adrian [Sun, 11 Mar 2012 19:08:56 +0000 (19:08 +0000)]
Upgrade the netgraph vlan node to support 802.1q, encapsulation type,
PCP and CFI fields.
* Ethernet_type for VLAN encapsulation is tunable, default is 0x8100;
* PCP (Priority code point) and CFI (canonical format indicator) is
tunable per VID;
* Tunable encapsulation to support 802.1q
* Encapsulation/Decapsulation code improvements
New messages have been added for this netgraph node to support the
new features.
However, the legacy "vlan" id is still supported and compiled in by
default. It can be disabled in a future release.
TODO:
* Documentation
* Examples
PR: kern/161908
Submitted by: Ivan <rozhuk.im@gmail.com>
luigi [Sun, 11 Mar 2012 17:35:12 +0000 (17:35 +0000)]
- remove an extra parenthesis in a closing brace;
- add the macro NETMAP_RING_FIRST_RESERVED() which returns
the index of the first non-released buffer in the ring
(this is useful for code that retains buffers for some time
instead of processing them immediately)
marius [Sun, 11 Mar 2012 13:39:19 +0000 (13:39 +0000)]
Fix a bug introduced in r223938; on big-endian machines coping a 32-bit
quantum bytewise to the address of a 64-bit variable results in writing
to the "wrong" 32-bit half so adjust the address accordingly. This fix
is implemented in a hackish way for two reasons:
o in order to be able to get it into 8.3 with zero impact on the little-
endian architectures where this bug has no effect and
o to avoid blowing the x86 boot2 out of the water again when compiling
it with clang, which all sane versions of this fix tested do.
This change fixes booting from UFS1 file systems on big-endian machines.
jmallett [Sun, 11 Mar 2012 06:17:49 +0000 (06:17 +0000)]
Merge the Cavium Octeon SDK 2.3.0 Simple Executive code and update FreeBSD to
make use of it where possible.
This primarily brings in support for newer hardware, and FreeBSD is not yet
able to support the abundance of IRQs on new hardware and many features in the
Ethernet driver.
Because of the changes to IRQs in the Simple Executive, we have to maintain our
own list of Octeon IRQs now, which probably can be pared-down and be specific
to the CIU interrupt unit soon, and when other interrupt mechanisms are added
they can maintain their own definitions.
Remove unmasking of interrupts from within the UART device now that the
function used is no longer present in the Simple Executive. The unmasking
seems to have been gratuitous as this is more properly handled by the buses
above the UART device, and seems to work on that basis.
jmallett [Sun, 11 Mar 2012 06:11:31 +0000 (06:11 +0000)]
Disable -Winline on MIPS in preparation for the import of the latest version
of the Cavium Simple Executive, which violates large function growth rules
in such a way that simply increasing the large function growth parameter is
insufficient.
jmallett [Sun, 11 Mar 2012 00:34:14 +0000 (00:34 +0000)]
Fix promiscuous mode with if_octm:
o) The MAC set must occur before the multicast list is set up as the former
will enable the CAM unconditionally, while promiscuous mode disables it,
so if promiscuous mode is to be set this must occur after the MAC is
programmed.
o) The multicast list must be set up unconditionally as even if flags have
not changed, if the interface has gone through a reinitialization, the
state of the CAM as changed by the MAC initialization could be incorrect.
o) Call octm_init when flags change, even if the interface is already running.
netchild [Sat, 10 Mar 2012 23:10:18 +0000 (23:10 +0000)]
- add comments to syscalls.master and linux(32)_dummy about which linux
kernel version introduced the sysctl (based upon a linux man-page)
- add comments to sscalls.master regarding some names of sysctls which are
different than the linux-names (based upon the linux unistd.h)
- add some dummy sysctls
- name an unimplemented sysctl
mav [Sat, 10 Mar 2012 21:08:07 +0000 (21:08 +0000)]
ServerWorks HT1000 HPET reported to have problems with IRQs >= 16.
Lower (ISA) IRQs are working, but allowed mask is not set correctly.
Block both by default to allow HP BL465c G6 blade system to boot.
Reported by: Attila Nagy <bra@fsn.hu>
MFC after: 1 week
adrian [Sat, 10 Mar 2012 20:09:02 +0000 (20:09 +0000)]
Stick the if_drv_flags access (check and modify) behind the ifq lock.
Although access to the flags to check/set OACTIVE is racy due to how
the default if_start() function works, this should remove any races
with read/modify/write between threads.
adrian [Sat, 10 Mar 2012 19:58:23 +0000 (19:58 +0000)]
Fix a panic introduced in a previous commit - non-beaconing modes (eg STA)
don't setup the avp mcast queue.
This is a bit annoying though - it turns out the mcast queue isn't
initialised for STA mode but it's then touched to see whether anything
is in it. That should be fixed in a subsequent commit.
mav [Sat, 10 Mar 2012 18:56:16 +0000 (18:56 +0000)]
Revert r175376 and tune cpufreq(4) frequency comparison logic instead.
Instead of using 25MHz equality threshold, look for the nearest value when
handling dev.cpu.0.freq sysctl and for exact match when it is expected.
ACPI may report extra level with frequency 1MHz above the nominal to
control Intel Turbo Boost operation. It is not a bug, but feature:
dev.cpu.0.freq_levels: 2934/106000 2933/95000 2800/82000 ...
In this case value 2933 means 2.93GHz, but 2934 means 3.2-3.6GHz.
I've found that my Core i7-870 based system has Intel Turbo Boost disabled
by default and without this change it was absolutely invisible and hard
to control.
iwasaki [Sat, 10 Mar 2012 17:08:57 +0000 (17:08 +0000)]
Fix wrong asresp frame parsing in iwi_checkforqos().
After 8.0-RELEASE, iwi(4) doesn't send any data frames in infrastructure
mode.
Bacause of the condition `while (frm < efrm)', IEEE80211_VERIFY_LENGTH()
was checking item length beyond the ieee80211_frame region, and returned
from iwi_checkforqos() without setting flags, capinfo and associd.
In infrastructure mode associd is required, so this problem causes
discarding mbuf in ieee80211_start().
PR: kern/165819
Tested/Reviewed/Supported by: bschmidt and adrian
MFC after: 1 week
mav [Sat, 10 Mar 2012 14:57:21 +0000 (14:57 +0000)]
Idle ticks optimization:
- Pass number of events to the statclock() and profclock() functions
same as to hardclock() before to not call them many times in a loop.
- Rename them into statclock_cnt() and profclock_cnt().
- Turn statclock() and profclock() into compatibility wrappers,
still needed for arm.
- Rename hardclock_anycpu() into hardclock_cnt() for unification.
gavin [Sat, 10 Mar 2012 11:25:53 +0000 (11:25 +0000)]
Move determination of socket buffer sizes from startup to the first time a
socket is used. The previous code structure assumed that AF_INET sockets
were always available, which is an invalid assumption on IPv6-only systems.
This merges the fololowing revisions from NetBSD:
src/usr.bin/ftp/main.c 1.120
src/usr.bin/ftp/util.c 1.156
kib [Sat, 10 Mar 2012 08:49:44 +0000 (08:49 +0000)]
Optimize tls_get_addr_common(). The change provides around 30% speedup
for TLS microbenchmark using global-dynamic TLS model on amd64 (which is
default for PIC dso objects).
Split the slow path into tls_get_addr_slow(), for which inlining is
disabled. This prevents the registers spill on tls_get_addr_common()
entry.
Provide static branch hint to the compiler, indicating that slow path
is not likely to be taken.
jmallett [Sat, 10 Mar 2012 06:54:37 +0000 (06:54 +0000)]
o) Remove some CPU_CNMIPS-related magical thinking about the status register's
contents for user programs.
o) Conditionalize the installation of an XTLB handler on ABI, not CPU family.
jmallett [Sat, 10 Mar 2012 05:38:04 +0000 (05:38 +0000)]
o) Bump INTRCOUNT_COUNT to 256, since Octeon already has >128.
XXX It would be good to use a better way to size intrcnt.
o) Fix literal 4s that are supposed to be sizeof (u_long).
XXX Why the * 2 here? Is this an artifact of a different system that this
code came from? We seem to allocate twice as much space for intrcnt
as we admit to in sintrcnt.
adrian [Sat, 10 Mar 2012 04:14:04 +0000 (04:14 +0000)]
Don't flood the cabq/mcastq with frames.
In a very noisy 2.4GHz environment (with HT/40 enabled, making it worse)
I saw the following occur:
* the air was considered "busy" a lot of the time;
* the cabq time is quite short due to staggered beacons being enabled;
* it just wasn't able to keep up TX'ing CABQ frames;
* .. and the cabq would swallow up all the TX ath_buf's.
This patch introduces a twiddle which allows the maximum cabq depth to be
set, forcing further frames to be dropped.
It defaults to the TX buffer count at the moment, so the default behaviour
isn't changed.
I've also started fleshing out a similar setup for the data path, so
it doesn't swallow up all the available TX buffers and preventing management
frames (such as ADDBA) out.
adrian [Fri, 9 Mar 2012 22:58:34 +0000 (22:58 +0000)]
Document that we may end up with some suboptimal handling of data
frames with stations in power saving mode.
I'm not (yet) sure how to handle TX'ing aggregates frames to stations
that are in power saving mode, or whether that's even a feasible thing
to do. So in order to (mostly) not forget, leave a couple of comments
in the code.
The code presently assumes that the aggregation TID state for an ath_node
is locked not by the ath_node lock or a node+TID lock, but behind the
hardware queue said TID maps to. This assumption is going to be
incorrect for stations in power saving mode as we'll be TX'ing frames
on the multicast queue.
dim [Fri, 9 Mar 2012 21:02:39 +0000 (21:02 +0000)]
Partially undo r228693, by removing NO_WFORMAT.clang in fsdb's Makefile,
and fixing the format string in sbin/fsdb/fsdbutil.c instead.
Note the remark "Work around a problem with format string warnings and
ntohs macros" was actually incorrect. The DIP(dp, di_nlink) macro
invocation actually returned an int, due to its ternary expression, even
though the di_nlink members of struct ufs1_dinode and struct ufs2_dinode
are both defined as int16_t.
dim [Fri, 9 Mar 2012 20:34:31 +0000 (20:34 +0000)]
Add casts to __uint16_t to the __bswap16() macros on all arches which
didn't already have them. This is because the ternary expression will
return int, due to the Usual Arithmetic Conversions. Such casts are not
needed for the 32 and 64 bit variants.
While here, add additional parentheses around the x86 variant, to
protect against unintended consequences.
jhb [Fri, 9 Mar 2012 19:42:48 +0000 (19:42 +0000)]
Allow a native i386 kernel to be built with 'nodevice atpic'. Just as on
amd64, if 'device isa' is present quiesce the 8259A's during boot and
resume from suspend.
While here, be more selective on amd64 about which kernel configurations
need elcr.c.
jhb [Fri, 9 Mar 2012 19:20:19 +0000 (19:20 +0000)]
MFamd64:
- Return failure for a suspend attempt if we have no wake address.
- Use intr_disable()/intr_restore() instead of ACPI_DISABLE_IRQS().
- Invoke intr_suspend() earlier and call intr_resume() if suspend
fails.
- Use pause in the loop waiting for CPU to suspend.
- Restore PAT MSR, switchtime, switchticks, and MTRRs on resume.
tijl [Fri, 9 Mar 2012 11:48:56 +0000 (11:48 +0000)]
Clean up x86 endian.h:
- Remove extern "C". There are no functions with external linkage here. [1]
- Rename bswapNN_const(x) to bswapNN_gen(x) to indicate that these macros
are generic implementations that can take non-constant arguments. [1]
- Split up __GNUCLIKE_ASM && __GNUCLIKE_BUILTIN_CONSTANT_P and deal with
each separately.
- Replace _LP64 with __amd64__ because asm instructions are machine
dependent, not ABI dependent.
adrian [Thu, 8 Mar 2012 23:53:38 +0000 (23:53 +0000)]
Correctly initialise the TXQ link pointer to the last descriptor in
the last buffer in the list.
The current behaviour (due to me, so pointy hat is firmly on my head here)
was incorrect - it was setting the link pointer to the last descriptor
of the _first_ buffer in the TXQ. Instead, it should have set it to the
last descriptor in the _last_ buffer in the TXQ.
This showed up as occasional TX stalls with frames in the TXQ but no
TX progress being made. Further inspection showed the TXQ looked like
it contained multiple "lists" of frames - there'd be a list of correct
frames, then a NULL link pointer, but there'd be a next buffer in the
list.
Since this code is only called upon an interface reset, it's likely
this only began showing up when I started doing stress testing
in environments which annoy the radios enough to cause lockups.
I've not yet any TX stalls with this patch applied.
kan [Thu, 8 Mar 2012 21:09:34 +0000 (21:09 +0000)]
Save more of config space for PCI Express and PCI-X devices.
Expand pci_save_state and pci_restore_state to save more of
the config state for PCI Express and PCI-X devices. Various
writable control registers are present in PCI Express that
can potentially be lost over suspend/resume cycle.
This change is modeled after similar functionality in Linux.
pfg [Thu, 8 Mar 2012 21:06:05 +0000 (21:06 +0000)]
Add support for ns timestamps and birthtime to the ext2/3 driver.
When using big inodes there is sufficient space in ext3 to
keep extra resolution and birthtime (creation) timestamps.
The appropriate fields in the on-disk inode have been approved
for a long time but support for this in ext3 has not been
widely distributed.
In preparation for ext4 most linux distributions have enabled
by default such bigger inodes and some people use nanosecond
timestamps in ext3. We now support those when the inode is big
enough and while we do recognize the EXT4F_ROCOMPAT_EXTRA_ISIZE,
we maintain the extra timestamps even when they are not used.
An additional note by Bruce Evans:
We blindly accept unrepresentable tv_nsec in VOP_SETATTR(), but
all file systems have always done that. When POSIX gets around
to specifying the behaviour, it will probably require certain
rounding to the fs's resolution and not rejecting the request.
This unfortunately means that syscalls that set times can't
really tell if they succeeded without reading back the times
using stat() or similar and checking that they were set close
enough.
jhb [Thu, 8 Mar 2012 19:41:05 +0000 (19:41 +0000)]
Add a new sched_clear_name() method to the scheduler interface to clear
the cached name used for KTR_SCHED traces when a thread's name changes.
This way KTR_SCHED traces (and thus schedgraph) will notice when a thread's
name changes, most commonly via execve().
glebius [Thu, 8 Mar 2012 09:20:00 +0000 (09:20 +0000)]
Merge from OpenBSD:
revision 1.146
date: 2010/05/12 08:11:11; author: claudio; state: Exp; lines: +2 -3
bzero() the full compressed update struct before setting the values.
This is needed because pf_state_peer_hton() skips some fields in certain
situations which could result in garbage beeing sent to the other peer.
This seems to fix the pfsync storms seen by stephan@ and so dlg owes me
a whiskey.
I didn't see any storms, but this definitely fixes a useless memory
allocation on the receiving side, due to non zero scrub_flags field
in a pfsync_state_peer structure.
emaste [Thu, 8 Mar 2012 02:00:52 +0000 (02:00 +0000)]
Work around broken BIOS memory reporting
Andrzej has a machine with 32GB of RAM, but only 16GB is reported by the
smbios.memory.enabled. Thus, use the greater of hw.realmem and the smbios
value.
Reported by: Andrzej Tobola <ato of iem pw edu pl>
emaste [Thu, 8 Mar 2012 01:48:44 +0000 (01:48 +0000)]
Remove unactionable message about label geometry
It's not clear to a user what they should do after seeing the "geometry
does not match label" kernel message, and it does not appear to present
a problem in practice. Thus, just remove the messages.
hrs [Thu, 8 Mar 2012 01:47:12 +0000 (01:47 +0000)]
- Clean up extra ${.OBJDIR}.
- Add ${IMAGE} for the supported image files. This fixes the install target
on FreeBSD/pc98.
- Use "mkdir -p" instead of "-mkdir" consistently.
jhb [Wed, 7 Mar 2012 18:50:33 +0000 (18:50 +0000)]
Simplify the PCI bus dma tag code a bit. First, don't create a tag at
all for platforms that only have 32-bit bus addresses. Second, remove
the 'tag_valid' flag from the softc. Instead, if we don't create a
tag in pci_attach_common(), just cache the value of our parent's tag
so that we always have a valid tag to return.
emaste [Wed, 7 Mar 2012 18:46:21 +0000 (18:46 +0000)]
Improve multi-line kernel ident parsing
Instead of blindly grabbing the line with 'Version string' and the
following one from the core info file, take all lines after 'Version
string' until the one one that matches the field format in the core info
file. This provides compatibility with VendorBSD modifications that have
a different kernel ident format.
pluknet [Wed, 7 Mar 2012 09:42:19 +0000 (09:42 +0000)]
Update ifa_rtrequest() description after post-4.4BSD change made in r85074.
3rd argument of ifa->ifa_rtrequest is now ``rt_addrinfo *'' instead of
``sockaddr *''. While here, un-document RTM_RESOLVE cmd argument for
ifa_rtrequest() that became a stub after separating L2 tables in r186119.