Add a newline to a printf message.

(simonb)

Apply FreeBSD svn r373278 fix for ZFS corruption.  Fix for NetBSD
PR kern/58111 .

It would be extremely unlikely to trip this bug on NetBSD, as we don't
expose SEEK_DATA and SEEK_HOLE and you need to call ioctl(2) with
FIOSEEKDATA and FIOSEEKHOLE directly which no currently known code does,
and even then be unlucky enough to trip a race condition.

With a reproducer based on that in https://www.illumos.org/issues/16087,
I saw 11 groups of failures over 8 hours.  With this patch, no
failures in 10 hours.  The repro for NetBSD will be attached to
https://gnats.netbsd.org/58111 .

Original FreeBSD commit message:
--------------------------------
dnode_is_dirty: check dnode and its data for dirtiness

Over its history this the dirty dnode test has been changed between
checking for a dnodes being on `os_dirty_dnodes` (`dn_dirty_link`) and
`dn_dirty_record`.

It turns out both are actually required.

In the case of appending data to a newly created file, the dnode proper
is dirtied (at least to change the blocksize) and dirty records are
added.  Thus, a single logical operation is represented by separate
dirty indicators, and must not be separated.

The incorrect dirty check becomes a problem when the first block of a
file is being appended to while another process is calling lseek to skip
holes. There is a small window where the dnode part is undirtied while
there are still dirty records. In this case, `lseek(fd, 0, SEEK_DATA)`
would not know that the file is dirty, and would go to
`dnode_next_offset()`. Since the object has no data blocks yet, it
returns `ESRCH`, indicating no data found, which results in `ENXIO`
being returned to `lseek()`'s caller.

This change simply updates the dirty check to check both types of dirty.
If there's anything dirty at all, we immediately go to the "wait for
sync" stage, It doesn't really matter after that; both changes are on
disk, so the dirty fields should be correct.

Sponsored by:  Klara, Inc.
Sponsored by:  Wasabi Technology, Inc.

(simonb)

Updates for less - add me as co-maintainer, update to latest version.

(simonb)

Revert mistaken "no obmem0" that I added when enabling UFS_DIRHASH.

Thanks tsutsui@ for pointing this out.

(simonb)

Remove funny blank line.

(simonb)

Add Cavium CN68xx to list of known CPUs.

(simonb)

Fix nanosecond math in gfrtc_gettime().

(simonb)

Note some large xstate stack objects what Somebody(tm) should look at
when they find some round tuits.

(simonb)

Fix some comments.

(simonb)

Add some earmv4 kernels.

(simonb)

All other install kernels use a 6MB RAM disk.  Stop with the teeny
frequent bumps and jump straight to the size everything is using.

(simonb)

Catch up with recent changes in distrib/sets/lists/dtb/ad.earmv7hfeb

(simonb)

Catch up with recent changes in distrib/sets/lists/dtb/ad.earmv7hf.

(simonb)

Add
  #include <cdbr.h>
to the synopsis.

(simonb)

Align sysconf numbers.

(simonb)

Add _SC_AVPHYS_PAGES.
getconf(1) add this and SC_PHYS_PAGES.
libc: Use vm.uvmexp2 over vm.meter is it's twice as fast on my setup.
getconf.3: Tidy up wording for SC_PHYS_PAGES.

(simonb)

Correct a comment - 8 * 1 million is 8 million, not 10 million (!).

(simonb)

Remove extraneous comma after pool info

(simonb)

Debug printf tidy up, KNF comma,space nits.

(simonb)

Reduce upstream diffs - remove a gcc warnings " = 0" change.  This
was a problem in 2006, doesn't appear to be a problem anymore.

(simonb)

Reduce upstream diffs.

(simonb)

Reduce upstream diffs - tabs vs space.

(simonb)

Reduce upstream diffs.

(simonb)

Reduce upstream diffs - remove a gcc warnings " = 0" change.  This
was a problem in 2006, doesn't appear to be a problem anymore.

(simonb)

Reduce upstream diffs.

(simonb)

Reduce upstream diffs.

(simonb)

Reduce upstream diffs.

(simonb)

Tweaks.

(simonb)

Note update to less-643.

(simonb)

Clean up fluffed cvs import :/.

(simonb)

Many tweaks to 10-year-old script cherry-picking from many other
foo2netbsd scripts.

(simonb)

Handle new /include/ dir for defines.h, add extra .c files for less, lesskey.

(simonb)

Updated autoconf generated files for less-643.

(simonb)

Resolve conflicts from less-643 import.

(simonb)

Whitespace nit.

(simonb)

For vmstat -mW, add a total KB consumed column and widen a couple of
other fields.

(simonb)

Make mstats() decl match prototype so this compiles with -DMSTATS.

(simonb)

Pull git commit way-etc-20140221.tar.bz2 by Xin Li from FreeBSD:

    Remove unneeded checks for prelen.

    In order to determine the type of a compressed file, we have to read
    in the first four bytes which may also be important for decompression
    purposes, to do that we would pass the buffer that we have already
    read in, along with the size of it.

    Rename header1 to fourbytes to make that explicit, and remove all
    checks for prelen.

(simonb)

Alignment nit.

(simonb)

#define<tab>

(simonb)

In "devlist" mode, exit with a 0 return code if any nvme devices are
found, rather than exiting with 1 return code always.

(simonb)

Adjust userspace comments in db_read_bytes() and db_write_bytes() to
match current reality.

(simonb)

Enable UFS_DIRHASH (missed with previous change) and UFS_EXTATTR
(mistakenly commented out with previous change).

(simonb)

Note that strchrnul() first appeared in glibc 2.1.1 and appeared in
FreeBSD 10 before it appeared in NetBSD 8.

(simonb)

KNF nit: opening brace of a function on next line.

(simonb)

Welcome to 2023.  Wrap lines so the years fit in to 80 columns with
a leading kernel log timestamp.

(simonb)

Use updated defines for the user-mode sstatus value.

(simonb)

Document lots of bits.
Remove bits no longer in the RISC-V supervisor spec.
Update defines for the user-mode sstatus value.

(simonb)

Use better types and struct member names in the clockframe.
Avoid a magic number in CLKF_USERMODE().

(simonb)

Enable CD9660; noticed by he@.

(simonb)

Include GENERIC.local if it exists.

(simonb)

Use similar macro-magic to aarch64 armreg.h to add per-csr
read/write/set-bits/clear-bits inline functions.  Keep the
open-coded 32-bit version of riscvreg_cycle_read() than reads
a 64-bit cycle counter values.

Added benefit of fixing these so that the inline asm uses __volatile
and aren't opmtimised to nops by the compiler.

(simonb)

Add NFS_BOOT_DHCP option.

(simonb)

The supervisor status register is the native word width, not fixed
at 32 bits.

(simonb)

Parentheses police.

(simonb)

Add cause register trap types, and some macros to access cause register
fields.

(simonb)

Add a #define for XLEN, the RISC-V native base integer ISA width.

(simonb)

Fix tyop in __BITS for the MIPSNN_MTI_CFG7_PREF_MASK macro.

(simonb)

Define the mount program for zfs.

(simonb)

Use PRIx64 in a debug print macro to print a uint64_t.

(simonb)

In bus_space_write_{1,2,4,8} store the correct register in write to device.

(simonb)

Fix tyop in END for generic_bs_r_8.

(simonb)

Provide an empty ASSERT macro ifdef _STANDALONE.

(simonb)

Whitespace nit.

(simonb)

Fix (invalid) gcc uninitialised warning with i386 gcc using -Os.

(simonb)

Only need TIMESPEC_OVERFLOW() to check against INT32_{MIN,MAX} on FreeBSD.

(simonb)

Add a -q option for pgrep to not print anything; similar to grep -q.

(simonb)

Remove extraneous "d" in a printf format string in UVM history dump
(leftover int format specifier from conversion to FMTd32?).

(simonb)

Consistency nit: use "__volatile" instead of "volatile" with asm()s.

(simonb)

Add SATP modes for bare, SV57 and SV64.

(simonb)

Use __volatile so the compiler doesn't optimise out successive calls
to cpu_counter().
Add a 64-bit cycle counter on _LP64.

(simonb)

#define<tab>

(simonb)

"options<space><tab>"

(simonb)

Nuke funny trailing whitespace.

(simonb)

Set RISCV_DIRECTMAP_SIZE to 2^64-PAGESIZE, since 2^64 is effectively 0
for a 64bit constant.
Bump VM_PHYSSEG_MAX from 1 to 16.

(simonb)

Fix a tyop regname in a comment.

(simonb)

NKMEMPAGES_MIN_DEFAULT is in pages not bytes (hint is in the name).
Also set NKMEMPAGES_MAX_UNLIMITED while we're here.

(simonb)

Regen.

(simonb)

Regen.

(simonb)

UFS/LFS dirhash:
- Enable UFS_DIRHASH if the architecture or kernel model specific config
  file can use 128MB of RAM or more.
- Remove experimental tag from UFS_DIRHASH; it's been with RUMP kernel
  and by a number of NetBSD developers for years.
- Add LFS_DIRHASH if LFS was enabled.
- Be somewhat consistent with FS options order.

(simonb)

If UFS or LFS dirhash is enabled in the kernel, set the dirhash cache
size dependant on memory size.  If less than 128MB of memory, default
to no cache.  With 128MB of memory or more, use a maximum cache size of
1/64th of memory; cap maximum default cache size to 32MB (for systems
with 2GB of memory or more).

The dirhash cache sizes are still explicityly setable by sysctl(8) or
by adding relevant entry(s) to sysctl.conf(5).

(simonb)

Add some file system options present in GENERIC kernels.

(simonb)

Removed unused opt_wapbl.h include.

(simonb)

When operating on core files or /dev/mem when using the -M option,
use 64-bit math to calculate pool sizes.  Fixes overflow errors for
pools larger than 4GB and gives the correct output with "vmstat -m"
for in use, total allocation and utilisation numbers.

(simonb)

Use 64-bit math to calculate pool sizes.  Fixes overflow errors for
pools larger than 4GB and gives the correct output for kernel pool pages
in "vmstat -s" output.

(simonb)

Move the call to getnlist() to after we check if kvm_openfiles(3) succeeded.
Avoids a coredump when called with "vmstat -M /dev/mem".

(simonb)

Add basic ZFS support to fstat(1).
XXX: Use a local zfs_znode.h for a znode_t definition because the ZFS
kernel code makes it too hard to include headers from userland.

(simonb)

Bump some column widths for "vmstat -mW".

(simonb)

White space KNF nits.

(simonb)

Add the increasingly spreading __HIDE_DELAY kludge to evbmips.

(simonb)

In biosdisk_findpartition() check if part_name isn't NULL before
assigning *part_name.

Thanks to manu@ for the pointer.

(simonb)

Fix a tyop.

(simonb)

Revert rev 1.12 of devopen.c.  This had the unintented side effect
of breaking opens on non-root filesystems (eg trying to open/read
"esp:/EFI/NetBSD/boot.cfg" on the EFI system partition).

Ok manu@.  Original problem to be re-addressed.

(simonb)

In zfs_setattr() don't recheck the auth policy for a "nodump" flags
change. zfs_netbsd_setattr() has already checked if this request is
authorised, and our secpolicy_xvattr() doesn't check kauth chflags.

XXX: Fix this propery when we migrate to openzfs.

riastradh@: Seems reasonable.

(simonb)

Explicitly use -I$S/external/bsd/dwc2/dist for dwc2 instead of possibly
getting that include path via DRM.  Fixes evbmips cavium kernel build.

(simonb)

In vattr_init_mask() check for va_flags and if set add to the mask of
to the things that setattr is interested in.
Fixes chflags(2) on ZFS.

Thanks riastradh@ for fix.

(simonb)

Update -m description to reflect reality.  Kernel dynamic memory tracking
was removed long ago; only pool usage is reported.

(simonb)

In zfs_statvfs(), set f_bresvd and f_fresvd to 0.  Fixes random kernel
accounting suspend/resumes with erroneous values leaking out.

Note: no userland leakage as statvfs(2) handler memset 0's the buffer.

XXX: Should be fixed with a memset in VFS_STATVFS().

(simonb)

Add /etc/sshd/ssh_host_ed25519_key* .

(simonb)

Set totals.f_frsize to DEV_BSIZE instead of totals.f_bsize so that
addstat() uses an initialised value for total size calculations.
Fixes core dump for "df -c".

(simonb)

Add some { } around an if body that is followed by an "else {".

(simonb)

Use 50Hz ticks so that time works accurately on hosts with 100Hz ticks.

Note: The 20Kc CPU isn't handled correctly by QEMU, use the 5Kc CPU for
a 64-bit guest for accurate time.

(simonb)

Use CPU frequencies defined in QEMU for mipssim - 6 MHz for 64-bit
guests and 12 MHz for 32-bit guests.

(simonb)

Use the register define MIPS_HWR_ULR instead of a magic number.

(simonb)

Use the architecture documented name ULR for the RDHWR user local
register.

(simonb)

Add some comments for the RDHWR register numbers.

(simonb)

Only need one #define for MIPS_HWR_CPUNUM.

(simonb)

Fix tyop in a comment.

(simonb)

options<space><tab>

(simonb)

Hacks to get zdb working on NetBSD:
- Force accessing raw device but we're passed the block device.
- Deal with wedges not returning their size on a stat(2) and use
  the DIOCGMEDIASIZE ioctl instead.

Ok chs@.

(simonb)

TAB police.

(simonb)

When adjusting the max ASID count, check if ti->ti_asid_max == 0 as
well.  This defaults to 0 for the non-PMAP_TLB_NUM_PIDS case, so would
skip the updated test.

Fix for port-pmax/56466 (which affects all MIPS).

ok srkll@

(simonb)

Whitespace nits.

(simonb)

Spinkle some KNF spaces after commas.

(simonb)

Remove funny straggling blank line.

(simonb)

Clamp zfs nlinks at UINT32_MAX (nlink_t max) instead of (much) smaller
16-bit LINK_MAX until we bump LINK_MAX.  Fixes fts(3) problems with
"rm -rf" on zfs directories with > 32766 subdirectories.

Thanks mlelstv@ and mrg@ for helping debug this.

(simonb)

#define<tab> consistency.

(simonb)

Update to reflect current reality for template characters - the process
ID is not longer used and we replace each "X" with one of 64 values not
26 values.

(simonb)

If we're only doing a count-only kern.buf sysctl, just return the number
of active members in the pool cache (plus some slop) instead of looking
in all the free buffer list.  Should reduce CPU usage for "systat vm"
to << 1% especially for machines with a larger number of buffers.

(simonb)

Add accessor functions to get the number of gets and puts on pools and
pool caches.

(simonb)

Instead of getting an estimate of the number of kernel buffers,
allocating memory for them, requesting all the metadata contents of
these buffers (and repeating in the unlikely case of the number of
buffers increasing too much since the estimate) and then straight away
throwing all the contents out just to count how many buffers there were,
just get the initial estimate from the kernel and subtract the slop.

Reduces system CPU usage of "systat vm" by approx 80% for any system
with a reasonable number of buffers.

(simonb)

Expose KERN_BUFSLOP in <sys/sysctl.h>.

(simonb)

Pad out the slop for kern.buf based on the passed in element size,
rather than a size of an unrelated struct.

(simonb)

Call check_siginfo() in compression loop to provide SIGINFO compression
progress.
Suggested by mrg@.

(simonb)

Remove an unused #define.

(simonb)

Return ENOENT if the hashstat sysctl was called to query a specific hash
name and that hash name doesn't exist.

(simonb)

Remove the warning about autoselecting nfs based on : or @.

Ok jmcneill@, mrg@.

(simonb)

s/leading spaces/tabs/

(simonb)

Include "opt_param.h" (ifdef _KERNEL_OPT) everywhere that MSGBUFSIZE is
referenced since some sources include <machine/param.h>.

(simonb)

Use <sys/param.h> in the synopsis rather than <machine/param.h>; an MI
interface should be pulled in via <sys/...>.

(simonb)

MIPS n64 dynamic binaries have worked for a while now, remove -static
for these.

(simonb)

Update the FPU register names and bit definitions to something somewhat
modern (MIPS32/MIPS64) and convert to __BIT/__BITS.

(simonb)

Put the MSGBUFSIZE option in opt_param.h so that changes of its setting
in kernel config files will be picked up automatically without needing
a "make clean".  Added to opt_param.h since ports define their default
MSGBUFSIZE settings in their <arch/include/param.h> (or equivalent).

(simonb)

Update R6000 description.

(simonb)

Rename the unhelpfully named mips_emul_lwc0() and mips_emul_swc0() to
mips_emul_ll() and mips_emul_sc(); make these static to mips_emul.c.

(simonb)

Print the CP0 status register too in the debug trap code.

(simonb)

Remove nfs_putpages() prototype; it's not defined anywhere.

(simonb)

Move the send queue checking to a new function, and also call this
in the rx interrupt path.  Measureable improvement on a NFS "create
small files" test.

(simonb)

Schedule the send cleanup function for next tick in cnmac_start().  In
the send cleanup function, schedule for the next tick instead of waiting
for HZ ticks if there are still send requests outstanding.  Greatly
increases NFS throughput, perhaps other types of network traffic.

Use callout_setfunc() after callout_init() instead of callout_reset().

(simonb)

Alignment nit.

(simonb)

Whitespace alignment nit.

(simonb)

The MIPS O64 ABI uses full 64-bit FP regs.

(simonb)

Fix a missed bitmask to __SHIFTOUT conversion in rev 1.12.

Fixes negotiation problems on non-gige switches.  Problem discovered and
tested by riastradh@.

(simonb)

Update ISA for some "L" variant instructions after checking the R4400 UM.

(simonb)

Note which ISA the unimplemented instructions belong to.

(simonb)

If we're going to print a number in hex, at least put a 0x in front of it.

(simonb)

Whitespace nit.

(simonb)

Use "static" in the function intro if the function is static.

(simonb)

Revert rev 1.5 - put the NOFPU option back.  Older cnMIPS cores don't
have an FPU.

(simonb)

Column alignment.

(simonb)

Sprinkle some static.

(simonb)

The watchdog setup code isn't used (yet), so don't compile in the
octeon_nmi_stack that it uses either.

(simonb)

Add links for the other functions documented in arc4random(3).

(simonb)

Whitespace police.

(simonb)

Don't force all MALTA kernels to be compiled with -mips64 when they
also support MIPS32-only CPU configs.

(simonb)

Fix another misplaced label for cvt_s_w() but use a named local label
and redo fix for cvt_d_w() in rev 1.52 the same way.

(simonb)

Move a comment slighty so that it's before two #ifdef blocks that do the
same thing instead of in the middle of them.

(simonb)

Move a branch target in cvt_d_w() to where it will be hit for either
case of an #ifdef block of code.  Fixes an FP emulation problem if
compiled with -mips32 or -mips64.

(simonb)

Merge the .text.* sections into the .text section on MIPS, SPARC and
SPARC64 as is done on all other ports.  Mostly costmetic, but does mean
that main() now appears before the _etext symbol instead of after it.

(simonb)

Add CVS ID line.

(simonb)

Delete the NOFPU option.  These CPUs do have an FPU.  Allows o32
binaries to work now.

(simonb)

Use __register_t instead of uregister_t - this is available to all ports
and both userland and kernel.

(simonb)

Cast the fill value to unsigned char so that the "fill" value used for
full-word fills isn't garbage.

(simonb)

Disable the larger/faster code path.  While the optimised code path was
indeed quicker, it nonetheless failed to actually fill all the requested
memory with the specified value much of the time if a non-aligned start
address was used.

(simonb)

Compile in compat32 syscall names all the time, as we do for native
syscall names.  Needed for dtrace.

(simonb)

Print target addresses similar to aarch64 as "address <sym+off>"
instead of "<sym+off> [addr:address]".  Uses less columns, a bit
easier on the eyes.

(simonb)

Build the netbsd32_syscall module if mips64.
Don't build the linux_syscall module if mips*

(simonb)

Sort includes.

(simonb)

tc_init(): KASSERT that the timecounter hasn't previously been
initialised by checking the tc_next field.  Prevents nastly endless
loops in the timecounter code.

(simonb)

Update usage message to reflect reality.

(simonb)

Add a # to a %x printf format to get some 0x hex number prefixes.

(simonb)

Revert the change to not build and install /usr/lib/dtrace/drti.o; this
is used by the ruby30-base package when compiled with the dtrace option
(enabled by default).

Problem found by ryoon@.

(simonb)

Basic dtrace trap support.

Mostly from FreeBSD.

(simonb)

Use NAL (nop-and-link) to get the current PC.

(simonb)

The location arg of db_disasm() is a db_addr_t, not a vaddr_t.

(simonb)

Remove duplicate db_disasm() prototype.

(simonb)

Include <ddb/db_interface.h> to get ddb's db_disasm() prototype.

(simonb)

Fix a typo, remove a useless blank line.

(simonb)

Revert rev 1.278 which tried to call ctfmerge only once - this has
caused kernels with embedded symbol tables to explode in size.

(simonb)

If we're going to print a number in hex, at least put a 0x in front of
it so we can cut'n'paste it into gdb directly.  This has only annoyed
me for 25 or so years...

Wrap a long long while here.

(simonb)

sizeof(*var) instead of sizeof(type)

(simonb)

sizeof(*var) instead of sizeof(type)
Call kmem_free() with size of data being freed.

(simonb)

Call kmem_free() with the size of the thing we are freeing.  Fixes panic
when unloading the ktrace module on MIPS.

(simonb)

Check the CTF magic number in an endian friendly way (and don't use
magic numbers to check the magic number!).

(simonb)

Define the CTF ELF section header magic number.

(simonb)

Fix the column width calculation for the lstart column if an empty
column header is specified.

Fixes bug pointed out by Ted Spradley in
https://mail-index.netbsd.org/netbsd-users/2021/04/05/msg026808.html .

(simonb)

We only need -Wno-format-y2k for print.c .

(simonb)

Some QED instructions are included in MIPS32 and MIPS64 instruction sets.
Update a few comments.

(simonb)

Fix cut'n'paste typo - OP_CVM_DMUL is dmul, not baddu.

(simonb)

Tidy up NOP disassembly, handle "pause" as well.

(simonb)

gcc/gas also emits "or ...,zero" as well as "addu/daddu ...,zero" for a
"move" pseudo instruction.  Disassemble the "or" case as a "move" too.

(simonb)

Allow disassembly in XKSEG for LP64 kernels.  Can now x/i on modules
with an N64 kernel.

(simonb)

For bc{0,1,2}{t,f} check for the TRUE value not the MASK value (even
though they're the same).

(simonb)

Move the ctfmerge call from SYSTEM_LD_TAIL (used by both STAGE1 and STAGE2)
to SYSTEM_LD_TAIL_STAGE2 only.  ctfmerge is slow enough that just calling
it once is enough.

(simonb)

Don't call strncmp() where the length arg is the size of a pointer
instead of the max string length.  Switch to strcmp() - both strings
are known nul-terminated.

Thanks tnn@ (and clang!).

(simonb)

Add a -L option to follow all symbolic links.  Useful if you have symlinks
in a makefs directory tree but want to refer to the actual file.

(simonb)

Deal with the case when a device gets recreated (eg when mtree is called by
MAKEDEV) and the stat buffer contains information about the original device
before checking/setting the owner/group/mode of the new device file.

(simonb)

Centralise the setitimer() timer type validation in dosetitimer() as is
done with dogetitimer().

(simonb)

Print the "preposterous TOD clock time" message only if the RTC is
before very late 2020 (1/1/2021 minus a few leap days) instead of
late 1994.  Someone(TM) should remember to adjust this in somewhat
less than 26 years.

(simonb)

CTASSERT that NSIG <= 128.  There are many hard-coded assumptions that
there are <= 4 x 32bit signal mask bits.

(simonb)