royger [Mon, 26 Jan 2015 08:42:47 +0000 (08:42 +0000)]
amd64: allow base memory segment to start at address different than 0
Current code requires that the first physical memory segment starts at 0,
but this is not really needed. We only need to make sure the bootstrap code
and page tables for APs are allocated below 4GB.
This patch removes this requirement and allows booting a Dell R710 from
UEFI, where the first physical memory segment starts at 0x10000.
gonzo [Sun, 25 Jan 2015 23:58:34 +0000 (23:58 +0000)]
- Increase default i2c bus timeout to 5 seconds from 1 second. Sometimes
1 second is not enugh for TDA19988 HDMI framer (e.g. on Beaglebone Black)
- Add per-device i2c_timout sysctl (dev.iichb.X.i2c_timeout) to control
I2C bus timeout manually
- Pass softc instead of device_t to all sysctl handlers
jhb [Sun, 25 Jan 2015 20:37:32 +0000 (20:37 +0000)]
natd(8) will work with an unconfigured interface and effectively not do
anything until the interface is assigned an address. This fixes
ipfw_nat to do the same by using an IP of INADDR_ANY instead of
aborting the nat setup if the requested interface is not yet configured.
jhb [Sun, 25 Jan 2015 20:16:45 +0000 (20:16 +0000)]
If the boot-time memory test is enabled, output a dot ('.') for
each GB of RAM tested so people watching the console can see that
the machine is making progress and not hung.
PR: 196650
Submitted by: Ravi Pokala <rpokala@panasas.com>
Suggestions from: Eric van Gyzen <eric@vangyzen.net>
MFC after: 2 weeks
jhb [Sun, 25 Jan 2015 19:45:44 +0000 (19:45 +0000)]
Use an sbuf to generate the output of the net.inet.tcp.hostcache.list
sysctl to avoid a possible buffer overflow if the cache grows while the
text is being generated.
markj [Sun, 25 Jan 2015 00:47:06 +0000 (00:47 +0000)]
gr_equal(): Fix a crash that could occur if the first group's member list
was longer than the second's. There is no need to compute and compare the
member list lengths in a separate pass, since we now just return false when
comparing member names if the list lengths are not equal.
ian [Sat, 24 Jan 2015 20:18:37 +0000 (20:18 +0000)]
Reimplement fdt_clock_register_provider() correctly. It turns out you
can't use OF_xref_from_device() to implement the function that registers
the xref association with the device.
will [Sat, 24 Jan 2015 17:32:45 +0000 (17:32 +0000)]
Add vm.panic_on_oom sysctl, which enables those who would rather panic than
kill a process, when the system runs out of memory. Defaults to off.
Usually, this is most useful when the OOM condition is due to mismanagement
of memory, on a system where the applications in question don't respond well
to being killed.
In theory, if the system is properly managed, it shouldn't be possible to
hit this condition. If it does, the panic can be more desirable for some
users (since it can be a good means of finding the root cause) rather than
killing the largest process and continuing on its merry way.
As kib@ mentions in the differential, there is also protect(1), which uses
procctl(PROC_SPROTECT) to ensure that some processes are immune. However,
a panic approach is still useful in some environments. This is primarily
intended as a development/debugging tool.
rstone [Sat, 24 Jan 2015 16:59:38 +0000 (16:59 +0000)]
vmspace_release() may sleep if the last reference is being released,
so add a WITNESS_WARN() to catch cases where it is called with a
non-sleepable lock held.
kib [Sat, 24 Jan 2015 15:33:42 +0000 (15:33 +0000)]
Avoid calling vmspace_free() while owning the process lock. Freeing
of an vm space may require obtaining sleepable locks. Hold the
process to keep the pointer valid, and change trylock to lock, since
there is no longer two process locks owned simultaneously in
vm_pageout_oom().
Note that after the process lock is dropped, process might exec, and
no longer qualify as the owner of biggest vm space.
In collaboration with: rstone
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
br [Sat, 24 Jan 2015 13:07:07 +0000 (13:07 +0000)]
Add support for audio transmitting, include drivers for:
o Digital Audio Multiplexer (AUDMUX)
o Smart Direct Memory Access Controller (SDMA)
o Synchronous Serial Interface (SSI)
Disable by default as it depends on SDMA firmware.
kib [Sat, 24 Jan 2015 12:51:15 +0000 (12:51 +0000)]
Remove Giant from /dev/mem and /dev/kmem. It is definitely not needed
for i386, and from the code inspection, nothing in the
arm/mips/sparc64 implementations depends on it.
Discussed with: imp, nwhitehorn
Sponsored by: The FreeBSD Foundation
MFC after: 3 weeks
luigi [Sat, 24 Jan 2015 08:48:05 +0000 (08:48 +0000)]
do not strip /stand
This is a temporary workaround until the elftoolchain's version
of strip is fixed:
The previous (GNU) strip, when acting on a file with multiple links,
would modify the one and only file in place (which means creating
a new stripped copy, and then writing it back to the original).
The new version from elftoolchain creates the new file and then
unlinks the old one and renames the new.
With multiple hard links, the original remains alive. In the /stand
directory, this ends up creating 80+ copies of the same file.
will [Sat, 24 Jan 2015 00:43:02 +0000 (00:43 +0000)]
When creating or updating a node, use vfs_timestamp() for "now" instead
of gethrestime(), to allow the administrator to decide the appropriate
timestamp precision instead of always using nanosecond precision.
will [Fri, 23 Jan 2015 23:53:56 +0000 (23:53 +0000)]
Fix panic in firewire and creation of invalid config ROM.
(This change was supposed to be included in r277508.)
sys/boot/i386/libfirewire/firewire.c:
Fix configuration ROM generation count wrapping logic
so that the generation count is never outside of
allowed limits (0x2 -> 0xF).
emaste [Fri, 23 Jan 2015 21:34:08 +0000 (21:34 +0000)]
Restore addr2line to cross tools
Addr2line is not required for the build, and a per-arch binary is no
longer required with the switch to the ELF Tool Chain. However, building
these tools during the cross tools stage can be useful for developers
who cross build HEAD from stable/10, and adds very little to the build
time.
jilles [Fri, 23 Jan 2015 21:07:08 +0000 (21:07 +0000)]
Add futimens and utimensat system calls.
The core kernel part is patch file utimes.2008.4.diff from
pluknet@FreeBSD.org. I updated the code for API changes, added the manual
page and added compatibility code for old kernels. There is also audit and
Capsicum support.
A new UTIME_* constant might allow setting birthtimes in future.
gjb [Fri, 23 Jan 2015 21:04:59 +0000 (21:04 +0000)]
When iterating through VMFORMATS, the VMBASE file is
not removed or truncated to a zero-size file, which
if used to create more than one disk image format, can
result in accidental pollution of the target formatted
disk image.
Instead of using a single VMBASE image (vm.img, by
default), use a single base file for each format, named
as VMFORMAT.img, which produces VMBASE.VMFORMAT as the
final formatted image.
Reported by: cperciva
MFC after: 1 month
X-MFC-with: r277458, r277536
Sponsored by: The FreeBSD Foundation
cperciva [Fri, 23 Jan 2015 20:24:13 +0000 (20:24 +0000)]
Provide a recipe of "true" for building the "release" target once its
prerequisites are satisfied, in order to avoid having an implicit rule
triggered by the presence of release.sh.
ian [Fri, 23 Jan 2015 16:05:47 +0000 (16:05 +0000)]
Add pinctrl driver support for the encoded input register config words
that the linux guys made up on the fly (but didn't document) last August.
This type of encoded config now appears in the imx6 fdt data.
will [Fri, 23 Jan 2015 16:03:02 +0000 (16:03 +0000)]
Allow lists for "nooption" and "nooptions" keywords.
usr.sbin/config/config.y
According to config(5), the "device", "devices",
"nodevice", "nodevices", "option", "options",
"nooption", and "nooptions" keywords can all take a
comma-separated list of values. However, the yacc code
did not allow lists for "nooption" and "nooptions", only
single values. This commit fixes the yacc code to allow
comma separated values for all the above keywords.
will [Fri, 23 Jan 2015 15:55:03 +0000 (15:55 +0000)]
Print transfer times for read, write, & overall independently.
Round up calculated values for iops and average time per io to avoid a
shifting display if there are 1000+ (or even 10000+) iops, or if an
average time per io column is 1000+ ms.
Reviewed by: ken
MFC after: 1 week
Sponsored by: Spectra Logic
MFSpectraBSD: 1089947 on 2014/09/26 1093625 on 2014/09/29 1093650 on 2014/09/29 1095662 on 2014/10/09
ae [Fri, 23 Jan 2015 13:26:35 +0000 (13:26 +0000)]
After r270929 RAW IP code assumes that all IP fields are in network
byte order. Fix ping(8) to pass an IP header with converted ip_off
and ip_len fields, when IP_HDRINCL socket option used.
nwhitehorn [Fri, 23 Jan 2015 07:36:51 +0000 (07:36 +0000)]
Use relocation-safe methods to determine the sizes of the exception handlers.
A "size" symbol with its address set to the length of handler would be
shifted forward with all other addresses when relocations are processed.
Instead, just note the end and do the subtraction at runtime.
danfe [Fri, 23 Jan 2015 07:30:57 +0000 (07:30 +0000)]
Fix usage example in kvprintf(9) and its copy in libstand(3): trailing '\n'
in bitfield argument is wrong, as it will be treated as bit 10, causing any
code printing >=10 bits with bit 10 on as having a trailing comma.
Newline (intended one) should be part of the format string (already present
in the examples).
Also fix grammar and kill EOL whitespace in comment while here.
will [Fri, 23 Jan 2015 00:06:35 +0000 (00:06 +0000)]
Improve the distribution of LAGG port traffic.
I edited the original change to retain the use of arc4random() as a seed for
the hashing as a very basic defense against intentional lagg port selection.
The author's original commit message (edited slightly):
sys/net/ieee8023ad_lacp.c
sys/net/if_lagg.c
In lagg_hashmbuf, use the FNV hash instead of the old
hash32_buf. The hash32 family of functions operate one octet
at a time, and when run on a string s of length n, their output
is equivalent to :
The problem is that the last five bytes of input don't get
multiplied by sufficiently many powers of 33 to rollover 2^32.
That means that changing the last few bytes (but obviously not
the very last) of input will always change the value of the
hash by a multiple of 33. In the case of lagg_hashmbuf() with
ipv4 input, the last four bytes are the TCP or UDP port
numbers. Since the output of lagg_hashmbuf is always taken
modulo the port count, and 3 is a common port count for a lagg,
that's bad. It means that the UDP or TCP source port will
never affect which lagg member is selected on a 3-port lagg.
At 10Gbps, I was not able to measure any difference in CPU
consumption between the old and new hash.
gjb [Thu, 22 Jan 2015 19:43:05 +0000 (19:43 +0000)]
Rename the 'release' target to 'real-release', and
add two targets, 'vm-release' and 'cloudware-release',
that are invoked if WITH_VMIMAGES and WITH_CLOUDWARE
are not empty.
This fixes an issue where 'make release' would not
build the cloud provider targets because CLOUDWARE was
not yet set. [1]
Move the WITH_VMIMAGES and WITH_CLOUDWARE targets to
Makefile.vm.
Note: There is no 'cloudware-install' target yet, since
some of the disk image names may need to be specific to
the provider, so this is probably best handled by the
build scripts.
Reported by: cperciva [1]
MFC after: 1 month
X-MFC-with: r277458
Sponsored by: The FreeBSD Foundation
emaste [Thu, 22 Jan 2015 17:53:30 +0000 (17:53 +0000)]
Update reported elftoolchain version
The last elftoolchain update is upstream rev 3136. Update our reported
version and add M (Modified) suffix as our elftoolchain contains
additional fixes not yet committed upstream.