ru [Fri, 24 Aug 2012 06:55:16 +0000 (06:55 +0000)]
A workaround in r238563 was for makes (notably bmake) without the
internal knowledge that "cd" is a shell's built-in. Such makes
may attempt to exec() "cd" that in turn will fail on systems that
lack the "cd" executable.
adrian [Fri, 24 Aug 2012 00:09:49 +0000 (00:09 +0000)]
Add some new flags:
* mfp support;
* 4.9ghz support in the HAL;
* device type - specifically, the bus type and whether it's a HB63
NIC (which requires some subtle chainmask handling differences
in the AR5416 HAL.)
jamie [Thu, 23 Aug 2012 19:39:23 +0000 (19:39 +0000)]
Partially roll back r239601 - keep parameter strings both length-delimited
and null-terminated at the same time, because they're later passed to
libjail as null-terminated. That means I also need to add a nul byte when
comma-combining array parameters.
mm [Thu, 23 Aug 2012 19:32:57 +0000 (19:32 +0000)]
Merge recent vendor changes:
3086 unnecessarily setting DS_FLAG_INCONSISTENT on async destroyed datasets
3090 vdev_reopen() during reguid causes vdev to be treated as corrupt
3102 vdev_uberblock_load() and vdev_validate() may read the wrong label
imp [Thu, 23 Aug 2012 04:35:55 +0000 (04:35 +0000)]
The check for MAXPHYS doesn't make sense, so remove it.
Report errors indicated by the transport. If this is too chatty, I'll
throw it behind a debug write.
Remove commented out debugs that are no longer useful.
jamie [Thu, 23 Aug 2012 01:43:22 +0000 (01:43 +0000)]
Pre-separate IP addresses passed on the command line, so they can be
properly parsed for interface prefixes and netmask suffixes. This was
already done for the old-style (fixed) command line, but missed for
the new-style.
jamie [Thu, 23 Aug 2012 01:43:01 +0000 (01:43 +0000)]
Remember that I'm using length-defined strings in parameters:
Remove a bogus null terminator when stripping the netmask from
IP addresses. This was causing later addresses in a comma-separated
string to disappear.
Use memcpy instead of strcpy. This could just cause Bad Things.
obrien [Wed, 22 Aug 2012 23:37:24 +0000 (23:37 +0000)]
* Reinstate r128059's consumption of our best entropy first.
r128060 for "hardware-supplied entropy" reversed this without reason,
seems a typo.
* Isolate "better than nothing" implementation to a function.
gonzo [Wed, 22 Aug 2012 22:48:50 +0000 (22:48 +0000)]
Do not change "cachable" attribute for DMA memory allocated with
BUS_DMA_COHERENT attribute
The minimum unit for changing "cachable" attribute is page, so call
to pmap_change_attr effectively disable cache for all pages that newly
allocated DMA memory region spans on. The problem is that general-purpose
memory could reside on these pages too and disabling cache might affect
performance. Moreover ldrex/strex operators raise Data Abort exception
when accessing memory on page with "cachable" attribute off.
BUS_DMA_COHERENT does nto require memory to be coherent. It just suggests
to do best effort for reducing synchronization overhead.
jimharris [Wed, 22 Aug 2012 20:22:55 +0000 (20:22 +0000)]
Remove unncessary atomic operation when reading process flags in
PMC_PROC_IS_USING_PMCS macro.
Invocations of this macro are not synchronized with setting/clearing
of P_HWPMC flag, so the atomic operation here isn't needed. Removing
the atomic operation provides noticeable improvement (5-6%) on
some scheduler-intensive workloads with HWPMC_HOOKS enabled on an
8C Sandy Bridge Xeon system.
kib [Wed, 22 Aug 2012 20:05:34 +0000 (20:05 +0000)]
Provide some compat32 shims for sysctl vfs.conflist. It is required
for getvfsbyname(3) operation when called from 32bit process, and
getvfsbyname(3) is used by recent bsdtar import.
Reported by: many
Tested by: David Naylor <naylor.b.david@gmail.com>
MFC after: 5 days
obrien [Wed, 22 Aug 2012 18:49:02 +0000 (18:49 +0000)]
Depend on the new 'postrandom' instead of random.
We need to limit the amount of time between consuming the entropy seeds
and removing it in case of a kernel panic.
jhb [Wed, 22 Aug 2012 13:53:37 +0000 (13:53 +0000)]
Revert r239356 and use an alternate algorithm.
First, don't exit when the link goes down on an interface. Instead,
teach dhclient to track changes in link state and to enter the reboot
state when the link on an interface goes up causing dhclient to attempt
to renew its existing lease.
Second, remove the change I added to clear the old lease when dhclient
exits due to an error (such as ifconfig down). If an interface is
using autoconfiguration it should keep its autoconfiguration as much as
possible. If the next time it needs a configuration it is able to reuse
the previous autoconfiguration, then leaving the settings intact allows
existing connections to survive temporary outages, etc.
jimharris [Tue, 21 Aug 2012 22:28:14 +0000 (22:28 +0000)]
Fix/add support for SCSI UNMAP to ATA DSM translation.
This addresses kernel panic observed when sending SCSI UNMAP
commands to SATA disks attached to isci(4).
1) Flesh out callback routines to allocate/free buffers needed for
translating SCSI UNMAP data to ATA DSM data.
2) Add controller-level pool for storing buffers previously allocated
for UNMAP translation, to lessen chance of no buffer available
under memory pressure.
3) Ensure driver properly handles case where buffer pool is empty
and contigmalloc returns NULL.
Sponsored by: Intel
Reported by: Maksim Yevmenkin <max at netflix dot com>
Discussed with: scottl
MFC after: 3 days
np [Tue, 21 Aug 2012 22:23:17 +0000 (22:23 +0000)]
Deal with the case where a syncache entry added by the TOE driver is
evicted from the syncache but a later syncache_expand succeeds because
of syncookies. The TOE driver has to resort to more direct means to
install its hooks in the socket in this case.
np [Tue, 21 Aug 2012 18:09:33 +0000 (18:09 +0000)]
Correctly handle the case where an inp has already been dropped by the time
the TOE driver reports that an active open failed. toe_connect_failed is
supposed to handle this but it should be provided the inpcb instead of the
tcpcb which may no longer be around.
dim [Tue, 21 Aug 2012 17:58:30 +0000 (17:58 +0000)]
Support the WITH_SHARED_TOOLCHAIN setting that was introduced in r234782
for the clang executable. Build it statically by default, like the gcc
executables, which should improve performance a little bit.
hrs [Tue, 21 Aug 2012 17:49:20 +0000 (17:49 +0000)]
- Calculate the I2C baud rate to keep them <100 kHz under different TCLK
frequencies. The maximum freqency is 100 kHz according to the datasheet.
- Add child device probing support based on the device tree. It now tries to
find i2c-address property in the tree and attach the device with given slave
address to iicbus.
jchandra [Tue, 21 Aug 2012 09:37:23 +0000 (09:37 +0000)]
Add correct range parameter in XLP DTS
r239274 added support for ranges. Update XLP DTS to provide the correct
range parameter for the XLP SoC bus. Also fix bus_space_map method
for XLP bus space.
Submitted by: Sreekanth M. <sreekanth.molagavalli@broadcom.com>
adrian [Tue, 21 Aug 2012 06:31:26 +0000 (06:31 +0000)]
Initial support for running FreeBSD on the Nintendo Wii. We're able to
reach single user mode using a memory disk device as the file system.
This port includes the framebuffer driver, the PIC driver, a platform
driver and the GPIO driver. The IPC driver (to talk to IOS kernels) is
not yet written but there's a placeholder for it.
There are still some MMU problems and to get a working system you need to
patch locore32.S. Since we haven't found the best way yet to address that
problem, we're not committing those changes yet. The problem is related to
the different BAT layout on the Wii and to the fact that the Homebrew
loader doesn't clean up the special registers (including the 8 BATs)
before passing control to us.
You'll need a Wii with Homebrew loader and a TV that can do NTSC (for now).
ache [Tue, 21 Aug 2012 06:14:08 +0000 (06:14 +0000)]
According to resolvconf.conf(5) manpage and sources, there is no
'nameservers' option which used in examples in resolvconf.conf(5),
it spelled 'name_servers', so fix examples.
kan [Mon, 20 Aug 2012 23:43:03 +0000 (23:43 +0000)]
Do not call process_nodelete with NULL object pointer.
The place where the function is called can be reached if object loading
and relocation fails too, in which case obj pointer will be NULL. Do not
call process_nodelete then, or crash will follow.
pluknet [Mon, 20 Aug 2012 20:40:14 +0000 (20:40 +0000)]
Avoid segfault in the 'smpphylist' subcommand.
Initialize devlist.dev_queue tail queue early enough before its any
potential traversal in freebusdevlist() when in smpphylist error path.
Reported by: Pavel Polyakov <bsd kobyla org> (on irc)
Reviewed by: ken
MFC after: 5 days