alfred [Mon, 8 Jan 2007 23:21:06 +0000 (23:21 +0000)]
Add the following functions to abstract away the creation of task threads
for usb. I hope that this will eventually be used for generic devices
that need full fledged blocking threads for event processing.
Create a taskqueue:
void usb_ether_task_init(device_t, int, struct usb_taskqueue *);
Enqueue a task:
void usb_ether_task_enqueue(struct usb_taskqueue *, struct task *);
Wait for all tasks queued to complete:
void usb_ether_task_drain(struct usb_taskqueue *, struct task *);
Destroy the taskqueue:
void usb_ether_task_destroy(struct usb_taskqueue *);
rwatson [Mon, 8 Jan 2007 22:30:39 +0000 (22:30 +0000)]
Factor out my copyrights + licenses from Charles D. Cranor and
University of Washington copyrights, which include the
advertising clause. Move $NetBSD$ into standard location for
FreeBSD source files, and normalize formatting.
rwatson [Mon, 8 Jan 2007 22:14:00 +0000 (22:14 +0000)]
Factor out UCB and my copyrights from copyrights of Mike Mitchell;
the former use a three-clause BSD license (per UCB authorization
letter), whereas he uses a four-clause BSD license.
imp [Mon, 8 Jan 2007 21:21:45 +0000 (21:21 +0000)]
Replace (name) with Henric Vestergaard Draboel since it was clear that
this was just cut N paste and '(name)' doesn't make any sense. If
someone knows how to contact Mr. Draboel, please let me know.
rwatson [Mon, 8 Jan 2007 20:35:13 +0000 (20:35 +0000)]
Resort copyrights and licenses in kern_acct.c: per UCB letter,
the UCB license now excludes the advertising clause. I'm not
interested in it either, so move my copyright. This leaves
only a CGD copyright with the advertising clause.
imp [Mon, 8 Jan 2007 18:25:58 +0000 (18:25 +0000)]
Per Olivier Houchard, use the proper license for this file. He
bogusly used the kvm_powerpc.c file as a template for the license, but
then either wrote the code himself, or cribbed it from the kvm_i386
file. The only thing from the kvm_powerpc.c file was the license.
Correct this mistake with his blessing.
sam [Mon, 8 Jan 2007 18:23:43 +0000 (18:23 +0000)]
Fix potential node refcnt leak. If mbufs are q'd on ic_mgtq when
the state machine clocks to INIT, node references are not reclaimed.
Add a new routine ieee80211_drain_ifq that does this and use it
instead of IF_DRAIN.
rwatson [Mon, 8 Jan 2007 17:49:59 +0000 (17:49 +0000)]
Canonicalize copyrights in some files I hold copyrights on:
- Sort by date in license blocks, oldest copyright first.
- All rights reserved after all copyrights, not just the first.
- Use (c) to be consistent with other entries.
sam [Mon, 8 Jan 2007 17:24:51 +0000 (17:24 +0000)]
Correct several issues with rate set negotiation:
o add IEEE80211_F_JOIN flag to ieee80211_fix_rate to indicate a station
is joining a BSS; this is used to control whether or not we over-write
the basic rate bit in the calculated rate set
o fix ieee80211_fix_rate to honor IEEE80211_F_DODEL when IEEE80211_F_DONEGO
is not specified (e.g. when joining an ibss network)
o on sta join always delete unusable rates from the negotiated rate set,
this was being done only ibss networks but is also needed for 11g bss
with mixed stations
o on sta join delete unusable rates from the bss node's rate set, not the
scan table entry's rate set
o when calculating a rate set for new neighbors in an ibss caculate a
negotiated rate set so drivers are not presented with rates they should
not use
marius [Mon, 8 Jan 2007 01:26:47 +0000 (01:26 +0000)]
o Changes to psycho_attach(): [1]
- Clear the PCI AFSR and status error bits as previous errors still
might be indicated.
- Set up the PCI control and diagnostic registers according to the
capabilities, workarounds, etc of/for specific revisions of the
supported bridges. This includes no longer setting Hummingbird-/
Sabre-specific bits in the PCI control register but preserving
what the firmware has initialized them to like OpenSolaris does.
Previously we were setting these bits according to the example in
the Sabre documentation, which I doubt is appropriate for all
Sabre based designs and especially not for Hummingbirds. This
also includes not enabling bus parking unless the firmware tells
us to.
- Set the PCI latency timer register as this isn't always done by
the firmware.
o Remove a redundant argument from psycho_set_intr() and in this
function check the return value of bus_setup_intr(). [2]
o Let psycho_setup_intr() return ENOMEM instead of 0 when it can't
allocate memory for the interrupt wrapper stub and EINVAL instead
of 0 if it can't find the interrupt vector in the interrupt map.
o Add a workaround for a bug of the Sabre-APB-combination where it
doesn't drain DMA write data for devices behind additional PCI-PCI
bridges underneath the APB PCI-PCI bridge. This workaround (do
things necessary in order to achieve a manual drain when coherency
is required) is currently implemented in psycho_setup_intr() and
psycho_intr_stub() (for easy MFC'ing) and therefore is only applied
for interrupt handlers. This should be moved to psycho(4)-specific
bus_dma_tag_create() and bus_dmamap_sync() methods, respectively,
once this driver is converted to make use of BUS_GET_DMA_TAG(), so
the workaround is also applied for polling(4) callbacks. [3]
o Fix some minor style issues.
marius [Mon, 8 Jan 2007 01:08:24 +0000 (01:08 +0000)]
In ofw_pcibus_attach() skip dupe PCI devices reported by the
firmware (mainly 'pmu' and its 'lomp' dupe found in a couple of
later USII{e,i}-based machines) by checking whether a device with
the same triple of bus number, slot and function already has been
added. This is the simple yet effective approach introduced in
OpenBSD some time ago, but which has the flaw that it assumes
that the device and its dupe(s) found in the OFW device tree are
equal or at least the one encountered first is in some way the
more important one (this is the case with 'pmu' and 'lomp'; the
'pmu' node has couple of properties and children while the 'lomp'
one misses most of these). If there's ever a device/dupe pair
where we don't encounter the more important node first, we'll
probably need to introduce a quirk list in order to add the
desired device but prevent its dupe(s) from being added.
yongari [Mon, 8 Jan 2007 00:58:00 +0000 (00:58 +0000)]
Don't rely on GM_GP_CTRL register contents when mii(4) layer reports
link state changes. Instead, build new speed/duplex/flow-control
settings from the values reported from PHY.
This should fix speed/duplex/flow-control mismatches between GMAC and
PHY which resulted in very poor Rx performance due to lots of
out-of-order packet delivery.
Reported by: Arno J. Klaassen <arno AT heho DOT snv DOT jussieu DOT fr>
Tested by: Arno J. Klaassen <arno AT heho DOT snv DOT jussieu DOT fr>
pjd [Sun, 7 Jan 2007 23:06:53 +0000 (23:06 +0000)]
Be more specific in ENXIO description:
- O_NONBLOCK flag has to be set, if it is not set, open(2) will wait for
another process opening the fifo for reading,
- Use O_WRONLY which implies that the file has to be opened _only_ for write.
njl [Sun, 7 Jan 2007 21:53:42 +0000 (21:53 +0000)]
Re-work Cx handling to be per-cpu and asymmetrical, fixing support on
modern dual-core systems as well.
- Parse the _CST packages for each cpu and track all the states individually,
on a per-cpu basis.
- Revert to generic FADT/P_BLK based Cx control if the _CST package
is not present on all cpus. In that case, the new driver will
still support per-cpu Cx state handling. The driver will determine the
highest Cx level that can be supported by all the cpus and configure the
available Cx state based on that.
- Fixed the case where multiple cpus in the system share the same
registers for Cx state handling. To do that, added a new flag
parameter to the acpi_PkgGas and acpi_bus_alloc_gas functions that
enable the caller to add the RF_SHAREABLE flag. This flag could also be
useful to other callers (acpi_throttle?) in the tree but this change is
not yet made.
- For Core Duo cpus, both cores seems to be taken out of C3 state when
any one of the cores need to transition out. This broke the short sleep
detection logic. It is disabled now if there is more than one cpu in
the system for now as it fixed it in my case. This quirk may need to
be re-enabled later differently.
- Added support to control cx_lowest on a per-cpu basis. There is still
a generic cx_lowest to enable changing cx_lowest for all cpus with a single
sysctl and for ease of use. Sample output for the new sysctl:
pjd [Sun, 7 Jan 2007 17:32:16 +0000 (17:32 +0000)]
Prefer "to be modified" over "to be opened for writing".
This is quite tricky situation, because we allow to open a file with
O_RDONLY|O_TRUNC. O_TRUNC modifies a file, but we actually don't open
it for writing. EISDIR is also returned when we try to open a directory
O_RDONLY|O_TRUNC, which is correct.
POSIX says that "The result of using O_TRUNC with O_RDONLY is undefined.",
we choose to accept it (Solaris did the same), that's why "to be modified"
seems more accurate to me.
rwatson [Sat, 6 Jan 2007 22:28:28 +0000 (22:28 +0000)]
When returning early from audit_arg_file() due to so->so_pcb being NULL
(due to an early reset or the like), remember to unlock the socket lock.
This will not occur in 7-CURRENT, but could in theory occur in 6-STABLE.
dougb [Sat, 6 Jan 2007 22:04:46 +0000 (22:04 +0000)]
More minor cleanups:
1. Fix small typo "retorted ." -> "retorted."
2. Remove from fortunes a story that is duplicated in fortunes2-o
3. Remove from fortunes and fortunes2-o Zippy the Pinhead quotes
that are already in the zippy file.
4. ... therefore remove zippy from fortunes.sp.ok
5. Remove a duplicate in the zippy file.
das [Sat, 6 Jan 2007 21:46:23 +0000 (21:46 +0000)]
Fix a problem relating to fesetenv() clobbering i387 register stack.
Details: As a side-effect of restoring a saved FP environment,
fesetenv() overwrites the tag word, which indicates which i387
registers are in use. Normally this isn't a problem because
the calling convention requires the register stack to be empty
on function entry and exit. However, fesetenv() is inlined, so we
need to tell gcc explicitly that the i387 registers get clobbered.
das [Sat, 6 Jan 2007 21:22:38 +0000 (21:22 +0000)]
Correctly handle inf/nan. This routine is currently unused because we
seem to have assembly versions for all architectures, but it can't
hurt to fix it.
rodrigc [Sat, 6 Jan 2007 20:46:02 +0000 (20:46 +0000)]
When performing a mount update to change a mount from read-only to read-write,
do not call markvoldirty() until the mount has been flagged as read-write.
Due to the nature of the msdosfs code, this bug only seemed to appear for
FAT-16 and FAT-32.
This fixes the testcase:
#!/bin/sh
dd if=/dev/zero bs=1m count=1 oseek=119 of=image.msdos
mdconfig -a -t vnode -f image.msdos
newfs_msdos -F 16 /dev/md0 fd120m
mount_msdosfs -o ro /dev/md0 /mnt
mount | grep md0
mount -u -o rw /dev/md0; echo $?
mount | grep md0
umount /mnt
mdconfig -d -u 0
netchild [Sat, 6 Jan 2007 19:08:39 +0000 (19:08 +0000)]
Sync with NetBSD:
revision 1.98 is NOT merged, because FreeBSD does not support this
syntax.
revision 1.99 is NOT merged, "const poisoning" part is not applicable
to FreeBSD. There is no variable shadowing, GCC can't find
this one (but there are others)
revision 1.100 is NOT merged, because it was null patch (no changes)
revision 1.101 is NOT merged, there is no BIT() macro in FreeBSD
revision 1.102 is merged
revision 1.103 is partially merged. There is no ai.ifaceh in FreeBSD
revision 1.104 is NOT merged
revision 1.105 is merged
revision 1.106 is not merged, because of rev. 1.107
revision 1.107 is a backuout of 1.106
netchild [Sat, 6 Jan 2007 18:59:35 +0000 (18:59 +0000)]
Extend the emu10kx driver. With the words of the author:
---snip---
New features:
1. Optional multichannel recording (32 channels on Live!, 64 channels
on Audigy).
All channels are 16bit/48000Hz/mono, format is fixed.
Half of them are copied from sound output, another half can be
used to record any data from DSP. What should be recorded is
hardcoded in DSP code. In this version it records dummy data, but
can be used to record all DSP inputs, for example..
Because there are no support of more-than-stereo sound streams
multichannell stream is presented as one 32(64)*48000 Hz 16bit mono
stream.
Channel map:
SB Live! (4.0/5.1)
offset (words) substream
0x00 Front L
0x01 Front R
0x02 Digital Front L
0x03 Digital Front R
0x04 Digital Center
0x05 Digital Sub
0x06 Headphones L
0x07 Headphones R
0x08 Rear L
0x09 Rear R
0x0A ADC (multi-rate recording) L
0x0B ADC (multi-rate recording) R
0x0C unused
0x0D unused
0x0E unused
0x0F unused
0x10 Analog Center (Live! 5.1) / dummy (Live! 4.0)
0x11 Analog Sub (Live! 5.1) / dummy (Live! 4.0)
0x12..-0x1F dummy
Audigy / Audigy 2 / Audigy 2 Value / Audigy 4
offset (words) substream
0x00 Digital Front L
0x01 Digital Front R
0x02 Digital Center
0x03 Digital Sub
0x04 Digital Side L (7.1 cards) / Headphones L (5.1 cards)
0x05 Digital Side R (7.1 cards) / Headphones R (5.1 cards)
0x06 Digital Rear L
0x07 Digital Rear R
0x08 Front L
0x09 Front R
0x0A Center
0x0B Sub
0x0C Side L
0x0D Side R
0x0E Rear L
0x0F Rear R
0x10 output to AC97 input L (muted)
0x11 output to AC97 input R (muted)
0x12 unused
0x13 unused
0x14 unused
0x15 unused
0x16 ADC (multi-rate recording) L
0x17 ADC (multi-rate recording) R
0x18 unused
0x19 unused
0x1A unused
0x1B unused
0x1C unused
0x1D unused
0x1E unused
0x1F unused
0x20..0x3F dummy
Fixes:
1. Do not assign negative values to variables used to index emu_cards
array. This array was never accessed when index is negative, but
Alexander (netchild@) told me that Coverity does not like it.
After this change emu_cards[0] should never be used to identify
valid sound card.
2. Fix off-by-one errors in interrupt manager. Add more checks there.
3. Fixes to sound buffering code now allows driver to use large playback
buffers.
4. Fix memory allocation bug when multichannel recording is not
enabled.
5. Fix interrupt timeout when recording with low bitrate (8kHz).
Hardware:
1. Add one more known Audigy ZS card to list. Add two cards with
PCI IDs betwen old known cards and new one.
Other changes:
1. Do not use ALL CAPS in messages.
Incomplete code:
1. Automute S/PDIF when S/PDIF signal is lost.
This commits enables a little bit of debugging output when the driver is
loaded as a module. I did a cross-build test for amd64.
The code has some style issues, this will be addressed later.
The multichannel recording part is some work in progress to allow playing
around with it until the generic sound code is better able to handle
multichannel streams.
This is supposed to fix
CID: 171187
Found by: Coverity Prevent
netchild [Sat, 6 Jan 2007 15:58:34 +0000 (15:58 +0000)]
MFi386 rev 1.56:
Bring the linux mmap code more into line with how linux (2.4.x) behaves.
Tested by: Scot Hetzel <swhetzel@gmail.com> on amd64 without PROT_EXEC
Additionally to the i386 version always use PROT_EXEC in the mapping like the
previous version of the amd64 code did. We need to examinate this further to
decide what the right thing to do is. For now this fixes several problems in
the LTP test runs and should behave regarding PROT_EXEC like before.
jeff [Sat, 6 Jan 2007 12:33:43 +0000 (12:33 +0000)]
- Don't let SCHED_TICK_TOTAL() return less than hz. This can cause integer
divide faults in roundup() later if it is able to return 0. For some
reason this bug only shows up on my laptop and not my testboxes.
jeff [Sat, 6 Jan 2007 08:44:13 +0000 (08:44 +0000)]
- Fix the sched_priority() invalid priority bugs. Use roundup() instead
of max() when computing the divisor in SCHED_TICK_PRI(). This prevents
cases where rounding down would allow the quotient to exceed
SCHED_PRI_RANGE.
- Garbage collect some unused flags and fields.
- Replace TDF_HOLD with sched_pin_td()/sched_unpin_td() since it simply
duplicated this functionality.
- Re-enable the rebalancer by default and fix the sysctl so it can be
modified.
jeff [Sat, 6 Jan 2007 02:34:23 +0000 (02:34 +0000)]
- Don't IPI unless we're going to interrupt something exiting in the kernel.
otherwise we can afford the latency. This makes a significant performance
improvement.
jeff [Fri, 5 Jan 2007 23:45:38 +0000 (23:45 +0000)]
- Fix a comparison in sched_choose() that caused cpus to be constantly
marked idle, thus breaking cpu load balancing.
- Change sched_interact_update() to fix cases where the stored history
has expanded significantly rather than handling them in the callers. This
fixes a case where sched_priority() could compute a bad value.
- Add a sysctl to disable the global load balancer for experimentation.
jkim [Fri, 5 Jan 2007 22:55:19 +0000 (22:55 +0000)]
- Clean up Aperture Access Global Enable (APEN) bit access.
- Rename confusing AGP_INTEL_I845_MCHCFG to AGP_INTEL_I845_AGPM.
- Move E7205 and E7505 from i8x5 to i8x0 family. It probably worked
because the actual offset is the same.
In fact, all three families have the bit at the exact same place. Only
differences are name and width of the registers, i.e., NBXCFG (0x50, dword),
RDCR (0x51, byte), AGPM (0x51, byte), MCHCFG (0x50, word) depending on
the family of the chipsets.
jhb [Fri, 5 Jan 2007 21:04:37 +0000 (21:04 +0000)]
Add code to parse the utrace(2) entries generated by malloc(3) in a more
human-readable format. Note that we report 'realloc(p, 0)' as 'free(p)'
since both cases are encoded the same way and 'free()' is more common
than a realloc() to 0.
jhb [Fri, 5 Jan 2007 19:59:46 +0000 (19:59 +0000)]
- Close a race between enumerating UNIX domain socket pcb structures via
sysctl and socket teardown by adding a reference count to the UNIX domain
pcb object and fixing the sysctl that enumerates unpcbs to grab a
reference on each unpcb while it builds the list to copy out to userland.
- Close a race between UNIX domain pcb garbage collection (unp_gc()) and
file descriptor teardown (fdrop()) by adding a new garbage collection
flag FWAIT. unp_gc() sets FWAIT while it walks the message buffers
in a UNIX domain socket looking for nested file descriptor references
and clears the flag when it is finished. fdrop() checks to see if the
flag is set on a file descriptor whose refcount just dropped to 0 and
waits for unp_gc() to clear the flag before completely destroying the
file descriptor.
MFC after: 1 week
Reviewed by: rwatson
Submitted by: ups
Hopefully makes the panics go away: mx1
jhb [Fri, 5 Jan 2007 19:09:01 +0000 (19:09 +0000)]
- Add a new function uma_zone_exhausted() to see if a zone is full.
- Add a printf in swp_pager_meta_build() to warn if the swapzone becomes
exhausted so that there's at least a warning before a box that runs out
of swapzone space before running out of swap space deadlocks.
cperciva [Fri, 5 Jan 2007 16:20:21 +0000 (16:20 +0000)]
If append_archive fails while writing an archive header, output the error
message from the archive being written (not the message from the archive
being read, where no error has occurred).
rodrigc [Fri, 5 Jan 2007 14:06:42 +0000 (14:06 +0000)]
Simplify code in union_hashins() and union_hashget() functions. These
functions now more closely resemble similar functions in nullfs.
This also eliminates some errors.
Submitted by: daichi, Masanori OZAWA <ozawa ongs co jp>
cperciva [Fri, 5 Jan 2007 10:48:18 +0000 (10:48 +0000)]
Change the client skipper API to use off_t instead of size_t/ssize_t; but
wrap this within #if/#else/#endif so that it will only take effect once
ARCHIVE_API_VERSION is increased (which should happen on HEAD some time
between now and when RELENG_7 is branched).
jeff [Fri, 5 Jan 2007 08:50:38 +0000 (08:50 +0000)]
- ftick was initialized to -1 for init and any of it's children. Fix this by
setting ftick = ltick = ticks in schedinit().
- Update the priority when we are pulled off of the run queue and when we
are inserted onto the run queue so that it more accurately reflects our
present status. This is important for efficient priority propagation
functioning.
- Move the frequency test into sched_pctcpu_update() so we don't repeat it
each time we'd like to call it.
- Put some temporary work-around code in sched_priority() in case the tick
mechanism produces a bad priority. Eventually this should revert to an
assert again.
ticso [Fri, 5 Jan 2007 01:18:32 +0000 (01:18 +0000)]
MFp4: Read access require PDC to be setup first otherwise we might get
overrun errors.
Write access however need cmd first, so keep the existing order
for them.