mem.c: mem_range_softc undefined reference

[Conrad J. Sabatier]

On 06-Aug-2004 Sean McNeil wrote:
> On Thu, 2004-08-05 at 20:32, Conrad J. Sabatier wrote:
>> On 06-Aug-2004 Sean McNeil wrote:
>> > On Thu, 2004-08-05 at 20:02, Conrad J. Sabatier wrote:
>> >> On 04-Aug-2004 Ceri Davies wrote:
>> >> > On Wed, Aug 04, 2004 at 01:57:36PM -0700, Sean McNeil wrote:
>> >> >> Anyone else getting this?  I've compared my kernel config with
>> >> >> GENERIC
>> >> >> and I have the extra devices that were added recently.
>> >> > 
>> >> > markm just committed a fix for this.
>> >> 
>> >> I don't think so.  Where do you see this?  I'm still getting:
>> >> 
>> >> linking kernel
>> >> amd64_mem.o(.text+0xc1c): In function `amd64_mem_drvinit':
>> >> : undefined reference to `mem_range_softc'
>> >> mem.o(.text+0x24f): In function `memioctl':
>> >> : undefined reference to `mem_range_softc'
>> >> mem.o(.text+0x25a): In function `memioctl':
>> >> : undefined reference to `mem_range_softc'
>> >> mem.o(.text+0x364): In function `dev_mem_md_init':
>> >> : undefined reference to `mem_range_softc'
>> >> mem.o(.text+0x378): In function `dev_mem_md_init':
>> >> : undefined reference to `mem_range_softc'
>> >> memutil.o(.text+0x3): more undefined references to
>> >> `mem_range_softc'
>> >> follow
>> >> *** Error code 1
>> >> 
>> >> Stop in /usr/obj/usr/src/sys/CUSTOM.
>> > 
>> > This is because your CUSTOM does not have options SMP.  Try adding
>> > it. 
>> > UP is still broken.  I think a fix for i386 has already been made,
>> > but amd64 didn't get fixed.
>> 
>> Yes, I did try running an SMP kernel earlier today, but it caused
>> some problems with my NIC (fxp).  Had to remove DEVICE_POLLING to
>> build for SMP, and for some reason, I started seeing device timeouts
>> on fxp0.  Strange.
>> 
>> Why is DEVICE_POLLING incompatible with SMP anyway?
>> 
>> As if all this annoyance weren't enough, I'm *still* getting the pcm
>> timeouts.  Beginning to wonder if that will ever be fixed.
>> 
>> If it weren't for the fact that I really need to be running current
>> to benefit from the latest amd64 developments, honestly, I think I
>> would just switch back to RELENG_5_2, or even 4.x.  And if this
>> sound problem doesn't get fixed, I'm not sure what I'll do. 
>> Switch to Linux?  (yuck)
>> 
>> Yeah, I'm a little disgusted at the moment.
> 
> I sympathize.  I have the sound issue, kernel panic most of the time
> on boot because of ATAPI_RESET, hard lockups most likely due to ATA
> code, plus I can't burn a CD without a kernel panic.  To top it all
> off, there isn't a Java solution available for FreeBSD/amd64 to be
> found.

I suppose I should consider myself relatively fortunate.  I've never
had the kernel panics on boot due to ATAPI_RESET (knock on wood), hard
lockups and instant reboots have subsided since preemption was turned
off, and my CD burner is functioning very well.

Yeah, it's a bummer having no easy Java solution, and no Linux
emulation (at least not officially).
 
> The sound issue appears to be related to PREEMPTION and will one day
> get debugged and turned back on.  There are people working hard on
> it. 

I was hoping that was the case, but even without PREEMPTION, the pcm
device times out and just plain dies.  No more sound without a reboot. 
:-(

> My ATA issues will hopefully be fixed when Soren commits his changes
> to the drivers.

I see he's done some interesting commits recently.  Now, when I get the
ad0 timeouts, the channel is reinited automatically.  Nice.

> I have no experience with the Linux amd64 environment, so I cannot
> comment on that.  My reaction, though, mirrors yours (yuck).

A guy I communicate a lot with on my ISP's private newsgroups (Cox) has
become a total Gentoo freak, after running Mandrake for the longest. 
We trade war stories and whatnot, but I'm not hearing anything that
makes me want to trash my FreeBSD setup on this amd64 box and checkout
Gentoo.  At least, not yet.  :-)
 
> I just hope all these issues are resolved before 5.3 is blessed.

Yes, let us hope.  :-)

-- 
Conrad J. Sabatier <conrads at cox.net> -- "In Unix veritas"