https://www.illumos.org/issues/10554
During the port of MMP (illumos bug 10499) from ZoL, I found this
earlier ZoL project is a prerequisite. Here is the original
description. This addition will enable us to sync an open TXG to the
main pool on demand. The functionality is similar to 'sync(2)' but
'zpool sync' will return when data has hit the main storage instead of
potentially just the ZIL as is the case with the 'sync(2)' cmd.
Toomas Soome [Thu, 7 Nov 2019 11:17:03 +0000 (11:17 +0000)]
loader: implement fallback efi_devpath_to_name()
UEFI 1.10 on macs does not seem to provide devpath to name translation,
provide our own (limited) version, so we can get information about commmon
devices.
Dimitry Andric [Thu, 7 Nov 2019 06:59:09 +0000 (06:59 +0000)]
Merge commit 8e34dd941 from llvm git (by Sanjay Patel):
[x86] avoid crashing when splitting AVX stores with non-simple type
(PR43916)
The store splitting transform was assuming a simple type (MVT), but
that's not necessarily the case as shown in the test.
This should fix 'Assertion failed: (isSimple() && "Expected a
SimpleValueType!")' when building the security/openssl111 port targeting
a CPU that supports AVX, but not AVX2, such as sandybridge.
Phil Shafer [Thu, 7 Nov 2019 03:57:04 +0000 (03:57 +0000)]
Import libxo-1.3.0:
- move from "oxtradoc" to RST/Sphinx documentation
- new "csv" encoder, which allows path and leaf lists
- address warnings from PVS-Stdio tool
- add "xolint" detected errors to the documentation
Gleb Smirnoff [Thu, 7 Nov 2019 00:27:23 +0000 (00:27 +0000)]
TCP timers are executed in callout context, so they need to enter network
epoch to look into PCB lists. Mechanically convert INP_INFO_RLOCK() to
NET_EPOCH_ENTER(). No functional change here.
Gleb Smirnoff [Wed, 6 Nov 2019 23:45:43 +0000 (23:45 +0000)]
If vm_pager_get_pages_async() returns an error synchronously we leak wired
and busy pages. Add code that would carefully cleanups the state in case
of synchronous error return. Cover a case when a first I/O went on
asynchronously, but second or N-th returned error synchronously.
Conrad Meyer [Wed, 6 Nov 2019 23:44:38 +0000 (23:44 +0000)]
clang: Enable unwind tables on !amd64
There doesn't seem to be much sense in defaulting "on" unwind tables on
amd64 and not on other arches. It causes surprising differences between
platforms, such as the PR below.
Prior to this change, FreeBSD inherited the default implementation of the
method from the Gnu.h Generic_Elf => Generic_GCC parent class, which
returned true only for amd64 targets. Override that and opt on always,
similar to, e.g., NetBSD.
PR: 241562
Reported by: lwhsu
Reviewed by: dim
Discussed with: emaste
MFC after: I'm not going to, but you should feel free
Relnotes: yes
Differential Revision: https://reviews.freebsd.org/D22252
Bjoern A. Zeeb [Wed, 6 Nov 2019 22:40:19 +0000 (22:40 +0000)]
m_pulldown(): Change an if () panic() into a KASSERT().
If we pass in a NULL mbuf to m_pulldown() we are in a bad situation
already. There is no point in doing that check for production code.
Change the if () panic() into a KASSERT.
Toomas Soome [Wed, 6 Nov 2019 21:13:10 +0000 (21:13 +0000)]
loader.efi: HARDDRIVE_DEVICE_PATH may have subpaths
The macos does create Vendor Media devices on top of APFS container
(like partition table inside the partition), so we need to collect such
devices into respective device tree.
Make linux(4) create /dev/shm. Linux applications often expect
a tmpfs to be mounted there, and because they like to verify it's
actually a mountpoint, a symlink won't do.
Reviewed by: dchagin (earlier version)
MFC after: 2 weeks
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D20333
Ed Maste [Wed, 6 Nov 2019 19:51:40 +0000 (19:51 +0000)]
Introduce if_mgb driver for Microchip LAN743x PCIe NIC
The Microchip LAN7430 is a PCIe 10/100/1000 Ethernet MAC with integrated
PHY, and the LAN7431 is a MAC with RGMII interface.
To be connected to the build after further testing and review.
Committing now so that changes like r354345 (adding a common
ETHER_IS_ZERO macro) will update this driver too.
Submitted by: Gerald ND Aryeetey <aryeeteygerald_rogers.com>
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D20079
Brooks Davis [Wed, 6 Nov 2019 19:44:44 +0000 (19:44 +0000)]
libstats: Improve ABI assertion.
On platforms where pointers are larger than 64-bits, struct statsblob
may be harmlessly padded out such that opaque[] always has some included
space. Make the assertion more general by comparing to the offset of
opaque rather than the size of struct statsblob.
Warner Losh [Wed, 6 Nov 2019 18:54:04 +0000 (18:54 +0000)]
Remove obsolete addition of MACHINE_CPUARCH
In the past, we would add symbolic links for MACHINE_CPUARCH when it differed
from MACHINE. This was for pc98 only, however. All other architectures didn't
need this and it was really due to pc98 pulling from i386 rather than something
more intrinsic. At the time, we had the split we did to mimic what NetBSD did
for its 68k ports where many different kernels were possible for the same
architecture. Since then, both projects have moved away from this convention to
having a more generic MACHINE for each architecture. FreeBSD's new arm64/aarch64
breaks this old notion and so was an exception to the rule. So, we no longer
need to create this link for any old machine or any new machine, delete it
entirely.
Mark Johnston [Wed, 6 Nov 2019 16:59:16 +0000 (16:59 +0000)]
Fix a race in release_page().
Since r354156 we may call release_page() without the page's object lock
held, specifically following the page copy during a CoW fault.
release_page() must therefore unbusy the page only after scheduling the
requeue, to avoid racing with a free of the page. Previously, the
object lock prevented this race from occurring.
Add some assertions that were helpful in tracking this down.
Reported by: pho, syzkaller
Tested by: pho
Reviewed by: alc, jeff, kib
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D22234
Emmanuel Vadot [Wed, 6 Nov 2019 14:58:25 +0000 (14:58 +0000)]
arm: allwinner: aw_mmc: Check if the regulator support the voltage
Don't blindy say that we support both 3.3V and 1.8V.
If we have a regulator for the data lines, check that the voltage is
supported before adding the signaling caps.
If we don't have a regulator, just assume that the data lines are 3.3V
This unbreak eMMC on some allwinner boards.
Emmanuel Vadot [Wed, 6 Nov 2019 14:55:24 +0000 (14:55 +0000)]
regulator: Add regulator_check_voltage function
This function will call the regnode_check_voltage method for a given regulator
and check if the desired voltage in reachable by it.
Also adds a default method that check the std_param and which should be enough
for most regulators and add it as the method for axp* rk805 and fixed regulators.
Bjoern A. Zeeb [Wed, 6 Nov 2019 14:30:56 +0000 (14:30 +0000)]
ipv6 tests: cleanup
Remove mentions of fragmentation tests from extension header test.
Remove setting an MTU > IF_MAXMTU from the test cases to avoid warnings;
this was only possible in a local research tree.
https://www.illumos.org/issues/10554
During the port of MMP (illumos bug 10499) from ZoL, I found this
earlier ZoL project is a prerequisite. Here is the original
description. This addition will enable us to sync an open TXG to the
main pool on demand. The functionality is similar to 'sync(2)' but
'zpool sync' will return when data has hit the main storage instead of
potentially just the ZIL as is the case with the 'sync(2)' cmd.
Portions contributed by: Jerry Jelinek <jerry.jelinek@joyent.com>
Author: Alek Pinchuk <apinchuk@datto.com>
Justin Hibbits [Wed, 6 Nov 2019 04:40:12 +0000 (04:40 +0000)]
powerpc/booke: Fix pmap_mapdev_attr() for multi-TLB1 entry mappings
Also, fix pmap_change_attr() to ignore non-kernel mappings.
* Fix a masking bug in mmu_booke_mapdev_attr() which caused it to align
mappings to the smallest mapping alignment, instead of the largest. This
caused mappings to be potentially pessimally aligned, using more TLB
entries than necessary.
* Return existing mappings from mmu_booke_mapdev_attr() that span more than
one TLB1 entry. The drm-current-kmod drivers map discontiguous segments
of the GPU, resulting in more than one TLB entry being used to satisfy the
mapping.
* Ignore non-kernel mappings in mmu_booke_change_attr(). There's a bug in
the linuxkpi layer that causes it to actually try to change physical
address mappings, instead of virtual addresses. amd64 doesn't encounter
this because it ignores non-kernel mappings.
With this it's possible to use drm-current-kmod on Book-E.
Andriy Gapon [Tue, 5 Nov 2019 14:34:59 +0000 (14:34 +0000)]
fix up r354333, make zfsproc visible to dtrace, rename to system_proc
I overlooked the fact that zfsproc is required by dtrace modules that
use illumos compatible taskq KPI. So, move the symbol definition to
the opensolaris module that provides compatibility support for both ZFS
and DTrace. Also, rename zfsproc to system_proc to reflect that it is
not specific to ZFS.
Bjoern A. Zeeb [Tue, 5 Nov 2019 10:59:33 +0000 (10:59 +0000)]
ip6 tests: Add very simplistic extension header tests
Add a simple test case which can exercise some of the IPv6 extension
header code paths. At the moment only a small set of extension headers
is implemented and no options to the ones which take them.
Also implements a "bad" case to make sure that error handling works.
The tests were used to test m_pullup() changes to the code paths while
removing the KAME PULLDOWN_TEST cases and related macros.
Bjoern A. Zeeb [Tue, 5 Nov 2019 10:53:55 +0000 (10:53 +0000)]
frag6 tests: set end to catch timeout as error
There are times when we have to wait for reply packets. There are
either an ICMPv6 (error) reply or the expiration timeout.
In these cases synchonous ICMPv6 replies should arrive, always,
unless the packet is lost. Due to errors experienced with the
test software sending an invlaid request on at least i386 (*) these
packets are not generated. That means we are waiting for a long time
for the replies or even timeout the test case.
Manually set the "End" flag on these test cases as well, so they do
fail rather than timeout as the sniffer timeout happens. This improves
debugging options, reflects the error properly, and saves time on each
test suit run.
(*) The real cause for that is still to be found (see the referenced PRs)
Andriy Gapon [Tue, 5 Nov 2019 06:34:20 +0000 (06:34 +0000)]
if_ixv: disable RSS configuration on 82599 and X540 VFs
It is reported that those VFs share their RSS configuration with PF and,
thus, they cannot be configured independently.
Also:
- add missing opt_rss.h to if_ixv.c, otherwise RSS kernel option could
not be seen
- do not enable IXGBE_FEATURE_RSS on the older VFs
- set flowid / hash type to M_HASHTYPE_NONE or M_HASHTYPE_OPAQUE_HASH
(based on what the hardware reports) if IXGBE_FEATURE_RSS is not set
Conrad Meyer [Tue, 5 Nov 2019 03:20:40 +0000 (03:20 +0000)]
Fix llvm-libunwind userspace build on ARM
GCC's libgcc exports a few ARM-specific symbols for ARM EABI, AEABI, or
EHABI or whatever it's called. Export the same ones from LLVM-libunwind's
libgcc_s, on ARM. As part of this, convert libgcc_s from a direct
Version.map to one constructed from component Symbol.map files. This allows
the ARM-specific Symbol.map to be included only on ARM.
Fix ARM-only oddities in struct name/aliases in LLVM-libunwind to match
non-ARM definitions and ARM-specific expectations in libcxxrt /
libcompiler_rt.
No functional change intended for non-ARM architectures.
This commit does not actually flip the switch for ARM defaults from libgcc
to llvm-libunwind, but makes it possible (to compile, anyway).
Kyle Evans [Tue, 5 Nov 2019 02:30:41 +0000 (02:30 +0000)]
Remove picobsd build scripts and the remaining tinyware
Postings were sent to -arch@ on 2019/09/13 and 2019/10/01, proposing and
confirming a removal of these scripts on 2019/10/31, due to significant work
needed to bring this into the modern world and nobody having done this work
in the past couple of years. No objections or proposed work was raised in
response to these postings. The tinyware may see a resurrection into a
separate repo for archival purposes if any users of it show interest in
doing so.
Eric Joyner [Tue, 5 Nov 2019 00:12:21 +0000 (00:12 +0000)]
net: add ETHER_IS_ZERO macro similar to ETHER_IS_BROADCAST
Some places in network code may need to verify that an ethernet address
is not the 'zero' address. Provide a standard macro ETHER_IS_ZERO for
this purpose, similar to the ETHER_IS_BROADCAST macro already available.
This patch also removes previous ETHER_IS_ZERO definitions in several
USB ethernet drivers, in favor of this centrally-located macro.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Submitted by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed by: erj@
Sponsored by: Intel Corporation
Differential Revision: https://reviews.freebsd.org/D21240
Eric Joyner [Mon, 4 Nov 2019 23:06:57 +0000 (23:06 +0000)]
iflib: properly release memory allocated for DMA
DMA memory allocations using the bus_dma.h interface are not properly
released in all cases for both Tx and Rx. This causes ~448 bytes of
M_DEVBUF allocations to be leaked.
First, the DMA maps for Rx are not properly destroyed. A slight attempt
is made in iflib_fl_bufs_free to destroy the maps if we're detaching.
However, this function may not be reliably called during detach. Indeed,
there is a comment "asking" if this should be moved out.
Fix this by moving the bus_dmamap_destroy call into iflib_rx_sds_free,
where we already sync and unload the DMA.
Second, the DMA tag associated with the ifr_ifdi descriptor DMA is not
released properly anywhere. Add a call to iflib_dma_free in
iflib_rx_structures_free.
Third, use of NULL as a canary value on the map pointer returned by
bus_dmamap_create is not valid. On some platforms, notably x86, this
value may be NULL. In this case, we fail to properly release the related
resources.
Remove the NULL checks on map values in both iflib_fl_bufs_free and
iflib_txsd_destroy.
With all of these fixes applied, the leaks to M_DEVBUF are squelched,
and iflib drivers now seem to properly cleanup when detaching.
Dimitry Andric [Mon, 4 Nov 2019 19:38:07 +0000 (19:38 +0000)]
Merge commit 97e362607 from llvm git (by Nemanja Ivanovic):
[PowerPC] Do not emit HW loop if the body contains calls to
lrint/lround
These two intrinsics are lowered to calls so should prevent the
formation of CTR loops. In a subsequent patch, we will handle all
currently known intrinsics and prevent the formation of HW loops if
any unknown intrinsics are encountered.
Andrew Gallatin [Mon, 4 Nov 2019 19:30:19 +0000 (19:30 +0000)]
Add tunable to allow interrupts on hyperthreaded cores
Enabling interrupts on htt cores has benefits to workloads which are primarily
interrupt driven by increasing the logical cores available for interrupt handling.
The tunable is named machdep.hyperthreading_intr_allowed
Gleb Smirnoff [Mon, 4 Nov 2019 17:35:37 +0000 (17:35 +0000)]
In nd6_timer() enter the network epoch earlier. The defrouter_del() may
call into leaf functions that require epoch. Since the function is already
run in non-sleepable context, it should be safe to cover it whole with epoch.
Andriy Gapon [Mon, 4 Nov 2019 13:30:37 +0000 (13:30 +0000)]
zfs: enable SPA_PROCESS on the kernel side
The purpose of this change is to group kernelthreads specific to a
particular ZFS pool under a kernel process. There can be many dozens of
threads per pool. This change improves observability of those threads.
This change consists of several subchanges:
1. illumos taskq_create_proc can now pass its process parameter to
taskqueue. Also, use zfsproc instead of NULL for taskq_create. Caveat:
zfsproc might not be initialized yet. But in that case it is still NULL,
so not worse than before.
2. illumos sys/proc.h: kthread id is stored in t_did field, not t_tid.
3. zfs: enable SPA_PROCESS on the kernel side. The change is a bit hairy
as newproc() is implemented privately to spa.c. I couldn't think of a
better way to populate process name than to poke inside the argument for
the process routine.
4. illumos thread_create: allow assigning thread to process other than
zfsproc.
5. zfs: expose spa_proc to other users, assign sync and quiesce threads
to it.
Pool-specific threads created using (relatively new) zthr mechanism are
still assigned to the zfskern process rather than to a respective
zpool-xxx process. I am going to address this a bit later.
Reviewed by: no one
MFC after: 5 weeks
Relnotes: perhaps
Differential Revision: https://reviews.freebsd.org/D9720
Kyle Evans [Mon, 4 Nov 2019 03:07:01 +0000 (03:07 +0000)]
patch(1): give /dev/null patches special treatment
We have a bad habit of duplicating contents of files that are sourced from
/dev/null and applied more than once... take the more sane (in most ways)
GNU route and complain if the file exists and offer reversal options.
This still falls short a little bit as selecting "don't reverse, apply
anyway" will still give you duplicated file contents. There's probably other
issues as well, but awareness is the first step to happiness.
Justin Hibbits [Mon, 4 Nov 2019 00:35:40 +0000 (00:35 +0000)]
powerpc/pmap: Make use of tlb1_mapin_region in pmap_mapdev_attr()
tlb1_mapin_region() and pmap_mapdev_attr() do roughly the same thing -- map
a chunk of physical address space(memory or MMIO) into virtual, but do it in
differing ways. Unify the code, settling on pmap_mapdev_attr()'s algorithm,
to simplify and unify the logic. This fixes a bug with growing the kernel
mappings in mmu_booke_bootstrap(), where part of the mapping was not getting
done, leading to a hang when the unmapped VAs were accessed.
Andrew Turner [Sun, 3 Nov 2019 22:17:49 +0000 (22:17 +0000)]
Move the struct debug_monitor_state out of _KERNEL.
Some userland libraries incude machine/pcb.h and this needs the full
definition of struct debug_monitor_state. To allow this to work move
stuct debug_monitor_state out of the _KERNEL guard.
Handle error bits of INTR_STAT and TX_ABORT registers.
Move interrupt clearing from interrupt handler to polling loop to get
common execution path with polled mode.
Do not clear interrupts with reading of IG4_REG_CLR_INTR register as
interrupts, triggered during the period from reg_read(IG4_REG_INTR_STAT)
to reg_read(IG4_REG_CLR_INTR) will be missed.
Instead, read each IG4_REG_CLR_* register separately.
[ig4] Convert polling loop from status-based to interrupt-based
INTR_STAT register exposes more useful informaton then STA register does
e.g. it exposes error and I2C bus STOP conditions. Make it a main source
of I2C transfer state.
[ig4] Set clock registers based on controller model
IC clock rates are varied between different controller models so we have
to adjust timing registers in each case individually. Borrow intresting
constants and formulas from Intel specs, i2c-designware and lpss_intel
drivers and apply them to FreeBSD supported controller models.
Implement fetching of timing data via ACPI methods execution if available.
[cyapa] Postpone start of the polling thread until sleep is available
After recent ig4 changes cyapa driver can be attached before timers
initialization is completed. Start polling thread from config_intrhook
to avoid busy loops in that case.
[ig4] Do not wait until interrupts are enabled at attach stage
as the driver is fully functional on a cold boot through utilization of
polled mode.
As a side effect, ig4 children probe and attach methods can be called
earlier in the boot sequence, so now it is up to the child drivers
to wait for a kernel initialization completion if it is required.
[ig4] Allow enabling of polled mode from iicbus allocation callback
If controller is allocated with IIC_NOWAIT option ig4 enables polled mode
for a period of allocation that makes possible to start I2C transfers
from the contexts where sleeping is not allowed e.g. from ithreads or
callouts.
Currently ig4 internally depends on it's own interrupts and uses mtx_sleep()
to wait for them. That means it can not be used from any context where
sleeping is disallowed e.g. on cold boot, from DDB/KDB, from other device
driver's interrupt handlers and so on.
This change replaces sleeps with busy loops in cold boot and DDB cases.
[ig4] We actually need to set the Rx threshold register one smaller.
Setting the IG4_REG_RX_TL register to 1 was actually generating an
interrupt after 2 bytes were available in the Rx fifo. We need to set the
register to 0 to get an interrupt for 1 byte already.
Now io_lock is used as condition variable to synchronize active process with
the interrupt handler. It is not used for tasks other than waiting for
interrupt and passing parameters to and from it's handler.
There is no need to read all controller's RX FIFO data to clear RX_FULL
bit in interrupt handler as interrupts are masked permanently since
previous commit.
projects / FreeBSD / FreeBSD.git / log