oga [Sun, 17 Aug 2008 16:13:10 +0000 (16:13 +0000)]
drm_load() and drm_unload() are a remnant of code meant to be shared
between the bsds. Just put them in attach/detach instead. Kills a little
bit of dead code, too.
kettenis [Sun, 17 Aug 2008 15:55:55 +0000 (15:55 +0000)]
Scale down cpu_clockrate[1] when we scale down the cpu clock frequency such
that delay(9) continues to do the right thing.
oga [Sun, 17 Aug 2008 15:31:39 +0000 (15:31 +0000)]
Kill some redundant ifdefs, they're taken care of elsewhere.
krw [Sun, 17 Aug 2008 15:14:44 +0000 (15:14 +0000)]
Force fdisk to use the same geometry as the miniroot2.5M disktab
entry, the one that the disklabel will use, rather than the vnd
default geometry. Lets the generated miniroot44.fs boot again.
Problem found by Diana Eichart. Suggestions from drahn@.
ok deraadt@
oga [Sun, 17 Aug 2008 14:37:31 +0000 (14:37 +0000)]
Make sure we can't sleep with a spinlock held
kettenis [Sun, 17 Aug 2008 14:25:19 +0000 (14:25 +0000)]
Garbage collect stupid delay loop.
mglocker [Sat, 16 Aug 2008 18:56:07 +0000 (18:56 +0000)]
If dwMaxPayloadTransferSize doesn't exactly match to an existing
endpoint, pick the next higher endpoint bandwidth.
mglocker [Sat, 16 Aug 2008 16:47:32 +0000 (16:47 +0000)]
Spacing.
mglocker [Sat, 16 Aug 2008 16:37:30 +0000 (16:37 +0000)]
Sync ehci isochronous part with NetBSD:
- Fix check for maximum bInterval value.
- Calculate frames/microframes values slightly different (but with mostly
same result finally).
espie [Sat, 16 Aug 2008 12:26:07 +0000 (12:26 +0000)]
fix
espie [Sat, 16 Aug 2008 12:25:21 +0000 (12:25 +0000)]
revert eval.c
espie [Sat, 16 Aug 2008 12:23:50 +0000 (12:23 +0000)]
not yet, committed too much
espie [Sat, 16 Aug 2008 12:21:46 +0000 (12:21 +0000)]
argument parsing should only skip spaces outside of parenthesis.
Inside matching parenthesis, keep spaces as is (use chrsave instead of
pbstr, since there's no way it can be a further macro expansion).
Fixes a long-standing issue with autoconf ( --option -> --option),
matches other m4 than gnum4
okay millert@, fries@
espie [Sat, 16 Aug 2008 12:19:49 +0000 (12:19 +0000)]
be more liberal in include handling, namely we only error out if it's
the end of the last included file AND we still have a macro to expand.
autoconf uses this in wine, and it turns out other m4 also do things that
way (not only gnu m4).
okay fries@, millert@
espie [Sat, 16 Aug 2008 10:02:32 +0000 (10:02 +0000)]
another issue for which I have a patch.
Issue reported by Vortechz Anderson <utg_vrtz@yahoo.se>
espie [Sat, 16 Aug 2008 09:57:12 +0000 (09:57 +0000)]
a new regression test, I has a diff for it.
oga [Sat, 16 Aug 2008 01:53:05 +0000 (01:53 +0000)]
Apply a light paddling with the knf stick. No binary change.
krw [Sat, 16 Aug 2008 00:26:26 +0000 (00:26 +0000)]
bios_dev -> ba_name and bios_apmp -> ba_apmp in DPRINTF statements.
i386 kernels compile with BIOS_DEBUG again.
Spotted by Mathias Schmocker.
miod [Fri, 15 Aug 2008 22:50:25 +0000 (22:50 +0000)]
Make the DZ_{READ,WRITE}_{BYTE,WORD} macros visible outside of dz.c, for
other dz attachments to be able to use them (soon).
miod [Fri, 15 Aug 2008 22:47:45 +0000 (22:47 +0000)]
Fix array index computation in the machine check handler. Gives fatal bus errors
a chance to be reported correctly...
miod [Fri, 15 Aug 2008 22:44:12 +0000 (22:44 +0000)]
In dzcninit(), do not forget to enable the receiver in addition to the
transmitter. On machines with glass console for which we do not have
a driver yet, and fall back to serial console, the PROM might not have
enabled it.
miod [Fri, 15 Aug 2008 22:41:46 +0000 (22:41 +0000)]
Move the vxt-specific clock functions frow clock.c to the cpu-specific support
file, fewer #ifdef this way.
miod [Fri, 15 Aug 2008 22:41:02 +0000 (22:41 +0000)]
Correct disassembly of the start of a function: fix display layout and skip
only the 2 bytes of the procedure entry mask, not 4 bytes.
miod [Fri, 15 Aug 2008 22:39:59 +0000 (22:39 +0000)]
Remove unused and incomplete vax_bus_t enum, and unused vaxbus_dma_get_tag()
macro and related function pointers.
miod [Fri, 15 Aug 2008 22:38:20 +0000 (22:38 +0000)]
Rename the cpu_dep hook ``steal_pages'' to ``init'', as it serves as an
early initialization routine (to enable caches, etc) while still running
physical, and does not allocate memory anymore.
(The irony in this is that forthcoming KA60 support actually steals pages
in its init function...)
miod [Fri, 15 Aug 2008 19:42:03 +0000 (19:42 +0000)]
hmac.c is needed by softraid crypto now, even on non-wlan machines.
naddy [Fri, 15 Aug 2008 15:49:08 +0000 (15:49 +0000)]
Prefer unaligned accesses over copying the rx buffer on all archs that can
handle it, not only i386. ok brad@
djm [Fri, 15 Aug 2008 14:57:19 +0000 (14:57 +0000)]
Add resolv.conf(5) option to force lookups by TCP: "options tcp"
Also Extend "nameserver" declaration syntax to support port
numbers. To avoid ambiguity these are only parsed when the address
is enclosed in square brackets, e.g. "nameserver [127.0.0.1]:5353"
Together these changes make forwarding DNS over a SSH tunnel very
easy, but unfortunately some programs in ports/ implement their
own resolvers (e.g. firefox). These will need to be modified to
support these options separately.
fixes jsing@ reyk@
ok deraadt@ millert@ krw@ + "I like it" from lots
jsing [Fri, 15 Aug 2008 14:41:21 +0000 (14:41 +0000)]
Add -s option to usage().
Prompted by jmc@
ok jmc@
djm [Fri, 15 Aug 2008 11:18:10 +0000 (11:18 +0000)]
use new shared HMAC_SHA1 code in crypto/hmac.[ch] rather than local
copy; ok hshoexer@
damien [Fri, 15 Aug 2008 08:15:27 +0000 (08:15 +0000)]
use LLC_SNAPFRAMELEN here too.
henning [Thu, 14 Aug 2008 22:29:43 +0000 (22:29 +0000)]
only update stats when we actually wrote sth, relevant for -n, ok djm
From: Pierre Riteau <pierre.riteau@gmail.com>
claudio [Thu, 14 Aug 2008 19:39:40 +0000 (19:39 +0000)]
Kill the _MCLDEREFERENCE() macro it was only used once and it should be only
used once -- in m_free(). Removed so that people don't get stupid ideas.
OK thib@
jsing [Thu, 14 Aug 2008 17:41:20 +0000 (17:41 +0000)]
If the AML SizeOf() method is called with an object reference it should
return the size of the referenced object. This fixes a panic on the ASUS
eeePC 1000H when attaching the acpiasus(4) driver.
Issue reported and fix tested by robert@
ok marco@
jsing [Thu, 14 Aug 2008 17:10:29 +0000 (17:10 +0000)]
Allow the sector size to be specified by the user when configuring a
vnd(4) device, via a new -s option to vnconfig/mount_vnd. This allows us
to create disklabels and file systems that are suitable for use on
devices that have a non-512 byte sector size (eg. CDROMs).
With help from krw@ and feedback from pedro@.
ok krw@, pedro@
damien [Thu, 14 Aug 2008 16:14:53 +0000 (16:14 +0000)]
ieee80211_derive_pmkid() is not used either
damien [Thu, 14 Aug 2008 16:10:01 +0000 (16:10 +0000)]
#ifdef notyet ieee80211_kdf() as it is not used yet (shrink kernel a bit)
damien [Thu, 14 Aug 2008 16:07:58 +0000 (16:07 +0000)]
move the HostAP bridge code and the delivery of data frames to ether_input
to a separate ieee80211_deliver_data() function.
this will later be called by the A-MSDU deaggregation routine.
damien [Thu, 14 Aug 2008 16:02:23 +0000 (16:02 +0000)]
in ieee80211_input(), call if_start() after enqueuing frames in if_snd
when acting as an access point instead of having each driver doing the
job.
tested by krw@ (ral AP) and me with several drivers.
damien [Thu, 14 Aug 2008 15:51:43 +0000 (15:51 +0000)]
put ieee80211_dump_pkt() under #ifdef IEEE80211_DEBUG
ratchov [Thu, 14 Aug 2008 15:25:16 +0000 (15:25 +0000)]
fix comments suggested by eric@, remove stupid DPRINTF()
robert [Thu, 14 Aug 2008 14:31:31 +0000 (14:31 +0000)]
add missing semicolon; ok mbalmer@, marco@
espie [Thu, 14 Aug 2008 13:06:04 +0000 (13:06 +0000)]
let's make this half useful to recover from crashed disks with wrong info.
martin [Thu, 14 Aug 2008 11:41:30 +0000 (11:41 +0000)]
convert the last traces of btoc/ctob macros
ok miod@
espie [Thu, 14 Aug 2008 10:18:34 +0000 (10:18 +0000)]
simplify S definition to just srcdir, since it appears there's no special
case anymore.
Note that you can always override it from the command line if you need it.
after discussion with deraadt@ and tedu@
ratchov [Thu, 14 Aug 2008 10:02:10 +0000 (10:02 +0000)]
in abuf_rgetblk() the ``start'' pointer may take a value beyond the
end of the buffer because periodic boundary conditions of the FIFO are
not met
ok jakemsr
ratchov [Thu, 14 Aug 2008 09:58:55 +0000 (09:58 +0000)]
move all device related stuff from aucat.c to a new dev.c file.
The new dev_xxx() routines expose a "high level" self-contained
interface to the device. At initialization, the device is opened
and two chains of aproc structures are created:
* a playback chain that exposes a (initially) empty mix
aproc to which the rest of the code can attach new
streams to be played
* record chain that exposes a (initially) empty sub aproc
to which the rest of the code can attach new stream to
to record
The rest of the code, has just to use dev_attach() routine to
attach streams. While we're at it, add a ``devops'' structure
containing pointers to the device-specific routines. This will
allow later to add support for other type of device than the
Sun API.
Also, write the .wav headers in file_del(), so put all header
related data in the file strucuture. This allows to close() the
file, as soon as wpipe_xxx() aproc terminates. This will be
useful for the server, because it will need to close() descripts
of closed connections immediately.
add mix_pushzero() routine to fill the mixer with silence. It
will be used to avoid the mixer to underrun when there are no
input streams. Since we always have at least one input stream
there's no behaviour change.
ok jakemsr
ratchov [Thu, 14 Aug 2008 09:48:50 +0000 (09:48 +0000)]
in file.c, before dereferencing pointers to in(), out(), eof(),
hup() routines of the aproc strucure check that the aproc
structure has not desappeared. This never happens currently, but
will be allowed later. No behaviour change.
ok jakemsr
ratchov [Thu, 14 Aug 2008 09:47:51 +0000 (09:47 +0000)]
add an "AUTOQUIT" flag to mix and sub aprocs. If the flag is
set, the mix aproc will exit once there are no more input
streams, similarly the sub aproc will exit once there are no
more ouput streams. If the flag is not set, the mix aproc will
generate silence, and the sub aproc will drop samples. By
default this flag is set, so no behaviour change.
ok jakemsr
ratchov [Thu, 14 Aug 2008 09:46:36 +0000 (09:46 +0000)]
add a bytes counter to abuf structures to count the number of
bytes that entered the FIFO. The counter may overflow, so it
should be used with "modulo 2^32" arithmetic. The counter will
be used later to synchronize playback to record. No behaviour
change.
ok jakemsr
ratchov [Thu, 14 Aug 2008 09:45:23 +0000 (09:45 +0000)]
add a xxx_done() method to aproc structures, that cleans up just
before free()ing the aproc structure, this is cleaner and will
allow to reuse wpipe_xxx() and rpipe_xxx() when defining new
aprocs. No behaviour change.
ok jakemsr
ratchov [Thu, 14 Aug 2008 09:44:15 +0000 (09:44 +0000)]
move mix/sub underrun/overrun handling code and data in the
generic abuf structure, so it can reused. Required for an audio
server. No behaviour change.
ok jakemsr
ratchov [Thu, 14 Aug 2008 09:39:16 +0000 (09:39 +0000)]
factor code that discards data from abuf strucure into a single
abuf_rdiscard() routine. Similarly add abuf_wcommit() routine for
writing. The purpose is to stop manupulating FIFO pointers in
various places outiside abuf.c, its too error prone.
No behaviour change.
ok jakemsr
guenther [Thu, 14 Aug 2008 05:57:06 +0000 (05:57 +0000)]
Match libpthread's behavior and make pthread_join(NULL, whatever) fail
instead of crashing
ok kurt@
first observed by Jung <moorang at gmail.com>
guenther [Thu, 14 Aug 2008 05:20:44 +0000 (05:20 +0000)]
Fix 5771/library: in pthread_exit(), delay the call to _sem_post() that
unblocks the pthread_join() for this thread until the exiting thread is
completely done with its thread structure, as the joining thread will
free it once unblocked. Also, don't bother to call _sem_post() if the
thread is detached.
ok kurt@
guenther [Thu, 14 Aug 2008 05:15:41 +0000 (05:15 +0000)]
If the initial thread calls pthread_exit(), don't overwrite its thread
structure, as the 'tid' member there is used by other parts of
librthread to determine whether the current thread is the initial thread
sthen [Thu, 14 Aug 2008 00:59:50 +0000 (00:59 +0000)]
Tools from /usr may not be used in netstart since it may be NFS-mounted
and not available at that time. Rewrite the hostname.if permission check
to use only /bin/ls and the shell. Requested by deraadt.
ok todd, "Twisted." deraadt
oga [Wed, 13 Aug 2008 21:56:50 +0000 (21:56 +0000)]
Make sure that we don't sleep with a lock held, some small cleanup while
i'm here.
oga [Wed, 13 Aug 2008 20:45:42 +0000 (20:45 +0000)]
Check for zero in drm_calloc() to prevent divide-by-zero. Also rearrange
the two calls to drm_calloc() that had nmemb and size the wrong way
around.
oga [Wed, 13 Aug 2008 20:38:26 +0000 (20:38 +0000)]
Kill file_priv->refs, it is always 1. Also move two things from the
softc into file_priv since otherwise the wrong value could rarely be
used.
mglocker [Wed, 13 Aug 2008 20:29:34 +0000 (20:29 +0000)]
Add VIDIOC_ENUM_FRAMESIZES ioctl. This permits applications to query the
available formats and resolutions of a device (e.g. "luvcview -L").
oga [Wed, 13 Aug 2008 20:27:58 +0000 (20:27 +0000)]
Change the drm_lock*() api to take a pointer to the struct as a
parameter, this makes things a little cleaner and removes the need to
pass in the softc for some functions.
oga [Wed, 13 Aug 2008 20:22:55 +0000 (20:22 +0000)]
idgen32 wasn't really right here. Change back to how it was, but
done a little nicer.
Advice from deraadt@.
oga [Wed, 13 Aug 2008 19:33:29 +0000 (19:33 +0000)]
Generate the magic id using idgen32(). While i'm here fix the locking a
bit so we don't sleep with a spinlock.
ok djm@.
oga [Wed, 13 Aug 2008 19:30:11 +0000 (19:30 +0000)]
Kill a bunch of the drm context code that nothing uses anymore, the
xserver only actually uses a small part of it nowadays.
Kill a few other dead bits from the drmP.h while i'm here.
kettenis [Wed, 13 Aug 2008 19:30:02 +0000 (19:30 +0000)]
Initial stab at a driver for the integrated random number generator of the
UltraSPARC T2.
damien [Wed, 13 Aug 2008 17:44:45 +0000 (17:44 +0000)]
one should never use sizeof(struct llc) since 802.2 LLC headers
have a variable length. use LLC_SNAPFRAMELEN instead (which happen
to be equal to sizeof(struct llc)) as we only deal with LLC+SNAP
headers.
some indentation tweaks while i'm here.
no binary change.
damien [Wed, 13 Aug 2008 17:38:02 +0000 (17:38 +0000)]
do not increment is_rx_eapol_key twice for each frame.
fix a test on the mbuf length and one on the descriptor version.
remove redundant checking of GTK KDE length.
sobrado [Wed, 13 Aug 2008 17:24:54 +0000 (17:24 +0000)]
sync synopsis/usage; fix double "usage:"; KNF.
weingart [Wed, 13 Aug 2008 16:01:08 +0000 (16:01 +0000)]
This tab had bugged me forever.
art [Wed, 13 Aug 2008 15:46:21 +0000 (15:46 +0000)]
Disable the fantastics mis-feature on some newer Turion CPUs called C1E.
This "power saving" disables the apic when both cpu cores hit the hlt
instruction which kills our timer.
From FreeBSD.
(poked by kettenis)
thib [Wed, 13 Aug 2008 15:45:30 +0000 (15:45 +0000)]
remove a bunch of bogus ARGSUSED
art [Wed, 13 Aug 2008 15:44:44 +0000 (15:44 +0000)]
Disable the fantastics mis-feature on some newer Turion CPUs called C1E.
This "power saving" disables the apic when both cpu cores hit the hlt
instruction which kills our timer.
From FreeBSD.
av [Wed, 13 Aug 2008 12:21:19 +0000 (12:21 +0000)]
Old drives (reported Giovanni Bechis <bigionews@snb.it>) don't report CD-RW
write feature. Add additional check for media type before blanking.
The patch fixes problem for Giovanni.
ok fgsch
jmc [Wed, 13 Aug 2008 06:20:48 +0000 (06:20 +0000)]
-S currently does nothing useful, so remove it from
SYNOPSIS and usage(); after some discussion with canacar
brad [Wed, 13 Aug 2008 03:59:16 +0000 (03:59 +0000)]
Add more Config mode registers.
From FreeBSD
brad [Wed, 13 Aug 2008 03:47:16 +0000 (03:47 +0000)]
Add the 8139D PCI id.
brad [Wed, 13 Aug 2008 03:46:30 +0000 (03:46 +0000)]
regen
brad [Wed, 13 Aug 2008 03:45:55 +0000 (03:45 +0000)]
Add another PCI id for the Realtek 8139D chipset.
From FreeBSD
brad [Wed, 13 Aug 2008 03:18:19 +0000 (03:18 +0000)]
Replace the use of a separate softc field to track the link
state with the newly minted link flag in the softc's rl_flags
field.
Tested by fkr@
oga [Tue, 12 Aug 2008 23:10:41 +0000 (23:10 +0000)]
Kill dmesg spam. Since we can't guarantee that a device is capable of
having a drm driver, don't print when we fail to attach one.
stops the "drm at vga? not supported" spam people are seeing.
deraadt [Tue, 12 Aug 2008 22:48:31 +0000 (22:48 +0000)]
Fix multiple addition error recently introduced for when extended partitions
are just before an OpenBSD partition in a MBR; done with kettenis & otto
ok reyk
damien [Tue, 12 Aug 2008 20:07:13 +0000 (20:07 +0000)]
adopt the integrity group cipher of the AP at association time.
damien [Tue, 12 Aug 2008 19:59:09 +0000 (19:59 +0000)]
nuke unused fields.
damien [Tue, 12 Aug 2008 19:56:59 +0000 (19:56 +0000)]
add the code to encrypt/decrypt management frames, retrieve key id
from MMIE etc...
this code can't be triggered as no drivers claim MFP capability yet.
damien [Tue, 12 Aug 2008 19:51:19 +0000 (19:51 +0000)]
reserves the correct number of bytes in ieee80211_send_4way_msg2().
damien [Tue, 12 Aug 2008 19:50:39 +0000 (19:50 +0000)]
clear Tx_Rx protection bits properly.
damien [Tue, 12 Aug 2008 19:45:22 +0000 (19:45 +0000)]
do not drop frames that have an invalid IE, just stop processing
more IEs.
damien [Tue, 12 Aug 2008 19:42:37 +0000 (19:42 +0000)]
populate ieee80211_recv_action(), does nothing yet.
damien [Tue, 12 Aug 2008 19:40:13 +0000 (19:40 +0000)]
add a couple of new definitions (action frames, new subtypes for control
frames, new IEs etc...)
damien [Tue, 12 Aug 2008 19:34:54 +0000 (19:34 +0000)]
missing SHA-256 bits.
damien [Tue, 12 Aug 2008 19:34:35 +0000 (19:34 +0000)]
v3 descriptors use AES Key Wrap too.
damien [Tue, 12 Aug 2008 19:29:07 +0000 (19:29 +0000)]
new SHA-256 based AKMPs.
damien [Tue, 12 Aug 2008 19:21:04 +0000 (19:21 +0000)]
add/process group integrity cipher suite in RSN IEs.
add support for MFP negotiation during association.
damien [Tue, 12 Aug 2008 19:05:39 +0000 (19:05 +0000)]
add IEEE80211_RSNIE_MAXLEN and IEEE80211_WPAIE_MAXLEN definitions
instead of hard-coding values.
for RSN IE, request space for Group Integrity Cipher Suite.
damien [Tue, 12 Aug 2008 18:48:35 +0000 (18:48 +0000)]
add support for EAPOL-Key v3 descriptors (similar to v2 except that the
MIC is computed using AES-128-CMAC instead of HMAC-SHA1).
add a SHA-256 based key derivation function (not used yet).
damien [Tue, 12 Aug 2008 18:41:18 +0000 (18:41 +0000)]
the only integrity group cipher currently supported is AES-128-CMAC.
damien [Tue, 12 Aug 2008 18:37:23 +0000 (18:37 +0000)]
generate a random IGTK in HostAP mode if we're MFP-capable.
damien [Tue, 12 Aug 2008 18:30:56 +0000 (18:30 +0000)]
add IEEE80211_F_MFPR config flags (management frame protection required)
and IEEE80211_C_MFP capability bit (driver supports MFP).
damien [Tue, 12 Aug 2008 18:30:02 +0000 (18:30 +0000)]
add definitions for RSN capabilities field (RSN IE).