Tomohiro Kusumi [Wed, 2 Feb 2022 15:59:42 +0000 (00:59 +0900)]
sys/vfs/msdosfs: Sanity check sector count from BPB
taken-from FreeBSD
ba2c98389b78b548aedac0be53121df909c3fe2f
Tomohiro Kusumi [Wed, 2 Feb 2022 15:10:09 +0000 (00:10 +0900)]
sys/vfs/msdosfs: Do no allow lookup to return vdp except for dot lookups
taken-from FreeBSD
aec97963cd03f10e04083537ed449a84a5e42f87
Tomohiro Kusumi [Tue, 1 Feb 2022 16:49:57 +0000 (01:49 +0900)]
sys/vfs/msdosfs: Handle a case when non-dot lookup returned dvp
This means that filesystem is corrupted, there is a loop.
taken-from FreeBSD
1319c433f4e5d67f8ea6fd2369c604069f733c94
Tomohiro Kusumi [Tue, 1 Feb 2022 16:23:17 +0000 (01:23 +0900)]
sys/vfs/msdosfs: Take inusemap inconsistency as an error, not invariants violation
In other words, stop silently accepting freeing free cluster in
non-debug kernels, but return the error to the caller. Modify callers
to handle errors from usemap_free().
taken-from FreeBSD
2c9a1c22c323b069d1f4883170349545c94b7b8d
Tomohiro Kusumi [Mon, 31 Jan 2022 15:23:08 +0000 (00:23 +0900)]
sys/vfs/msdosfs: mountmsdosfs() some style
taken-from FreeBSD
04fd468da0d0baea535da418b92df74101a9659d
Matthew Dillon [Sun, 30 Jan 2022 18:42:11 +0000 (10:42 -0800)]
kernel - Fix namecache issue that can slow systems down over time
* Fix a serious dead-record issue with the namecache.
It is very common to create a blah.new file and then rename it over
an existing file, say, blah.jpg, in order to atomically replace the
existing file. Such rename-over operations can cause up to
(2 * maxvnodes) dead dead namecache records to build up on a
single hash slot's list.
* Over time, this could result in over a million records on a single
hash slot's list which is often scanned during namecache lookups,
causing the kernel to turn into a sludge-pile.
This was not a memory leak per-say, the kernel still cleans excess
structures (above 2 * maxvnodes) up, but that just maintains the
status-quo and leaves the system in a slow, poorly-responsive state.
* Fixed by proactively deleting matching dead entries during namecache
lookups. The 'live' record is typically at the beginning of the list.
So to fix, the namecache lookup now scans the list for the hash slot
backwards and attempts to dispose of dead records.
Tomohiro Kusumi [Sun, 30 Jan 2022 15:07:01 +0000 (00:07 +0900)]
usr.sbin/makefs: Remove set but not used variables
These were leftovers from the port from NetBSD (where they are used).
taken-from FreeBSD
093cf790569775b80662926efea6d9d3464bde94
Tomohiro Kusumi [Sun, 30 Jan 2022 14:30:07 +0000 (23:30 +0900)]
msdosfs: clusterfree() is used only in error handling cases
Change its return type to void, because its result is ignored in both
call sites. Remove oldcnp argument as well, it is NULL always.
taken-from FreeBSD
65990b68a2cd89a08f0350e187df1968b16f4255
Sascha Wildner [Sun, 30 Jan 2022 01:18:39 +0000 (02:18 +0100)]
libc: Add strerror_l() (taken from FreeBSD).
Reported-by: zrj
Sascha Wildner [Sun, 30 Jan 2022 00:36:28 +0000 (01:36 +0100)]
<sys/param.h>: Add missing whitespace.
Sascha Wildner [Sat, 29 Jan 2022 23:26:52 +0000 (00:26 +0100)]
libc: Fix building without YP.
Sascha Wildner [Sat, 29 Jan 2022 23:15:08 +0000 (00:15 +0100)]
libc: Remove libc checks around manual pages from the Makefile.inc's.
Handle this better by setting NOMAN for libc_rtld.
Tomohiro Kusumi [Thu, 27 Jan 2022 16:03:32 +0000 (01:03 +0900)]
sys/vfs/hammer2: Add hammer2_ prefixed spin_[un]lock_update()
like any other spin lock variants.
Tomohiro Kusumi [Thu, 27 Jan 2022 16:02:36 +0000 (01:02 +0900)]
sys/vfs/hammer2: Use hammer2_ prefixed spin_init() for portability
Sascha Wildner [Wed, 26 Jan 2022 10:48:17 +0000 (11:48 +0100)]
<sys/elf_common.h>: Use standard types, like the rest of the elf headers.
Matthew Dillon [Tue, 25 Jan 2022 22:33:55 +0000 (14:33 -0800)]
kernel - Fix sysv semaphore panic
* Fix an exit race related to lingering sysv semaphores which can sometimes
cause a kernel panic.
Reported-by: zrj
Tomohiro Kusumi [Mon, 24 Jan 2022 16:10:10 +0000 (01:10 +0900)]
sys/vfs/hammer2: Add missing hammer2_spin_init() for chain core
This happened to be harmless, as all it does is zero initialize
some fields that are already zero.
Other spinlocks are explicitly initialized.
Tomohiro Kusumi [Sun, 23 Jan 2022 17:00:18 +0000 (02:00 +0900)]
sys/vfs/hammer2: Remove wrong comments in hammer2_chain_core_init()
No such arguments *trans, *nchain, *ochain since
da6f36f44bb9c76612e31b56cdfe4d787e53d61f in 2014 and
23c7c7dd4e59da5bdb6ffa7fe2074c5b7ffe5e4f in 2015.
Matthew Dillon [Mon, 24 Jan 2022 06:02:56 +0000 (22:02 -0800)]
kernel - Fix ktrace's handling of system call return values
* Distinguish between int and long returns, properly recording
long returns (e.g. lseek(), mmap(), read(), write(), etc).
kdump will then display the correct return value instead of a truncated
return value for the affected system calls.
* Involves including a return value type width in the sysent and adjusting
the generator to calculate it.
Matthew Dillon [Mon, 24 Jan 2022 06:01:50 +0000 (22:01 -0800)]
libc - Increase opendir/readdir buffer size
* Increase the directory scan buffer size from 4096 bytes to 16384
bytes, significantly reducing the number of discrete system calls
which must be made when scanning very large directories.
Matthew Dillon [Mon, 24 Jan 2022 05:55:44 +0000 (21:55 -0800)]
tmpfs - Fix readdir() races
* Fix multi-threaded deletion races against readdir(). These races
can cause a directory scan in one thread to return EINVAL if the
file representing the chaining cookie is deleted by another thread.
Fix the issue by allowing tmpfs_dir_lookupbycookie() to return the
nearest directory entry with a cookie >= the requested cookie.
* Use a better cookie value for EOF
* Allow readdir() to fail on "." or "..". Instead just iterate to
the next entry. This can occur when a directory is deleted out
from under a scan that has chdir()'d into it.
* tmpfs was previously rescanning all entries to locate a cookie,
which is rather dumb given that we have a RB tree. Do a proper
iterative recursive search instead.
Reported-by: zrj, others
Matthew Dillon [Fri, 14 Jan 2022 20:55:36 +0000 (12:55 -0800)]
hammer2 - Fix excess chain structure allocations during bulkfree (2)
Try to fix this again. The last attempt failed. Basically the
problem is that when the depth-first traversal gets too deep, it
has to defer the node at that point by adding it to a list.
* Change the order to insert deferrals on the head instead of the
tail in order to exhaust the most recently deferred sub-tree as
quickly as possible (LIFO order). We previously exhausted deferred
sub-trees in FIFO order but this often resulted in multiple levels
of the radix tree building up on the list, which could blow-up
memory use.
* Once the list limit is reached (well, really 1/4 of the list limit),
we previously continued to accumulate deferral as long as the
relative depth was less then 3. Unfortunately, the maximum radix
tree table size of roughly 512 elements could cause the number
of deferrals even after the limit was reached to be up to
134 million chains (512 * 512 * 512).
This is fixed by removing any depth test when the list limit is
exhausted.
* This combination of fixes should prevent the number of deferrals
from blowing out hammer2's kmalloc pool. Theoretically the
worst-case possible deferrals should now be 512 chains x max depth
of radix tree. A very reasonable number.
Sascha Wildner [Thu, 20 Jan 2022 21:43:54 +0000 (22:43 +0100)]
sysctl.3: Adjust for pageout daemon algo changes from last year.
Taken-from: dillon's commit message
Sascha Wildner [Wed, 19 Jan 2022 15:13:12 +0000 (16:13 +0100)]
Update the pciconf(8) database.
January 13, 2022 snapshot from https://pci-ids.ucw.cz
Sascha Wildner [Wed, 19 Jan 2022 05:51:20 +0000 (06:51 +0100)]
vnconfig.8: Mention comment lines in the configuration file.
Antonio Huete Jimenez [Mon, 17 Jan 2022 22:09:55 +0000 (23:09 +0100)]
build - Update pkg-static to version 04
- Should fix a problem with the metadata versioning.
pkg-static: repository meta has wrong version 2
pkg-static: Repository Avalon load error: meta cannot be loaded
Undefined error: 0
YONETANI Tomokazu [Fri, 5 Nov 2021 00:37:39 +0000 (09:37 +0900)]
Add the auto-bound IPv6 address to the wildcard hash
Dragonfly-bug: https://bugs.dragonflybsd.org/issues/3294
Matthew Dillon [Sat, 8 Jan 2022 00:41:35 +0000 (16:41 -0800)]
hammer2 - Fix excess chain structure allocations during bulkfree
* On H2 filesystems with a very large number of inodes, such as
those used for backups or which contain many snapshots (e.g.
tens of millions of inodes), the bulkfree process may attempt
to track more hammer2_chain structures than the kmalloc subsystem
allows.
* This issue might also have contributed to reported lockups during
bulkfree (if the bulkfree wound up eating too much non-pagable kernel
memory), though the more typical outcome is a kernel panic when
the kmalloc limit is exceeded for the pool.
* During bulkfree, H2 must control recursion depth during the scan.
It does this by deferring deep nodes in the topology by recording
a pointer to a referenced hammer2_chain structure on a list.
Once the recursion backs out, H2 then processes these saved chains.
Topologies greater than 32 nodes deep can wind up being deferred
more than once to cover the entire depth.
It is possible for an excessive number of chains to accumulate on the
deferral list during the scan, each structure burning up kernel memory.
* Emplace a limit on the number of chains which can be deferred,
vfs.hammer2.limit_saved_chains, with a reasonable default. When
the limit is reached, the bulkfree simply records a chain higher-up
in the recursion instead of going deeper, thus limiting the number
of chains that will be deferred.
This has the effect is capping memory use during the bulkfree scan.
The cost is slightly less scan efficiency, but it should mostly be in
the noise.
Sascha Wildner [Thu, 6 Jan 2022 19:48:42 +0000 (20:48 +0100)]
libc/isc: Stop using self assignment for unused variables.
Some checkers complain about this.
Taken-from: FreeBSD
Matthew Dillon [Mon, 3 Jan 2022 20:13:18 +0000 (12:13 -0800)]
dsynth - Add config var for Meta_version and make the default 2
* Add the 'Meta_version' configuration variable.
* Make the default 2 (it was previously 1), to remain compatible
with future pkg versions.
Requested-by: zrj
Aaron LI [Mon, 3 Jan 2022 09:21:34 +0000 (17:21 +0800)]
stand: Fix a typo in comment
While there, adjust header inclusion a bit.
zrj [Mon, 3 Jan 2022 06:18:15 +0000 (06:18 +0000)]
libc: Fix environment passing in posix_spawnp(3)
The execvPe() was always passing the parent process environment to
children (a long standing bug).
While there, replace the deprecated index(3) with strchr(3).
Reported-by: falsifian (breakage in hs-git-annex)
Justin C. Sherrill [Mon, 3 Jan 2022 03:13:40 +0000 (22:13 -0500)]
6.3 version changes.
Matthew Dillon [Mon, 3 Jan 2022 00:51:22 +0000 (16:51 -0800)]
hammer2 - (when used for booting @BOOT) (2)
* Adjust the range to use HAMMER2_DIRHASH_LOMASK
Reported-by: aly
Matthew Dillon [Sun, 2 Jan 2022 20:21:44 +0000 (12:21 -0800)]
hammer2 - (when used for booting @BOOT)
* Fix the @BOOT lookup to allow the whole hash range for "BOOT".
This can come up if you delete the BOOT pfs and re-create it.
hammer2 incremental the name hash collision counter to ensure
that new PFSs are unique.
Reported-by: aly, mneumann
Sascha Wildner [Fri, 31 Dec 2021 18:29:14 +0000 (19:29 +0100)]
Add a getentropy(3) manual page.
Taken-from: FreeBSD
Sascha Wildner [Fri, 31 Dec 2021 17:54:49 +0000 (18:54 +0100)]
Bump copyrights. Happy new year!!
Matthew Dillon [Fri, 31 Dec 2021 00:43:25 +0000 (16:43 -0800)]
kernel - Fix if_sn, if_sbsh issue (not in kernel by default)
* Fix a security problem with a data field. Note that these drivers
are not loaded into the kernel by default.
Matthew Dillon [Fri, 31 Dec 2021 00:40:59 +0000 (16:40 -0800)]
kernel - Add SIOCGHWADDR, fix SIOCGIFADDR
* Add SIOCGHWADDR, this is more commonly used to obtain the
ether mac address of an interface these days.
* Fix SIOCGIFADDR which had a security problem with a data field.
Michael Neumann [Thu, 30 Dec 2021 23:26:05 +0000 (00:26 +0100)]
psm - Do not freeze mouse for 2 seconds upon encountering a single sync error
The original code would drop all PS/2 packets for the next 2 seconds
upon missing a *single* sync byte. Instead, just discard the byte and in
case we cannot find the sync byte within the next few bytes, fall back
to the old behaviour.
Fixes bug: #3309
Matthew Dillon [Thu, 30 Dec 2021 21:15:44 +0000 (13:15 -0800)]
kernel - Add DIOCGMEDIASIZE and DIOCGSECTORSIZE
* Add DIOCGMEDIASIZE and DIOCGSECTORSIZE for 'go' and other dports.
At the moment this requires #include <sys/diskslice.h>.
Requested-by: zrj (for dports)
Matthew Dillon [Thu, 30 Dec 2021 21:13:55 +0000 (13:13 -0800)]
libc - Add getentropy()
* Add getentropy(). This function uses getrandom() internally.
Requested-by: zrj, for dports
Sascha Wildner [Thu, 30 Dec 2021 14:38:51 +0000 (15:38 +0100)]
Remove the old <sys/dir.h> compat header. Nothing needs it anymore.
The world has since switched over to <dirent.h>. Thanks to zrj for
confirming that with a full dports bulk build.
The libssh and RPC examples adjustments are just cosmetics. libssh
includes it only on NeXTSTEP and the RPC examples don't build.
Tomohiro Kusumi [Tue, 28 Dec 2021 16:14:03 +0000 (01:14 +0900)]
sys/vfs/hammer2: Use HAMMER2_METH_DEFAULT
hammer2_chain_create() callers use HAMMER2_METH_DEFAULT, not '-1'.
Michael Neumann [Sun, 26 Dec 2021 15:38:25 +0000 (16:38 +0100)]
drm/i915: Enable WhiskeyLake GPUs
* The PCI ids were added in commit
a8138aa83 but the ids were never
added to `pciidlist` so that the i915 driver did not recognize
WhiskeyLake GPUs.
* Successfully tested with WhiskeyLake-U GT2 [UHD Graphics 620]
(device=0x3ea0).
* As in Linux 5.8, they use the same configuration as CoffeeLake GPUs.
Sascha Wildner [Sat, 25 Dec 2021 22:41:21 +0000 (23:41 +0100)]
Add a small ksscanf(9) manual page.
Sascha Wildner [Sat, 25 Dec 2021 22:40:10 +0000 (23:40 +0100)]
kprintf.9: Fix function argument name.
Matthew Dillon [Sat, 25 Dec 2021 20:30:52 +0000 (12:30 -0800)]
nvme - Back-off if driver lies about reported queue limits
* Apparently some low-rent nvme controllers lie about how many
queues they support.
* If the nvme driver lies about queue support and a queue
create command fails, attempt to back-off to fewer queues
first, rather than giving up immediately. Complain mightily
on the console.
Matthew Dillon [Sat, 25 Dec 2021 06:45:17 +0000 (22:45 -0800)]
hammer2 - Do not panic on freemap block I/O error
* During bulkfree, do not panic if a freemap block has an I/O error.
Reported-by: aly
Sascha Wildner [Sat, 25 Dec 2021 10:09:32 +0000 (11:09 +0100)]
Add some more printf(1) tests.
Taken-from: FreeBSD
Sascha Wildner [Thu, 23 Dec 2021 17:33:17 +0000 (18:33 +0100)]
libc: Add a clearenv() function.
Seems useful for dports, as pointed out by zrj.
Taken-from: FreeBSD
Sascha Wildner [Sat, 18 Dec 2021 22:41:28 +0000 (23:41 +0100)]
gcc: Define ElfW() if it is not defined.
Better than commenting it out.
Suggested-by: zrj
Sascha Wildner [Sat, 18 Dec 2021 19:16:36 +0000 (20:16 +0100)]
edk2: Sync our TianoCore EDK II headers with the edk2-stable202111 tag.
Sascha Wildner [Fri, 17 Dec 2021 21:23:17 +0000 (22:23 +0100)]
Sync ACPICA with Intel's version
20211217.
* S4/S5 related fixes.
* Fix a buffer overflow.
* Add support for PCC Opregion special context data.
* Various improvements, such as adding support for various tables,
in iasl(8).
For a detailed list, please see sys/contrib/dev/acpica/changes.txt.
Sascha Wildner [Thu, 16 Dec 2021 21:16:00 +0000 (22:16 +0100)]
kernel: Remove a.out and gzipped a.out executable support.
This isn't useful for anything anymore. Likely never was.
Also point out that our current executable format is ELF on the
a.out(5) manual page, which we keep along with <sys/a.out.h> because
ports need it.
Sascha Wildner [Thu, 16 Dec 2021 20:45:07 +0000 (21:45 +0100)]
kernel/proc0_init: Do not reuse the a.out sysvec.
Also use correct types now.
In preparation for removing a.out executable support.
Sascha Wildner [Thu, 16 Dec 2021 20:40:36 +0000 (21:40 +0100)]
kernel: Remove an unnecessary include.
Sascha Wildner [Wed, 15 Dec 2021 15:46:46 +0000 (16:46 +0100)]
Update the pciconf(8) database.
December 10, 2021 snapshot from https://pci-ids.ucw.cz
Sascha Wildner [Mon, 13 Dec 2021 20:09:33 +0000 (21:09 +0100)]
gcc8: Fix the fix.
Sascha Wildner [Mon, 13 Dec 2021 20:04:04 +0000 (21:04 +0100)]
gccs had ElfW() already. Comment out for now.
Sascha Wildner [Mon, 13 Dec 2021 19:45:17 +0000 (20:45 +0100)]
<sys/elf_generic.h>: Add ElfW() macro, for dports' benefit.
Taken-from: FreeBSD
Reported-by: zrj
Matthew Dillon [Thu, 9 Dec 2021 18:18:03 +0000 (10:18 -0800)]
kernel - Fix posix_fallocate()
* Our newly minted posix_fallocate() system call was ignoring the
offset field.
* Should fix samba, however any tdb databases that were created are
probably corrupt so you may need to rm -rf /var/db/samba4, restart
samba, and re-create your smb passwords.
Reported-by: Tim Darby
Sascha Wildner [Wed, 8 Dec 2021 21:12:15 +0000 (22:12 +0100)]
Remove a number of duplicate struct declarations in header files.
Sascha Wildner [Wed, 8 Dec 2021 15:53:55 +0000 (16:53 +0100)]
Remove the ext2fs_freebsd.ko module via 'make upgrade'.
Sascha Wildner [Wed, 8 Dec 2021 15:51:55 +0000 (16:51 +0100)]
kernel/acpi: Improve an error message a bit.
Sascha Wildner [Tue, 7 Dec 2021 22:49:49 +0000 (23:49 +0100)]
kernel/acpica: Clean up the module's .PATH a little bit more.
Sascha Wildner [Tue, 7 Dec 2021 20:35:19 +0000 (21:35 +0100)]
kernel/acpica: Remove some unneeded -I and .PATH from the Makefile.
Sascha Wildner [Tue, 7 Dec 2021 17:52:40 +0000 (18:52 +0100)]
Bring in FreeBSD's ext2fs(5) manual page.
With some modifications as suggested by tkusumi.
Sascha Wildner [Tue, 7 Dec 2021 16:56:38 +0000 (17:56 +0100)]
kernel/ext2fs: Add some line breaks in SRCS in the Makefile.
Sascha Wildner [Tue, 7 Dec 2021 15:35:33 +0000 (16:35 +0100)]
Some more adjustments regarding the recent switch to sys/vfs/ext2fs.
* Remove /usr/include/gnu via 'make upgrade'.
* Remove a no longer needed #undef in kdump and friends' ioctl.c.
Tomohiro Kusumi [Sun, 5 Dec 2021 16:23:39 +0000 (01:23 +0900)]
ext2fs: Remove sys/gnu/vfs/ext2fs and make sys/vfs/ext2fs the default
This commit removes the old+unstable GPL ext2 implementation,
and makes the new FreeBSD based ext2 implementation
(since
cfe603905713d4e92a7956678970d5dff8e913f2) the default.
- Remove sys/gnu/vfs.
- Rename sys/vfs/ext2fs/ext2fs_freebsd.ko -> ext2fs.ko.
- Modify unusual userspace program which includes kernel struct.
- Bump __DragonFly_version to 600107.
Sascha Wildner [Mon, 6 Dec 2021 20:23:44 +0000 (21:23 +0100)]
bus_dma.9: Fix typo.
Tomohiro Kusumi [Sun, 5 Dec 2021 13:00:23 +0000 (22:00 +0900)]
sys/vfs/ext2fs: Check for e2fs_first_dblock in ext2_compute_sb_data()
This prevents a kernel panic on a damaged ext2 superblock.
taken-from FreeBSD
3dd3a395ba975d0fbe13320e6e69fb85b037da5e
Sascha Wildner [Sun, 5 Dec 2021 07:08:01 +0000 (08:08 +0100)]
Makefile.inc1: Fix a typo.
Sascha Wildner [Wed, 1 Dec 2021 19:48:02 +0000 (20:48 +0100)]
callout.9: Add callout_init_mp() to the SYNOPSIS.
Matthew Dillon [Tue, 30 Nov 2021 17:45:17 +0000 (09:45 -0800)]
drm - Hack i915 to workaround startup crash
* Work around a startup crash (at least on my i5-6500) caused by
a ring timeout and request-not-completed. Replace the BUG_ON
with a warning and document the hack.
Matthew Dillon [Mon, 29 Nov 2021 20:46:07 +0000 (12:46 -0800)]
kernel - Do not block indefinitely in exit1() when draining controlling tty
* exit1() tries to drain the controlling terminal. This can block
indefinitely and signals are no longer operational. Force a
1-second timeout for drain attempts.
This is a bit of a hack but it works.
Reported-by: zach, pikrzyszt, piecuch, bug #3239
Sascha Wildner [Mon, 29 Nov 2021 20:41:25 +0000 (21:41 +0100)]
kernel/drmP.h: Use Linux' __printf() macro.
Sascha Wildner [Sat, 27 Nov 2021 23:15:44 +0000 (00:15 +0100)]
acpi.4: Fix typo, ACPI_POWER -> ACPI_POWERRES.
Sascha Wildner [Thu, 25 Nov 2021 22:17:56 +0000 (23:17 +0100)]
kernel: Rename BUILDING_WITH_KERNEL to KERNBUILDDIR.
Fits better with e.g. KERNCONFDIR.
Add a small paragraph in kmod.mk which was taken from FreeBSD.
Sascha Wildner [Thu, 25 Nov 2021 20:54:21 +0000 (21:54 +0100)]
kernel: Staticize some variables.
Sascha Wildner [Thu, 25 Nov 2021 16:45:40 +0000 (17:45 +0100)]
nrelease: Switch to makefs(8) for the ISO's EFI system partition image.
Tested-by: tuxillo
Sascha Wildner [Thu, 25 Nov 2021 16:43:47 +0000 (17:43 +0100)]
<sys/param.h>: Bump __DragonFly_version for makefs(8)'s msdosfs support.
Sascha Wildner [Thu, 25 Nov 2021 16:42:24 +0000 (17:42 +0100)]
nrelease: Add a $WORLD_VERSION variable.
Can be useful when using new features in the release build.
Antonio Huete Jimenez [Thu, 25 Nov 2021 00:15:21 +0000 (01:15 +0100)]
nrelease: DPorts not needed for binpkgs/nopkgs
Sascha Wildner [Wed, 24 Nov 2021 19:21:32 +0000 (20:21 +0100)]
initrd/mkinitrd.sh: Switch the initrd generation from vn(4) to makefs(8).
We lose the 'df -h' output because nothing is mounted anymore.
Sascha Wildner [Wed, 24 Nov 2021 16:49:35 +0000 (17:49 +0100)]
mkinitrd.sh: Fix typo.
Sascha Wildner [Sun, 21 Nov 2021 20:08:02 +0000 (21:08 +0100)]
<machine/stdint.h>: Add platform specific __double_t and __float_t types.
It's mostly just cosmetics but it reduces differences with OpenBSD's
<math.h>.
Also, looks like at least www/firefox is wrongly using __float_t.
Reported-by: tuxillo
Tomohiro Kusumi [Sun, 21 Nov 2021 15:23:40 +0000 (00:23 +0900)]
sys/vfs/hammer2: Remove unused local variables in freemap
Tomohiro Kusumi [Sun, 21 Nov 2021 12:02:19 +0000 (21:02 +0900)]
sys/vfs/hammer2: Rename **bresp -> **brefp
hammer2_blockref_t pointer is *bref, then
the double pointer should be **brefp.
Sascha Wildner [Sat, 20 Nov 2021 14:26:42 +0000 (15:26 +0100)]
libedit: Remove unused Symbol.map/Versions.def.
Sascha Wildner [Sat, 20 Nov 2021 10:45:24 +0000 (11:45 +0100)]
libc: Adjust the private namespace's parent to 6.2.
I forgot this in
c89c8747ca2bf3b7292cc93bbdbefdfbfed7c821.
Matthew Dillon [Fri, 19 Nov 2021 18:10:29 +0000 (10:10 -0800)]
Fix select()/kqueue() bug with pipe()s
* This causes select() / kqueue() was not returning a writable condition
on a pipe() whos other end was closed early.
If a select/kqueue test is performed prior to the other end closing,
then future calls will later properly detect when the other end closes.
* Should fix numerous niggling issues, including csh sometimes failing
to recognize early exits of children.
Reported-by: tmunro, deef, bug #3307
Antonio Huete Jimenez [Thu, 18 Nov 2021 12:37:25 +0000 (13:37 +0100)]
dfregress: Add missing copyright
Antonio Huete Jimenez [Wed, 17 Nov 2021 23:28:58 +0000 (00:28 +0100)]
.gitignore: Ignore dfregress output files
Antonio Huete Jimenez [Wed, 17 Nov 2021 18:22:54 +0000 (19:22 +0100)]
Revert "bin/test: Temporarily break world"
This reverts commit
655933d6558d16623b4af8e48844c5c37cff5ab3.
Sorry for the noise!
Antonio Huete Jimenez [Wed, 17 Nov 2021 18:17:53 +0000 (19:17 +0100)]
bin/test: Temporarily break world
Antonio Huete Jimenez [Wed, 17 Nov 2021 17:19:10 +0000 (18:19 +0100)]
test/test: dummy commit
Tomohiro Kusumi [Mon, 15 Nov 2021 16:12:06 +0000 (01:12 +0900)]
sys/vfs/hammer2: Fix wrong comment on freemap zone/rotation
Freemaps are in every 1GB, and there are 8 rotations in each freemap.
It was 2GB and 4 when this was written in
1a7cfe5ae3c897f704a358fd3e556a55e430dcb1 in 2013.
Sascha Wildner [Mon, 15 Nov 2021 14:04:06 +0000 (15:04 +0100)]
Update the pciconf(8) database.
November 11, 2021 snapshot from https://pci-ids.ucw.cz