O2 Micro product 0x7114 (rev. 0x20)
cbb0: NOT USED because of unconfigured interrupt
cbb1 at pci0 dev 9 function 1: O2 Micro product 0x7114 (rev. 0x20)
cbb1: NOT USED because of unconfigured interrupt

Any thoughts?

(Continue reading)

Richard Rauch | 3 Mar 07:22 2005

Jerky system behavior.

About August or so of last year, I upgraded from one -current to
another on my AMD64 box and suddenly started seeing a jerkiness to the
system.  Others reported it and I figured that it was a short-lived
kernel issue.  I recently did a new CVS update and built a fresh
kernel.  The problem is still there.

(I only recall AMD64 people remarking about it, so I assume that
it's AMD64-specific.)

Basically, somewhere on the order of each minute, one or more processes
freeze for about half a second, or perhaps a full second.

The system stability does not seem to be affected.

Not everything gets frozen.

Examples of things that have been affected:

 * bzflag (a game from pkgsrc).
 * emacs (text input, menus, making the cursor solid to indicate
   that it is the current window---this is under X, of course).
 * Remote access seems to also be affected sometimes (e.g.,
   shell commands, or a shell itself, while logged in via ssh).

By and large, the mouse cursor does not seem to be affected, so when
a process "freezes", I can still mouse around.  I just can't do anything
until things "unfreeze".

It *usually* happens when I am changing between windows or launching
a new shell command.  (But, as I noted, I also observe what appears
(Continue reading)

Andy Ruhl | 3 Mar 14:09 2005
Picon

Re: Jerky system behavior.

On Thu, 3 Mar 2005 00:22:39 -0600, Richard Rauch <rkr <at> olib.org> wrote:
> About August or so of last year, I upgraded from one -current to
> another on my AMD64 box and suddenly started seeing a jerkiness to the
> system.  Others reported it and I figured that it was a short-lived
> kernel issue.  I recently did a new CVS update and built a fresh
> kernel.  The problem is still there.
> 
> (I only recall AMD64 people remarking about it, so I assume that
> it's AMD64-specific.)
> 
> Basically, somewhere on the order of each minute, one or more processes
> freeze for about half a second, or perhaps a full second.
> 
> The system stability does not seem to be affected.
> 
> Not everything gets frozen.
> 
> Examples of things that have been affected:
> 
>  * bzflag (a game from pkgsrc).
>  * emacs (text input, menus, making the cursor solid to indicate
>    that it is the current window---this is under X, of course).
>  * Remote access seems to also be affected sometimes (e.g.,
>    shell commands, or a shell itself, while logged in via ssh).
> 
> By and large, the mouse cursor does not seem to be affected, so when
> a process "freezes", I can still mouse around.  I just can't do anything
> until things "unfreeze".
> 
> It *usually* happens when I am changing between windows or launching
(Continue reading)

Richard Rauch | 3 Mar 21:50 2005

Re: Jerky system behavior.

On Thu, Mar 03, 2005 at 06:09:45AM -0700, Andy Ruhl wrote:
> On Thu, 3 Mar 2005 00:22:39 -0600, Richard Rauch <rkr <at> olib.org> wrote:
 [...]
> > Basically, somewhere on the order of each minute, one or more processes
> > freeze for about half a second, or perhaps a full second.
 [...]
> I don't know if this helps, but I'm pretty sure I don't see this on
> the release-2 branch. Doesn't cron check the tab once a minute?
> Anything else that happens once a minute?  Is it that regular though?

It's more like every 15 seconds, when I checked last night.  The "on the
order of each minute" was a rough estimate; I didn't want to overestimate
the frequency and had never made any effort to determine the frequency.
(Not having a stopwatch handy, I simply counted seconds off in my head
while playing a little bzflag.)

--

-- 
  "I probably don't know what I'm talking about."  http://www.olib.org/~rkr/

Wolfgang Solfrank | 4 Mar 17:46 2005
Picon

Re: Jerky system behavior.

Hi,

> About August or so of last year, I upgraded from one -current to
> another on my AMD64 box and suddenly started seeing a jerkiness to the
> system.  Others reported it and I figured that it was a short-lived
> kernel issue.  I recently did a new CVS update and built a fresh
> kernel.  The problem is still there.
> 
> (I only recall AMD64 people remarking about it, so I assume that
> it's AMD64-specific.)

Hmm, since I see this on an i386 box with an MPACPI kernel, too, I currently
suspect the ACPI interpreter taking too much time.  Investigating this
further is somewhere on my todo list, but didn't make it to the top yet ;-(.

Ciao,
Wolfgang
--

-- 
ws <at> TooLs.DE                            Wolfgang Solfrank, TooLs GmbH

Richard Rauch | 4 Mar 20:08 2005

Re: Jerky system behavior.

On Fri, Mar 04, 2005 at 05:46:17PM +0100, Wolfgang Solfrank wrote:
> Hi,
> 
> >About August or so of last year, I upgraded from one -current to
> >another on my AMD64 box and suddenly started seeing a jerkiness to the
> >system.  Others reported it and I figured that it was a short-lived
> >kernel issue.  I recently did a new CVS update and built a fresh
> >kernel.  The problem is still there.
> >
> >(I only recall AMD64 people remarking about it, so I assume that
> >it's AMD64-specific.)
> 
> Hmm, since I see this on an i386 box with an MPACPI kernel, too, I currently
> suspect the ACPI interpreter taking too much time.  Investigating this
> further is somewhere on my todo list, but didn't make it to the top yet ;-(.

I see this even without MPACPI enabled:

$ grep -i mpacpi socrates
#options        MPACPI
#options        MPACPI_SCANPCI          # MPBIOS configures PCI roots

(Yes.  Flying in the face of convention and stirring up great tides
of controversy, no doubt, I use lower-cased kernel config files.
(^&)

--

-- 
  "I probably don't know what I'm talking about."  http://www.olib.org/~rkr/

(Continue reading)

Samuel Kesterson | 7 Mar 05:37 2005

Re: amd64 and PCMCIA

Thus spake Sami Kantoluoto (sami.kantoluoto <at> embedtronics.fi) [2005-03-02 20:13] :

> On Tue, Mar 01, 2005 at 08:25:20PM +0100, Frank van der Linden wrote:
> > That's not hard to do.. So far it hasn't been added because noone has
> > tested it yet, but it's likely to "just work".
> >
> > What you need to do is to take the cardbus+pcmcia stuff out of
> > sys/arch/i386/conf/files.i386, and put it in sys/arch/amd64/conf/files.amd64,
> > at the same spot. Then reconfigure the kernel.
> > 
> > If this works for you, let me know, and I'll enable it by default.
> 
> This alone was not just enough but I copied
> sys/arch/i386/i386/rbus_machdep.c and sys/arch/i386/include/rbus_machdep.h
> to appropriate places and with these I got it compiled (NOTE: I don't have
> any understanding about those files). But:
> 

	-{Snip}-

	Ok, here's where I am. 

	I copied the section for PCMCIA out of files.i386 into
files.amd64 and copied rbus_machdep.c and rbus_machdep.h into the amd64
tree. I had to comment the include for opt_pcibios.h out of
rbus_machdep.c, and then I was able to config/make and get up on the new
kernel. I've posted the dmesg at
http://www.k-labs.com/~samuelk/dmesg.amd64_PCMCIA and my kernel config
at http://www.k-labs.com/~samuelk/GENERIC_PCMCIA

(Continue reading)

Samuel Kesterson | 7 Mar 14:48 2005

Re: amd64 and PCMCIA

Thus spake Samuel Kesterson (netbsd <at> k-labs.com) [2005-03-06 22:38] :

> 	Now the system sees the cardslot, but there's still a problem.
> I'm trying to use an old prism2 card (Compaq HNW-100). When the card's
> inserted, the kernel logs this:
> 
> cbb0: wait took 0.015387s
> pcmcia0: card appears to have bogus CIS
> 
> 	I found this in the mail archive and am rebuilding now to see if
> it gets me up and running.
> http://mail-index.netbsd.org/netbsd-bugs/2005/01/20/0013.html
> 
> 	Do the resident experts have any other information on this that
> I should take into consideration?
> 

	No, this patch didn't change anything. I still get the bogus CIS
message. 

Samuel Kesterson | 8 Mar 06:08 2005

ACPI on amd64 laptops

	Forgive me if I'm rehashing old conversations, but I read back
throught the 2004 archives and only saw this touched on.

	What's the status of ACPI on the amd64 port? I don't know if
it's something with this laptop or acpi in general, but I enabled all
the acpi devices (but, lid, acad, etc), and while it looks like they're
found, they all say "not configured". See dmesg
http://www.k-labs.com/~samuelk/dmesg.amd64_AKITA and kernel config at
http://www.k-labs.com/~samuelk/AKITA

	The power management stuff would be a nice plus, but I
mostly want to get this thing cranking on all 1800 MHz. I tried to get
it wedged in myself, but I'm just not famaliar enough with kernel land 
to be trying things like that.

	Can anyone help me get my CPU running full speed (a writable
sysctl knob would be *awesome*)? I'm considering installing Linux in
another partition on this machine so I can get the full use out of it
(FreeBSD *still* doesn't boot on this laptop, and Open's amd64 shares
problems with the Net port), but the thought of Linux on this just makes
me feel dirty ...

TIA


Gmane