Tidy up and make it look like the other drivers.

(christos)

Remove one that won't happen

(skrll)

Some updates

(skrll)

correct spkr softc.

(christos)

Remove binutils.old as everything has switched.

(skrll)

no more constant pitch, now we can use all video memory

(macallan)

1320-1322

(snj)

The t_clock_subr test doesn't require rump.

(rjs)

Pull up following revision(s) (requested by bouyer in ticket #1321):
sys/dev/usb/uchcom.c: revision 1.17
Followling an advice in a linux forum, don't update LCR1/LCR2.
With this change this CH340 usb/serial based device:
https://www.olimex.com/Products/Breadboarding/BB-CH340T/open-source-hardware
(the chip is written H340T)
works as expected. As I'm not sure if this is needed for older device,
make this change for sc_version 0x30 or newer only.
While there, match USB_PRODUCT_WINCHIPHEAD2_CH341_2 too.

(snj)

regen for ticket 1320

(snj)

Pull up following revision(s) (requested by bouyer in ticket #1320):
sys/dev/usb/usbdevs: revision 1.728
Rename "CH341 serial/parallel" to "CH341 USB-Serial Bridge", and
add a second device id for this chip.
From FreeBSD.

(snj)

Pull up following revision(s) (requested by riastradh in ticket #1322):
sys/external/bsd/drm2/drm/drmfb.c: revision 1.3
sys/external/bsd/drm2/i915drm/intelfb.c: revision 1.14
sys/external/bsd/drm2/include/drm/drmfb.h: revision 1.2
sys/external/bsd/drm2/nouveau/nouveaufb.c: revision 1.4
add da_fb_linebytes to drmfb_attach_args and use it to pass linebytes
from nouveau code to drmfb. keep the same linebytes logic for i915.
nvidia hardware needs 256 byte alignment, so aligning to just 64 was
not enough.
fixes broken console with a width of 1440px (PR kern/51181)
ok riastradh

(snj)

remove local lwp_eprio() definition that broke the clang build.

(mrg)

appease clang

(macallan)

use host blits for image uploads on pm2

(macallan)

Add "define ia64Architecture" for ia64 case; probably more changes needed later.

(scole)

Make ia64 considered 64bit for the _XSERVER64 define so a release with x11 will at least build.

(scole)

Change the freelists to lrulists, all vnodes are always on one
of the lists.  Speeds up namei on cached vnodes by ~3 percent.

Merge "vrele_thread" into "vdrain_thread" so we have one thread
working on the lrulists.  Adapt vfs_drainvnodes() to always wait
for a complete cycle of vdrain_thread().

(hannken)

Move vnode members "v_freelisthd" and "v_freelist" from "struct vnode"
to "struct vnode_impl" and rename to "vi_lrulisthd" and "vi_lrulist".

No functional change intended.

Welcome to 7.99.48

(hannken)

Remove the "target" argment from vfs_drainvnodes() as it is
always equal to "desiredvnodes" and move its definition
from sys/vnode.h to sys/vnode_impl.h.

Extend vfs_drainvnodes() to also wait for deferred vrele to flush
and replace the call to vrele_flush() with a call to vfs_drainvnodes().

(hannken)

Add a HISTORY section
Move information about exit status to EXIT STATUS section
Add more xrefs

ok wiz@

(abhinav)

Don't hold softc lock when calling ucom_read

PR/51714: uftdi (on xhci): mutex_vector_enter: locking against myself

(skrll)

Comment out spkr at audio, as audio is commented out too

(martin)

Comment out spkr at pcppi, as pcppi is commented out too.

(martin)

Simplify (and no need for sub-shells).
Same functionality, same method, simpler exposition.

(kre)

Sort the unsorted sets lists (they should all be correctly sorted now.)

(kre)

fix race of gif_softc->gif_ro when we send multiple flows over gif on NET_MPSAFE enabled kernel.

make gif_softc->gif_ro percpu as well as ipforward_rt to resolve this race.
and add future TODO comment for etherip(4).

(knakahara)

Ammend #1305 for additional changes

(martin)

Pull up revisions 1.11 and 1.15, requested by snj in #1305:

don't look for libdrm_nouveau unless we've built it.
only build the radeon and nouveau drivers on i386/amd64/evbarm,
where they're useful.  this saves almost 2MB in text/data on shark.

(martin)

Fix ip[46]csum-tx doesn't work other than TCP and UDP.

(msaitoh)

1324

(snj)

Apply patch (requested by kre in ticket #1324):
Update tz info.

(snj)

update 1324

(snj)

Apply patch (requested by kre in ticket #1324):
Update tz info.

(snj)

Add tests for flushing prefix and default router entries

(ozaki-r)

Reduce return points

No functional change intended.

(ozaki-r)

Update TODO.ptrace

Add new entries:
- add support for PT_STEP, PT_GETREGS, PT_SETREGS, PT_GETFPREGS,
  PT_SETFPREGS in all ports
- integrate all ptrace(2) features in gdb
- add ptrace(2) NetBSD support in LLDB

Sponsored by <The NetBSD Foundation>

(kamil)

Define -D_KERNTYPES in CPPFLAGS unconditionally to fix MKRUMP=no build

Reported by Robert Swindells

Sponsored by <The NetBSD Foundation>

(kamil)

Use macro to iterate on the nd_prefix list

(ozaki-r)

Make functions static

(ozaki-r)

add wait_for_disconnected to run_test() as well as run_test6().

Before commited MP-safe patch, IPv4 test can run in time without
wait_for_disconnected. Currently, wait_for_disconnected is required
because of locking overhead.

(knakahara)

Rename dump because it's used in net_common.sh

(ozaki-r)

More changes for creds, mutex, etc.

(pgoyette)

Mostly mechanical updates for caddr_t, thread, mutex, and credentials

(pgoyette)

Disable duplicate typedef

(pgoyette)

More cleanup of #includes

(pgoyette)

More cleanup of #includes

(pgoyette)

Adapt code for our location of the nfs headers, and some updates for
differences in kernel options(4) handling

(pgoyette)

Update for location of newnfs headers, clean up some other #include

(pgoyette)

Update locations for new nfs header files.

XXX Some #includes are just disabled (via #if 0) and will need to be
XXX resolved before further progress can be made.

(pgoyette)

One more option for the opt_ file.

(pgoyette)

fix format warning.

(christos)

wildcard speaker attachments, now that we can handle many of them.

(christos)

Restructure speaker devices so that there can be multiple of them, and have
proper softc's.

(christos)

back to passing device

(christos)

back to passing device_t

(christos)

ping is not a reliable means of testing if connectivity to download sets
actually works, so remove it.
Hopefully fixes PR kern/51531.

(roy)

Add regs1 in arch/i386/t_ptrace_wait*

regs1:
    Call PT_GETREGS and iterate over General Purpose registers

Sponsored by <The NetBSD Foundation>

(kamil)

Add rescan for audio children.

(christos)

Remove dbregs* in arch/amd64/t_ptrace_wait*

CPU Debug Registers won't be exposed as is to userland.

Hardware Watchpoints will be exported to userland dedicated interface
through the ptrace(2) interface.

Sponsored by <The NetBSD Foundation>

(kamil)

Add regs1 in arch/amd64/t_ptrace_wait*

regs1:
    Call PT_GETREGS and iterate over General Purpose registers

Sponsored by <The NetBSD Foundation>

(kamil)

Define in CPPFLAGS symbol _KERNTYPES in order to get PRIxREGISTER

This type will be used in t_ptrace_wait* for the printf(3) function.

Sponsored by <The NetBSD Foundation>

(kamil)

Remove dbregs[12] from t_ptrace_wait{,3,4,6,id,pid}

CPU Debug Registers won't be exposed as is to userland.

Hardware Watchpoints will be exported to userland dedicated interface
through the ptrace(2) interface.

Sponsored by <The NetBSD Foundation>

(kamil)

options(4): Remove KSTACK_CHECK_DR0.

(kamil)

Bump date for previous.

(wiz)

KSTACK_CHECK_DR0 is no longer available

It used to detect stackoverflow on the i386 port.

Sponsored by <The NetBSD Foundation>

(kamil)

Torn down KSTACK_CHECK_DR0, i386-only feature to detect stack overflow

This feature was intended to detect stack overflow with CPU Debug Registers
(x86). It was never ported to other ports, neither amd64 and should be
adapted for SMP...

Currently there might be better ways to detect stack overflows like page
mapping protection. Since the number of Debug Registers is restricted
(4 on x86), torn it down completely.

This interface introduced helper functions for Debug Registers, they will
be replaced with the new <x86/dbregs.h> interface.

KSTACK_CHECK_DR0 was disabled by default and won't affect ordinary users.

Sponsored by <The NetBSD Foundation>

(kamil)

Switch x86 CPU Debug Register types from vaddr_t to register_t

This is more opaque and appropriate type, as vaddr_t is meant to be used
for vitual address value. Not all DR on x86 are used to represent virtual
address (DR6 and DR7 are definitely not).

No functional change intended.

Change suggested by <christos>

Sponsored by <The NetBSD Foundation>

(kamil)

- Change to use 2500Base-KX correctly on C2000(I354). It worked, but the
output of ifconfig and if_baudrate was not good. Tested by nils@:
  - The STATUS_TBIMODE bit in the STATUS register is deleted since 82575,
    so check for 82575 and newer first and then check for old devices.
  - Check the 2P5_SKU and 2P5_SKU_OVER bit for KX.
  - Set IFM_2500_SX instead of IFM_1000_SX for 2.5G.
- Check SERDES's speed directly from the PCS layer (PCS_LSTS register) for old
  devices.
- Style fix.

(msaitoh)

Add a bit for Mobile IPv6.

(rjs)

Add initial TODO list for the ptrace(2) (and procfs) interface.

Sponsored by <The NetBSD Foundation>

(kamil)

Remove unnecessary inclusions of nd6.h

(ozaki-r)

1323, 1324

(snj)

1323, 1324

(snj)

Pull up following revision(s) (requested by kre in ticket #1324):
distrib/sets/lists/base/mi: 1.1125, 1.1128, 1.1132, 1.1137, 1.1139, 1.1140
external/public-domain/tz/dist/CONTRIBUTING: up to 1.1.1.3
external/public-domain/tz/dist/LICENSE: up to 1.1.1.2
external/public-domain/tz/dist/Makefile: up to 1.1.1.16
external/public-domain/tz/dist/NEWS: up to 1.1.1.16
external/public-domain/tz/dist/README: up to 1.1.1.5
external/public-domain/tz/dist/TZDATA_VERSION: up to 1.6
external/public-domain/tz/dist/Theory: up to 1.1.1.9
external/public-domain/tz/dist/africa: up to 1.1.1.9
external/public-domain/tz/dist/antarctica: up to 1.1.1.7
external/public-domain/tz/dist/asia: up to 1.1.1.15
external/public-domain/tz/dist/australasia: up to 1.1.1.10
external/public-domain/tz/dist/backward: up to 1.1.1.6
external/public-domain/tz/dist/backzone: up to 1.1.1.10
external/public-domain/tz/dist/etcetera: up to 1.1.1.2
external/public-domain/tz/dist/europe: up to 1.1.1.16
external/public-domain/tz/dist/factory: up to 1.1.1.2
external/public-domain/tz/dist/iso3166.tab: up to 1.1.1.4
external/public-domain/tz/dist/leap-seconds.list: up to 1.1.1.6
external/public-domain/tz/dist/leapseconds: up to 1.1.1.7
external/public-domain/tz/dist/northamerica: up to 1.1.1.14
external/public-domain/tz/dist/southamerica: up to 1.1.1.10
external/public-domain/tz/dist/version: up to 1.1.1.3
external/public-domain/tz/dist/zone.tab: up to 1.1.1.10
external/public-domain/tz/dist/zone1970.tab: up to 1.1.1.12
external/public-domain/tz/share/zoneinfo/Makefile: up to 1.3
external/public-domain/tz/tzdata2netbsd: up to 1.11
Update tzdata to 2016j.

(snj)

Pull up following revision(s) (requested by kre in ticket #1324):
distrib/sets/lists/base/mi: 1.1125, 1.1128, 1.1132, 1.1137, 1.1139, 1.1140
external/public-domain/tz/dist/CONTRIBUTING: up to 1.1.1.3
external/public-domain/tz/dist/LICENSE: up to 1.1.1.2
external/public-domain/tz/dist/Makefile: up to 1.1.1.16
external/public-domain/tz/dist/NEWS: up to 1.1.1.16
external/public-domain/tz/dist/README: up to 1.1.1.5
external/public-domain/tz/dist/TZDATA_VERSION: up to 1.6
external/public-domain/tz/dist/Theory: up to 1.1.1.9
external/public-domain/tz/dist/africa: up to 1.1.1.9
external/public-domain/tz/dist/antarctica: up to 1.1.1.7
external/public-domain/tz/dist/asia: up to 1.1.1.15
external/public-domain/tz/dist/australasia: up to 1.1.1.10
external/public-domain/tz/dist/backward: up to 1.1.1.6
external/public-domain/tz/dist/backzone: up to 1.1.1.10
external/public-domain/tz/dist/etcetera: up to 1.1.1.2
external/public-domain/tz/dist/europe: up to 1.1.1.16
external/public-domain/tz/dist/factory: up to 1.1.1.2
external/public-domain/tz/dist/iso3166.tab: up to 1.1.1.4
external/public-domain/tz/dist/leap-seconds.list: up to 1.1.1.6
external/public-domain/tz/dist/leapseconds: up to 1.1.1.7
external/public-domain/tz/dist/northamerica: up to 1.1.1.14
external/public-domain/tz/dist/southamerica: up to 1.1.1.10
external/public-domain/tz/dist/version: up to 1.1.1.3
external/public-domain/tz/dist/zone.tab: up to 1.1.1.10
external/public-domain/tz/dist/zone1970.tab: up to 1.1.1.12
external/public-domain/tz/share/zoneinfo/Makefile: up to 1.3
external/public-domain/tz/tzdata2netbsd: up to 1.11
Update tzdata to 2016j.

(snj)

Pull up following revision(s) (requested by kre in ticket #1323):
lib/libc/time/private.h: patch
lib/libc/time/zic.c: patch
Make zic properly parse newer tzdata files.

(snj)

Pull up following revision(s) (requested by kre in ticket #1323):
lib/libc/time/private.h: patch
lib/libc/time/zic.c: patch
Make zic properly parse newer tzdata files.

(snj)

Constify ifp of if_is_deactivated

(ozaki-r)

MP-safe pppoe(4).

Nearly all parts is implemented by Shoichi YAMAGUCHI<s-yamaguchi@IIJ>, thanks.

(knakahara)

Bump for drm2 da_fb_linebytes

Welcome to 7.99.47

(maya)

wtf tfw

Courtesy of i3enedek.

(riastradh)

Handle functions that use the offset from either kinfo_proc2/kinfo_lwp properly.

(christos)

add da_fb_linebytes to drmfb_attach_args and use it to pass linebytes
from nouveau code to drmfb. keep the same linebytes logic for i915.

nvidia hardware needs 256 byte alignment, so aligning to just 64 was
not enough.

fixes broken console with a width of 1440px (PR kern/51181)
ok riastradh

(maya)

mknative, binutils 2.27, and ia64.

(mrg)

Clean-up (remove redundant declarations, concentrate checks).

(flxd)

Followling an advice in a linux forum, don't update LCR1/LCR2.
With this change this CH340 usb/serial based device:
https://www.olimex.com/Products/Breadboarding/BB-CH340T/open-source-hardware
(the chip is written H340T)
works as expected. As I'm not sure if this is needed for older device,
make this change for sc_version 0x30 or newer only.
While there, match USB_PRODUCT_WINCHIPHEAD2_CH341_2 too.

(bouyer)

Regen for USB_PRODUCT_WINCHIPHEAD2_CH341_2

(bouyer)

Rename "CH341 serial/parallel" to "CH341 USB-Serial Bridge", and
add a second device id for this chip.
From FreeBSD.

(bouyer)

simple path fixes

(christos)

acknowleg -> acknowledg, proceedure -> procedure.
only comments were changed.

from miod

(maya)

Fix path to files.audio.

(nat)

Whitespace

(skrll)

WIP MPification

(skrll)

Add missing files in distribution file lists to fix evbarm64-aarch64 build

Add ./usr/libdata/ldscripts/aarch64elf{,b}.xd{,c,w}

Tested with MKLLVM=yes and HAVE_LLVM=yes.

(kamil)

Update audio man page with regard to audio changes.
OK wiz@

(nat)

Mark bell_thread as static and dead.

(joerg)

sig_atomic_t does not include volatile. Prevent static analysis from
understanding that the test function is dead.

(joerg)

Update for ticket 1304.

(msaitoh)

Improve wording. Fix typo. Remove superfluous Pp.

(wiz)

New sentence, new line.
Sort SEE ALSO.

(wiz)

Fix xref.

(wiz)

Whitespace.

(wiz)

Add three miissing files for ticket #1305 requested by snj.

(msaitoh)

fix accidentally if_pppoe atf failure depends on cpu workload.

advised by s-yamaguchi@IIJ, thanks.

(knakahara)

r-sync-to-head for ticket #1304 requested by snj.

(msaitoh)

Ticket 1280, 1304 and 1305.

(msaitoh)

Pull up following revision(s) (requested by snj in ticket #1305):
distrib/sets/lists/xbase/md.amd64: revision 1.15 via patch
distrib/sets/lists/xbase/md.i386: revision 1.29
distrib/sets/lists/xcomp/md.amd64: revision 1.43 via patch
distrib/sets/lists/xcomp/md.i386: revision 1.48 via patch
distrib/sets/lists/xcomp/mi: revision 1.175
distrib/sets/lists/xdebug/md.amd64: revision 1.22 via patch
distrib/sets/lists/xdebug/md.i386: revision 1.21 via patch
distrib/sets/lists/xserver/md.amd64: revisions 1.85-1.87
distrib/sets/lists/xserver/md.i386: revision 1.103-1.105
external/mit/xorg/lib/Makefile: revision 1.36
external/mit/xorg/lib/gallium/Makefile: revision 1.8, 1.9
external/mit/xorg/lib/libdrm/drm/Makefile: revision 1.7
external/mit/xorg/lib/libdrm_nouveau/Makefile: revision 1.1
external/mit/xorg/lib/libdrm_nouveau/shlib_version: revision 1.1
external/mit/xorg/server/drivers/Makefile: revision 1.70
external/mit/xorg/server/drivers/xf86-video-nouveau/Makefile: revisions 1.1, 1.2 via patch
share/mk/bsd.own.mk: revision 1.875 via patch
Build xf86-video-nouveau and libdrm_nouveau on x86.

(msaitoh)

Pull up following revision(s) (requested by snj in ticket #1304):
external/mit/xorg-server/dist/include/list.h: revision 1.3
add list_append() macro that xf86-video-nouveau 1.0.11 wants.

(msaitoh)

Pull up following revision(s) (requested by snj in ticket #1280):
sys/external/bsd/drm2/ttm/ttm_bus_dma.c: revision 1.2
sys/external/bsd/drm2/include/drm/drm_wait_netbsd.h: revision 1.14
sys/external/bsd/drm2/dist/drm/nouveau/nouveau_dp.c: revision 1.2
sys/external/bsd/drm2/dist/drm/ttm/ttm_tt.c: revision 1.7
sys/external/bsd/drm2/dist/drm/nouveau/core/engine/device/nouveau_engine_device_base.c: revision 1.11
sys/external/bsd/drm2/dist/drm/nouveau/core/engine/fifo/nouveau_engine_fifo_nv40.c: revision 1.2
sys/external/bsd/drm2/dist/include/drm/ttm/ttm_bo_driver.h: revision 1.3
sys/external/bsd/drm2/dist/drm/radeon/radeon_ttm.c: revision 1.8
sys/external/bsd/drm2/dist/drm/nouveau/nouveau_connector.c: revision 1.3
sys/external/bsd/drm2/dist/drm/nouveau/nouveau_gem.c: revision 1.4
sys/external/bsd/drm2/dist/drm/nouveau/core/subdev/devinit/nouveau_subdev_devinit_nv04.c: revision 1.3
sys/external/bsd/drm2/dist/drm/nouveau/nouveau_bo.c: revision 1.7
sys/external/bsd/drm2/dist/drm/nouveau/core/core/nouveau_core_object.c: revision 1.3
sys/external/bsd/drm2/dist/drm/i915/i915_gpu_error.c: revision 1.4
sys/external/bsd/drm2/dist/drm/nouveau/core/subdev/mxm/nouveau_subdev_mxm_mxms.c: revision 1.2
sys/external/bsd/drm2/dist/drm/nouveau/core/subdev/mxm/nouveau_subdev_mxm_nv50.c: revision 1.3
sys/external/bsd/drm2/dist/drm/nouveau/core/engine/disp/nouveau_engine_disp_nvd0.c: revision 1.3
sys/external/bsd/drm2/dist/drm/nouveau/core/engine/fifo/nouveau_engine_fifo_base.c: revision 1.5
sys/external/bsd/drm2/include/drm/ttm/ttm_page_alloc.h: revision 1.2
sys/external/bsd/drm2/drm/drm_vma_manager.c: revision 1.4
sys/external/bsd/drm2/nouveau/nouveau_pci.c: revision 1.8
- fix non-debug build
- Use %"PRIx8", not %hx, for uint8_t.  %hx is for unsigned short.
- Flush unused function. Found by joerg.
- Make sure rbtrees are empty on desctruction.
  If related to PR kern/51076, might help catch the bug a bit earlier.
- revert rev 1.6.  we don't have access to those files as normal files
  and possible will get them directly elsewhere soon.
  (if i hack my nouveau to ignore fireware failure, it end up having
  a vaguely working console and slowly working X, which is better than
  the current situation.)
- Avoid possible null pointer dereference. Found by Coverity, CID 709895.
- Mark fallthrough to suppress Coverity complaints.
  CID 143119
  CID 143120
  CID 143121
  CID 143122
  CID 143123
  CID 143124
- Sanity-check that the encoder we found is not null.
  I think the previous code guarantees that finding this encoder should
  work, so this should be a moot point. CID 145720.
- Mark intended fallthrough to suppress Coverity CID 201378.
- Mark fallthrough with a comment to appease Coverity, CID 703385.
- Make it clearer to Coverity that there's no array overrun. CID 989067.
- Rework ttm tt swapin/swapout logic.
  Rather than handling `swapping in/out' here, per se, we let uvm do
  that, we interpret `swap out' as `deactivate pages', and we add
  generic ttm operations to wire and unwire pages, for the ttm_tt
  driver to use.
  This fixes certain graphics buffer eviction logic, which enables
  nouveau to suspend/resume on one of my machines.  (The machine
  doesn't resume overall for other reasons, but the nouveau device
  suspends and resumes in isolation.)
- Use bus_space_subregion to get fifo channels out of mmio registers.
  Evidently it is not enough to just map them separately.  Ran out of
  time to investigate why, last time I poked at this and confirmed this
  change works.
- Avoid taking locks during interrupts and explain why we are doing it this way.
- We now use cpu_intr_p() all the time.

(msaitoh)

tickets 1301-1303, 1306-1308

(snj)

tickets 1303, 1306, and 1308

(snj)

Pull up following revision(s) (requested by riastradh in ticket #1308):
usr.sbin/mld6query/mld6.c: revision 1.15
PR/51353: Patrick Mackey: mld6query  "-r" option is not recognised
XXX: pullup 7

(snj)

Pull up following revision(s) (requested by riastradh in ticket #1308):
usr.sbin/mld6query/mld6.c: revision 1.15
PR/51353: Patrick Mackey: mld6query  "-r" option is not recognised
XXX: pullup 7

(snj)

Pull up following revision(s) (requested by riastradh in ticket #1307):
lib/libterminfo/curterm.c: revisions 1.11, 1.12
lib/libterminfo/term.c: revisions 1.18-1.20
lib/libterminfo/termcap.c: revisions 1.18, 1.19
lib/libterminfo/terminfo.3: revision 1.13
lib/libterminfo/tparm.c: revision 1.16
terminfo.3: fix ti_puts prototype
--
PR/50092: Fix memory leak.
--
PR/50092: Rin Okuyama: Fix memory leak.
--
We have the max length; use snprintf.
--
From PR/50092:
- handle calling _ti_readterm with an existing initialized terminal
- simplify free code
Also:
- fix an inconsistency in userdefs count computation
--
Always copy the area buffer, even when the length was the same
(from Rin Okuyama)
--
- if we are freeing cur_term, set it to NULL.
- preserve and free "last" properly.
--
off-by-one in memcpy. Found by ASAN (Carsten Kunze)

(snj)

Pull up following revision(s) (requested by ozaki-r in ticket #1306):
sys/kern/subr_xcall.c: revision 1.19
Fix a race condition of low priority xcall
xc_lowpri and xc_thread are racy and xc_wait may return during/before
executing all xcall callbacks, resulting in a kernel panic at worst.
xc_lowpri serializes multiple jobs by a mutex and a cv. If all xcall
callbacks are done, xc_wait returns and also xc_lowpri accepts a next job.
The problem is that a counter that counts the number of finished xcall
callbacks is incremented *before* actually executing a xcall callback
(see xc_tailp++ in xc_thread). So xc_lowpri accepts a next job before
all xcall callbacks complete and a next job begins to run its xcall callbacks.
Even worse the counter is global and shared between jobs, so if a xcall
callback of the next job completes, the shared counter is incremented,
which confuses wc_wait of the previous job as all xcall callbacks of the
previous job are done and wc_wait of the previous job returns during/before
executing its xcall callbacks.
How to fix: there are actually two counters that count the number of finished
xcall callbacks for low priority xcall for historical reasons (I guess):
xc_tailp and xc_low_pri.xc_donep. xc_low_pri.xc_donep is incremented correctly
while xc_tailp is incremented wrongly, i.e., before executing a xcall callback.
We can fix the issue by dropping xc_tailp and using only xc_low_pri.xc_donep.
PR kern/51632

(snj)

Pull up following revision(s) (requested by ozaki-r in ticket #1306):
sys/kern/subr_xcall.c: revision 1.19
Fix a race condition of low priority xcall
xc_lowpri and xc_thread are racy and xc_wait may return during/before
executing all xcall callbacks, resulting in a kernel panic at worst.
xc_lowpri serializes multiple jobs by a mutex and a cv. If all xcall
callbacks are done, xc_wait returns and also xc_lowpri accepts a next job.
The problem is that a counter that counts the number of finished xcall
callbacks is incremented *before* actually executing a xcall callback
(see xc_tailp++ in xc_thread). So xc_lowpri accepts a next job before
all xcall callbacks complete and a next job begins to run its xcall callbacks.
Even worse the counter is global and shared between jobs, so if a xcall
callback of the next job completes, the shared counter is incremented,
which confuses wc_wait of the previous job as all xcall callbacks of the
previous job are done and wc_wait of the previous job returns during/before
executing its xcall callbacks.
How to fix: there are actually two counters that count the number of finished
xcall callbacks for low priority xcall for historical reasons (I guess):
xc_tailp and xc_low_pri.xc_donep. xc_low_pri.xc_donep is incremented correctly
while xc_tailp is incremented wrongly, i.e., before executing a xcall callback.
We can fix the issue by dropping xc_tailp and using only xc_low_pri.xc_donep.
PR kern/51632

(snj)

Pull up following revision(s) (requested by mrg in ticket #1303):
lib/libc/stdio/funopen.c: revision 1.15
PR/51572: calling funopen on each printw call is a tad excessive, but this
is not a reason for funopen to leak. Always call the close function since we
are indirecting through our cookie and we need to free it.

(snj)

Pull up following revision(s) (requested by mrg in ticket #1303):
lib/libc/stdio/funopen.c: revision 1.15
PR/51572: calling funopen on each printw call is a tad excessive, but this
is not a reason for funopen to leak. Always call the close function since we
are indirecting through our cookie and we need to free it.

(snj)

Pull up following revision(s) (requested by msaitoh in ticket #1302):
sys/dev/mii/igphyreg.h: revisions 1.7-1.10
sys/dev/mii/ikphyreg.h: revisions 1.3
sys/dev/mii/inbmphyreg.h: revisions 1.4-1.9
sys/dev/mii/mii.h: revisions 1.19-1.20
sys/dev/pci/if_wm.c: revisions 1.390, 1.392-1.395, 1.397, 1.419-1.425, 1.427-1.428, 1.430-1.435, 1.437-1.453 via patch
sys/dev/pci/if_wmreg.: revisions 1.89-1.93 via patch
sys/dev/pci/if_wmvar.h: revisions 1.31-1.32
Update wm(4) up to if_wm.c rev. 1.453 except MSI/MSI-X, multiqueue and
NET_MPSAFE:
- Add I219 support. It's not stable so it's disabled by default.
- wm_gate_hw_phy_config_ich8lan() is for younger than PCH2.
- Drop the host wakeup bit after resetting PHY on PCH and newer
  devices.
- Increase delay while toggling LANPHYPC
- Move call of wm_reset() in wm_attach() after setting PHY and NVM
  related flags because those flags are used in wm_reset().
- Use mutex for NVM access on ICH8 and newer devices. Same as FreeBSD.
- Rewrite PHY related lock stuff. Almost the same as FreeBSD.
  This change will fix a bug that PHY read/write fail on some cases.
- Increase delay in wm_phy_resetisblocked(). Same as FreeBSD.
- Use semaphore in wm_hv_phy_workaround_ich8lan() and
  wm_k1_gig_workaround_hv()
- Use wm_gii_mdic_readreg/writereg() in wm_access_phy_wakeup_reg_bm()
  because these functions are called with taking lock.
- 82567V_3 is BME1000_E_2(bm). Tested with Advantech AIMB-212 1st
  Ethernet port.
- Use wm_gmii_82544_{read,write}reg() on non-82567 ICH8, 9 and 10.
- Remove an 82578 workaround which was for PCH rev < 3. FreeBSD
  removed this workaround in r228386.
- Add an 82578 workaround which is for PHY rev < 2. From FreeBSD and
  Linux.
- Fix wm(4) input drop packet counter. WMREG_RNBC is incremented when
  there is no available buffers in host memory. However, ethernet
  controller can receive packets in such case if there is space in
  phy's FIFO. That is, ethernet controller drops packet only if there
  is no available buffers *and* there is no space in phy's FIFO. So,
  the number of dropped packets should be added WMREG_MPC only.
- Use MII_ADDRMASK.
- Define WMPHY_I217, WMPHY_VF and WMPHY_210.
- Use BME1000_PHY_PAGE_SELECT in wm_gmii_bm_{read,write}reg(). This
  change has no effect because GG82563_PHY_PAGE_SELECT and
  BME1000_PHY_PAGE_SELECT have the same value.
- Fix PHY access on  82567(ICH8 or ICH10), 82574 and 82583:
  - Use wm_gmii_bm_{read,write}reg() on 82574 and 82573.
  - Issue page select correctly on BM PHYs.
- Fix workaround which did dummy read BM_WUC register. This code was
  changed to drop BM_WUC_HOST_WU_BIT of BM_PROT_GEN_CFG register in
  FreeBSD r228386. The code was added rev. 1.149, but the location was
  not the best.
- wm_gmii_hv_{read/write}reg*(): USE PHY address 1 for some special
  registers.
- Add check code for an 82578 workaround. Not completed yet.
- wm_release_hw_control(): Remove extra line. No any effect.
- Add "10/100" into non-gigabit devices' name.
- Call wm_enable_wakeup() in wm_detach() and wm_suspend(). Now wake on
  lan works on Thinkpad X61(ICH8).
- Fix wm_access_phy_wakeup_reg_bm(). This change has no effect because
  this function is used for WUC register and our driver currenlty
  doesn't access to it.
- Call wm_enable_phy_wakeup() on PCH2 and newer, too. Now these devices
  can do WOL. Tested with Thinkpad X220(PCH2).
- Set CTRL_MEHE correctly (PCH_{LPT,SPT} only).
- Add three workarounds for PCH_{LPT,SPT}.
- Fix a bug that 8257[56], 82580, I35[04] and I21[01] didn't use
  wm_{get,release}_hw_control() correctly.
- Sync wm_smbustopci() with Linux and FreeBSD. This change effects PCH
  and newer devices.
- Move the location of wm_smbustopci() call.
- Fix flag check in wm_get_wakeup()
  - 8254[17]* and 8257[124] should not set WM_F_ARC_SUBSYS_VALID.
  - Add missing WM_T_82541_2 and WM_T_82547_2.
- Fix WOL related setting of the WUC register for other than PCH* in
  wm_enable_wakeup(). Tested with 82567V(ICH8) and 82583V.
- Use common MII_ADDRMASK.
- igphy(4): No binary change:
  - s/IGPPHY/IGPHY/
  - Fix the definition of PLHR_VALID_CHANNEL_*
  - Fix the definition of MSE_CHANNEL_*
  - Add MII_IGPHY_POWER_MGMT.
- Add some KASSERT.
- Add comment. Modify comment.
- Add debug code.

(snj)

Pull up following revision(s) (requested by msaitoh in ticket #1301):
sys/dev/mii/inphy.c: revision 1.55
sys/dev/mii/makphy.c: revision 1.42
sys/dev/mii/ikphy.c: revision 1.12
sys/dev/mii/atphy.c: revision 1.18
sys/dev/mii/ihphy.c: revision 1.10
Set mii_mpd_{oui,model,rev}.
--
Set mii_mpd_{oui,model,rev}.
--
Set mii_mpd_{oui,model,rev}.
--
Set mii_mpd_{oui,rev}.

(snj)

fix placement of lint comment

(christos)

Bump for MP-safe routing table

Welcome to 7.99.46

(ozaki-r)

Make the routing table and rtcaches MP-safe

See the following descriptions for details.

Proposed on tech-kern and tech-net

Overview
--------

We protect the routing table with a rwock and protect
rtcaches with another rwlock. Each rtentry is protected
from being freed or updated via reference counting and psref.

Global rwlocks
--------------

There are two rwlocks; one for the routing table (rt_lock) and
the other for rtcaches (rtcache_lock). rtcache_lock covers
all existing rtcaches; there may have room for optimizations
(future work).

The locking order is rtcache_lock first and rt_lock is next.

rtentry references
------------------

References to an rtentry is managed with reference counting
and psref. Either of the two mechanisms is used depending on
where a rtentry is obtained. Reference counting is used when
we obtain a rtentry from the routing table directly via
rtalloc1 and rtrequest{,1} while psref is used when we obtain
a rtentry from a rtcache via rtcache_* APIs. In both cases,
a caller can sleep/block with holding an obtained rtentry.

The reasons why we use two different mechanisms are (i) only
using reference counting hurts the performance due to atomic
instructions (rtcache case) (ii) ease of implementation;
applying psref to APIs such rtaloc1 and rtrequest{,1} requires
additional works (adding a local variable and an argument).

We will finally migrate to use only psref but we can do it
when we have a lockless routing table alternative.

Reference counting for rtentry
------------------------------

rt_refcnt now doesn't count permanent references such as for
rt_timers and rtcaches, instead it is used only for temporal
references when obtaining a rtentry via rtalloc1 and rtrequest{,1}.
We can do so because destroying a rtentry always involves
removing references of rt_timers and rtcaches to the rtentry
and we don't need to track such references. This also makes
it easy to wait for readers to release references on deleting
or updating a rtentry, i.e., we can simply wait until the
reference counter is 0 or 1. (If there are permanent references
the counter can be arbitrary.)

rt_ref increments a reference counter of a rtentry and rt_unref
decrements it. rt_ref is called inside APIs (rtalloc1 and
rtrequest{,1} so users don't need to care about it while
users must call rt_unref to an obtained rtentry after using it.

rtfree is removed and we use rt_unref and rt_free instead.
rt_unref now just decrements the counter of a given rtentry
and rt_free just tries to destroy a given rtentry.

See the next section for destructions of rtentries by rt_free.

Destructions of rtentries
-------------------------

We destroy a rtentry only when we call rtrequst{,1}(RTM_DELETE);
the original implementation can destroy in any rtfree where it's
the last reference. If we use reference counting or psref, it's
easy to understand if the place that a rtentry is destroyed is
fixed.

rt_free waits for references to a given rtentry to be released
before actually destroying the rtentry. rt_free uses a condition
variable (cv_wait) (and psref_target_destroy for psref) to wait.

Unfortunately rtrequst{,1}(RTM_DELETE) can be called in softint
that we cannot use cv_wait. In that case, we have to defer the
destruction to a workqueue.

rtentry#rt_cv, rtentry#rt_psref and global variables
(see rt_free_global) are added to conduct the procedure.

Updates of rtentries
--------------------

One difficulty to use refcnt/psref instead of rwlock for rtentry
is updates of rtentries. We need an additional mechanism to
prevent readers from seeing inconsistency of a rtentry being
updated.

We introduce RTF_UPDATING flag to rtentries that are updating.
While the flag is set to a rtentry, users cannot acquire the
rtentry. By doing so, we avoid users to see inconsistent
rtentries.

There are two options when a user tries to acquire a rtentry
with the RTF_UPDATING flag; if a user runs in softint context
the user fails to acquire a rtentry (NULL is returned).
Otherwise a user waits until the update completes by waiting
on cv.

The procedure of a updater is simpler to destruction of
a rtentry. Wait on cv (and psref) and after all readers left,
proceed with the update.

Global variables (see rt_update_global) are added to conduct
the procedure.

Currently we apply the mechanism to only RTM_CHANGE in
rtsock.c. We would have to apply other codes. See
"Known issues" section.

psref for rtentry
-----------------

When we obtain a rtentry from a rtcache via rtcache_* APIs,
psref is used to reference to the rtentry.

rtcache_ref acquires a reference to a rtentry with psref
and rtcache_unref releases the reference after using it.
rtcache_ref is called inside rtcache_* APIs and users don't
need to take care of it while users must call rtcache_unref
to release the reference.

struct psref and int bound that is needed for psref is
embedded into struct route. By doing so we don't need to
add local variables and additional argument to APIs.

However this adds another constraint to psref other than
reference counting one's; holding a reference of an rtentry
via a rtcache is allowed by just one caller at the same time.
So we must not acquire a rtentry via a rtcache twice and
avoid a recursive use of a rtcache. And also a rtcache must
be arranged to be used by a LWP/softint at the same time
somehow. For IP forwarding case, we have per-CPU rtcaches
used in softint so the constraint is guaranteed. For a h
rtcache of a PCB case, the constraint is guaranteed by the
solock of each PCB. Any other cases (pf, ipf, stf and ipsec)
are currently guaranteed by only the existence of the global
locks (softnet_lock and/or KERNEL_LOCK). If we've found the
cases that we cannot guarantee the constraint, we would need
to introduce other rtcache APIs that use simple reference
counting.

psref of rtcache is created with IPL_SOFTNET and so rtcache
shouldn't used at an IPL higher than IPL_SOFTNET.

Note that rtcache_free is used to invalidate a given rtcache.
We don't need another care by my change; just keep them as
they are.

Performance impact
------------------

When NET_MPSAFE is disabled the performance drop is 3% while
when it's enabled the drop is increased to 11%. The difference
comes from that currently we don't take any global locks and
don't use psref if NET_MPSAFE is disabled.

We can optimize the performance of the case of NET_MPSAFE
on by reducing lookups of rtcache that uses psref;
currently we do two lookups but we should be able to trim
one of two. This is a future work.

Known issues
------------

There are two known issues to be solved; one is that
a caller of rtrequest(RTM_ADD) may change rtentry (see rtinit).
We need to prevent new references during the update. Or
we may be able to remove the code (perhaps, need more
investigations).

The other is rtredirect that updates a rtentry. We need
to apply our update mechanism, however it's not easy because
rtredirect is called in softint and we cannot apply our
mechanism simply. One solution is to defer rtredirect to
a workqueue but it requires some code restructuring.

(ozaki-r)