Only do the pci_conf_write if PCI_COMMAND_MASTER_ENABLE isn't already set.

(matt)

Only probe the upper BAR of a mem64 if the bit31 of the lower BAR isn't
writable.

(matt)

If the default path doesn't result in a match, bail out early instead of
running into a segmentation fault. Based on patch by Abhinav Upadhyay.

(joerg)

Display current mount point, rather than previous one, when printing
the "replaying log to disk" message.

OK dholland@

Fixes PR kern/39609

(pgoyette)

Set the total memory length to be ALL the space *before* the kernel.

(reinoud)

Use M_WAITOK instead of M_NOWAIT for the stack allocations

(reinoud)

Rename the debug printf's to use a thunk_ prefix to avoid confusion.

(reinoud)

Fix logic that checks if its the kernel or a userland process that causes the
trap. Signal sending still needs to be addressed.

(reinoud)

Use device_t instead of 'struct device *'.
Call aprint_* in auto-config time.

(kiyohara)

Fix the CPU #ifdefs

(skrll)

Tidy up the M[FT]CPU_* hv-specific instructions some more.

Same code before and after change.

(skrll)

Welcome to 5.1.1!

(snj)

If MIPS64R2_XLP, just let MIPSX(exception) do the work on cache exceptions.
(this might be generalized to all mipsNN with more testing).

(matt)

support anti-aliased fonts in 32bit colour
enable with options RADEONFB_DEPTH_32

(macallan)

1707

(snj)

Pull up following revision(s) (requested by is in ticket #1707):
xfree/xc/programs/Xserver/mi/misprite.c: revision 1.2
xfree/xc/programs/Xserver/mi/miarc.c: revision 1.2
xfree/xc/programs/Xserver/fb/fbpict.c: revision 1.5
This fix is taken from xorg-server 1.9.2.
mod(a,b) used to be defined with a - in front of naked a, such that uses
of mod with certain arithmetic expressions as a led to surprising results,
namely the one in Xrender praised in CVE-2010-1166.

(snj)

Note that the files in ticket 1707 are xsrc.

(snj)

1707

(snj)

Pull up following revision(s) (requested by is in ticket #1707):
xfree/xc/programs/Xserver/mi/misprite.c: revision 1.2
xfree/xc/programs/Xserver/mi/miarc.c: revision 1.2
xfree/xc/programs/Xserver/fb/fbpict.c: revision 1.5
This fix is taken from xorg-server 1.9.2.
mod(a,b) used to be defined with a - in front of naked a, such that uses
of mod with certain arithmetic expressions as a led to surprising results,
namely the one in Xrender praised in CVE-2010-1166.

(snj)

1707

(snj)

Pull up following revision(s) (requested by is in ticket #1707):
xfree/xc/programs/Xserver/mi/misprite.c: revision 1.2
xfree/xc/programs/Xserver/mi/miarc.c: revision 1.2
xfree/xc/programs/Xserver/fb/fbpict.c: revision 1.5
This fix is taken from xorg-server 1.9.2.
mod(a,b) used to be defined with a - in front of naked a, such that uses
of mod with certain arithmetic expressions as a led to surprising results,
namely the one in Xrender praised in CVE-2010-1166.

(snj)

Remove dead variable

(liamjfoy)

* Remove PGO_RECLAIM during lfs_putpages()' call to genfs_putpages(),
  to avoid a live lock in the latter when reclaiming a vnode with
  dirty pages.

* Add a new segment flag, SEGM_RECLAIM, to note when a segment is
  being written for vnode reclamation, and record which inode is being
  reclaimed, to aid in forensic debugging.

* Add a new segment flag, SEGM_SINGLE, so that opportunistic writes
  can write a single segment's worth of blocks and then stop, rather
  than writing all the way up to the cleaner's reserved number of
  segments.

* Add assert statements to check mutex ownership is the way it ought
  to be, mostly in lfs_putpages; fix problems uncovered by this.

* Don't clear VU_DIROP until the inode actually makes its way to disk,
  avoiding a problem where dirop inodes could become separated
  (uncovered by a modified version of the "ckckp" forensic regression
  test).

* Move the vfs_getopsbyname() call into lfs_writerd.  Prepare code to
  make lfs_writerd notice when there are no more LFSs, and exit losing
  the reference, so that, in theory, the module can be unloaded.  This
  code is not enabled, since it causes a crash on exit.

* Set IN_MODIFIED on inodes flushed by lfs_flush_dirops.  Really we
  only need to set IN_MODIFIED if we are going to write them again
  (e.g., to write pages); need to think about this more.

Finally, several changes to help avoid "no clean segments" panics:

* In lfs_bmapv, note when a vnode is loaded only to discover whether
  its blocks are live, so it can immediately be recycled.  Since the
  cleaner will try to choose ~empty segments over full ones, this
  prevents the cleaner from (1) filling the vnode cache with junk, and
  (2) squeezing any unwritten writes to disk and running the fs out of
  segments.

* Overestimate by half the amount of metadata that will be required
  to fill the clean segments.  This will make the disk appear smaller,
  but should help avoid a "no clean segments" panic.

* Rearrange lfs_writerd.  In particular, lfs_writerd now pays
  attention to the number of clean segments available, and holds off
  writing until there is room.

(perseant)

Remove tests that are dubious and should be (re)moved anyway.

(reinoud)

Add UDF, NILFS and NFS to the standard file systems and boost the symbol table
size to accomodate for this.

(reinoud)

In lfs_cleanerd, check errno earlier, to avoid premature exit when the
real errno was EAGAIN.

When coalescing, check against NULL to avoid seg fault.

(perseant)

Remove unnecessary cast to caddr_t.

(dholland)

stdlib.h, not malloc.h

(dholland)

Don't include "xorg.cf" twice for NetBSD/sparc and NetBSD/sparc64.
This avoid a lot redefinition warnings from "imake" which are e.g.
triggered by pkgsrc's checks.

(tron)

Bump date for previous. Move RCS Id to top of file.

(wiz)

Remove trailing whitespace.

(wiz)

Use more markup. New sentence, new line. Sort sections.
Bump date for previous.

(wiz)

New sentence, new line. Bump date for previous.

(wiz)

From OpenBSD...

Fix encoding of DIAG instructions in the MFCPU_T and MTCPU_T macros. MTCPU_T
was encoded as a read instruction, while MFCPU_T was encoded as a
read-through-TH instruction, although the only uses of it concern registers
read through CH.

(You are not supposed to understand this, this is a dark corner of undocumented
rotten PA-RISC bowels).

Cache SID hashing will now correctly be disabled on PCXS, PCXT and PCXT'
processors.

Proper encoding from the old PA-RISC mklinux sources; tested on 720 (PCXS)
and 715/75 (PCXT); ok kettenis@

(skrll)

Clarify the behavior of exit(x) for out-of-range x. PR 45769

(dholland)

Correct the include path.

(rkujawa)

Do release all l2 pages on destroy

(reinoud)

Remove unnecessary quoting. From Abhinav Upadhyay.

(wiz)

Remove unnecessary quoting. From Abhinav Upadhyay.

(wiz)

Fix typo, from Abhinav Upadhyay.

(wiz)

fix range check for mmap

(jmcneill)

a typo and missing markup

(christos)

Fix typo in comment.

(riz)

Improve returns description from Solar Designer.

(christos)

Rewrite usermodes pmap to use indirect adressing significantly reducing pmap
memory usage. Typically a decimation. This is done using L1 and L2 mappings.

(reinoud)

When converting a floating value to an integer, properly raise the "invalid"
exception when the floating value is infinite or NaN or if the integral part
of the floating value exceeds the range of the integer type, as required by
the C99/IEEE754 standard.

Fixes Python 2.6 build on hppa.

From OpenBSD.

(skrll)

Note the fact that wsmoused will work only with some wsdisplay drivers.

(rkujawa)

Handle devices with two network interfaces (e.g. a DSM-G600 with stge(4)
and ath(4)).

(phx)

From Rainer Weikusat <rweikusat@mobileactivedefense.com>: Fix one byte too
short memory allocation in isakmp_unity.c:splitnet_list_2str().

(tteras)

From Rainer Weikusat <rweikusat@mobileactivedefense.com>: Fix one byte too
short memory allocation in isakmp_unity.c:splitnet_list_2str().

(tteras)

From Wolfgang Schmieder <wolfgang@die-schmieders.de>: Fix default NAT-T
port for listen { isakmp_natt } config directive.

(tteras)

From Wolfgang Schmieder <wolfgang@die-schmieders.de>: Fix various typos in
comments and log messages. Fix default port used in copy_ph1addresses().

(tteras)

Fix myaddr_getsport() to return -1 if no suitable address is found. This is
used in pfkey.c:pk_recvacquire() to check if IKE negotiation should be
started or not.

(tteras)

Fix the previous commit.

(tteras)

Add <sys/types.h> to <sys/quota.h> or the latter doesn't compile on
its own.

(dholland)

From Wolfgang Schmieder <wolfgang@die-schmieders.de>: Fix memory leaks from
configuration reading code, and clean up error handling.

(tteras)

Add the description for the "mem:<address>" bootname.

(phx)

Add Iomega StorCenter.

(phx)

Fix typo in comment

(reinoud)

According to the pmap manpage wired pages should not track RWM, so map them RW

(reinoud)

Tweak white space so I get my annual commit.

(wrstuden)

Welcome to 2012, Happy New Year!

Don't forget to adjust copyright date in new code.

(riz)

include stddef.h for offsetof

(christos)

According to most other OSes and documentation sources, an MBR has
one "extended" partition and the partitions inside that are "logical"
partitions.  Make fdisk(8) man page follow suit.

Patch from Julian Fagir in PR#45695.

(riz)

1705 was via patch.

(snj)

1705

(snj)

Pull up following revision(s) (requested by riz in ticket #1705):
sys/dev/vnd.c: revision 1.218
sys/dev/vndvar.h: revision 1.31
Make vnd(4) work on sparse files:
- Make the strategy decision a device flag and set VNF_USE_VN_RDWR for
  files known to be sparse.
- Change handle_with_rdwr() to use POSIX_FADV_NOREUSE advise to disable
  read ahead and keep the size of mapped pages below 1 MByte.
No objections on tech-kern@.

(snj)

Adjust prototypes.

(christos)

- fix syscall error returns
- fix clobbered errno
- use warn instead of perror
- use ssize_t consistently
- retry on EINTR and EAGAIN

(christos)

pass ${COPTS} to the build of thunk.c so that it builds on amd64.

(christos)

Belatedly note hannken's addition of sparse file support to vnd(4) from
June.

(riz)

1440

(snj)

Pull up following revision(s) (requested by christos in ticket #1440):
lib/libtelnet/encrypt.c: revision 1.15
Avoid buffer overflow, reported by Colin Percival at FreeBSD

(snj)

1440

(snj)

Pull up following revision(s) (requested by christos in ticket #1440):
lib/libtelnet/encrypt.c: revision 1.15
Avoid buffer overflow, reported by Colin Percival at FreeBSD

(snj)

- fix offsetof usage, and redundant defines
- kill pointer casts to 0

(christos)

Fix typos and style glitches.  Remove outdated info about only
Ethernet, SLIP, and PPP being supported, and no network types with
variable-length headers (such as 802.11).
Document BIOCSRSIG and BIOCGRSIG.

From Guy Harris in PR#40703.

(riz)

1704

(snj)

Pull up following revision(s) (requested by christos in ticket #1704):
lib/libtelnet/encrypt.c: revision 1.15
Avoid buffer overflow, reported by Colin Percival at FreeBSD

(snj)

1704

(snj)

Pull up following revision(s) (requested by christos in ticket #1704):
lib/libtelnet/encrypt.c: revision 1.15
Avoid buffer overflow, reported by Colin Percival at FreeBSD

(snj)

1704

(snj)

Pull up following revision(s) (requested by christos in ticket #1704):
lib/libtelnet/encrypt.c: revision 1.15
Avoid buffer overflow, reported by Colin Percival at FreeBSD

(snj)

+uipad

(jakllsch)

Move the thunk_rfb_send_pending() up to where it belonged. It could disconnect
AND event could be NULL while it still would need to send... i.e. the barrier
in vncfbcopyrows()

(reinoud)

Implement safe_recv() in analog to save_send() to make sure we don't miss
bytes along and mess up the protocol.

(reinoud)

Fix the usage/help for mach frame.

(skrll)

Switch to using IST_<foo> instead of private enums.

(matt)

Add IST_EDGE_RISING and IST_EDGE_FALLING.

(matt)

XLP8xx and XLP3xx have different IRT layouts.

(matt)

Consolidate and complete PCITAGs.
Print/Set BARs for AHCI and SRIO.

(matt)

Deal with the movement of some GPIO registers on the XPL3xx.

(matt)

Fix xlgpio_pin_ctl inversion problem.

(matt)

Regen.

(matt)

Add XLP RXE and SRIO product ids.

(matt)

Fix panic triggered by pressing the caps lock key:
http://c0ff33.net/drop/bt_caps_panic.jpg

OK'd mrg.

(rkujawa)

add uipad

(christos)

Add an ipad

(christos)

Regen

(christos)

Add a tiny driver that sends the magic command to an ipad instructing it
to charge.

(christos)

Add template code for a client driven display update request. This is
currently not possible since its event parameter is not allways passed.

(reinoud)

Avoid panic on DIAGNOSTIC kernels with ktrace -p <not-existing-process>
The old logic was:

error = ktrace_common(, fp);
if (fp)
if (error)
fd_abort(, fp, );
else
fd_abort(, NULL, );

The 'if (fp)' portion really means if the op is not KTROP_CLEAR,
since the logic above always sets up fp otherwise, so change the
code to test this directly.

ktrace_common() can return an error both on the kernel thread
creation failure, which means that we should be calling fd_abort()
with fp, since nobody used the file yet and we should clear it now.
But it can also return an error because later, after the thread
creation if the process or process group was not found. In this
second case, we should be calling fd_abort with NULL, since the fp
is now used by the thread and it is going to clean it later. So
instead of checking the error from ktrace_common() to decide if we
are going to call fd_abort() with a NULL fp or not, let krace_common()
decide for us.  So the new logic becomes:

error = ktrace_common(, &fp);
if (op != KTROP_CLEAR)
fd_abort(, fp, );

Since I am here, fix a freed memory access, by setting ktd to FALSE.

(christos)

add some register definitions

(macallan)

add wsmouse support

(jmcneill)

report rfb pointer events

(jmcneill)

add mmap support

(jmcneill)

provide pmap_phys_address

(jmcneill)

Never cut-paste code from email!
Use the right count (0 -> 2) of l3 unshared userland entries for per-cpu initialisation.

(cherry)

prevent kernel from writing more than userland passed.

(christos)

PR/45760: Henning Petersen: Add Adaptec29329LPE to ahd_pci.c

(christos)

Force pae l3 page allocation for new vcpus to be < 4G, so they fit in 32bits

(cherry)

Move the per-cpu l3 page allocation code to a separate MD function. Avoids code duplication for xen PAE

(cherry)

per-cpu shadow directory pages should be updated locally via cross-calls. Do this.

(cherry)

This fix is taken from xorg-server 1.9.2.

mod(a,b) used to be defined with a - in front of naked a, such that uses
of mod with certain arithmetic expressions as a led to surprising results,
namely the one in Xrender praised in CVE-2010-1166.

(is)

this should have been committed along with the previous changes in
radeonfb.c

(macallan)

no need to poll for pending drawing ops if no client is connected

(jmcneill)

add a barrier before copyrows(), and add an RRE based fillrect functino,
use it for eraserows and erasecols

(jmcneill)

Document that overflow in P[X+k:n] terminates a filter program.
Based on a discussion with Christos on source-changes-d.

(alnsn)

Implement VNC's copyrect sending and let the copyrows use the new
vncfb_copyrecs()

(reinoud)

support wskbd bell

(jmcneill)

Bump SYMTAB_SPACE so that the contents fits again.

(he)

Remove spurious (debug) printf()

(cherry)

map Fn keys, make vt switching work

(jmcneill)

add WSDISPLAY_COMPAT_USL to make 'wsconscfg -s' work

(jmcneill)

print key event info if RFB_DEBUG

(jmcneill)

Use the encoding values of the VNC spec and add a copyrect (not used yet)

(reinoud)

Add RFB type flag to indicate what kind of request is waiting

(reinoud)

use O_ASYNC + SIGIO instead of polling for input

(jmcneill)

Rename variable Bpp to bytes_per_pixel

(reinoud)

double SIGIO_MAX_HANDLERS

(jmcneill)

Use a wrapper around the send() to make sure it will transmit EVERYTHING and
not just a part of the data

(reinoud)

optimize rfb updates by allowing queueing of multiple update messages,
dropping duplicates, and transmitting them in batches

(jmcneill)

take nrows into account when calculating the update rectangle in vncfb_copyrows

(jmcneill)

send framebuffer updates for cursor changes too

(jmcneill)

Add PA8700.

(skrll)

Add the Usermode vnc framebuffer and (vnc) remote keyboard types.

(reinoud)

G/C pmap_page_remove_locked prototype.

(skrll)

Cope with userland mappings of unmanaged pages (i.e. device memory being
mmap'ed), by properly making these mappings uncached, and correctly accounting
their removal to not get stuck with leftover pte pages.

From OpenBSD.

(skrll)

Change verbose kernel printing of kernel memory layout to increasing addresses
for clarity.

(reinoud)

Add GPIO support for XLP.
Let NAND, MMC/SD, and SPI remove their pins from the GPIO available pin mask.

(matt)

Use new xl* device names.
Add GPIO.

(matt)

Use the new xl* device names.

(matt)

Initially mark all GPIO pins as available.

(matt)

Add a field to track available GPIO pins.

(matt)

Change devices name from rmixl_* to xl*.

(matt)

Cleanup USB byte swap support.

(matt)

PR/45759: Khanh-Dang Nguyen Thu Lam: slattach(8) is not compatible with
pseudo-terminals. Fixed by ignoring the error for the DTR ioctl.

(christos)

Apply same bounds checks for BPF_LD|BPF_B|BPF_IND as for
BPF_LD|BPF_H|BPF_IND and BPF_LD|BPF_W|BPF_IND.

From FreeBSD r48548, the original r45574 had a typo.

(alnsn)

Make usage() match man page.

(wiz)

No need for sync_caches - f[di]cache already issue the sync instruction.

(skrll)

Replace the SDL based genfb driver with a wsdisplay and wskbd driver that
implements the VNC (RFB) protocol.

To enable the VNC server, add 'vnc=640x480,5900' to the kernel command line
(where 640x480 is the desired fb resolution and 5900 is the TCP port).

Screenshot of it here: http://www.netbsd.org/~jmcneill/usermode.tiff

(jmcneill)

PR/45751: Alexander Nasonov: No overflow check in BPF_LD|BPF_ABS

(christos)

rework putchar():
- get rid of engine stalls when using the blitter to draw characters
- add a wrapper for non-accelerated putchar() so we only wait for the
  engine when we actually want to scribble into video memory
- rework accelerated putchar(), should work on R3xx now but needs testing

(macallan)

finish device_t-ification

(macallan)

Make msgfile optional like other OS's

(christos)

Replaced 16 by I8259_ICU in intr_establish().

(phx)

Redo previous;  remove all the casts I added, and use PRI* macros instead.
(by popular demand - makes sense, too)

(riz)

Cast time_t to intmax_t for printf purposes, and format with %j.  Fixes
build on amd64 and probably i386 as well.

(riz)

Remove trailing whitespace. Avoid xr to itself.

(wiz)

reflect current reality about return values.

(christos)

crypt does not return NULL anymore.

(christos)

Note viaide(4) VT6410 and sandpoint Iomega Storcenter support.

(phx)

Add a workaround for the VT6410 IDE controller on the Iomega Storcenter.
Its interrupt cannot be disabled and remains asserted during the whole
device probing procedure, causing an interrupt storm.
This was fixed by establishing an edge-triggered interrupt for it, so it
will trigger only once during probing.
This workaround makes the WDC_NO_IDS option obsolete, which was removed
from the GENERIC config.

(phx)

Add VT6410 support, heavily based on a patch by Takahiro Kambe:
  http://mail-index.netbsd.org/port-i386/2007/03/30/0000.html
The VT6410 is special, because it may exist without a VIA PCI-ISA bridge.

(phx)

Remove temporary variable definition that is unused in non DIAGNOSTIC builds.

(cherry)

Improve markup, remove tailing whitespace.

(wiz)

if the cursor row passed in a device property doesn't fit on the screen just
ignore it and clear the screen

(macallan)

add a somewhat smaller Droid font
this was generated with a newer version of ttf2wsfont so diacritics are no
longer cut off

(macallan)

make anti-aliased fonts with odd widths work
The drawing engine needs lines padded to 64bit when doing colour host blits
which is of course not mentioned anywhere in the manual.

(macallan)

Add a FUSE trace facility, with statistics collection. This should help
tracking bugs and performance issues

(manu)

add 2 clause BSD license

(macallan)

tool to generate anti-aliased fonts for wscons from TTF files

(macallan)

Reverse the order of printf formatting to make it easier for compilers to
check printf arguments.
From joerg@

(christos)

- reverse the order of printf formatting to make it easier for compilers to
  check printf arguments
- add compiler checking for printf formats and fix broken ones.
From joerg@

(christos)