marius [Wed, 31 Jan 2018 21:56:23 +0000 (21:56 +0000)]
Account for the fact that jemalloc 5.0.0 dropped STATIC_PAGE_SHIFT
in favor for using LG_PAGE directly and, thus, for the fact that
host and target don't necessarily use pages of the same sizes.
wulf [Wed, 31 Jan 2018 21:46:37 +0000 (21:46 +0000)]
psm(4): Reduce psm watchdog verbosity
Modern touchpads do not issue interrupts on inactivity so "lost interrupt"
message became annoying spam nowadays. This change quiets the message
if debug.psm.loglevel=5 (or less) is set in /boot/loader.conf
wulf [Wed, 31 Jan 2018 21:14:59 +0000 (21:14 +0000)]
psm(4): Add support for HP EliteBook 1040 ForcePads.
ForcePads do not have any physical buttons, instead they detect click
based on finger pressure. Forcepads erroneously report button click
if there are 2 or more fingers on the touchpad breaking multifinger
gestures. To workaround this start reporting a click only after
4 consecutive single touch packets has been received. Skip these packets
in case more contacts appear.
jhb [Wed, 31 Jan 2018 18:03:40 +0000 (18:03 +0000)]
Update limits on makecontext() arguments in the setcontext_link test.
sparc64 and riscv do not support 10 arguments, but MIPS now does.
While here, combine clauses for architectures that support the same
number of arguments to reduce duplication.
jhb [Wed, 31 Jan 2018 18:02:02 +0000 (18:02 +0000)]
Add a new set of simple tests for makecontext().
In contrast to the existing NetBSD setcontext_link test, these tests
verify that passing from 1 to 6 arguments through to the callback function
work correctly which can be useful for testing ABIs which split arguments
between registers and the stack.
jhb [Wed, 31 Jan 2018 18:00:23 +0000 (18:00 +0000)]
Remove limitation of 6 arguments for makecontext() on mips.
This implementation spills additional arguments on the stack so works
fine with more than 6 arguments. I believe the check was just copied
over from sparc64 (which doesn't support spilling onto the stack)
jhb [Wed, 31 Jan 2018 17:57:59 +0000 (17:57 +0000)]
Remove bogus checks against NCARGS.
NCARGS isn't a limit on the number of arguments to pass to a function,
but the number of bytes that can be consumed by arguments to exec. As
such, it is not suitable for a limit on the count of arguments passed
to makecontext().
jhb [Wed, 31 Jan 2018 17:36:39 +0000 (17:36 +0000)]
Consistently use 16-byte alignment for MIPS N32 and N64.
- Add a new <machine/abi.h> header to hold constants shared between C
and assembly such as CALLFRAME_SZ.
- Add a new STACK_ALIGN constant to <machine/abi.h> and use it to
replace hardcoded constants in the kernel and makecontext(). As a
result of this, ensure the stack pointer on N32 and N64 is 16-byte
aligned for N32 and N64 after exec(), after pthread_create(), and
when sending signals rather than 8-byte aligned.
kib [Wed, 31 Jan 2018 14:36:27 +0000 (14:36 +0000)]
IBRS support, AKA Spectre hardware mitigation.
It is coded according to the Intel document 336996-001, reading of the
patches posted on lkml, and some additional consultations with Intel.
For existing processors, you need a microcode update which adds IBRS
CPU features, and to manually enable it by setting the tunable/sysctl
hw.ibrs_disable to 0. Current status can be checked in sysctl
hw.ibrs_active. The mitigation might be inactive if the CPU feature
is not patched in, or if CPU reports that IBRS use is not required, by
IA32_ARCH_CAP_IBRS_ALL bit.
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
Differential revision: https://reviews.freebsd.org/D14029
avg [Wed, 31 Jan 2018 11:14:26 +0000 (11:14 +0000)]
vmm/svm: post LAPIC interrupts using event injection, not virtual interrupts
The virtual interrupt method uses V_IRQ, V_INTR_PRIO, and V_INTR_VECTOR
fields of VMCB to inject a virtual interrupt into a guest VM. This
method has many advantages over the direct event injection as it
offloads all decisions of whether and when the interrupt can be
delivered to the guest. But with a purely software emulated vAPIC the
advantage is also a problem. The problem is that the hypervisor does
not have any precise control over when the interrupt is actually
delivered to the guest (or a notification about that). Because of that
the hypervisor cannot update the interrupt vector in IRR and ISR in the
same way as real hardware would. The hypervisor becomes aware that the
interrupt is being serviced only upon the first VMEXIT after the
interrupt is delivered. This creates a window between the actual
interrupt delivery and the update of IRR and ISR. That means that IRR
and ISR might not be correctly set up to the point of the
end-of-interrupt signal.
The described deviation has been observed to cause an interrupt loss in
the following scenario. vCPU0 posts an inter-processor interrupt to
vCPU1. The interrupt is injected as a virtual interrupt by the
hypervisor. The interrupt is delivered to a guest and an interrupt
handler is invoked. The handler performs a requested action and
acknowledges the request by modifying a global variable. So far, there
is no VMEXIT and the hypervisor is unaware of the events. Then, vCPU0
notices the acknowledgment and sends another IPI with the same vector.
The IPI gets collapsed into the previous IPI in the IRR of vCPU1. Only
after that a VMEXIT of vCPU1 occurs. At that time the vector is cleared
in the IRR and is set in the ISR. vCPU1 has vAPIC state as if the
second IPI has never been sent.
The scenario is impossible on the real hardware because IRR and ISR are
updated just before the interrupt handler gets started.
I saw several possibilities of fixing the problem. One is to intercept
the virtual interrupt delivery to update IRR and ISR at the right
moment. The other is to deliver the LAPIC interrupts using the event
injection, same as legacy interrupts. I opted to use the latter
approach for several reasons. It's equivalent to what VMM/Intel does
(in !VMX case). It appears to be what VirtualBox and KVM do. The code
is already there (to support legacy interrupts).
Another possibility was to use a special intermediate state for a vector
after it is injected using a virtual interrupt and before it is known
whether it was accepted or is still pending.
That approach was implemented in https://reviews.freebsd.org/D13828
That method is more complex and does not have any clear advantage.
Please see sections 15.20 and 15.21.4 of "AMD64 Architecture
Programmer's Manual Volume 2: System Programming" (publication 24593,
revision 3.29) for comparison between event injection and virtual
interrupt injection.
adrian [Wed, 31 Jan 2018 07:36:51 +0000 (07:36 +0000)]
[arswitch] Fix ATU flushing on AR8216/AR8316 and most of the later chips.
The switch hardware requires this bit to be set in order to kick start the
actual ATU update. This was being masked on some chips by the learning
programming (what to do when a MAC address moves, hash table collision, etc)
which is currently inconsistent between chips.
imp [Wed, 31 Jan 2018 05:07:43 +0000 (05:07 +0000)]
Update stand.h for changes for strto*l
Move prototypes to proper section now that we don't have modified
versions of strtol and strtoul in libsa. Add prototypes for new
strtoll and strtoull. Use prototypes copied from stdlib.h instead of
the old hand-rolled ones.
(I forgot to move this file form my lua branch in r328613)
imp [Wed, 31 Jan 2018 04:29:00 +0000 (04:29 +0000)]
Kill copies of strtol and strtoul. Use the ones that are in libc,
since they suffice. Create xlocale_private.h which provides the most
minimal locale implementation we can get away with. Add strtoll and
strtoull from libc.
imp [Wed, 31 Jan 2018 03:05:14 +0000 (03:05 +0000)]
Move strtold wrapper from strtol.c to its own strtold.c. This code
was written by theraven@ (David Chisnall) entirely, there's no
original Berkeley code left here so just copy his copyright over.
mav [Wed, 31 Jan 2018 01:04:36 +0000 (01:04 +0000)]
Try to preallocate receive memory early.
We may not have enough contiguous memory later, when NTB connection get
established. It is quite likely that NTB windows are symmetric and this
allocation remain, but even if not, we will just reallocate it later.
jhb [Tue, 30 Jan 2018 23:01:37 +0000 (23:01 +0000)]
Export tcp_always_keepalive for use by the Chelsio TOM module.
This used to work by accident with ld.bfd even though always_keepalive
was marked as static. LLD honors static more correctly, so export this
variable properly (including moving it into the tcp_* namespace).
asomers [Tue, 30 Jan 2018 21:25:43 +0000 (21:25 +0000)]
zfsd: Don't spare a vdev that's being replaced
If a zfs pool contains a replacing vdev (either created manually by "zpool
replace" or by zfsd(8) via autoreplace by physical path) and then new spares
get added to the pool, zfsd shouldn't use one to replace the drive that is
already being replaced. That's a waste of resources that just slows down
the rebuild.
hselasky [Tue, 30 Jan 2018 12:38:06 +0000 (12:38 +0000)]
Move the mlx5 core device pointer first in the mlx5en priv. This help simplify
checks to recognize own network devices when using mlx5ib. This patch fixes
an issues where mlx5ib fails to recognize mceX network devices for use with
RoCE.
trasz [Tue, 30 Jan 2018 10:08:11 +0000 (10:08 +0000)]
Make the handler routine for the hw.usb.template sysctl trigger the USB
host to reprobe the bus by switching the USB pull up resistors off and
back on. In other words - when FreeBSD is configured as a USB device,
changing the sysctl will be immediately noticed by the machine it's
connected to.
Reviewed by: hselasky@
MFC after: 2 weeks
Sponsored by: The FreeBSD Foundation
manu [Tue, 30 Jan 2018 09:59:52 +0000 (09:59 +0000)]
nfsstat: Add libxo output
Add libxo output support
Merge exp41_intpr and exp_intpr function. The only difference is to print
NFSV4.1 operations in exp41, add a third arguement to control that.
printtitle was set to 1 and don't have a switch, add a -q options to control it.
pfg [Mon, 29 Jan 2018 22:38:23 +0000 (22:38 +0000)]
libedit: sort the Makefile in line with NetBSD's version.
NetBSD's libedit has been been cleaned-up considerably so the
non--widecharacter version is no longer an option. Re -sorting the
Makefile should make it easier for some brave soul trying to update it.
kevans [Mon, 29 Jan 2018 18:50:45 +0000 (18:50 +0000)]
Remove t_grep:mmap_eof_not_eol test
The test was marked as an expected failure in r320414 after r319971's import
of a newer jemalloc removed an essential feature (opt.redzone) for
reproducing the behavior it was testing. Since then, no way has been found
or demonstrated to reliably test the behavior, so remove the test.
imp [Mon, 29 Jan 2018 18:07:14 +0000 (18:07 +0000)]
Do the book-keeping on release before we release the reference. The
periph was going away on final release, and then returning and we
started dancing in free memory.
benno [Mon, 29 Jan 2018 17:32:30 +0000 (17:32 +0000)]
Remove some duplicated sys/conf/files* entries.
net80211/ieee80211_ageq.c was present twice in sys/conf/files so leave the
correctly sorted one. dev/wpi/if_wpi.c was present in sys/conf/files as well
as sys/conf/files.amd64 and sys/conf/files.i386 so prefer the sys/conf/files
entry.
vangyzen [Mon, 29 Jan 2018 16:12:26 +0000 (16:12 +0000)]
ND6: Set the correct state for new neighbor cache entries
Restore state 6. Many of the UNH tests end up exercising this
state, where we have a new neighbor cache entry and a new link-layer
entry is being created for it. The link-layer address is currently
unknown so the initial state of the "llentry" should remain initialized
to ND6_LLINFO_NOSTATE so that the ND code will send a solicitation.
Setting this to ND6_LLINFO_STALE implies that the link-level entry
is valid and can be used (but needs to be refreshed via the Neighbor
Unreachability state machine).
kevans [Mon, 29 Jan 2018 14:15:44 +0000 (14:15 +0000)]
awk(1): Don't install tests at all
Tests were disconnected so that running `make check` in usr.bin/awk did not
have any effect, but CI runs use installed tests. Fully disconnect tests/
from the build for the time being as a short term solutio
kevans [Mon, 29 Jan 2018 14:00:33 +0000 (14:00 +0000)]
libregex: Mark gnuext test as an expected fail
The test was added prematurely as a goal to reach with the GNU extension
functionality, but the functionality has not yet been introduced. Mark it as
an expected fail until that point.
emaste [Mon, 29 Jan 2018 13:55:50 +0000 (13:55 +0000)]
lld: Put the header in the first PT_LOAD even if that PT_LOAD has a LMAExpr
The root problem is that we were creating a PT_LOAD just for the header.
That was technically valid, but inconvenient: we should not be making
the ELF discontinuous.
The solution is to allow a section with LMAExpr to be added to a PT_LOAD
if that PT_LOAD doesn't already have a LMAExpr.
LLVM PR: 36017
Obtained from: LLVM r323625 by Rafael Espindola
emaste [Mon, 29 Jan 2018 13:54:51 +0000 (13:54 +0000)]
lld: Move LMAOffset from the OutputSection to the PhdrEntry. NFC.
If two sections are in the same PT_LOAD, their relatives offsets,
virtual address and physical addresses are all the same.
[Rafael] initially wanted to have a single global LMAOffset, on the
assumption that every ELF file was in practiced loaded contiguously in
both physical and virtual memory.
Unfortunately that is not the case. The linux kernel has:
The delta for all but the third PT_LOAD is the same:
0xffffffff80000000. [Rafael] thinks the 3rd one is a hack for implementing
per cpu data, but we can't break that.
ae [Mon, 29 Jan 2018 11:03:29 +0000 (11:03 +0000)]
Do not skip scope zone violation check, when mbuf has M_FASTFWD_OURS flag.
When mbuf has M_FASTFWD_OURS flag, this means that a destination address
is our local, but we still need to pass scope zone violation check,
because protocol level expects that IPv6 link-local addresses have
embedded scope zone indexes. This should fix the problem, when ipfw is
used to forward packets to local address and source address of a packet
is IPv6 LLA.
ae [Mon, 29 Jan 2018 10:33:55 +0000 (10:33 +0000)]
Assign IPv6 link-local address to loopback interfaces whith unit > 0.
When an interface has IFF_LOOPBACK flag in6_ifattach() tries to assing
IPv6 loopback address to this interface. It uses in6ifa_ifpwithaddr()
to check, that interface doesn't already have given address and then
uses in6_ifattach_loopback(). If in6_ifattach_loopback() fails, it just
exits and thus skips assignment of IPv6 LLA.
Fix this using in6ifa_ifwithaddr() function. If IPv6 loopback address is
already assigned in the system, do not call in6_ifattach_loopback().
wma [Mon, 29 Jan 2018 09:24:28 +0000 (09:24 +0000)]
loader: support for mixed-endianness ELF/loader and POWER8
On POWER8 with current petitpoot, the loader.kboot might be
run as little-endian application. The FreeBSD kernel is
always big-endian, so the load_elf_* routines must be aware
of proper endianness of all fields.
wma [Mon, 29 Jan 2018 09:15:38 +0000 (09:15 +0000)]
PPC64: use hwref instead of cpuid
On CHRP and PowerNV, use the interrupt server number in the cpuref and pcpu
hwref field instead of the device-tree phandle and make the CPU IDs reported
to the scheduler dense and with the BSP at 0.
wma [Mon, 29 Jan 2018 08:10:03 +0000 (08:10 +0000)]
PPC64: cleanup APs startup routines
Cleaning up AP startup routines. This is a mix of changes
required to make PowerNV running and to modify the code
to be more robust. Previously, some races were seen if more
than 90CPUs were online.
nwhitehorn [Mon, 29 Jan 2018 04:33:41 +0000 (04:33 +0000)]
Remove hard-coded trap-handling logic involving the segmented memory model
used with hashed page tables on AIM and place it into a new, modular pmap
function called pmap_decode_kernel_ptr(). This function is the inverse
of pmap_map_user_ptr(). With POWER9 radix tables, which mapping to use
becomes more complex than just AIM/BOOKE and it is best to have it in
the same place as pmap_map_user_ptr().
imp [Mon, 29 Jan 2018 00:14:39 +0000 (00:14 +0000)]
Create deprecation management functions.
gone_in(majar, msg); If we're running in FreeBSD major, tell
the user this code may be deleted soon.
If we're running in FreeBSD major - 1,
the the user is deprecated and will
be gone in major.
Otherwise say nothing.
gone_in_dev(dev, major, msg) Just like gone_in, except use device_printf.
New tunable / sysctl debug.oboslete_panic: 0 - don't panic,
1 - panic in major or newer , 2 - panic in major - 1 or newer
default: 0
if NO_OBSOLETE_CODE is defined, then both of these turn into compile
time errors when building for major. Add options NO_OBSOLETE_CODE to
kernel build system.
This lets us tag code that's going away so users know it will be gone,
as well as automatically manage things.
imp [Mon, 29 Jan 2018 00:00:52 +0000 (00:00 +0000)]
Use atomic load and stores to ensure that the compiler doesn't
optimize away these loops. Change boolean to int to match what atomic
API supplies. Remove wmb() since the atomic_store_rel() on status.done
ensure the prior writes to status. It also fixes the fact that there
wasn't a rmb() before reading done. This should also be more efficient
since wmb() is fairly heavy weight.
jhibbits [Sun, 28 Jan 2018 20:35:48 +0000 (20:35 +0000)]
Start building modules for QORIQ64
There's no reason not to build modules for 64-bit QorIQ devices. This
config has evolved to be analogous to the AIM GENERIC64 kernel, so will grow
to match it in more ways as well.
pfg [Sun, 28 Jan 2018 19:37:30 +0000 (19:37 +0000)]
Avoid implicit gcc nonnull attribute in vwarnx().
We removed the nonnull attributes from our headers long ago, but still
__printflike() includes it implicitly. This will cause the NULL check to
be optimized away in higher -O levels and it will also trigger a
-Wnonnull-compare warning.
jhibbits [Sun, 28 Jan 2018 19:18:40 +0000 (19:18 +0000)]
Consolidate trap instruction checks to a single function
Summary:
Rather than duplicating the checks for programmatic traps all over the code, put
it all in one function. This helps to remove some of the #ifdefs between AIM
and Book-E.
mmel [Sun, 28 Jan 2018 15:20:45 +0000 (15:20 +0000)]
diff --git a/sys/dev/extres/clk/clk.c b/sys/dev/extres/clk/clk.c
index c6a1f466ceb..c3708a0ce27 100644
--- a/sys/dev/extres/clk/clk.c
+++ b/sys/dev/extres/clk/clk.c
@@ -642,10 +642,11 @@ clknode_adjust_parent(struct clknode *clknode, int idx)
if (clknode->parent_cnt == 0)
return;
if ((idx == CLKNODE_IDX_NONE) || (idx >= clknode->parent_cnt))
- panic("Invalid clock parent index\n");
+ panic("%s: Invalid parent index %d for clock %s",
+ __func__, idx, clknode->name);
if (clknode->parents[idx] == NULL)
- panic("%s: Attempt to set invalid parent %d for clock %s",
+ panic("%s: Invalid parent index %d for clock %s",
__func__, idx, clknode->name);
/* Remove me from old children list. */
@@ -674,8 +675,8 @@ clknode_init_parent_idx(struct clknode *clknode, int idx)
if ((idx == CLKNODE_IDX_NONE) ||
(idx >= clknode->parent_cnt) ||
(clknode->parent_names[idx] == NULL))
- panic("%s: Invalid clock parent index: %d\n", __func__, idx);
-
+ panic("%s: Invalid parent index %d for clock %s",
+ __func__, idx, clknode->name);
clknode->parent_idx = idx;
}
mmel [Sun, 28 Jan 2018 15:02:49 +0000 (15:02 +0000)]
Fix handling of I-cache sync operations
- pmap_enter_object() can be used for mapping of executable pages, so it's
necessary to handle I-cache synchronization within it.
- Fix race in I-cache synchronization in pmap_enter(). The current code firstly
maps given page to target VA and then do I-cache sync on it. This causes
race, because this mapping become visible to other threads, before I-cache
is synced.
Do sync I-cache firstly (by using DMAP VA) and then map it to target VA.
- ARM64 ARM permits implementation of aliased (AIVIVT, VIPT) I-cache, but we
can use different that final VA for flushing it. So we should use full
I-cache flush on affected platforms. For now, and as temporary solution,
use full flush always.
imp [Sun, 28 Jan 2018 05:13:08 +0000 (05:13 +0000)]
Avoid using \$. It's an unknown escape sequence. Some awks warn about
that. Instead, simply remove the sequence entirely because we never
commit the generated files.
The warning is bogus: GCC8 only looks at the size of the destination.
We shouldn't be fixing imaginary problems, so perhaps its better to deal
with this later on by disabling such warnings.
kevans [Sun, 28 Jan 2018 03:07:22 +0000 (03:07 +0000)]
stand/fdt: Check /compatible property on overlay if it exists
Example overlays seen in other places use a compatible property on root node
of an overlay to specify SOC compatibility. These don't get merged into base
FDT as they're not part of a fragment, but it's expected that consumers of
the overlay may want to check it.
If /compatible on the overlay is missing, just apply it. This is the "I know
what I'm doing" mode for those wanting to whip up a quick overlay and apply
it. An overlay intended for distribution should include /compatible so as
not to break a user's system.
If /compatible on the overlay exists, honor it and cross-check it with
/compatible on the base FDT. If /compatible on the base FDT is missing in
this case, don't apply the overlay rather than risk breaking the system.
Move the COPYOUT of overlay material to before we allocate space for
next_fdtp so that we can avoid the allocation and copy into next_fdtp if we
already know that the overlay can't apply.
This gives way to the possibility of autoloading overlays found in
/boot/overlays, since this provides a means of filtering out overlays not
applicable to the current board.
kevans [Sun, 28 Jan 2018 01:22:15 +0000 (01:22 +0000)]
stand/fdt: Consolidate overlay handling a little further
This should have been done as part of r327350, but due to lack of foresight
it came later. In the different places we apply overlays, we duplicate the
bits that check for fdt_overlays in the environment and supplement that with
any other places we need to check for overlays to load. These "other places"
will be loader specific and are not candidates for consolidation.
Provide an fdt_load_dtb_overlays to capture the common logic, allow passing
in an additional list of overlays to be loaded. This additional list of
overlays is used in practice for ubldr to pull in any fdt_overlays passed to
it from U-Boot environment, but it can be used for any other source of
overlays.
These additional overlays supplement loader.conf(5) fdt_overlays, rather
than replace, so that we're not restricted to specifying overlays in only
one place. This is a change from previous behavior where loader.conf(5)
supplied fdt_overlays would cause us to ignore U-Boot environment, and this
seems nonsensical- user should have sufficient control over both of these
aspects, or lack of control for good reasons.
A knob could be considered in the future to ignore U-Boot supplied overlays,
but the supplemental treatment seems like a good start.