add accumulative mode. "tprof top -a"

(ryo)

Revert com.c:r1.381 because i386/qemu cannot boot.  Pointed out by gson@n.o and martin@n.o.

(knakahara)

+AEAD, HKDF, NMR, SIV

(jschauma)

Fix hang up writing /dev/console rarely in specific environments.

Some BMC seems to require these syncronous operations.  If not,
it does not send transmit completion interrupts for some reason.

(knakahara)

Workaround for some hypervisor environments.

The environments cannot stop e1000 interrupt immediately.

(knakahara)

Fix: sp->lastused should be updated by time_uptime, and refactor a little.

(knakahara)

Fix: update lastused of ipsecif(4) IPv6 out SP.

(knakahara)

doc: Add note about tprof improbements.

(msaitoh)

Fix typo.

(knakahara)

Add AMD family 19h (zen3 and zen4) support to tprof.

(msaitoh)

Use lowercase consistently for hexadecimal numbers. No functional change.

(msaitoh)

3RDPARTY: note newer versions of acpica and gmake are out

(gutteridge)

make: clean up comments

(rillig)

Refactor ip_encap.[ch]

    - remove encap_attach() which is no longer used
    - remove USE_RADIX code in ip_encap.c, which is used for
      encap_attach() only
    - remove mask members in encaptab

(knakahara)

gif(4), ipsec(4) and l2tp(4) use encap_attach_addr().

(knakahara)

Implement encap_attach_addr() which is used by IP-encaped tunnels.

The tunnels attached by encap_attach() can process receiving packets
fastly as the softc is searched by radix-tree.  However, the tunnels
cannot use priority function which decides tunnel's softc by not only
source and destination but also other informations.
On the other hand, the tunnels attached by encap_attach_func() can
use priority function. However, the tunnels can be slow receiving
processing as the softc is searched by linear search (and uses each
priority function).

encap_attach_addr() can be used for tunnels which is fixed tunnel
source address and tunnel destination address.  The tunnels attached
by encap_attach_addr() is searched by thmap(9), so the receiving processing
can be fast.  Moreover, the tunnels can use priority function.

(knakahara)

refactor: use typedef for ip_encap priority function

(knakahara)

KNF. No functional change.

(msaitoh)

exsystem.c: revert ACPI warning about sleep values >10ms

acpica 20220331 introduced the following change:

For the ASL Sleep() operator, issue a warning if the sleep value is
greater than 10 Milliseconds. Quick boottime is important, so warn about
sleeps greater than 10 ms. Distribution Linux kernels reach initrd in 350
ms, so excessive delays should be called out. 10 ms is chosen randomly,
but three of such delays would already make up ten percent of the
boottime.

This can generate many console messages like:
[    26.055036] ACPI Warning: Firmware issue: Excessive sleep time (0x0000000000000064 ms > 10 ms) in ACPI Control Method (20220331/exsystem-239)
(60 of them in the latest boot on one older Ivy Bridge laptop.)

This has been reverted upstream and is not present in the next release.
https://github.com/acpica/acpica/commit/df896f452f945d498736c5775d4ee6629df31786
(Also previously reverted in FreeBSD and Linux.)

(gutteridge)

Tickets #1551 and #1552

(martin)

Pull up following revision(s) (requested by tsutsui in ticket #1552):

sys/arch/hp300/dev/rd.c: revision 1.120
sys/arch/hp300/dev/rd.c: revision 1.121
sys/arch/hp300/dev/rd.c: revision 1.122
sys/arch/hp300/dev/rd.c: revision 1.123
sys/arch/hp300/dev/rd.c: revision 1.124
sys/arch/hp300/dev/rdreg.h: revision 1.19
sys/arch/hp300/dev/rd.c: revision 1.111
sys/arch/hp300/dev/rd.c: revision 1.112
sys/arch/hp300/dev/rd.c: revision 1.113
sys/arch/hp300/stand/common/rd.c: revision 1.12
sys/arch/hp300/dev/rd.c: revision 1.114
sys/arch/hp300/stand/common/rd.c: revision 1.13
sys/arch/hp300/dev/rd.c: revision 1.115
sys/arch/hp300/stand/common/rd.c: revision 1.14
sys/arch/hp300/dev/rd.c: revision 1.116
sys/arch/hp300/stand/common/rd.c: revision 1.15
sys/arch/hp300/dev/rd.c: revision 1.117
sys/arch/hp300/dev/rd.c: revision 1.118
sys/arch/hp300/dev/rd.c: revision 1.119
sys/arch/hp300/dev/rdvar.h: revision 1.27
sys/arch/hp300/dev/rdreg.h: revision 1.20

Sync a list of HP-IB disk IDs with kernel's one.
Now bootloader can boot from 2202A, 7908A, 7911A, and 7941A.
Tested on 425t and HPDisk.

Make local variables static and read only ones const.

Use common macro for numbers of cylinders and blocks for HP-IB disks.
No binary changes.

Fix "rd(4) at punits not configured on HPDisk are misprobed" problem.
- check not only stat value returned by C_QSTAT command against
  each punit but also desc.d_name returned by C_DESC command
  because it looks HPDisk responds to commands against punits supported
  but not-configured punits at 1-3
- prepare rdreset_unit() function to manage reset punit ops during
  probe without valid softc

The stragety is taken from OpenBSD with several modification.
Tested on 425t with real 9122D and HPDisk emulating two 7937H drives.
Should be pulled up to netbsd-9.

Remove now unused buffer members from softc prepared only for rdreset().

Use proper integer types for HP-IB xfer parameters.
No visible regression on 425t and HPDisk.

Sync with rdreg.h integer type changes.

Remove unnecessary declarations.

Check bounds of RAW_PART by bounds_check_with_mediasize() as other drivers.
With this check, dd(1) without a count value against a raw partition
is terminated properly at the end of media.  Tested on 425t and HPDisk.
Should be pulled up to netbsd-9.

Check bounds of each partition by MI bounds_check_with_label(9).
The previous implementation was committed in rev 1.26 about 26 years ago
and it looks there is no quirk to use the old MD one.

Pass a correct dev_t arg to disk_ioctl().
Probably harmless.

Remove a duplicated static function declaration.

Set disk_geom parameters required by DIOCGPARTINFO in disk_ioctl().
With this change raid(4) no longer complains before mountroot:
RAIDframe: can't get disk size for dev rd0 (22)
Should be pulled up to netbsd-9.

Use proper RAW_PART macro rather than a magic number.

Remove more duplicated static function declarations.

Misc KNF and cosmetics.

Add empty DIOCCACHESYNC ioctl(2) for rd(4).
Without this raid(4) always complains on various raidctl(8) ops:
raid0: cache flush[0] to component 0 failed (22)
RAID1 configured by raid(4) just works HP-IB disks on hp300.

Explicitly include <sys/kernel.h> for hz(9) as man pages says.
XXX: it looks sys/param.h r1.615 and later also has extern int hz

(martin)

Pull up following revision(s) (requested by tsutsui in ticket #1551):

sys/arch/hp300/dev/topcat.c: revision 1.6

Fix silent bus error panic on 98543A topcat framebuffer on HP320 and HP360.

It looks DELAY(100)s before checking cmap_busy in old pre-wscons grf_tc.c
http://cvsweb.netbsd.org/bsdweb.cgi/src/sys/arch/hp300/dev/Attic/grf_tc.c?rev=1.42
are actually necessary on 68020 (HP320) and 68030 (HP360) machines,
while it works without them on 68040 (HP380).

Should be pulled up to netbsd-9.

(martin)

+ROP - return-oriented programming

(jschauma)

unix(4): Fix/add markup

Untabify examples, .Bd -literal -offset indent doesn't seem to mesh
with tabs nicely in the PostScript output.  The indent is probably not
a multiple a of literal tab width.  This causes a small vertical
misalignment for tab-indented lines, but what's worse - it can cause
an extra tab in some cases (and did here for the unpcbid definition).

(uwe)

Undo upstream commit b4a105d77232a87304b7b621e2f99e699a8eebd3, it breaks
colours on big endian machines. Also reported upstream.

(martin)

convert more seriously early output to ofprint
regular console output needs more of the kernel in working order now, and this
stuff happ0ens long before the banner

(macallan)

editrc(5): Markup fixes

(uwe)

getbsize(3): Use .Ev for BLOCKSIZE

(uwe)

make: clean up comments

No binary change.

(rillig)

make: inline LazyBuf_AddBytesBetween

No binary change.

(rillig)

make: inline local macro in GNodeFlags_ToString

No binary change.

(rillig)

Update check files due to reversion of a libcurses change.

(blymn)

Revert previous change, it is not correct.

(blymn)

Do not bother to set PSL_SE in l_md.md_flags - it is not checked anywhere.

(martin)

Do not copy l_md to the new lwp in cpu_lwp_fork - as discussed on
tech-kern we do not want the child to inherit any of the flags in there.

(martin)

If no more softints are pending on this cpu, clear ci_want_resched
(instead of just assingning ci_data.cpu_softints to it - the bitsets
are not the same).
Discussed on tech-kern "ci_want_resched bits vs. MD ci_data.cpu_softints bits".

(martin)

+CSPM - cloud security posture management

(jschauma)

Use the SZREG symbol instead of __SIZEOF_POINTER__ as its shorter.  Assert
that they're the same just in case.

Sprinkle some KNF whitespace while I'm here.

(skrll)

RISC-V is RELA

(skrll)

Simplify. Same code before and after.

(skrll)

Fix a comment.

(skrll)

fix build error with llvm

(ryo)

getnetpath(3): Sprinkle some markup

(uwe)

login_cap(3): Markup fixes

(uwe)

On VAX, compile gdtoa/misc.c at -O0 to avoid incorrect machine code
resulting in some doubles printing as garbage.

(jakllsch)

fcntl(2): Fix wrong markup in a couple of places

(uwe)

Consistently use the 'mv' instrucation (which is itself an alias) instead
of the (old) alternative alias 'move'.

(skrll)

re_format(7): Add subsection headings for ERE and BRE

The first paragraph could use some rewording.  While BRE may be
obsolete, it's still the default for regcomp(3) and the default for
grep(1), sed(1), etc.

(uwe)

Restore t5 and t6 from the correct locations in exception_kernexit.

From Simon.

(skrll)

ASSERT that md_astpending it zero for the new lwp.

(skrll)

Bump date for the fixed RETURN VALUES in previous

(uwe)

Tweak markup in _lwp_*(2) manual pages

Use .Rv where possible.  Some manual pages claimed that the error "is
returned", but RTFS and some quick testing indicate that this is wrong.

The commit message from 2003 says that:

  Note our current implementation mis-matches [man pages] slightly
  (error codes are stuffed into errno, where they should simply be
  returned by these calls).  This will be addressed shortly.

That hasn't happened in the 20 years, so we might as well make the man
pages reflect the reality.

(uwe)

lib: Mark up error names in man pages with .Er

(uwe)

flock(2): s/Eq/Er/ typo in the ERRORS section

(uwe)

pw_init(3): Use .Ev for EDITOR

(uwe)

Make power save mode configurable by ifconfig, now defaults to non-powersave
like other interfaces.
Remove cargo-culted check for manual roaming.

(mlelstv)

allow testing of files other than those in /etc

To test e.g., the file "/some/where/protocols" instead of "/etc/protocols",
set TEST_FILE=/some/where/protocols in your environment.

Note: this now compares the contents of the file you gave versus what
getprotoent(3)/getservent(3) uses (which still is /etc/protocols via
h_protoent.c / /etc/services or /var/db/services.cdb via h_servent.c).

When you have expected changes in the services or protocols file that
you're generating, this necessarily produces a difference.  To really
allow testing the file versus what the library function returns, you'd
have to install the file on the system running the test, but at least
with this change you can now generate the file and verify that it didn't
caused unexpected differences.

(jschauma)

whitespace

(jschauma)

Sections 2 and 3 have RETURN VALUES, not DIAGNOSTICS

(uwe)

getservent(3): Section 3 has RETURN VALUES, not DIAGNOSTICS

While here, don't claim that null pointer is zero.

We should probably just pick the wording from POSIX, but that yak is a
bit too big for me to shave at the moment.

(uwe)

reference /var/db/services.cdb / services_mkdb(8)

(jschauma)

move ARMv7 PMC register definitions to armreg.h from tprof_armv7.c

(ryo)

Explicitly include <sys/kernel.h> for hz(9) as man pages says.

XXX: it looks sys/param.h r1.615 and later also has extern int hz

(tsutsui)

Fix bug in protocol parser that often caused fatal 'checksum error'.
Defer power save setting to interface start.
More verbose on errors.
Allow build without FDT.

(mlelstv)

Trailing whitespace

(skrll)

Fix some comments

(skrll)

improve a comment

(skrll)

KNF a comment

(skrll)

leading whitespace... oops

(skrll)

Trailing whitespace

(skrll)

Trailing whitespace

(skrll)

Trailing whitespace

(skrll)

Whitespace. NFC.

(skrll)

Correct some pre-existing relocations and add some new ones.

(skrll)

Add empty DIOCCACHESYNC ioctl(2) for rd(4).

Without this raid(4) always complains on various raidctl(8) ops:
> raid0: cache flush[0] to component 0 failed (22)

RAID1 configured by raid(4) just works HP-IB disks on hp300.

(tsutsui)

Split the BMP280 / BME280 driver into common code and create I2C and
SPI attachments.

(brad)

PR/57095: Yilei Yang: Change readline's replace_history_entry to not make a copy
of the string to replace since H_REPLACE already makes a copy (fixes memory leak)

(christos)

Correct lost change in the merge (Anthony Mallet)

(christos)

fix clang build

(christos)

Misc KNF and cosmetics.

(tsutsui)

fix sign.

(christos)

Add a sanity check for the ip header length (from FreeBSD)

(christos)

reinstate $NetBSD$

Noted by msaitoh@

(jschauma)

tests: fix Makefile and lists for MKRUMP=no

Pointed out by Michael Scholz, thanks.

(ozaki-r)

fix build error of printf format

(ryo)

Use a better choice for the print format string.

(brad)

tprof module revamped. Welcome to 9.99.108

(ryo)

Correct the humidity conversion for the BME280

(brad)

add "top" subcommand to tprof(8)

(ryo)

split ksyms stuff into ksyms.[ch]

(ryo)

Improve tprof(8)

- Added "tprof count" subcommand to perform counts only.
- Event options (u,k) are now optional. The default value is both userland and kernel. (:uk)
- Event counters can be displayed with SIGINFO during `tprof monitor' or `tprof count'.

(ryo)

Improve tprof(4)

- Multiple events can now be handled simultaneously.
- Counters should be configured with TPROF_IOC_CONFIGURE_EVENT in advance,
  instead of being configured at TPROF_IOC_START.
- The configured counters can be started and stopped repeatedly by
  PROF_IOC_START/TPROF_IOC_STOP.
- The value of the performance counter can be obtained at any timing as a 64bit
  value with TPROF_IOC_GETCOUNTS.
- Backend common parts are handled in tprof.c as much as possible, and functions
  on the tprof_backend side have been reimplemented to be more primitive.
- The reset value of counter overflows for profiling can now be adjusted.
  It is calculated by default from the CPU clock (speed of cycle counter) and
  TPROF_HZ, but for some events the value may be too large to be sufficient for
  profiling. The event counter can be specified as a ratio to the default or as
  an absolute value when configuring the event counter.
- Due to overall changes, API and ABI have been changed. TPROF_VERSION and
  TPROF_BACKEND_VERSION were updated.

(ryo)

tprof_armv7 initializes on each CPUs, like tprof_armv8.

(ryo)

PMCR.E should not be disabled from tprof.

PMCR.E controls not only performance event counters but also the cycle
counter operation, and the cycle counter may be used for cpu_counter.
Similarly, the 31st bit in PMINTENCLR and PMCNTENCLR controls the cycle
counter, not performance event counters, and should not be modified.

(ryo)

don't call kpreempt_{disable,enable}() from an interrupt handler.

Fixed a problem in which the system would freeze if a high load (e.g., build.sh -j20)
was applied while running `tprof monitor -e LsNotHaltedCyc ...' on x86.

This almost eliminates the problem, but still is not enough. tprof_x86 uses NMI
interrupts, which are interrupted even in splhigh(), leaving the possibility of
being interrupted in the splhigh section of percpu_cpu_swap().

(ryo)

lint: add measurement unit to local variable names

No binary change, except for line numbers in assertions.

(rillig)

lint: make definitions of basic types simpler

No binary change.

(rillig)

Fix primary boot selection for FFSv2ea filesystems

(martin)

Remove more duplicated static function declarations.

(tsutsui)

Avoid explicitly naming the raw part device with a partition letter - use
the non-partition letter variant instead.

(martin)

Unfortunately rump does not provide the same magic as MAKEDEV does
for native /dev and create an alias for disk devices w/o partition
latter pointing at the raw partition, so for rump based tests we
actually have to calculate the concrete device name.

Use an idiom suggested by kre for this which also works for ports that
have kern.rawpartition > 4.

(martin)

Use proper RAW_PART macro rather than a magic number.

(tsutsui)

Set disk_geom parameters required by DIOCGPARTINFO in disk_ioctl().

With this change raid(4) no longer complains before mountroot:
> RAIDframe: can't get disk size for dev rd0 (22)

Should be pulled up to netbsd-9.

(tsutsui)

Remove a duplicated static function declaration.

(tsutsui)

Pass a correct dev_t arg to disk_ioctl().

Probably harmless.

(tsutsui)

Check bounds of each partition by MI bounds_check_with_label(9).

The previous implementation was committed in rev 1.26 about 26 years ago
and it looks there is no quirk to use the old MD one.

(tsutsui)

Check bounds of RAW_PART by bounds_check_with_mediasize() as other drivers.

With this check, dd(1) without a count value against a raw partition
is terminated properly at the end of media.  Tested on 425t and HPDisk.

Should be pulled up to netbsd-9.

(tsutsui)

Switch the default FFS type from FFSv2 to FFSv2ea - we want extended
attribute support to be exercised and tested.

If you want to share a new installed disk with older NetBSD
installations or (read only) with other OSes you need to explicitly set
the FS type to FFSv2 now.

(martin)

Add support for FFSv2ea in the partition type menus (internally setting
fs version to 3 for this, where 2 is FFSv2 and 1 is FFSv1)

(martin)

Fix silent bus error panic on 98543A topcat framebuffer on HP320 and HP360.

It looks DELAY(100)s before checking cmap_busy in old pre-wscons grf_tc.c
http://cvsweb.netbsd.org/bsdweb.cgi/src/sys/arch/hp300/dev/Attic/grf_tc.c?rev=1.42
are actually necessary on 68020 (HP320) and 68030 (HP360) machines,
while it works without them on 68040 (HP380).

Should be pulled up to netbsd-9.

(tsutsui)

tests: restore a line removed accidentally

(ozaki-r)

Simplify: avoid use of the explicit raw partition, MAKDEV does
all the needed magic for us already. Pointed out by kre

(martin)

Update the check files due to bug fix in libcurses.

(blymn)

When performing a scroll, check the last line of the region is the same
on virtscr and curscr because the indexes past are supposed to be
one *past* the last matching line (they may actually match if the line is
at the bottom of the screen).  Iff they don't match reduce the scroll
region size by one so we don't scroll non-matching lines, also check
if the region is then 0 after the decrement and just return if it was.

(blymn)

tests: build and install t_ip_reass.c

(ozaki-r)

tests: tweak t_ip_reass.c for NetBSD

The test is modified to run on rump kernels.

(ozaki-r)

tests: import ip_reass_test.c from FreeBSD as t_ip_reass.c

As of:
commit 9ed1e4ecd4e9eb3bde16f52a937a6fa86a971638
Author: Mark Johnston <markj@FreeBSD.org>
Date:  Tue Nov 20 18:13:18 2018 +0000

    Plug a trivial memory leak.

    CID:            1396911
    MFC with:      r340485

(ozaki-r)

man, bpf: support loopback interfaces for send

(ozaki-r)

bpf: support sending packets on loopback interfaces

Previously sending packets on a loopback interface via bpf failed
because the packets are treated as AF_UNSPEC by bpf and the loopback
interface couldn't handle such packets.

This fix enables user programs to prepend a protocol family (AF_INET or
AF_INET6) to a payload.  bpf interprets it and treats a packet as so,
not just AF_UNSPEC.  The protocol family is encoded as 4 bytes, host byte
order as per DLT_NULL in the specification(*).

(*) https://www.tcpdump.org/linktypes.html

Proposed on tech-net and tech-kern

(ozaki-r)

Trailing whitespace

(skrll)

make.1: clarify in which case an expression may omit braces

(rillig)

tests/make: document that the ':tl' modifier does not split words

(rillig)

socketpair(2): Formatting nits

(uwe)

document the only supported type protocol, document type flags

This uses language taken from OpenBSD's manual page.

(jschauma)

Avoid dependence on details of the MD synthesized disklabel for
the vnd after mounting the single filesystem image.

(martin)

be more specific about what /etc/protocols describes

The reference to the "DARPA Internet" seems not quite up to date in 2022,
so moving that to the HISTORY section.

(jschauma)

regen from IANA as of 2022-09-28

This time using the updated tools from the net/iana-etc package, ensuring
we don't revert previous changes and pass the tests from tests/lib/libc/net/.

(jschauma)

whoops, we don't want an extra space here

(erroneously introduced in my previous commit)

(jschauma)

whitespace

(jschauma)

revert previous

Automated test reports failure, reverting until I know why the test
failed.

(jschauma)

ch-ch-changes

(nia)

the UFS_EXTATTR option was supposed to affect only UFS1 file systems,
but when the UFS2 extattr code was merged, the UFS_EXTATTR option was
mistakenly changed to affect UFS2 file systems as well.  this commit
changes UFS_EXTATTR back to affecting only UFS1 file systems as originally
intended.  in UFS2 (or rather UFS2ea in NetBSD), extattrs are a
native feature and are always supported.

(chs)

regen from IANA 2022-11-22

(jschauma)

regen from IANA 2022-09-28

(jschauma)

reallocarray.3: minor grammar and style tweaks

(gutteridge)

Bump MSGBUFSIZE to 32k; 4k just about overflows "by default" with qemu
and mipssim-virtio machine type.

(he)

Misc KNF and cosmetics.

No binary change.

(tsutsui)

move the creation of machdep.smbios into platform_init().

"dmidecode -d /dev/smbios" now works on x86.  also tested on arm64 (lx2k.)

(mrg)

kern/vfs_lockf.c: Parenthesize to make arithmetic match check.

We verified start + (fl->fl_len - 1) would not overflow, but then
computed (start + fl->fl_len) - 1 instead, and it is possible for
start + fl->fl_len to overflow before we subtract 1.

Reported-by: syzbot+762480b00cb14085f63a@syzkaller.appspotmail.com
https://syzkaller.appspot.com/bug?id=d69313b2460a12715315f9f5f74fbe44b8bc38ba

(riastradh)

Remove unnecessary declarations.

(tsutsui)

Service mode switch should not be treated as an error.

Use aprint_normal(9) to avoid confusing dmesg.

(tsutsui)

Sync with rdreg.h integer type changes.

(tsutsui)

Use proper integer types for HP-IB xfer parameters.

No visible regression on 425t and HPDisk.

(tsutsui)

Add driver for SiFive FU540 PRCI clock controller.

(jmcneill)

Add ATF for unnumbered interfaces.

(knakahara)

ifconfig(8) support unnumbered flag.

(knakahara)

Welcome to 9.99.107

    - add IFF_UNNUMBERED flag

(knakahara)

Support explicit unnumbered interface.

Currently, NetBSD supports implicit unnumbered interface by setting
the same IP address to two interfaces.  However, such interface is not
treated as unnumbered when one of the interfaces is being changed and
has been changed IP address.  That behavior can be harmful for some
routing daemons.

(knakahara)

KNF. No functional change.

(msaitoh)

Mention where in the datasheet the compensation algorithms came from.

(brad)

match zen3 "cezanne" (ryzen 5000-series APU.)

(mrg)

clean up

(knakahara)

in ofprint() only append \r if the last character is \n

(macallan)

move ofprint() to powerpc/ofw_machdep.c and make it official
now that console output uses locks it needs more of the kernel to function, so
for now use direct OF calls for earliest debug output

(macallan)

mmap(2): Avoid arithmetic overflow in search for free space.

PR kern/56900

Reported-by: syzbot+3833ae1d38037a263d05@syzkaller.appspotmail.com
https://syzkaller.appspot.com/bug?id=e542bcf59b2564cca1cb38c12f076fb08dcac37e

(riastradh)

New sentence, new line.

(wiz)

Use kpause() instead of delay() in the measurement cycle.  Try and
derive the proper wait delay for the measurement based upon the over
sampling setting and allow the wait factor multiplier to be adjusted.

(brad)

Add links to libm (Phil Rulon)

(christos)

Remove now unused buffer members from softc prepared only for rdreset().

(tsutsui)

Fix "rd(4) at punits not configured on HPDisk are misprobed" problem.

- check not only stat value returned by C_QSTAT command against
  each punit but also desc.d_name returned by C_DESC command
  because it looks HPDisk responds to commands against punits supported
  but not-configured punits at 1-3
- prepare rdreset_unit() function to manage reset punit ops during
  probe without valid softc

The stragety is taken from OpenBSD with several modification.
Tested on 425t with real 9122D and HPDisk emulating two 7937H drives.

Should be pulled up to netbsd-9.

(tsutsui)

fix vax

(christos)

fix vax build

(christos)

Read the datasheet more closely and put in some delays.  The chip will
just return junk if the wait is not long enough to allow a measurement
to start.

(brad)