HEADS UP! Debugging, SMP changed in RELENG_6

[Michael C. Shultz]

On Saturday 17 September 2005 21:00, Scott Long wrote:
> All,
>
> I've turned off kernel deubgging (INVARIANTS, WITNESS, KDB) and malloc
> debugging (the default is now 'aj) in the RELENG_6 branch in preparation
> for 6.0-BETA5.  I've also turned off SMP in the i386 and amd64 GENERIC
> kernels and have added and SMP kernel config for those architectures.
> All of these changes are nearly identical to what is in RELENG_5.  All
> of this was done to facilitate performance measurement of the upcoming
> release and to recognize that our buglist is shrinking considerably and
> thus the release will be coming soon.
>
> ### IMPORTANT ###
>
> If you are using a stock GENERIC kernel on amd64 or i386 and expect
> multiple CPUs to be enabled, you must either switch to the SMP kernel
> config, or add the 'options SMP' line back to the GENERIC config.
> Multiple CPUs, including mutli-core and Hyperthreading, will only be
> enabled when this option is included.  Hyperthreading also requires
> that you add the line 'machdep.hyperthreading_allowed=1' to
> /boot/loader.conf.
>
> Given the growth of multicore offerings from AMD and Intel, the default
> on SMP may change in future releases as SMP systems become ubiquitous.
> Appropriate warning will be given on the mailing lists if this happens.
>
> Scott


I've just spent the past few days trying to get 6.0 to work with my SATA
drives with little results.  Here is a DMESG when I booit with 5.4:

ad0: 6149MB <WDC AC26400B/32.02S32> [13328/15/63] at ata0-master UDMA33
acd0: CDRW <TEAC DVD+RW DV-W58E/D.0J> at ata1-master UDMA33
ad4: 76319MB <WDC WD800JD-00HKA0/13.03G13> [155061/16/63] at ata2-master 
SATA150
ad6: 114473MB <WDC WD1200JD-00HBB0/08.02D08> [232581/16/63] at ata3-master 
SATA150

The drives work fine in 5.4, but with 6.0 ad4 errors out, you can see it with 
atacontrol but it can't be reinitialized and there is only /dev/ad4 but none 
of the slices. ad6 works ok during reads but anything more than very light
writing and the system reboots.  Given these drives work with 5.4 it seems
a little more work needs to go into 6.0 before doing new releases I would
hope.

-Mike