RPI3 swap experiments

[bob prohaska]

On Sun, Jul 08, 2018 at 09:32:32AM +1000, Trev wrote:
> Trev wrote on 04/07/2018 07:09:
> > Conclusion: The only way to buildworld on the RPi3B+ is to use an 
> > external USB mechanical hard disk (my hybrid Seagate 500G 2.5" drive 
> > required a powered USB hub). No issues on a RPi2B v1.1 with swap and 
> > /usr on a USB memory key.
> 
> One last experiment occurred to me: I placed a SanDisk 8GB Ultra micro 
> SDCard in a microSD card reader (this one: 
> https://www.amazon.co.uk/MICRO-SD-MICRO-SDHC-MEMORY-READER/dp/B002WKD0CQ 
> ), created a 2G swap partition and ran make -j4 buildworld SUCCESSFULLY! 
> Twice.
> 
It seems something is right on the edge. R335835 updated itself to
r336019 using 3 GB of USB swap on da0, along with /var/, /tmp/ and /usr/.
No powered hub was used. No errors of any kind. When I tried to mix swap 
from microSD OOMA killed buildworld. OOMA also killed buildworld when all 
swap was on microSD, which is a step backwards; that arrangement used to work. 
I haven't yet repeated those tests on  r336019.

For the last 21 hours the machine has been running stress2's all.sh script.
Top reports CPU:  1.1% user,  0.0% nice, 98.9% system,  0.0% interrupt,  0.0% idle
but the machine seems quite responsive to ssh inputs. Not sure I believe it. 

> uname -aKUp
> FreeBSD rpi3.sentry.org 12.0-CURRENT FreeBSD 12.0-CURRENT #0 r335854: 
> Mon Jul  2 16:03:04 AEST 2018 
> root at rpi3.sentry.org:/usr/obj/usr/src/arm64.aarch64/sys/RPI3-12.0  arm64 
> aarch64 1200069 1200069
> 
> 
> gpart show -l da0
> =>      40  15523760  da0  GPT  (7.4G)
>          40   4194304    1  usdswap  (2.0G)
>     4194344  11329456       - free -  (5.4G)
> 
> 
> # Custom /etc/fstab for FreeBSD embedded images
> /dev/ufs/rootfs   /       ufs     rw,noatime      1       1
> /dev/msdosfs/MSDOSBOOT /boot/msdos msdosfs rw,noatime 0 0
> md1 /tmp mfs rw,noatime,-s100m 0 0
> md2 /var/log mfs rw,noatime,-s15m 0 0
> md3 /var/tmp mfs rw,noatime,-s15m 0 0
> /dev/da0p1      none    swap    sw      0       0
> 
> 
> da0 at umass-sim0 bus 0 scbus0 target 0 lun 0
> da0: <SD/MMC Card  Reader 1.00> Removable Direct Access SCSI device
> da0: Serial Number 058F63356432
> da0: 40.000MB/s transfers
> da0: 7580MB (15523840 512 byte sectors)
> da0: quirks=0x2<NO_6_BYTE>
> 
> 
> The only unusual, but apparently harmless, log messages were:
> 
> Jul  7 15:39:01 rpi3 kernel: swap_pager: indefinite wait buffer: bufobj: 
> 0, blkno: 143089, size: 327
> 68
> Jul  7 17:46:35 rpi3 kernel: swap_pager: indefinite wait buffer: bufobj: 
> 0, blkno: 14712, size: 8192
> Jul  7 17:47:21 rpi3 kernel: swap_pager: indefinite wait buffer: bufobj: 
> 0, blkno: 17239, size: 2457
> 6
> Jul  7 17:47:21 rpi3 kernel: swap_pager: indefinite wait buffer: bufobj: 
> 0, blkno: 20737, size: 8192
> Jul  7 17:47:21 rpi3 kernel: swap_pager: indefinite wait buffer: bufobj: 
> 0, blkno: 511807, size: 409
> 6
> Jul  7 18:50:34 rpi3 kernel: swap_pager: indefinite wait buffer: bufobj: 
> 0, blkno: 988, size: 4096
> Jul  7 18:50:34 rpi3 kernel: swap_pager: indefinite wait buffer: bufobj: 
> 0, blkno: 318727, size: 655
> 36
> Jul  7 18:55:02 rpi3 kernel: swap_pager: indefinite wait buffer: bufobj: 
> 0, blkno: 354041, size: 286
> 72
> Jul  7 21:45:10 rpi3 kernel: swap_pager: indefinite wait buffer: bufobj: 
> 0, blkno: 164496, size: 491
> 52
> Jul  7 23:04:38 rpi3 kernel: swap_pager: indefinite wait buffer: bufobj: 
> 0, blkno: 372151, size: 655
> 36

Those messages seem an ill portent, but they didn't guarantee failure for me,
either.

thanks for reading,

bob prohaska