Abstraction leakage burning test DVD+R
--VbJkn9YxBvnuCH5J
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable
Hello,
After my dump/restore test failed back in May of 2010[1], I finally got=20
around to burning a test DVD after installing about 2.5 inches of=20
semi-rigid foam under the machine to dampen any local vibration.
In designing the test, I was under the mistaken impression that buffer=20
under-runs would reduce the amount of space for writing. Apparently,=20
DVD+R media supports loss-less linking[2].
Test Procedure (using ATAPI interface, rather than the newer ATAPI/CAM=20
interface). The man pages do not imply any major differences for this=20
test. The drive in question is a LG drive with the SuperMulti logo.
Test procedure:
1.# mkfifo auxout aux45G
2. In other terminals:
# cat auxout | md5
# dd if=3Daux45G count=3D2197266 bs=3D2048 | md5
3. Write test disk:
# dd if=3D/dev/urandom bs=3D2048 count=3D2295104 | \
tee auxout aux45G | dd of=3D/dev/acd0 bs=3D2048
Error:
dd:/dev/acd0: Input/output error
679+0 records in
678+0 records out
1388544 bytes transferred in 80.97 seconds=20
(step 4 not completed due to premature burn failure after 684=20
blocks, 80 seconds (from 'dd if=3Daux45G'...) (15743 bytes/sec))
Drive spun up at least twice during this time.
4. Read test (in another term 'cat aux45G | md5')
# dd if=3D/dev/acd0 bs=3D2048 count=3D2295104 | tee aux45G | md5
Questions:
Currently, the drive is locked. Before forcefully ejecting it and=20
risking an new disk, I want to know what may have gone wrong. Was=20
I expecting too much of FreeBSD on old hardware?=20
The hardware is a Pentium-II desktop machine with 256MB of RAM, with a=20
Promise ATA100 controller card. An extra-long 80-wire cable is in use=20
(to ad4).
acd0: DVDR <HL-DT-STDVD-RAM GSA-H55N/1.03> at ata-1-master UDMA33
ad4: 78167MB <Maxtor 6Y080L0 YAR41WB0> at ata-2-master UDMA100
dd if=3D/dev/urandom of=3D/home/test
Wrote at 4485315 bytes/second: faster than I was expecting (based on=20
linux's slow urandom function), but still slow enough to cause buffer=20
under-runs. So the second question is: do I have to do anything special=20
to enable buffer underrun protection? (with DVD+R's lossless linking=20
feature, no data should be lost.)
How important is running ATAPI/CAM? Obviously DVD burning was happening=20
before the release of version 8, as far as I know.
If running out of data is supposed to be abstracted away, did I find a=20
bug that only shows up on old hardware or heavy load? (Or possibly drive=20
firmware bug?)
Regards,
James Phillips
# uname -a
FreeBSD dusty.inet 8.0-RELEASE FreeBSD 8.0-RELEASE #0: Sat Nov 21=20
15:48:17 UTC 2009 =20
root@almeida.cse.buffalo.edu:/usr/obj/usr/src/sys/GENERIC i386
[1] dump/restore (to DVD+R) test failure
http://docs.freebsd.org/cgi/getmsg.cgi?fetch=3D290415+0+archive/2010/freebs=
d-questions/20100530.freebsd-questions
[2] Why DVD+R(W) is superior to DVD-R(W)
http://www.myce.com/article/Why-DVDRW-is-superior-to-DVD-RW-203/
PS: is signing messages on the mailing list a faux-pas?
--=20
OpenPGP Public Key: http://phillipsjk.ca/signature0611.txt
--VbJkn9YxBvnuCH5J
Content-Type: application/pgp-signature; name="signature.asc"
Content-Description: Digital signature
Content-Disposition: inline
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
iD8DBQFOq90FPPYMXhz9onsRAoymAJ4mf2X8f1Z88150dkoNuU1w5RpfkgCdHiih
FoG2UgpsoF/KZvZACxBubR0=
=vmrS
-----END PGP SIGNATURE-----
--VbJkn9YxBvnuCH5J--