cem [Mon, 16 Mar 2020 22:25:25 +0000 (22:25 +0000)]
Remove misleading / redundant bzero in callout_callwheel_init
The intent seems to be zeroing all of the cc_cpu array, or its singleton on
such platforms. The assumption made is that the BSP is always zero. The
code smell was introduced in r326218, which changed the prior explicit zero
to 'curcpu'. The change is only valid if curcpu continues to be zero,
contrary to the aim expressed in that commit message.
So, more succinctly, the expression could be: memset(cc_cpu,0,sizeof(cc_cpu)).
However, there's no point. cc_cpu lives in the data section and has a zero
initial value already. So this revision just removes the problematic
statement.
No functional change. Appeases a (false positive, ish) Coverity CID.
mav [Mon, 16 Mar 2020 19:03:10 +0000 (19:03 +0000)]
Fix infinite scan on a pool with only special allocations
Attempt to run scrub or resilver on a new pool containing only special
allocations (special vdev added on creation) caused infinite loop
because of dsl_scan_should_clear() limiting memory usage to 5% of pool
size, which it calculated accounting only normal allocation class.
Addition of special and just in case dedup classes fixes the issue.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov> Signed-off-by: Alexander Motin <mav@FreeBSD.org> Sponsored-By: iXsystems, Inc.
Closes #10106
Closes #8694
openzfs/zfs@fa130e010c2ff9b33aba11d2699b667e454b3ccb
pkelsey [Sat, 14 Mar 2020 20:11:46 +0000 (20:11 +0000)]
Adjust if_vmx default receive parameters for better out-of-box performance
These adjustments improve performance with jumbo frames and/or LRO
enabled (i.e., when there may be multiple descriptors per packet) by
increasing the default size of the receive queues and by always using
page-sized buffers for the body type receive ring.
This patch also adjust the initialization of the max frame size to
remove cases where certain configuration sequences would result in 2K
receive buffers being used instead of 4K ones when jumbo frames were
enabled.
pkelsey [Sat, 14 Mar 2020 20:08:04 +0000 (20:08 +0000)]
Fix if_vmx receive checksum offload bug and harden against the device skipping receive descriptors
This fixes a bug where the checksum offload status of received packets
was being taken from the first descriptor instead of the last, which
affected LRO packets.
The driver has been hardened against the device skipping receive
descriptors, although it is not believed that this can occur given the
way this implementation configures the receive rings.
Additionally, for packets received with the error indicator set, the
driver now forces the length of all fragments in that packet to zero
prior to passing it to iflib. Such packets should wind up being
discarded at some point in the stack anyway, but this removes any
questions by killing them in the driver.
Counters have been added (and exposed via sysctls) for skipped receive
descriptors, zero-length packets received, and packets received with
the error indicator set so that these conditions can be easily
observed in the field.
pkelsey [Sat, 14 Mar 2020 19:55:05 +0000 (19:55 +0000)]
Remove freelist contiguous-indexes assertion from rxd_frag_to_sd()
The vmx driver is an example of an iflib driver that might report
packets using non-contiguous descriptors (with unused descriptors
either between received packets or between the fragments of a received
packet), so this assertion needs to be removed.
For such drivers, the freelist producer and consumer indexes don't
relate directly to driver ring slots (the driver deals directly with
freelist buffer indexes supplied by iflib during refill, and reports
them with each fragment during packet reception), but do continue to
be used by iflib for accounting, such as determining the number of
ring slots that are refillable.
pkelsey [Sat, 14 Mar 2020 19:51:55 +0000 (19:51 +0000)]
Fix iflib zero-length fragment handling
The dmamap for zero-length fragments should not be unloaded, as doing
so breaks the the cluster-reuse logic in _iflib_fl_refill().
All zero-length fragments are now handled by the assemble_segments()
path so that the cluster-reuse logic there does not have to be
replicated in the small-single-fragment-packet path of
iflib_rxd_pkt_get().
Packets consisting entirely of zero-length fragments (which result in
a NULL mbuf pointer) are now properly tolerated. This allows drivers
(such as the vmx driver) to pass such packets to iflib when a
descriptor error occurs during packet reception, the advantage being
that the refill of descriptors associated with the error packet are
handled via the existing iflib machinery without having to duplicate
parts of that machinery in the driver to handle that error case.
imp [Sat, 14 Mar 2020 18:35:44 +0000 (18:35 +0000)]
Sparc64 and pc98 needed cam_machdep and/or ata_machdep for various
reasons. Those have now left the tree, and with them the need to have machdep
files. Places that called the routines in quesiton have been removed
previously. Remove these files from the Makefile to tidy up.
kevans [Sat, 14 Mar 2020 15:15:27 +0000 (15:15 +0000)]
libssp: don't compile with -fstack-protector*
This similarly matches what we do in libc; compiling libssp with
-fstack-protector* is actively harmful. For instance, if the canary ctor
ends up with a stack protector then it will trivially trigger a false
positive as the canary's being initialized.
This was noted by the reporter as irc/ircd-hybrid started crashing at start
after our libssp was MFC'd to stable/11, as its build will explicitly link
in libssp. On FreeBSD, this isn't necessary as SSP bits are included in
libc, but it should absolutely not trigger runtime breakage -- it does mean
that the canary will get initialized twice, but as this is happening early
on in application startup it should just be redundant work.
Reported by: Tod McQuillin <devin@sevenlayer.studio>
MFC after: 3 days
bz [Sat, 14 Mar 2020 14:04:55 +0000 (14:04 +0000)]
kern_jail: missing \0 termination check on osrelease parameter
If a user spplies a non-\0 terminated osrelease parameter reading it back
may disclose kernel memory.
This is a problem in case of nested jails (children.max > 0, which is not
the default). Otherwise root outside the jail has access to kernel memory
by other means and root inside a jail cannot create a child jail.
Add the proper \0 check at the end of a supplied osrelease parameter and
make sure any copies of the field will be \0-terminated.
Submitted by: Hans Christian Woithe (chwoithe yahoo.com)
MFC after: 3 days
delphij [Sat, 14 Mar 2020 05:57:22 +0000 (05:57 +0000)]
Remove unneeded checks for prelen.
In order to determine the type of a compressed file, we have to read
in the first four bytes which may also be important for decompression
purposes, to do that we would pass the buffer that we have already
read in, along with the size of it.
Rename header1 to fourbytes to make that explicit, and remove all
checks for prelen.
Reported by: cem
Reviewed by: cem
MFC after: 2 weeks
Differential Revision: https://reviews.freebsd.org/D24034
imp [Sat, 14 Mar 2020 02:36:45 +0000 (02:36 +0000)]
Add comment about how the deferred callback for AC_FOUND_DEVICE we
generate for a race where a device goes away, we start to tear down
the periph state for the device, and then the device suddently
reappears. The key that makes it work is removal of periph from the
drv list before calling the deferred callback.
imp [Fri, 13 Mar 2020 18:45:36 +0000 (18:45 +0000)]
Give some indication of the CCB that's in flight when we panic.
Print the pointer to ccb so we can find it (for what good it does)
as well as the type of operation in flight when the cam_path has
been freed out from under us. This helps both core analysis as well
as automated systems that collect panic strings but little else.
emaste [Fri, 13 Mar 2020 15:40:35 +0000 (15:40 +0000)]
flua: implement chmod
Lua does not provide a native way to change the permission of a file.
Submitted by: Yang Wang <2333@outlook.jp>
Reviewed by: kevans
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D24036
0mp [Fri, 13 Mar 2020 11:07:48 +0000 (11:07 +0000)]
CMSG_DATA.3: Fix formatting of printf escape sequences
Escape sequences like "\n" have to be escaped twice in examples in our
mdoc(7)-based manual pages in order to be displayed properly. The problem
is that otherwise they are interpreted by mdoc(7), which results in:
freqlabs [Thu, 12 Mar 2020 23:04:40 +0000 (23:04 +0000)]
libpmcstat: Try /boot/modules if module not found
Modules from ports/pkg are commonly installed to /boot/modules rather than to
the same directory the kernel resides in. Look there if a module is not found
next to the kernel.
Submitted by: mmacy
Reported by: Nick Principe <nap@iXsystems.com>
Approved by: mmacy (mentor)
MFC after: 2 weeks
Sponsored by: iXsystems, Inc.
tuexen [Thu, 12 Mar 2020 15:37:41 +0000 (15:37 +0000)]
Use KMOD_TCPSTAT_INC instead of TCPSTAT_INC for RACK and BBR, since
these are kernel modules. Also add a KMOD_TCPSTAT_ADD and use that
instead of TCPSTAT_ADD.
emaste [Thu, 12 Mar 2020 13:42:08 +0000 (13:42 +0000)]
Extend r358907 to explicitly remove stale lib32 dependency
After r325072 stale lib32 dependencies were not remooved. A more
holistic approach is needed to address this but for the immediate issue
(-DNO_CLEAN builds across r358851) just readd the explicit lib32 path.
Reported by: dim
Sponsored by: The FreeBSD Foundation
dim [Thu, 12 Mar 2020 11:39:04 +0000 (11:39 +0000)]
Allow -DNO_CLEAN build across r358851.
The openmp 10.0.0 import renamed one .c file to .cpp, and this is
something our dependency system does not handle correctly. Add another
ad-hoc cleanup to get rid of the stale dependency.
tsoome [Thu, 12 Mar 2020 06:45:08 +0000 (06:45 +0000)]
test if port does exist via using scratch register
The SCR, scratch register was not present on the 8250 and 8250B UART, so we
can use to test if we actually do have serial port.
We need this test because some systems will get long delays while attempting
to write to non-existing port and this will slow down the console IO
to extreme.
mckusick [Wed, 11 Mar 2020 21:00:40 +0000 (21:00 +0000)]
When mounting a UFS filesystem, return EINTEGRITY rather than EIO
when a superblock check-hash error is detected. This change clarifies
a mount that failed due to media hardware failures (EIO) from a mount
that failed due to media errors (EINTEGRITY) that can be corrected by
running fsck(8).
wulf [Wed, 11 Mar 2020 20:05:49 +0000 (20:05 +0000)]
wmt(4): Reapply r358872 (by hselasky) modified to use
maximal input report size instead of wMaxPacketSize.
If the USB frame length is set to 1024 bytes, WMT_BSIZE, the EETI controller
will pack multiple touch events in the packet and the current code will only
process the first touch event.
As a result some important events are lost like releasing the finger from the
touchscreen.
Use the maximal input report size as buffer size instead.
PR: 244718
Tested by: Oskar Holmlund <oskar.holmlund@ohdata.se>, wulf
MFC after: 3 days
Discussed with: hselasky
wulf [Wed, 11 Mar 2020 19:57:43 +0000 (19:57 +0000)]
wmt(4): revert r358872 (by hselasky)
Limiting frame size to maximum packet size breaks devices which have input
report size larger than wMaxPacketSize. Maximal input report size should be
used instead.
Revert the commit as it have not been MFC-ed yet.
imp [Wed, 11 Mar 2020 15:12:31 +0000 (15:12 +0000)]
Remove redundantly repetitive static __inline forward function
declarations.
We typically don't use them elsewhere in the kernel, and they aren't
needed here: the actual functions are a few lines away and aren't
mutually recursive.
hselasky [Wed, 11 Mar 2020 08:16:13 +0000 (08:16 +0000)]
If the USB frame length is set to 1024 bytes, WMT_BSIZE, the EETI controller
will pack multiple touch events in the packet and the current code will only
process the first touch event.
As a result some important events are lost like releasing the finger from the
touchscreen.
Use the maximum maximum packet size as buffer size instead.
Submitted by: Oskar Holmlund <oskar.holmlund@ohdata.se>
PR: 244718
MFC after: 3 days
Sponsored by: Mellanox Technologies
asomers [Wed, 11 Mar 2020 04:29:45 +0000 (04:29 +0000)]
fusefs: avoid cache corruption with buggy fuse servers
The FUSE protocol allows the client (kernel) to cache a file's size, if the
server (userspace daemon) allows it. A well-behaved daemon obviously should
not change a file's size while a client has it cached. But a buggy daemon
might. If the kernel ever detects that that has happened, then it should
invalidate the entire cache for that file. Previously, we would not only
cache stale data, but in the case of a file extension while we had the size
cached, we accidentally extended the cache with zeros.
PR: 244178
Reported by: Ben RUBSON <ben.rubson@gmx.com>
Reviewed by: cem
MFC after: 2 weeks
Differential Revision: https://reviews.freebsd.org/D24012
imp [Tue, 10 Mar 2020 23:59:58 +0000 (23:59 +0000)]
Eliminate xpt_copy_path.
It's used in exactly one place. In that place it's used so we can hold the lock
on the device associated with the path (since we do a xpt_path_lock and unlock
pair around the callback). Instead, inline taking and dropping the reference to
the device so we can ensure we can unlock the mutex after the callback finishes
if the path in the ccb that's queued to be processed by xpt_scanner_thread is
destroyed while being processed. We don't actually need the path itself for
anything other than dereferencing it to get the device to do the lock and
unlock.
This also makes the locking / use model for cam_path a little cleaner by
eliminating a case where we needlessly copy the object.
Reviewed by: chuck, chs, ken
Differential Revision: https://reviews.freebsd.org/D24008
imp [Tue, 10 Mar 2020 23:58:41 +0000 (23:58 +0000)]
Remove unused cam ccb flags
These flags have been unused for some time. Some of them were in the
CAM2 specification, but CAM has moved on a bit from that. Some were
used in the old Pluto VideoSpace (and AirSpace) systems which had the
video playback I/O scheduler in userspace, but have been unused since
then.
Reviewed by: chuck, ken
Differential Revision: https://reviews.freebsd.org/D24008
kib [Tue, 10 Mar 2020 20:25:03 +0000 (20:25 +0000)]
Fix signal delivery might be on sigfastblock clearing.
When clearing sigfastblock, either by sigfastblock(UNSETPTR) call or
implicitly on execve(2), kernel must check for pending signals and
reschedule them if needed.
E.g. on execve, all other threads are terminated, and current thread
fast block pointer is cleaned. If any signal was left pending, it can
now be delivered to the current thread, and we should prepare for
ast() on return to userspace to notice the signals.
Reported and tested by: pho
Sponsored by: The FreeBSD Foundation
kib [Tue, 10 Mar 2020 20:04:38 +0000 (20:04 +0000)]
Return reschedule_signals() to being static again.
It was used after sigfastblock_setpend() call in in ast() when current
thread fast-blocks signals. Add a flag to sigfastblock_setpend() to
request reschedule, and remove the direct use of the function from
subr_trap.c
Tested by: pho
Sponsored by: The FreeBSD Foundation
mr [Tue, 10 Mar 2020 16:53:49 +0000 (16:53 +0000)]
Untangle TPR shadowing and APIC virtualization.
This speeds up Windows guests tremendously.
The patch does:
Add a new tuneable 'hw.vmm.vmx.use_tpr_shadowing' to disable TLP shadowing.
Also add 'hw.vmm.vmx.cap.tpr_shadowing' to be able to query if TPR shadowing is used.
Detach the initialization of TPR shadowing from the initialization of APIC virtualization.
APIC virtualization still needs TPR shadowing, but not vice versa.
Any CPU that supports APIC virtualization should also support TPR shadowing.
When TPR shadowing is used, the APIC page of each vCPU is written to the VMCS_VIRTUAL_APIC field of the VMCS
so that the CPU can write directly to the page without intercept.
On vm exit, vlapic_update_ppr() is called to update the PPR.
dim [Tue, 10 Mar 2020 06:49:43 +0000 (06:49 +0000)]
Remove -mlong-calls vhen building arm libraries and llvm.
Clang from 9.0.0 onwards already has the necessary relocation range
extenders, so this workaround is no longer needed (it produces longer
and slower code). Tested on real hardware, and in cross-compile
environment.
jhibbits [Tue, 10 Mar 2020 03:30:11 +0000 (03:30 +0000)]
powerpc/booke: Split out 32- and 64- bit pmap details from main body
Summary:
This is largely a straight-forward cleave of the 32-bit and 64-bit page
table specifics, along with the mmu_booke_*() functions that are wholely
different between the two implementations.
The ultimate goal of this is to make it easier to reason about and
update a specific implementation without wading through the other
implementation details. This is in support of further changes to the 64-bit
pmap.
kib [Mon, 9 Mar 2020 21:55:26 +0000 (21:55 +0000)]
Preallocate pipe buffers on pipe creation.
Return ENOMEM if one of the buffer cannot be created even with the
minimal size. This should avoid subsequent spurious ENOMEM errors
from write(2) when buffer cannot be allocated on the fly, after we
reported that the pipe was create succesfully.
emaste [Mon, 9 Mar 2020 20:46:43 +0000 (20:46 +0000)]
retire amd(8)
autofs was introduced with FreeBSD 10.1 and is the supported method for
automounting filesystems. As of r296194 the amd man page claimed that it
is deprecated. Remove it from base now; the sysutils/am-utils port is
still available if necessary.
Discussed with: cy
Relnotes: Yes
Sponsored by: The FreeBSD Foundation
wulf [Mon, 9 Mar 2020 20:31:38 +0000 (20:31 +0000)]
iicbus(4): Add support for ACPI-based children enumeration
When iicbus is attached as child of Designware I2C controller it scans all
ACPI nodes for "I2C Serial Bus Connection Resource Descriptor" described
in section 19.6.57 of ACPI specs.
If such a descriptor is found, I2C child is added to iicbus, it's I2C
address, IRQ resource and ACPI handle are added to ivars. Existing
ACPI bus-hosted child is deleted afterwards.
The driver also installs so called "I2C address space handler" which is
disabled by default as nontested.
Set hw.iicbus.enable_acpi_space_handler loader tunable to 1 to enable it.