Nathan Whitehorn [Mon, 11 Nov 2013 17:37:50 +0000 (17:37 +0000)]
Follow up r223485, which made AIM use the ABI thread pointer instead of
PCPU fields for curthread, by doing the same to Book-E. This closes
some potential races switching between CPUs. As a side effect, it turns out
the AIM and Book-E swtch.S implementations were the same to within a few
registers, so move that to powerpc/powerpc.
Mark Johnston [Mon, 11 Nov 2013 16:57:57 +0000 (16:57 +0000)]
Consistently add the relocation offset only when the ELF type is not
ET_EXEC. This fixes several problems with the DTrace pid provider not
being able to match probes.
Nathan Whitehorn [Mon, 11 Nov 2013 16:14:25 +0000 (16:14 +0000)]
Rename the "bare" platform "mpc85xx", which is what it actually is, and
add actual platform probing based on PVR. Still needs a little more work:
in particular, the CCRS setup should move here.
Also turn "bare" into a truly bare platform that doesn't pretend to know how
to do anything except get the memory map. This should also be enhanced to
process the FDT reserved memory list, but that is for another day.
Nathan Whitehorn [Mon, 11 Nov 2013 15:43:21 +0000 (15:43 +0000)]
Make tsec work with the device tree present on the RB800. The previous code
assumed that the MDIO bus was a direct child of the Ethernet interface. It
may not be and indeed on many device trees is not. While here, add proper
locking for MII transactions, which may be on a bus shared by several MACs.
Nathan Whitehorn [Mon, 11 Nov 2013 15:23:35 +0000 (15:23 +0000)]
Allow OF_decode_addr() to also be able to map resources on big-endian
devices. To this end, make PCI device detection rely on the device_type
field rather than name, as per the standard.
Nathan Whitehorn [Mon, 11 Nov 2013 15:00:33 +0000 (15:00 +0000)]
Consolidate Apple firmware hacks and improve them by switching on the
presence of mac-io devices in the tree, which uniquely identifies Apple
hardware.
Fix panic with RADIX_MPATH, when RTFREE_LOCKED() called for already
unlocked route. Use in6_rtalloc() instead of in6_rtalloc1. This helps
simplify the code and remove several now unused variables.
Adrian Chadd [Mon, 11 Nov 2013 09:08:22 +0000 (09:08 +0000)]
If A-MPDU transmission fails entirely, then no BA is received from the
NIC and pushed up to the driver. Unfortunately this means there's
no rate control notification done. Thus, if the rate control code
makes a decision that hits a crappy rate that can't succeed, the
rate code would never lower the rate and packet loss would continue.
So, fake some rate control notification in this case.
Adrian Chadd [Mon, 11 Nov 2013 08:53:20 +0000 (08:53 +0000)]
Send EAPOL frames at the management rate, not the data rate.
Without this, a far away station with low signal strength would
associate using the management rate (by default the lowest rate)
and then the EAPOL frames would go out at the current AMRR best
guess. This would result in association failing authentication.
Devin Teske [Mon, 11 Nov 2013 02:13:47 +0000 (02:13 +0000)]
Check the partition scheme before blowing away disks, instead of after.
The effects of this patch would only be noticeable if you were purposefully
setting a bad value and trying to see what happens; and leaving the disks
intact if a bad value has been set seems fair.
Glen Barber [Mon, 11 Nov 2013 01:42:59 +0000 (01:42 +0000)]
Unbreak the installer on head/:
When bsdinstall(8) sources the bsdconfig(8) common.subr file,
PKG_ABI is set by calling 'pkg -vv' and searching for the ABI
pkg(8) will use.
When pkg(8) is run for the first time, the bootstrap process
is run, which prompts for 'y/N' input from stdin if running with
TERM set.
Since TERM is set and it is the first time pkg(8) is run, which
happens automatically, bsdinstall(8) hangs waiting for user input
which is never displayed since a specific line is expected by
awk(1), and stdin is expected by pkg(8).
Set ASSUME_ALWAYS_YES=1, which will cause pkg(8) to assume the
'-y' flag is also used for the bootstrap process, allowing
bsdinstall(8) to proceed to the keymap lookup, otherwise
bsdinstall(8) appears to hang after selecting 'Install' from the
menu on first boot from CDROM.
Alexander Motin [Sun, 10 Nov 2013 23:48:16 +0000 (23:48 +0000)]
Use relaxed (write-only) memory barriers when writing some of queue index
registers (for now on ISP2400+). We never read those registers back and
AFAIK their semantics does not require any immediate reaction on write.
Alexander Motin [Sun, 10 Nov 2013 23:34:32 +0000 (23:34 +0000)]
Some more registers access optimizations:
- Process ATIO queue only if interrupt status tells so;
- Do not update queue out pointers after each processed command, do it
only once at the end of the loop.
Bruce M Simpson [Sun, 10 Nov 2013 19:41:04 +0000 (19:41 +0000)]
Document the RB_FOREACH_FROM() and RB_FOREACH_REVERSE_FROM() macros.
These are largely syntactic sugar. However, they improve code
readability where an RB_FOREACH() or RB_FOREACH_REVERSE()
traversal has been interrupted and must be resumed. Performance
is improved by avoiding unnecessary traversal from the head node.
Alexander Motin [Sun, 10 Nov 2013 13:37:44 +0000 (13:37 +0000)]
Save one more register read per command by not reading rqstoutrp register
every time. The purpose of that register is unlikely output queue overflow
detection, so read it only when its last known (and probably stale now)
value signals overflow.
This reduces CPU load and lock congestion and rises bottleneck in CTL
while doing target mode via two 8Gbps ports from 100K to 120K IOPS.
Alexander Motin [Sun, 10 Nov 2013 12:16:09 +0000 (12:16 +0000)]
Some CAM locks polishing:
- Fix LOR and possible lock recursion when handling high-power commands.
Introduce new lock to protect left power quota and list of frozen devices.
- Correct locking around xpt periph creation.
- Remove seems never used XPT_FLAG_OPEN xpt periph flag.
Rick Macklem [Sat, 9 Nov 2013 21:24:56 +0000 (21:24 +0000)]
Fix an NFSv4.1 client specific case where a forced dismount would hang.
The hang occurred in nfsv4_setsequence() when it couldn't find an
available session slot and is fixed by checking for a forced dismount
in progress and just returning for this case.
If filesystem declares that it supports shared locking for writes, use
shared vnode lock for VOP_PUTPAGES() as well. The only such
filesystem in the tree is ZFS, and it uses
vnode_pager_generic_putpages(), which performs the pageout with
VOP_WRITE().
Reviewed by: alc
Discussed with: avg
Tested by: pho
Sponsored by: The FreeBSD Foundation
MFC after: 2 weeks
Both vn_close() and VFS_PROLOGUE() evaluate vp->v_mount twice, without
holding the vnode lock; vp->v_mount is checked first for NULL
equiality, and then dereferenced if not NULL. If vnode is reclaimed
meantime, second dereference would still give NULL. Change
VFS_PROLOGUE() to evaluate the mp once, convert MNTK_SHARED_WRITES and
MNTK_EXTENDED_SHARED tests into inline functions.
Reviewed by: alc
Tested by: pho
Sponsored by: The FreeBSD Foundation
MFC after: 2 weeks
Adrian Chadd [Sat, 9 Nov 2013 07:30:13 +0000 (07:30 +0000)]
Fix AMRR to correctly select the initial rate.
There were two bugs:
* If the initial lowest rate didn't go through the loop at least once,
the AMRR rate index would be the highest rate in the table
(eg the rix mapping to MCS15) but rate would stay at the default
value, namely 0.
This meant that the initial rate selection would be MCS15 _but_ the
node ni_txrate value would be MCS0.
* If the node is 11n, then break out of the loop correctly. Beforehand,
my initial 11n AMRR commit would immediately exit out as it would
fail the 11n check, then it would always fall through to the non-11n
rate which would then see if it was < 36mbit (ie, "72"), which would
always match. Hence, it'd always return MCS15.
Tested:
* Intel Centrino 2230 STA (local changes), STA mode
* Intel Wifi 5100, STA
Colin Percival [Sat, 9 Nov 2013 04:50:05 +0000 (04:50 +0000)]
Fix typo in r256646: We want to generate lists of directories in INDEX-OLD
and INDEX-NEW and compare them, not generate the same list of directories
from INDEX-OLD twice...
Pointy hats to: cperciva & everybody who didn't proofread EN-13:04 enough
Mark Johnston [Sat, 9 Nov 2013 04:38:16 +0000 (04:38 +0000)]
Don't try to use the 32-bit drti.o unless the data model is explicitly set
to ILP32. Otherwise dtrace -G will attempt to use it on amd64 if it can't
determine which data model to use, which happens when -64 is omitted and
no object files are provided, e.g. with
# dtrace -G -n BEGIN
This would result in a linker error, but now works properly.
Justin T. Gibbs [Sat, 9 Nov 2013 03:07:48 +0000 (03:07 +0000)]
On XenServer the "halt" message is used instead of "poweroff", which
makes FreeBSD halt but not poweroff (as expected when issuing a
shutdown from the VM manager). Fix this by using the same handler
for both "halt" and "poweroff".
NB: The "halt" signal seems to be used on XenServer only. The OSS
Xen toolstack (xl) uses "poweroff" instead.
Submitted by: Roger Pau Monné
Sponsored by: Citrix Systems R&D
Reviewed by: gibbs
MFC after: 2 days
Ian Lepore [Sat, 9 Nov 2013 00:15:36 +0000 (00:15 +0000)]
This change builds kernel tools based on the same assumption as building
the kernel itself: If building for the same architecture as the build host,
the kernel build assumes that the host toolchain is capable of building the
kernel. If it's not, "make kernel-toolchain" will bootstrap a new set of
tools that will work.
With this change the same assumptions are made for building kernel tools,
and the existing host toolchain is used to do the build (notably, the build
doesn't link the tools with the legacy libraries, which may not even exist).
If ever for some reason the host toolchain isn't capable of building the
kernel tools, then doing a "make kernel-toolchain" will bootstrap newer
tools to get the job done.
So when built as part of buildworld or kernel-toolchain, the kernel tools
are built using the XMAKE (via BMAKE) commands and environment. When built
as part of building just the kernel on a same-target host, the tools are
built using the new KTMAKE commands and environment. What doesn't jump
out at you in the diffs is that the difference between BMAKE and KTMAKE
is that BMAKE contains this magic line which changes how the build is done
because it changes what files get included for .include <bsd.prog.mk> and
other standard includes:
Hopefully this brings the "how to build aicasm with the right toolchain"
saga to a conclusion that works in all usage scenarios that have
historically been supported.
Glen Barber [Fri, 8 Nov 2013 17:27:38 +0000 (17:27 +0000)]
Remove generate-release.sh from head/.
The release.sh (based heavily on generate-release.sh) has been
used for the 9.2-RELEASE and 10.0-RELEASE cycles, so make sure
there is no confusion on what is currently being used by having
two similar scripts.
A big "thank you" to Nathan Whitehorn, the author of the
generate-release.sh script, for writing this utility.
No objection: nwhitehorn
MFC after: never
Sponsored by: The FreeBSD Foundation
Fix signal delivery for the iBCS2 binaries. The iBCS2 sysvec uses
current FreeBSD signal trampoline, but does not specifies
sv_sigcode_base, since shared page is not mapped. This results in the
zero %eip for the signal frame. Fall back to calculating %eip as
offset from the psstrings when sv_sigcode_base is not initialized.
Reported by: Rich Naill <rich@enterprisesystems.net>
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
Alan Cox [Fri, 8 Nov 2013 16:25:00 +0000 (16:25 +0000)]
As of r257209, all architectures have defined VM_KMEM_SIZE_SCALE. In other
words, every architecture is now auto-sizing the kmem arena. This revision
changes kmeminit() so that the definition of VM_KMEM_SIZE_SCALE becomes
mandatory and the definition of VM_KMEM_SIZE becomes optional.
Replace or eliminate all existing definitions of VM_KMEM_SIZE. With
auto-sizing enabled, VM_KMEM_SIZE effectively became an alternate spelling
for VM_KMEM_SIZE_MIN on most architectures. Use VM_KMEM_SIZE_MIN for
clarity.
Change kmeminit() so that the effect of defining VM_KMEM_SIZE is similar to
that of setting the tunable vm.kmem_size. Whereas the macros
VM_KMEM_SIZE_{MAX,MIN,SCALE} have had the same effect as the tunables
vm.kmem_size_{max,min,scale}, the effects of VM_KMEM_SIZE and vm.kmem_size
have been distinct. In particular, whereas VM_KMEM_SIZE was overridden by
VM_KMEM_SIZE_{MAX,MIN,SCALE} and vm.kmem_size_{max,min,scale}, vm.kmem_size
was not. Remedy this inconsistency. Now, VM_KMEM_SIZE can be used to set
the size of the kmem arena at compile-time without that value being
overridden by auto-sizing.
Update the nearby comments to reflect the kmem submap being replaced by the
kmem arena. Stop duplicating the auto-sizing formula in every machine-
dependent vmparam.h and place it in kmeminit() where auto-sizing takes
place.
Julio Merino [Fri, 8 Nov 2013 14:29:06 +0000 (14:29 +0000)]
Install category Kyuafiles from their category directories.
Move the installation of /usr/tests/lib/Kyuafile from src/tests/lib/
to src/lib/. This is to keep the src/tests/ hierarchy unaware of the
rest of the tree, which makes things clearer in general. In particular:
1) Everything related to the construction of /usr/tests/lib/ is kept
in src/lib/. There is no need to think about different directories
and how they relate to each other. (The same applies for libexec,
usr.bin, etc. but these are not yet handled.)
2) src/tests becomes the place to keep cross-functional test programs
and nothing else, which also helps in simplifying things.
Julio Merino [Fri, 8 Nov 2013 14:26:52 +0000 (14:26 +0000)]
Handle the removal of the test suite when WITHOUT_TESTS=yes.
Add all files from /usr/tests to the obsoleted files list when the
build of the tests is disabled via the WITHOUT_TESTS knob. Do this
automatically so that we do not have to suffer the pain of maintaining
such list by hand.