we don't need or want COMPAT_NETBSD32 in a 32bit kernel
should cure pkgsrc MACHINE_ARCH mismatches

(macallan)

PR/48843: Jarmo Jaakkola: dot commands mess up scope nesting tracking

Evaluation of commands goes completely haywire if a file containing
a break/continue/return command outside its "intended" scope is sourced
using a dot command inside its "intended" scope.  The main symptom is
not exiting from the sourced file when supposed to, leading to evaluation
of commands that were not supposed to be evaluated.  A secondary symptom
is that these extra commands are not evaluated correctly, as some of them
are skipped.  Some examples are listed in the How-To-Repeat section.

According to the POSIX standard, this is how it should work:
    dot:
        The shell shall execute commands from the file in the current
        environment.
    break:
        The break utility shall exit from the smallest enclosing for, while,
        or until loop, [...]
    continue:
        The continue utility shall return to the top of the smallest
        enclosing for, while, or until loop, [...]
    return:
        The return utility shall cause the shell to stop executing
        the current function or dot script.  If the shell is not currently
        executing a function or dot script, the results are unspecified.

It is clear that return should return from a sourced file, which
it does not do.  Whether break and continue should work from the sourced
file might be debatable.  Because the dot command says "in the current
environment", I'd say yes.  In any case, it should not fail in weird
ways like it does now!

The problems occur with return (a) and break/continue (b) because:
    1)  dotcmd() does not record the function nesting level prior to
        sourcing the file nor does it touch the loopnest variable,
        leading to either
    2  a) returncmd() being unable to detect that it should not set
          evalskip to SKIPFUNC but SKIPFILE, or
        b) breakcmd() setting evalskip to SKIPCONT or SKIPBREAK,
        leading to
    3)  cmdloop() not detecting that it should skip the rest of
        the file, due to only checking for SKIPFILE.
The result is that cmdloop() keeps executing lines from the file
whilst evalskip is set, which is the main symptom.  Because
evalskip is checked in multiple places in eval.c, the secondary
symptom appears.
>How-To-Repeat:
Run the following script:

    printf "break\necho break1; echo break2" >break
    printf "continue\necho continue1; echo continue2" >continue
    printf "return\necho return1; echo return2" >return

    while true; do . ./break; done

    for i in 1 2; do . ./continue; done

    func() {
        . ./return
    }
    func

No output should be produced, but instead this is the result:
    break1
    continue1
    continue1
    return1

The main symptom is evident from the unexpected output and the secondary
one from the fact that there are no lines with '2' in them.
>Fix:
Here is patch to src/bin/sh to fix the above problems.  It keeps
track of the function nesting level at the beginning of a dot command
to enable the return command to work properly.

I also changed the undefined-by-standard functionality of the return
command when it's not in a dot command or function from (indirectly)
exiting the shell to being silently ignored.  This was done because
the previous way has at least one bug: the shell exits without asking
for confirmation when there are stopped jobs.

Because I read the standard to mean that break and continue should have
an effect outside the sourced file, that's how I implemented it.  For what
it's worth, this also seems to be what bash does.  Also laziness, because
this way required no changes to loopnesting tracking.  If this is not
wanted, it might make sense to move the nesting tracking to the inputfile
stack.

The patch also does some clean-up to reduce the amount of global
variables by moving the dotcmd() and the find_dot_file() functions from
main.c to eval.c and making in_function() a proper function.

(christos)

add one more directory.

(christos)

new shell tests

(christos)

add new directory

(christos)

PR/48843: Jarmo Jaakkola: Test cses for break/continue/return broken
inside dot commands:

The test cases are named with the pattern
    dotscope_cmd_cmdscope,
where dotscope is the scope in which the dot command is and cmdscope
is the scope where cmd is in the sourced file.

(christos)

use weak symbols instead of linking against libpthread.

(christos)

- match up comment and interface identifiers
- use RFC5737 documentation prefixes
- use a variable for the RFC1918 private address ranges

(spz)

example for port remapping added

(spz)

Instead of running MAKEDEV inside the image content dir (which would only
work for root), make it emit a mtree spec file and pass that to makefs.
This should also work for unprivileged builds.
Problem pointed out by Izumi Tsutsui.

(martin)

Add a variable CDMAKEFSEXTRAOPTS to provide additional parameters to the
makefs invocation that do not belong in the image option list (CDMAKEFSOPTIONS)

(martin)

Regen for munmap(2) signature.

(njoly)

Fix munmap(2) signature.

(njoly)

- npf_nat_freepolicy: handle a race condition when a new connection might
  be associated with a NAT policy which is going away and npfctl reload
  would wait for its natural expiration (potentially long time).
- Remove npf_ruleset_natreload() by merging into npf_ruleset_reload().
- npf_ruleset_reload: eliminate a small time period when a valid NAT
  policy might be inactive during the reload operation.

(rmind)

Inline trivial helper function.

(joerg)

Remove debugging code introduced by accident in locore.s revision 1.357

(palle)

Remove duplicated lines.

(msaitoh)

If clang is requested for i386, amd64 or eARM, also build libc++.

(joerg)

Add note about the eARM flag day for C++.

(joerg)

Always use our exception handling code for eARM.

(joerg)

Switch eARM to using normal DWARF exception handling.

(joerg)

Remove platform support list for libc unwinder, just keep comment about
ia64 state.

(joerg)

Update build system for LLVM/Clang snapshot r209886. This brings in the
merged AArch64 backend and the move of eARM to DWARF exception handling
for NetBSD. C++11 is required for building.

(joerg)

regen

(nonaka)

Added Realtek devices.

(nonaka)

len argument is of unsigned type (size_t) and thus cannot be negative;
remove corresponding statement in ERRORS.

(njoly)

Pre-populate /dev on the CD - with rockridge extensions we can do this and
it saves us one tmpfs (or mfs) instance, which helps quite a bit on low
memory machines.

(martin)

The current vax /boot is super slow unziping kernels - and this is a CD, wich
has space enough, so unzip the kernel on it to speed up the boot process.
XXX The /boot bug should nevertheless be fixed.

(martin)

Make it deal with /dev being the result of a "MAKDEV all" instead of
"MAKEDEV init" by mounting a ptyfs if needed.

(martin)

PR kern/48852 (which should have been bin/ in retrospect): apply patch
from Thomas Schmitt to fix rockridge encoding of device nodes.

(martin)

Drop undocumented and redundant 0 argument to .ent.

(joerg)

Change NFS from rbtree to vcache.

(hannken)

vfs_vnode_iterator_next(): if a vnode is reclaiming (VI_XLOCK) skip
the filter.  Vget() will wait until the vnode disappeared.  No more
"dangling vnode" panics on unmount.

(hannken)

msdosfs_reclaim(): add missing fstrans and protect change
of v_data with v_interlock as msdosfs_sync() now needs it.

(hannken)

Testing "v_usecount == 1" for exclusive reference will not always
work -- remove and test only readonly.

(hannken)

Convert mtree spec path characters with sed(1) to avoid tool'ifying vis(1).

(uebayasi)

Add 5.1.4, 5.2.2.  6.0.4 and 6.1.3.

(msaitoh)

Sync with FreeBSD:
- Add DragonFly 3.6.0 and 3.6.1
- Add FreeBSD 9.2 and 10.0
- Add Mac OS X 10.9
- Add OpenBSD 5.4
- Change the master's URL.
Add some releases:
- Add NetBSD 5.1.4 and 5.2.2
- Add OpenBSD 5.5

(msaitoh)

Add 6.0.[45] and 6.1.[34]

(msaitoh)

- Remove some obsoleted comments.
- KNF.

(msaitoh)

- Add PCI-X capability stuff.
- remove extra ':' in pci_conf_print_pcie_cap()
- Add comments.

(msaitoh)

Regen.

(msaitoh)

Add Intel 3200 Host and PCIe.

(msaitoh)

Kill __P. Use static.

(joerg)

Remove !NetBSD code.

(joerg)

tcp_signature_getsav: handle !ipsec_used case and fix the build (hi christos!).

(rmind)

Use __CTASSERT() in the header.

(rmind)

Include cdefs.h earlier for NetBSD.

(rmind)

No need for libkvm here.

(joerg)

netstat(1) and ifmcstat(8): ifnet was renamed to ifnet_list, PR/48850.

(rmind)

Explain ipsec.enabled better.

(christos)

Introduce 2 new variables: ipsec_enabled and ipsec_used.
Ipsec enabled is controlled by sysctl and determines if is allowed.
ipsec_used is set automatically based on ipsec being enabled, and
rules existing.

(christos)

mention new ipsec sysctls.

(christos)

Make IGMP and multicast group management code MP-safe.  Use a read-write
lock to protect the hash table of multicast address records; also, make it
private and eliminate some macros.  In the long term, the lookup path ought
to be optimised.

(rmind)

Work around broken GTT size detection in agp_i810 code.

This will do until I work out the twisty maze of registers, all
different, to reliably determine the size of the GTT (and hence the
GPU's virtual address space) separately from the size of the AGP
aperture.

(riastradh)

hashinit: replace loop with a formula.

(rmind)

\%s is not an escape sequence, and we want %s

(plunky)

Remove unused files.

(skrll)

Don't grab libstdc++-v3/config.h it's not used.

(skrll)

enable IPSEC

(christos)

enable IPSEC

(christos)

Comment out nfrac as it's unused.

(skrll)

Don't try to listen on UDP sockets.

(christos)

Regen for utimes(2).

(njoly)

Add utimes(2) support.

(njoly)

Regen for utimes(2).

(njoly)

Ajust for compat linux utimes(2) change.

(njoly)

Regen for utimes(2) change.

(njoly)

For utimes(2), use compat_50_sys_utimes() instead of local version.

(njoly)

Remove unused variable.

(skrll)

Bump date for previous.

(wiz)

implement "no ident".

(mrg)

commonize the netbsd local version string.  (the native one wasn't
updated, and now that won't be a problem again.)

(mrg)

Add missing braces. From Henning Petersen in PR 48846.

(wiz)

make LINK_EH_SPEC match other usage.  fixes powerpc build.

(mrg)

Add a hack entry for lib/libcurses/Makefile workaround for lib/48827

(abs)

Add missing __diagused in rump code

(justin)

CID 12164{49,51}: Remove bogus ifp == NULL checks; if ifp was really NULL,
we would have been dead a few lines before the tests.

(christos)

tzcode 2014d

(christos)

welcome to 2014d; zic big bang fixes:
some programs did not handle large negative time_t values well (gnome)

(christos)

move the -std=gnu++0x code from Makefile to Makefile.common.
fixes the remaining 2/3 link issues.

(mrg)

add -pthread to compiler/linker flags.  fixes 1/3 of the link issues.

(mrg)

Explain the fields of struct intel_gtt for future reference.

(riastradh)

Cosmetic changes require testing too...oops.

(riastradh)

tzcode2014d/tzdata2014d has been released.

There's no need to updata tzdata this time, because it's identical
to the previous version.

(apb)

Fix bogus GTT total size calculation.  Omit gtt_bsh for `agp' gtt.

(riastradh)

Cosmetics: long lines, needless variables, PCI parameter verbiage.

(riastradh)

CID 976378: Avoid incorrect computation with ngrps == 0

(christos)

CID 975117: check listen(2) return .

(christos)

CID 97511{4,5}: Ignore bindresvport return. It only works for root.

(christos)

CID 975113: Ignore __rpc_setnodelay error, it is not fatal

(christos)

CID 975112: Ignore bindresvport() return. This is can only succeed for root

(christos)

CID 975111: Ignore return value from initgroups(). If we failed, then
the failure would be that we could not open the .rhosts file so we fail
closed.

(christos)

CID 272959: NULL deref

(christos)

Regen.

(msaitoh)

- Add some Intel devices (Sandy Bridge, 6702PXH, X38 and Pineview)
- Add ITE IT8213

(msaitoh)

Remove unused file.

(skrll)

OpenLDAP 2.4.39 was imported.

(tron)

Update for OpenLDAP 2.4.39.

(tron)

Update for building OpenLDAP 2.4.39.

(tron)

Enable swap during sysinst as early as possible

(martin)

Re-generate after import of OpenLDAP 2.4.39.

(tron)

Change field "layerm_tag" to correct type "enum vtagtype".

CID 1216449:  Mixing enum types

(hannken)

Apply patch to fix CVE-2013-4449. This shouldn't be necessary as we don't
build the server or its plug-ins. But it is better to be safe in case
this changes.

(tron)

Resolve conflicts from last import.

(tron)

Fix typo in comment.

(msaitoh)

add new htmintrin.h and htmxlintrin.h files for GCC 4.8.3.

(mrg)

re-run mknative-gcc after threading fixes by skrll and myself.

(mrg)

regen.

(msaitoh)

Add Intel X540-AT2.

(msaitoh)

two fixes skrll@ pointed out:
- pass --enable-libstdcxx-threads to enable a bunch of tests and thus
  get the right answers
- MKNATIVE_CONFIG_TARGET_LIBS needs to do libgcc before libstdc++-v3,
  for the GCC 4.8 case.

with these we now get the right threading options in the headers.

(mrg)

apply patch from skrll@:  allow netbsd host=target to configure
GLIBCXX_IS_NATIVE.

(mrg)

apply patch from skrll@:  allow netbsd host=target to configure
GLIBCXX_IS_NATIVE.

(mrg)

rerun mknative-gcc for GCC 4.8.3 release.

XXX:  still not 100% right, but we're getting close.

(mrg)

Reject unaligned PCI config register ioctl requests before we kassert.

(riastradh)

Rationalize agp_i810 aprint output.  Fix agp_i810_init error branches.

(riastradh)

Oops -- gtt_off is relative to mmadr, not absolute.

(riastradh)

need esp

(christos)

just allow l2tp not regular ipsec.

(christos)

In shmrealloc(), add missing condvar initialisations for segments
copied from previous location.

(njoly)

Ensure a disklabel is present since sun4v OpenBoot requires a valid disklabel in the disk file image. With this change the miniroot.fs can be used as a virtual disk when using sun4v LDOMs. OK martin@

(palle)

Add missing word

(wiz)

I like riastradh's kmem/pool explanation.

(christos)

Print "range:  not set" if the decode window isn't set.

(msaitoh)

- Add some register definition for MSI and MSI-X
- print MSI-X capability

(msaitoh)

Fix typo in a comment.

(msaitoh)

use -Wno-traditional only for GCC rather than !CLANG

(plunky)

we use C99 syntax but with some GCC extensions, so signal that depending
on the compiler in use

(plunky)

- Fix incorrect calculation in PCI_MSIX_CTL_TBLSIZE().
- The PCI_MSIX_CTL_TBLSIZE bit field is in N-1, so add +1.

(msaitoh)

Optimise the BUS_DMASYNC_PREREAD operation by only using wbinv for partial
cachelines. Full cachelines are now invalidated only.

From the same change in various mips ports not least cobalt.

(skrll)

Related to PR kern/48533
Prism GT 1/2 is Conexant products.
Noted by skrll@, thank you.

(ryoon)

Regen.

(ryoon)

Related to PR kern/48533
Prism GT is Conexant products.

(ryoon)

Suppress uninitialized variable warnings in older and dumber GCC.

(riastradh)

Note 4.8.3 gcc

(skrll)

Note gcc 4.8.3 import.

(skrll)

Update with-pkgversion date for gcc 4.8.3 import.

(skrll)

Resolve conflicts

(skrll)

Move forward read pointer to the next line in the buffer
to prevent corrupting the most old line.

(msaitoh)

If -v is set and unknown cacheinfo desc is found, print it.

(msaitoh)

Rework agp_i810 attachment code a little.

- Fix up error branches in agp_i810_attach.

- Use a separate bus space handle for the GTT, whether it is in a
separate BAR or a subregion of the MMIO device registers, so that

(a) agp_i810_write_gtt_entry and agp_i810_post_gtt_entry are easier to
follow, and

(b) we can map the GTT prefetchable eventually.

(riastradh)

From dyoung@ -- bump max TX DMA size to avoid pathological condition with TSO.

(tls)

Don't check lfs flags on ffs vnodes.

(dholland)

mvphy is not configured, so don't try to build it

(pooka)

Fix previous. Anyone have a brown paper bag?

(dholland)

Use correct address for 64-bit flush page config register.

(riastradh)