zrj [Sat, 2 Mar 2019 09:14:04 +0000 (11:14 +0200)]
upgrade: Clarify deprecated files message.
Suggested-by: swildner
François Tigeot [Sun, 3 Mar 2019 20:10:07 +0000 (21:10 +0100)]
drm/linux/rbtree.h: Rename RB_ROOT to LINUX_RB_ROOT
Preventing clashes with the more general RB_ROOT kernel definition
Sascha Wildner [Sun, 3 Mar 2019 18:25:41 +0000 (19:25 +0100)]
Remove some IPSec remains via 'make upgrade'.
Reported-by: zrj
Sascha Wildner [Sun, 3 Mar 2019 14:41:48 +0000 (15:41 +0100)]
kernel/ixgbe: Fix buildkernel, re-apply
9e2814118aabe1f24cf00485.
Sepherosa Ziehau [Mon, 25 Feb 2019 15:03:32 +0000 (23:03 +0800)]
ix: Sync with Intel ixgbe-3.3.6
- Honor VLAN_HWTAGGING setting when setup RX VLAN tag stripping.
- Add workaround for possible FW hang on X550.
Matthew Dillon [Sat, 2 Mar 2019 22:45:33 +0000 (14:45 -0800)]
kernel - Add callout debugging
* We still have a few issues where the callout lists become corrupt.
Add debugging to store the originator of a callout by file/line in
the structure itself as a debugging aid.
Sascha Wildner [Sat, 2 Mar 2019 22:25:00 +0000 (23:25 +0100)]
<semaphore.h>: Fix guard comment.
Sascha Wildner [Sat, 2 Mar 2019 20:34:21 +0000 (21:34 +0100)]
kernel: Remove numerous #include <sys/thread2.h>.
Most of them were added when we converted spl*() calls to
crit_enter()/crit_exit(), almost 14 years ago. We can now
remove a good chunk of them again for where crit_*() are
no longer used.
I had to adjust some files that were relying on thread2.h
or headers that it includes coming in via other headers
that it was removed from.
Sascha Wildner [Sat, 2 Mar 2019 10:27:24 +0000 (11:27 +0100)]
kernel/sound: Remove local KOBJMETHOD_END defines.
Sascha Wildner [Sat, 2 Mar 2019 10:25:00 +0000 (11:25 +0100)]
<sys/systm.h>: No need to include <sys/callout.h> here.
zrj [Sat, 2 Mar 2019 08:21:47 +0000 (10:21 +0200)]
Makefile.inc1: Minor fixes for CTOOLS handling.
Simplify logic a bit, remove no longer needed variables and duplicates.
Fix gnu/lib/Makefile (WORLD_CCVER should not be used in tree context).
Aaron LI [Sat, 2 Mar 2019 02:17:33 +0000 (10:17 +0800)]
gpt(8): "show" command support to print partition GUID
The "show" subcommand gains the "-g" option to print the partition GUID
instead of the partition type/label.
This helps to identify the GPT partition corresponding to the UUIDs
shown in "/dev/slice-by-uuid".
Aaron LI [Sat, 2 Mar 2019 02:12:13 +0000 (10:12 +0800)]
gpt(8): Improve "show" subcommand
* Mark unused space with "Unused".
* Catch unknown partition types and print "Unknown"
* Show active MBR partition with the "(active)" mark. This can be
helpful for a hybrid GPT/MBR setup.
Obtained-from: NetBSD
Sascha Wildner [Fri, 1 Mar 2019 15:38:39 +0000 (16:38 +0100)]
Adjust various things for the recent brk(2) removal.
Sascha Wildner [Fri, 1 Mar 2019 13:30:07 +0000 (14:30 +0100)]
kernel/ena: Fix another error path.
Release the MSI-X vector if we fail after it's been allocated but
before we can use the goto.
Sascha Wildner [Fri, 1 Mar 2019 13:24:13 +0000 (14:24 +0100)]
kernel/ena: Fix an uninitialized variable issue in an error path.
At this point in initialization, we just return ENXIO on failure.
Reported-by: zrj
François Tigeot [Thu, 28 Feb 2019 21:23:40 +0000 (22:23 +0100)]
drm/linux: Add rcu_assign_pointer()
Sascha Wildner [Thu, 28 Feb 2019 16:10:48 +0000 (17:10 +0100)]
libc: Remove old MLINKS for fpclassify.3.
I forgot to remove them in
b6b35a1b4c682b25f983fc039f153dcd04094189.
Reported-by: zrj
zrj [Thu, 28 Feb 2019 15:17:17 +0000 (17:17 +0200)]
GC: Remove old clang 3.8 that required extern sources.
In preparations for libc++ import.
zrj [Thu, 28 Feb 2019 15:07:53 +0000 (17:07 +0200)]
gcc50: Mark as deprecated.
If gcc50 will not be further used as alternative compiler and it's
private dynamic libraries are no longer needed, it can be safely
removed from the base system using: make upgrade -DREMOVE_DEPRECATED
zrj [Thu, 28 Feb 2019 13:27:17 +0000 (15:27 +0200)]
Revert "kernel - Temporarily change WORLD_ALTCOMPILER default in master"
This reverts commit
6ef7b2b34b7cd49e3e26cf7316d8217e102e6b75.
Issue should be resolved by now. Build only gcc47 as alternative compiler
by default (for time being gcc50 can still be hooked into the build).
This reduces buildworld time by ~6 minutes.
Aaron LI [Thu, 28 Feb 2019 15:34:02 +0000 (23:34 +0800)]
nrelease: Minor cleanups to the Makefile
Aaron LI [Thu, 28 Feb 2019 15:17:32 +0000 (23:17 +0800)]
nrelease: More unambiguous label based on gitrev.sh
Use the git revision to compose the label to make it more unambiguous.
Use the same label as the .iso volume ID, but truncated to be <= 32
characters. Thank zrj for the make variable operation :D
Discusssed-with: swildner
Aaron LI [Thu, 28 Feb 2019 07:59:00 +0000 (15:59 +0800)]
nrelease: Use part-by-label to identify root in .img
Previously, the DragonFly .img hardcoded "da8" as the USB device, but
this may not be the case on some systems and requires user interaction
to boot.
Now that DragonFly has "/dev/part-by-label", use this new mechanism
(i.e., the label/name of the disklabel64(5) slice in the .img file) to
better identify the USB device with DragonFly installation image on it.
The label name is "DragonFly_${KERNCONF}" by default.
Matthew Dillon [Thu, 28 Feb 2019 06:36:27 +0000 (22:36 -0800)]
kernel - Add /dev/part-by-label
* The DragonFly disklabel has a 'label' field which accepts up to
a 63 byte ascii label.
* When this field is not empty, the kernel will now create a
"/dev/part-by-label/<label>.<part>" alias.
For example, if the label is FUBAR and you create a label with
partitions 'a', 'b', and 'd', then you will get:
# ls /dev/part-by-label
FUBAR.a FUBAR.b FUBAR.d
* WARNING! Duplicate labels will replace each other, so the
label that shows up will be non-deterministic.
Suggested-by: Aaron LI
Matthew Dillon [Thu, 28 Feb 2019 05:42:25 +0000 (21:42 -0800)]
kernel - Slightly change the behavior of MADV_NOSYNC and MAP_NOSYNC
* When files pages are mapped with the NOSYNC flag, the user program
is requesting that the kernel not periodically flush the dirty pages
to the filesystem under normal operation. Such pages can still be
flushed due to memory pressure.
* However, if the pages are not set back to MADV_AUTOSYNC, this behavior
can persist indefinitely (until umount or vnode recyclement), even
after all the related processes have closed/unmapped the file. This
results in very non-deterministic (and likely unexpected) flush behavior.
* Change the behavior to flush such dirty pages during the normal periodic
filesystem sync if the file is found to no longer be mapped and also
no longer open for writing. If the program wants to throw the data away
permanently, the program should delete or truncate the file.
In otherwords, the requested NOSYNC behavior is only applicable when the
file is open for writing or memory mapped.
Matthew Dillon [Thu, 28 Feb 2019 05:41:40 +0000 (21:41 -0800)]
kernel - Document possible dead code
* Document possible dead code due to the removal of B_MALLOC
buffers, but do not replace it with an assertion yet.
Matthew Dillon [Tue, 26 Feb 2019 06:58:24 +0000 (22:58 -0800)]
kernel - Reduce vm_page_list_find2() stalls
* Reduce stalls in vm_page_list_find2() which can occur in low-memory
situations, as well as in other situations. The problem is two fold.
First, that potentially all cpu cores can wind up waiting for a
single vm_page's spin-lock to be released.
Second, that a long-held vm_page spin-lock can cause the VM system
to stall unnecessarily long.
* Change vm_page_list_find() and vm_page_list_find2() to no longer
unconditionally spinlock a vm_page candidate and then retry if it
is found to be on the wrong queue.
Instead the code now spinlocks the queue, then iterates vm_page
candidates using spin_trylock(), skipping any pages whos spinlocks
cannot be immediately acquired. This is lock-order-reversed but
is ok because we use trylock. Also, by locking the queue first we
guarantee that a successfully spinlocked vm_page will be on the
correct queue.
* Should also reduce IPIQ drain stalls reported to the console as
shown below. The %rip sample is often found in vm_page_list_find2().
send_ipiq X->Y tgt not draining (STALL_SECONDS)
Tomohiro Kusumi [Mon, 25 Feb 2019 16:07:28 +0000 (08:07 -0800)]
sys/vfs/autofs: Drop .vop_getpages/.vop_putpages
Autofs only supports getattr/lookup/mkdir/readdir for
inmemory directory inodes.
Matthew Dillon [Sun, 24 Feb 2019 19:43:11 +0000 (11:43 -0800)]
kernel - Implement sbrk(), change low-address mmap hinting (2)
* Fix sysproto.h for the sbrk() change. 'incr' is size_t now.
(make sysent hadn't been run)
Reported-by: swildner
Tomohiro Kusumi [Sun, 24 Feb 2019 20:02:42 +0000 (12:02 -0800)]
sys/vm: Fix debug message in generic VOP_GETPAGES() handler
Sepherosa Ziehau [Sat, 16 Feb 2019 11:42:32 +0000 (19:42 +0800)]
netif/re: Sync with Realtek's re-1.95.
Mainly fix some register settings.
Sascha Wildner [Sun, 24 Feb 2019 01:35:33 +0000 (02:35 +0100)]
libc/sysvipc: Remove duplicate include.
Sascha Wildner [Sat, 23 Feb 2019 23:45:23 +0000 (00:45 +0100)]
<semaphore.h>: Whitespace cleanup.
Sascha Wildner [Sat, 23 Feb 2019 20:24:54 +0000 (21:24 +0100)]
<sys/ioctl.h>: Remove old SunOS compatibility definitions.
Nothing needs this anymore.
Thanks to tuxillo for build-testing dports.
Sascha Wildner [Sat, 23 Feb 2019 09:11:56 +0000 (10:11 +0100)]
kernel/bktr: Clean up unused code and the Makefile.
There should be no functional change.
Sascha Wildner [Fri, 22 Feb 2019 21:16:09 +0000 (22:16 +0100)]
Remove <sys/semaphore.h> so that we have <semaphore.h> only.
There is no point for us to go through the kernel for getting at
SEM_VALUE_MAX from sysconf(3), so return it directly and remove
the associated p1003_1b.sem_value_max sysctl.
Beyond that, the kernel didn't need this header. I've found no
ports that need it, either.
For safety, bump __DragonFly_version anyway.
Antonio Huete Jimenez [Thu, 21 Feb 2019 20:21:07 +0000 (21:21 +0100)]
nrelease - Fix target for Makefile.usr
- src-create-repo was merged to src-create in
ad8b70e08d. This happens only
for the GUI build.
François Tigeot [Tue, 19 Feb 2019 22:32:57 +0000 (23:32 +0100)]
drm: Reduce ioctl differences with Linux 4.7.10
Add a pair of missing generic drm ioctls.
Matthew Dillon [Tue, 19 Feb 2019 17:34:28 +0000 (09:34 -0800)]
libc - Use sbrk() system call, remove brk() (2)
* Change the initial TLS allocation to use mmap() instead of sbrk()
in order to maintain compatibility with older kernels.
* Using sbrk() will work with a fully updated system, but will not
work with an old kernel because it returns an error and sets errno,
but setting errno requires the TLS to already be installed. So
even though we have back-off code in libc, it seg-faults before
it gets to it.
* Only affects a few programs in base such as 'chflags' so not a
disaster. But still, the intent was to maintain compatibility
with older kernels.
Sascha Wildner [Tue, 19 Feb 2019 12:23:36 +0000 (13:23 +0100)]
LINT64: Add KTR_LAPIC.
Sascha Wildner [Mon, 18 Feb 2019 14:07:56 +0000 (15:07 +0100)]
<sys/cdio.h>: Remove definitions that are related to unimplemented ioctls.
None of this is handled in the kernel.
Sascha Wildner [Mon, 18 Feb 2019 10:52:33 +0000 (11:52 +0100)]
<libutil.h>: Remove an orphaned prototype.
Reported-by: tuxillo
Sascha Wildner [Mon, 18 Feb 2019 07:45:50 +0000 (08:45 +0100)]
ktr.4: Add a few words about KTR_LAPIC.
Matthew Dillon [Sun, 17 Feb 2019 00:41:25 +0000 (16:41 -0800)]
kernel - Implement sbrk(), change low-address mmap hinting
* Change mmap()'s internal lower address bound from dmax (32GB)
to RLIMIT_DATA's current value. This allows the rlimit to be
e.g. reduced and for hinted mmap()s to then map space below
the 4GB mark. The default data rlimit is 32GB.
This change is needed to support several languages, at least
lua and probably another one or two, who use mmap hinting
under the assumption that it can map space below the 4GB
address mark. The data limit must be lowered with a limit command
too, which can be scripted or patched for such programs.
* Implement the sbrk() system call. This system call was already
present but just returned EOPNOTSUPP and libc previously had its
own shim for sbrk() which used the ancient break() system call.
(Note that the prior implementation did not ENOSYS or signal).
sbrk() in the kernel is thread-safe for positive increments and
is also byte-granular (the old libc sbrk() was only page-granular).
sbrk() in the kernel does not implement negative increments and
will return EOPNOTSUPP if asked to. Negative increments were
historically designed to be able to 'free' memory allocated with
sbrk(), but it is not possible to implement the case in a modern
VM system due to the mmap changes above.
(1) Because the new mmap hinting changes make it possible for
normal mmap()s to have mapped space prior to the RLIMIT_DATA resource
limit being increased, causing intermingling of sbrk() and user mmap()d
regions. (2) because negative increments are not even remotely
thread-safe.
* Note the previous commit refactored libc to use the kernel sbrk()
and fall-back to its previous emulation code on failure, so libc
supports both new and old kernels.
* Remove the brk() shim from libc. brk() is not implemented by the
kernel. Symbol removed. Requires testing against ports so we may
have to add it back in but basically there is no way to implement
brk() properly with the mmap() hinting fix
* Adjust manual pages.
Matthew Dillon [Sat, 16 Feb 2019 19:40:04 +0000 (11:40 -0800)]
libc - Use sbrk() system call, remove brk()
* Attempt to use the kernel sbrk() system call. If it fails
with EOPNOTSUPP, back-off to the older libc sbrk() emulation
code which uses still-present but obsolete break() system call.
* Changes support both old and new kernels. Older kernels do not
implement the sbrk() system call, causing the call to return
-1/EOPNOTSUPP. Note that older kernels did not generate an
ENOSYS/signal for sbrk().
* sbrk() cannot support negative increments. Allow only 0 and positive
increments.
* Fix MPSAFE issues (requires new kernels) while we are here.
* Remove brk() support entirely. This was also emulated in older
kernels using break().
* These changes are necessary in order to support mmap() hinting above
RLIMIT_DATA's current value. Older kernels only support hinting above
the system maximum value (above the 32GB mark). Several languages in
dports really want to mmap() addresses below 4GB.
Sascha Wildner [Sun, 17 Feb 2019 09:32:21 +0000 (10:32 +0100)]
Update the pciconf(8) database.
February 13, 2019 snapshot from https://pci-ids.ucw.cz
Sascha Wildner [Wed, 6 Feb 2019 18:05:26 +0000 (19:05 +0100)]
install(1): Fix usage(), we don't have -T.
Sascha Wildner [Sat, 16 Feb 2019 21:27:50 +0000 (22:27 +0100)]
Sync ACPICA with Intel's version
20190215.
* Add support for the ACPI specification version 6.3.
* Fix skipping erroneous extended opcodes.
* Some improvements in iasl(8).
* Remove the legacy AcpiGbl_ExecuteTablesAsMethods global and all
associated code, i.e. the debug.acpi.group_module_level_code tunable
and acpiexec(8)'s -dp option.
For detailed list, please see sys/contrib/dev/acpica/changes.txt.
Sascha Wildner [Sat, 16 Feb 2019 20:08:51 +0000 (21:08 +0100)]
manpath.config: Add ravenports paths.
While here, remove an obsolete line.
Reported-by: marino
François Tigeot [Sun, 17 Feb 2019 09:25:59 +0000 (10:25 +0100)]
drm/linux: Add asm/processor.h
François Tigeot [Sun, 17 Feb 2019 08:06:17 +0000 (09:06 +0100)]
drm: Fix Linux/DragonFly PAGE_MASK confusion
* Both Linux and the DragonFly kernel use a PAGE_MASK constant,
which is unfortunately defined differently in each one
* Define a LINUX_PAGE_MASK macro and use it when applicable in the
drm code to fix this confusion
* One or two drm/radeon bugs were possibly caught during the process
* This commit based on a patch by tuxillo
Spotted-by: tuxillo
Fixes: https://bugs.dragonflybsd.org/issues/3171
Sascha Wildner [Sat, 16 Feb 2019 09:52:56 +0000 (10:52 +0100)]
fstab.5: Remove trailing whitespace from
828221af1c1fe3514.
Tomohiro Kusumi [Sat, 16 Feb 2019 09:27:25 +0000 (01:27 -0800)]
sys/vfs/autofs: Drop <sys/bus.h>
Looks like this is no longer needed just to compile.
Justin C. Sherrill [Sat, 16 Feb 2019 02:57:58 +0000 (21:57 -0500)]
Move some accidentally separated text back to the correct spot.
Aaron LI [Thu, 14 Feb 2019 16:12:06 +0000 (00:12 +0800)]
dd(1): Sync with FreeBSD
Some highlights:
* Add the "fillchar" operand.
* Add the "speed" operand to limit the speed.
* Support to specify the output data parity in the "conv" operand.
* Add tests, which can be run with "make test".
I adjusted the device names in the examples of the man page. Some minor
tweaks (e.g., styles, function prototypes, comments) were also applied.
Aaron LI [Thu, 14 Feb 2019 12:34:21 +0000 (20:34 +0800)]
dd(1): Implement GNU's extension of "status" operand
* status=none: do not print status information
* status=noxfer: do not print transfer statistics
This feature is useful because the statistics information can
sometimes be annoying, and redirecting stderr to /dev/null would also
lost error messages.
Obtained from FreeBSD (revision 264059)
* status=progress: report the current status on a single line every
second.
Obtained from FreeBSD (revisions 337505,337865,338646)
Aaron LI [Thu, 14 Feb 2019 13:38:21 +0000 (21:38 +0800)]
dd(1): Rework the time interval calculation
dd(1) uses gettimeofday(2) to compute the throughput statistics.
However, gettimeofday returns the system clock, which may jump forward
or back, especially if NTP is in use. If the time jumps backwards, then
dd will see negative elapsed time, round it up to 1usec, and print an
absurdly fast transfer rate.
The solution is to use clock_gettime(2) with CLOCK_MONOTONIC as the
clock_id. That clock advances steadily, regardless of changes to the
system clock.
Don't sacrifice precision by converting the output of clock_gettime() to
a double and then comparing the results. Instead, subtract the values
of the two clock_gettime() calls, then convert to double.
Obtained-from: FreeBSD (revisions 265472,265698)
Aaron LI [Thu, 14 Feb 2019 13:06:26 +0000 (21:06 +0800)]
dd(1): Make signal handler async safe
Obtained-from: FreeBSD (revision 250469)
Tomohiro Kusumi [Fri, 15 Feb 2019 00:27:25 +0000 (16:27 -0800)]
sys/sys/mutex2: Fix mutex(9) man page
"mtxex" does no exist after
cabfc9f6989fa1e4dbccc61162309bd2ec6891c9.
Matthew Dillon [Tue, 12 Feb 2019 20:28:11 +0000 (12:28 -0800)]
rtld-elf - Allow dynamic (late) relocations to relro section (2)
* Add missing header file adjustment
Bill Yuan [Tue, 12 Feb 2019 14:17:11 +0000 (14:17 +0000)]
ipfw3: allow the ipfw_insn with only the basic action
Matthew Dillon [Mon, 11 Feb 2019 21:47:22 +0000 (13:47 -0800)]
rtld-elf - Allow dynamic (late) relocations to relro section
* Normally the relro section is mprotect()ed to read-only after normal
load relocations.
* It appears that some programs can issue dynamic relocations at
run-time to such sections.
* If the relro mprotect has been done on an object, temporarily mprotect
the object back to RW to execute the relocation, then mprotect it back
to RO.
Reported-by: marino
Sascha Wildner [Mon, 11 Feb 2019 16:31:55 +0000 (17:31 +0100)]
bsd-family-tree: Sync with FreeBSD (add macOS 10.14).
Sascha Wildner [Sun, 10 Feb 2019 09:01:52 +0000 (10:01 +0100)]
makefs(8): DragonFly porting.
Some notes:
* UFS2 support has been disabled for now. It would require more UFS2
related pieces in our UFS kernel headers. I might look at enabling
this in the future.
* Our CGSIZE() macro causes issues when used in makefs(8), which is
why I'm using FreeBSD's currently.
* makefs(8) uses a local daddr_t type which is 64 bits wide, since
ours is 32 bits.
* I added cg_blks() and cg_blktot() calls in a few places because
our fsck(8) checks them in pass 5. I'm not sure if the associated
fields are actually needed anymore.
Sascha Wildner [Sun, 10 Feb 2019 08:11:11 +0000 (09:11 +0100)]
Bring in most of makefs(8) from FreeBSD unchanged.
To create filesystem images without kernel involvement. We can use it
to create our initrd without needing vn(4).
Originally from NetBSD.
Aaron LI [Mon, 4 Feb 2019 05:33:39 +0000 (13:33 +0800)]
Fix pkg "-f" flag in the "make upgrade" message
Aaron LI [Mon, 4 Feb 2019 05:22:02 +0000 (13:22 +0800)]
rc.d/sysctl: Only set a sysctl when give a new value
For a sysctl, compare the given value with its current value, and only
update the value if the new value is different. This suppresses
unhelpful messages on the console.
Reported-by: sephe
Matthew Dillon [Sat, 2 Feb 2019 22:29:40 +0000 (14:29 -0800)]
kernel - Fix panic in mlock() and munlock()
* Fix a degenerate case when mlock() or munlock() is called
with a size of 0. Silently allow the case and return success
(0).
* Fixes vm_map panic.
Sascha Wildner [Sat, 2 Feb 2019 18:07:58 +0000 (19:07 +0100)]
gcore(1): Some cleanup.
* Update usage(). While here, remove void casts.
* Sentences in manual pages should start on a new line.
* Flags without arguments should be specified first in the SYNOPSIS
grouped together.
* Bump manual page date.
Matthew Dillon [Sat, 2 Feb 2019 17:57:54 +0000 (09:57 -0800)]
gcore - Fix /proc map reading, add verbose option, add limit option
* Our /proc/*/map code allows incremental reads so the old
code that detects an insufficient buffer size no longer works.
Fix the code so processes with long maps are properly processed.
* Add -v to verbosely print the segments being dumped.
* Add -l <limit>{,k,m,g,t} to ignore segments larger than the
specified limitm.
Tomohiro Kusumi [Sat, 2 Feb 2019 16:57:25 +0000 (08:57 -0800)]
sys/vfs/autofs: Drop unused autofs_mount/autofs_softc fields
Drop unused ones which originate from FreeBSD.
Matthew Dillon [Sat, 2 Feb 2019 08:15:16 +0000 (00:15 -0800)]
gcore - Handle segments > 2GB
* Fix an improper 'int' plus a little cleanup. This allows gcore to
operate on processes which map segments greater than 2GB.
Tomohiro Kusumi [Thu, 31 Jan 2019 17:45:43 +0000 (09:45 -0800)]
sys/vfs/autofs: Whitespace cleanups
Use space rather than tab (originates from FreeBSD).
Matthew Dillon [Thu, 31 Jan 2019 19:35:40 +0000 (11:35 -0800)]
kernel - Sync usb3.0 pci ids w/FreeBSD
* Add missing PCI ids for Sunrise Point, Lewisburg, Union Point,
and Cavium ThunderX.
Reported-by: dualcyclo
Tomohiro Kusumi [Wed, 30 Jan 2019 23:05:14 +0000 (15:05 -0800)]
sys/vfs/autofs: Rename autofs_readdir_one() -> autofs_write_dirent()
Tomohiro Kusumi [Wed, 30 Jan 2019 22:50:41 +0000 (14:50 -0800)]
sys/vfs/autofs: Don't calculate dirent reclen twice
when autofs_readdir() may call autofs_dirent_reclen() after
returned from autofs_readdir_one().
Sepherosa Ziehau [Sun, 27 Jan 2019 14:06:08 +0000 (22:06 +0800)]
atomic: Alway use inline functions.
Originally only builtin functions use the inline version of the
atomic ops, while modules only use real function call, which is
designed for binary compatibility. However, this kind of
compatibility is not necessary for DragonFly.
François Tigeot [Sun, 27 Jan 2019 10:15:31 +0000 (11:15 +0100)]
drm/linux: Add EXPORT_SYMBOL_GPL()
François Tigeot [Sat, 26 Jan 2019 09:04:34 +0000 (10:04 +0100)]
drm/linux: Add preempt.h
Sascha Wildner [Sat, 26 Jan 2019 08:51:08 +0000 (09:51 +0100)]
pcm.4: Use .Dx
Sascha Wildner [Mon, 21 Jan 2019 10:50:09 +0000 (11:50 +0100)]
efirt.9: Comment out a non-existant function.
François Tigeot [Sun, 20 Jan 2019 20:11:44 +0000 (21:11 +0100)]
drm: Fix excessive wired memory usage
* This is a partial revert of
7ca4ece0a85bea451332580015344d3f12333419:
"drm: Rename struct drm_gem_object.vm_obj to filp".
* That commit caused wired memory usage to become excessive over time.
François Tigeot [Sat, 19 Jan 2019 19:11:34 +0000 (20:11 +0100)]
drm/gem: de-inline the ref/unref functions
Reducing differences with Linux 4.7.10
François Tigeot [Wed, 16 Jan 2019 19:43:47 +0000 (20:43 +0100)]
drm/linux/workqueue: the internal function is named func
Some driver code expects to be able to access it.
Aaron LI [Wed, 16 Jan 2019 09:00:47 +0000 (17:00 +0800)]
fgetln.3: Recommend to use the safer POSIX getline()
Obtained-from: OpenBSD
Aaron LI [Wed, 16 Jan 2019 08:57:05 +0000 (16:57 +0800)]
fgetln.3: Add the CAVEATS section with an example
Add the CAVEATS section to warn about the behavior of fgetln(). An
example is included to demonstrate how to properly use it.
Obtained-from: NetBSD, with the ferror() check from OpenBSD
Aaron LI [Wed, 16 Jan 2019 08:51:41 +0000 (16:51 +0800)]
getline.3: Note to distinguish between error and EoF
Note that the getline() and getdelim() functions do not distinguish
between end-of-file and error, so the callers must use feof() or
ferror() to determine which occurred.
Update the example to check the error.
Obtained-from: OpenBSD
Matthew Dillon [Tue, 15 Jan 2019 22:17:26 +0000 (14:17 -0800)]
tmpfs - Fix periodic syncer cpu-bound stalls w/tmpfs on big-mem boxes
* A bug in tmpfs leaves dirty tmpfs vnodes (which is basically all
of them) on the syncer list, causing the syncer to scan the entire
list every few seconds. When a large number of vnodes are involved,
this can lead to noticable cpu-bound stalls on the cpu the syncer is
running on.
* Generally speaking this should significantly improve applications
which use tmpfs a lot, such as synth runs, but was probably not noticed
on machines with less than 32GB of ram due to the lower kern.maxvnodes
default.
* Solved by unconditionally removing the vnode from the syncer list
in tmpfs_fsync(), but otherwise leaving the vnode marked VISDIRTY and
usually also VOBJDIRTY. Tmpfs's reclaim code properly handles the
disconnect regardless of the syncer state of the vnode.
Removing the vnode ensures that only one syncer pass is performed on
it, instead of repeated passes every few seconds.
Sascha Wildner [Tue, 15 Jan 2019 06:37:48 +0000 (07:37 +0100)]
Update the pciconf(8) database.
January 9, 2019 snapshot from https://pci-ids.ucw.cz
Sascha Wildner [Tue, 15 Jan 2019 03:44:42 +0000 (04:44 +0100)]
kernel: Remove puffs(4), putter(9) and associated libs and utilities.
All of these were originally pushed and hooked into the build in non-
working condition so that they would not go stale break building. The
hope was that someone would pick up the work and fix the remaining
issues, which never materialized.
This has led to more harm than good, with people occasionally assuming
that this code is functional. Also, the existence of /usr/include/fuse.h
might lead configure checks into believing that we have a working
implementation.
Sascha Wildner [Mon, 14 Jan 2019 17:59:58 +0000 (18:59 +0100)]
kernel: Hide the sysctl.debug sysctl in the SYSCTL_DEBUG kernel option.
The output is quite excessive and was previously too easily triggered,
like with "sysctl name=value" (instead of just "name=value") in
/etc/sysctl.conf.
Taken-from: FreeBSD
Sascha Wildner [Sun, 13 Jan 2019 14:54:35 +0000 (15:54 +0100)]
libcrypto: Switch over to libc's reallocarray().
Tomohiro Kusumi [Sun, 13 Jan 2019 22:40:19 +0000 (14:40 -0800)]
sbin/hammer2: Silence -Wformat-truncation= warning on Linux (gcc 7.3)
filename[HAMMER2_INODE_MAXNAME] may not be sufficient if
snprintf source pfs.name string is large enough. To silence
the warning, filename[] needs another 16 bytes.
This doesn't make difference on runtime, because bytes beyond
sizeof(pfs.name) won't be used even if strlen(pfs.name) is
large enough that filename[] contents exceeds sizeof(pfs.name).
(i.e. end result of pfs.name may be incomplete with or without
this commit)
--
cmd_snapshot.c: In function 'cmd_pfs_snapshot':
cmd_snapshot.c:84:9: warning: '%04d' directive output may be truncated writing between 4 and 11 bytes into a region of size between 0 and 255 [-Wformat-truncation=]
"%s.%04d%02d%02d.%02d%02d%02d",
^~~~
cmd_snapshot.c:83:3: note: 'snprintf' output between 17 and 324 bytes into a destination of size 256
snprintf(filename, sizeof(filename),
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
"%s.%04d%02d%02d.%02d%02d%02d",
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
pfs.name,
~~~~~~~~~
tp->tm_year + 1900,
~~~~~~~~~~~~~~~~~~~
tp->tm_mon + 1,
~~~~~~~~~~~~~~~
tp->tm_mday,
~~~~~~~~~~~~
tp->tm_hour,
~~~~~~~~~~~~
tp->tm_min,
~~~~~~~~~~~
tp->tm_sec);
~~~~~~~~~~~
Sascha Wildner [Sun, 13 Jan 2019 03:34:23 +0000 (04:34 +0100)]
libc: Adjust Versions.def properly.
Sascha Wildner [Sun, 13 Jan 2019 00:07:03 +0000 (01:07 +0100)]
i386 removal, part 70/x: Remove some i386 specific bits in the kernel.
Sascha Wildner [Thu, 3 Jan 2019 22:25:39 +0000 (23:25 +0100)]
i386 removal, part 69/x: Clean up sys/dev/netif.
According to comments from sephe.
Sascha Wildner [Sat, 12 Jan 2019 22:39:00 +0000 (23:39 +0100)]
Make libssh and mandoc use reallocarray() from libc.
Sascha Wildner [Sat, 12 Jan 2019 22:37:36 +0000 (23:37 +0100)]
libc: Add reallocarray() from OpenBSD.
It is useful for dhcpcd, mandoc and libssh. Also some dports might pick
it up. A few other dports have workarounds which, once reallocarray() is
in release too, can be removed.
Taken-from: OpenBSD (manual page from FreeBSD)
Pointed-out-by: aly
François Tigeot [Sat, 12 Jan 2019 12:05:56 +0000 (13:05 +0100)]
drm/linux: Add struct device_driver