2002-Feb-14

Well, its been around 12 hours since the CVS was finished, but connection is still in FIN_WAIT_2. So I killed it:

cvs [update aborted]: received termination signal

real    769m45.687s
user    0m4.172s
sys     0m7.646

I'll just assume it is all good.

Now to build new kernel:

$ cd usr.sbin/config/
$ make
$ su root -c 'make install'

(I don't think the above was needed; it hadn't and didn't change. It is suggested for -current though.)

$ cd /usr/src/sys/arch/i386/conf/
$ config JCR-1.5-20020213
../../../../arch/i386/conf/files.i386:92: cannot open ../../../../dev/i2o/files.i2o for reading: No such file or directory

../../../../dev/pci/files.pci:41: attaching undefined device `iop'
../../../../arch/i386/conf/files.i386: device `iop' used but not defined
*** Stop.

I asked about this on the mailing list; it was suggested that I try cvs update again. I also tried to get new cvs package, but I see I have latest.

I tried to cvs again from a different cvs server, but after it was done and sitting in FIN_WAIT_2, I saw it had used the same server! I had set the CVSROOT and CVS_IGNORE_REMOTE_ROOT, but cvs(1) seemed to ignore it. So I forced it to use the different CVSROOT with "-d $CVSROOT". After about a minute it spewed:

Protocol error: 'Directory' missingE Protocol error: 'Directory' missingE Protoc...
ol error: directory '/cvsroot/basesrc' not within root '/cvs/cvsroot

Now reading through cvs(1) manual and at NetBSD site, I see I forgot "-d" and "-P" options to get new directories and remove empty directories. So trying again!

One hour and 19 minutes later is finished successfully. (And cvs closed successfully.)

So back to building kernel:

$ cd sys/arch/i386/conf/
$ config JCR-1.5-20020213
$ cd ../compile/JCR-1.5-20020213/
$ make depend
$ make # took about 13 minutes

Then as root, I backed up old kernel and copied netbsd to /netbsd.

So after reboot, to check SanDisk (see yesterday 13/Feb/2002). Now the logs had extra details:

scsibus0 at umass0: 2 targets, 1 luns per target
scsibus0: waiting 2 seconds for devices to settle...
sd0 at scsibus0 target 1 lun 0:  SCSI2 0/direct removable
sd0: drive offline

So (as root) I tried to mount it:

$ mount -t msdos /dev/sd0d /mnt
mount_msdos: /dev/sd0d on /mnt: Invalid argument

And the kernel reported:

Feb 14 13:59:04 rainier /netbsd: sd0: mode sense (4) returned nonsense; using
fictitious geometry
Feb 14 13:59:04 rainier /netbsd: sd0: no disk label

Oh, I should have used sd0e instead! The disklabel for sd0 has:

e:    62560       32      MSDOS                        # (Cyl.    0*- 30*)

The mounted device has jpegs under /mnt/dcim/100hp315/. mount(8) showed:

/dev/sd0e        62041     3034    59007     4%    /mnt

(The images were taken on a Hewlett Packard PhotoSmart 315.) I was able to view and deleted the images. When done a umounted the device.

So now to build new userland:

$ cd /usr/src
$ make obj

Then as root:

$ time make UPDATE=1 build

(The build target does the install too. UPDATE means to only install updated software. (The last time I updated all source was 07/Apr/2001.) This build lasted from 15:05 to 17:34. time reported:

real    149m30.045s
user    68m58.805s
sys     13m12.217s