Daniel Fojt [Sat, 25 Jul 2020 18:02:43 +0000 (20:02 +0200)]
Add sanity check to "remove-deprecated-files" target.
Import helper awk script to detect objects linked to particular
library.
Separate the check to a new target "check-deprecated-files" in
etc/Makefile and utilize the script to report libraries that are
still in use.
Default is to check directories under /usr/local/. Additional places
to search may be specified via CHECK_DEPRECATED_DIRS variable, like:
make CHECK_DEPRECATED_DIRS="/opt/lib /opt/bin" upgrade
Suggested and reviewed by: Sascha Wildner
Sascha Wildner [Thu, 30 Jul 2020 20:27:47 +0000 (22:27 +0200)]
libstand: Add include guards to headers and cleanup.
Daniel Fojt [Thu, 30 Jul 2020 10:10:05 +0000 (12:10 +0200)]
Upgrade make(1). 2/2
- update README.{DELETED,DRAGONFLY}
- adapt Makefile
- regenerate config.h
Daniel Fojt [Thu, 30 Jul 2020 10:09:09 +0000 (12:09 +0200)]
Upgrade make(1). 1/2
Merge branch 'vendor/BMAKE' into master.
Daniel Fojt [Thu, 30 Jul 2020 03:55:49 +0000 (05:55 +0200)]
vendor/bmake: upgrade from
20160818 to
20200710
Summary of notable changes:
- add :range and :_
- support for .DELETE_ON_ERROR
- allow specifying the utc value used by :{gm,local}time
- allow :_=var and avoid use of special context
- use -C arg "as is" if it contains no relative component
- when setting .OBJDIR, ignore '$' in paths
- add -v option to always fully expand values
- pass SIGINT etc onto child and wait for it to exit
- when target is out-of-date per normal make rules record
value of .OODATE in meta file
- add :Or for reverse sort
- improvements to avoid unnecessary evaluation use enums
for flags
- debugging output for :@
- fix for compare_expression when doEval=0
For detailed list see ChangeLog.
Matthew Dillon [Wed, 29 Jul 2020 21:10:19 +0000 (14:10 -0700)]
dsynth - Log the portdir for all bulk scan errors
* All bulk scan errors are now able to log the portdir for the errors,
and will also log the errors to 00_last_results.log.
* Implemented by setting stderr to a unix-domain socket during the run
and using SO_PASSCRED.
* Requires a kernel which supports SO_PASSCRED. Kernels which do not
support SO_PASSCRED will still log the errors, but without the
portdir prefix.
Matthew Dillon [Wed, 29 Jul 2020 20:58:54 +0000 (13:58 -0700)]
kernel - Implement SO_PASSCRED
* Implement SO_PASSCRED. This allows the receiving end of a local socket
to request creds even if the sending end did not send any.
* Allows the use of a common local socket with multiple end points, e.g.
for stderr, such that the receiving entity can distinguish between
message sources.
Taken-from: Linux
Tomohiro Kusumi [Wed, 29 Jul 2020 16:39:57 +0000 (01:39 +0900)]
sys/vfs/ext2fs: Disable calling ext2_htree_add_entry()
4c3802f8a89fcb53ad1112cc71dedbc9a01c3384
disabled creation of a new htree dirent index, but also disable adding
entries to an existing index created at elsewhere as this could also
possibly cause the issue mentioned in the comment right above.
Daniel Fojt [Wed, 29 Jul 2020 08:53:17 +0000 (10:53 +0200)]
libressl: cleanup the build
Drop our custom CRYPTO_{free,malloc}.c and use libressl's malloc-wrapper.c
instead.
Discussed with: Rimvydas Jasinskas
Daniel Fojt [Wed, 29 Jul 2020 05:32:07 +0000 (07:32 +0200)]
Update share/skel/dot.cshrc.
Include also current working directory in default prompt.
Suggested by: Peeter Must
Approved by: Sascha Wildner, Pierre-Alain TORET
Matthew Dillon [Wed, 29 Jul 2020 02:40:58 +0000 (19:40 -0700)]
kernel - Cleanup recent DMAP hacks for EFI framebuffers
* Adjust the EFI framebuffer code to test against DMapMaxAddress
instead of DMAP_MAX_ADDRESS.
* Remove the temporary hack that burned memory with a too-large
minimum DMAP size.
Matthew Dillon [Tue, 28 Jul 2020 20:38:22 +0000 (13:38 -0700)]
kernel - Improve debugging of spurious interrupts
* Report spurious T_RESERVED interrupt vectors / trap numbers. Report
the actual trap number and try to ignore it. If this occurs it
helps with debugging as a cold boot 'vmstat -i -v' can be matched
up against the spurious interrupt number (usually spurious# - 0x20)
to locate the PCI device causing the problem.
Matthew Dillon [Tue, 28 Jul 2020 20:34:02 +0000 (13:34 -0700)]
kernel - Disable MSIX mode in pci_teardown_msix()
* Disable MSIX mode when pci_teardown_msix() is called. This fixes
issues on some BIOSes which do not fully clear the PCI hardware
configuration when warm-booting.
In particular, the NVME controller appears to have this issue with
some laptops (HP Intel gen 9 and gen 10 laptops) whereby the MSI
interrupt winds up being stuck on reboot.
Matthew Dillon [Tue, 28 Jul 2020 07:41:04 +0000 (00:41 -0700)]
kernel - Hack the DMAP size
* Make the minimum DMAP size much larger. This is a hack to accomodate
EFI framebuffers that have been mapped to very high physical addresses.
Due to the early initialization, they must use the DMAP.
* Fixes booting on modern intel-based HP laptops.
Matthew Dillon [Tue, 28 Jul 2020 07:34:40 +0000 (00:34 -0700)]
kernel - Fix bug in resource_list_alloc() for addresses > 4GB
* Fix a bug in resource_list_alloc() that was preventing DFly from
mapping PCI memory above 4GB. Basically we had a legacy 'max()'
function use in there that should have been 'ulmax()'.
* Fixes USB device probe on modern Intel laptops.
* Fixes some VMs which use high addresses to map fake devices.
Matthew Dillon [Tue, 28 Jul 2020 00:40:04 +0000 (17:40 -0700)]
boot - Fix 'shift key generates space' in boot loader
* Fix an issue where shift, control, and other non-keystroke keys
generate a space.
The keyboard read code assumed that there would not be two
back-to-back EFI_NOT_READY status returns from ReadKeyStroke.
Pierre-Alain TORET [Mon, 27 Jul 2020 11:42:24 +0000 (13:42 +0200)]
libc/sys : add SIGCKPT and SIGCKPTEXIT to sigaction(2) manual page
Pierre-Alain TORET [Mon, 27 Jul 2020 10:28:57 +0000 (12:28 +0200)]
libc/gen : display the table properly in signal(3) manual page
Sascha Wildner [Mon, 27 Jul 2020 11:35:23 +0000 (13:35 +0200)]
jail.8: Fix typo.
While here, add references to jexec(8) and jls(8).
Reported-by: Peeter Must
François Tigeot [Mon, 27 Jul 2020 06:54:28 +0000 (08:54 +0200)]
drm: Update to Linux 4.12.14
* Initial Geminilake support
* Support improvements and model-specific bug fixes for Broxton, Haswell,
Broadwell, Skylake and Kabylake hardware
* Various generic i915 bug fixes, including page flipping and memory
corruption issues
* Minor radeon bug fixes
François Tigeot [Mon, 27 Jul 2020 06:48:34 +0000 (08:48 +0200)]
drm/linux: Add pagevec API from Linux 4.15+
François Tigeot [Mon, 27 Jul 2020 06:37:56 +0000 (08:37 +0200)]
drm/linux: Add kthread parking functions
Sascha Wildner [Sun, 26 Jul 2020 18:20:07 +0000 (20:20 +0200)]
kernel/vmm: Clean up a bit (staticize, un-expose debug code).
Matthew Dillon [Sun, 26 Jul 2020 17:52:19 +0000 (10:52 -0700)]
kernel - Also include file size in VOP_GETATTR_QUICK()
* Originally VOP_GETATTR_QUICK() was only meant to be used by the
nlookup code but we are now also using it with lseek().
* Fix hammer2_vop_getattr_quick() to also provide the file size.
* Fixes bug introduced when lseek() was changed to use this
VOP.
Matthew Dillon [Sun, 26 Jul 2020 16:16:41 +0000 (09:16 -0700)]
kernel - Fix type-o
* Fix type-o in comment
Reported-by: mjg
Matthew Dillon [Sun, 26 Jul 2020 16:12:15 +0000 (09:12 -0700)]
kernel - more cleanup of syscall2()
* Implement an actual SYS___nosys system call.
* Convert one conditional to something that can use CMOV, using
the new SYS___nosys system call code.
* Get rid of special checks for SYS_syscall and SYS___syscall.
Instead, provide real vectors for these functions. This also
cleans up a few other bits of code in syscall2().
François Tigeot [Sun, 26 Jul 2020 08:35:50 +0000 (10:35 +0200)]
drm/linux: Fix complete() and complete_all()
Completion counters are supposed to be bounded by UINT_MAX.
Sascha Wildner [Sat, 25 Jul 2020 20:14:37 +0000 (22:14 +0200)]
Fix the vkernel build after recent syscall changes.
Matthew Dillon [Sat, 25 Jul 2020 19:11:13 +0000 (12:11 -0700)]
kernel - Add P_SWAPPEDOUT flag back in for deprecated compat
* Add this flag back in for backwards compatibility with some
ports. The flag doesn't do anything any more.
Matthew Dillon [Sat, 25 Jul 2020 17:02:12 +0000 (10:02 -0700)]
kernel - Optimize lseek() SEEK_END
* Use VOP_GETATTR_QUICK() instead of VOP_GETATTR_FP() to omit
unnecessary locking.
Submitted-by: mjg
Matthew Dillon [Sat, 25 Jul 2020 05:57:11 +0000 (22:57 -0700)]
kernel - Remove P_SWAPPEDOUT flag and paging mode
* This code basically no longer functions in any worthwhile or
useful manner, remove it.
The code harkens back to a time when machines had very little
memory and had to time-share processes by actually descheduling
them for long periods of time (like 20 seconds) and paging out
the related memory.
In modern times the chooser algorithm just doesn't work well
because we can no longer assume that programs with large
memory footprints can be demoted.
* In modern times machines have sufficient memory to rely almost
entirely on the VM fault and pageout scan. The latencies caused
by fault-ins are usually sufficient to demote paging-intensive
processes while allowing the machine to continue to function.
If functionality need to be added back in, it can be added back
in on the fault path and not here.
Matthew Dillon [Sat, 25 Jul 2020 05:06:08 +0000 (22:06 -0700)]
kernel - remove old debugging code in syscall2()
* Remove old debugging code that is no longer applicable.
Matthew Dillon [Sat, 25 Jul 2020 05:01:03 +0000 (22:01 -0700)]
kernel - optimize regcnt assignment a bit in syscall2()
* Adjust the regcnt assignment a bit to improve code flow
Matthew Dillon [Sat, 25 Jul 2020 04:25:07 +0000 (21:25 -0700)]
kernel - Refactor in-kernel system call API to remove bcopy()
* Change the in-kernel system call prototype to take the
system call arguments as a separate pointer, and make the
contents read-only.
int sy_call_t (void *);
int sy_call_t (struct sysmsg *sysmsg, const void *);
* System calls with 6 arguments or less no longer need to copy
the arguments from the trapframe to a holding structure. Instead,
we simply point into the trapframe.
The L1 cache footprint will be a bit smaller, but in simple tests
the results are not noticably faster... maybe 1ns or so
(roughly 1%).
Sascha Wildner [Sat, 25 Jul 2020 18:12:22 +0000 (20:12 +0200)]
<machine/*.h>: Use our __aligned() macro in various cases.
No functional change.
François Tigeot [Sat, 25 Jul 2020 08:59:55 +0000 (10:59 +0200)]
drm/linux: Add __list_del() and __list_del_entry()
Obtained-from: FreeBSD
François Tigeot [Sat, 25 Jul 2020 07:41:12 +0000 (09:41 +0200)]
drm/linux: dma_fence_get_rcu_safe()
Obtained-from: OpenBSD
François Tigeot [Sat, 25 Jul 2020 07:39:18 +0000 (09:39 +0200)]
drm/linux: Add various rbtree functions
Obtained-from: OpenBSD
Daniel Fojt [Fri, 24 Jul 2020 07:13:02 +0000 (09:13 +0200)]
libfetch: sync with FreeBSD
- document fetchReqHTTP() (revision 351573)
- avoid invalid escape sequences in URL (revision 357579)
- add SOCKS5 support (revisions 357968, 357977-9, 358227)
Daniel Fojt [Fri, 24 Jul 2020 06:30:17 +0000 (08:30 +0200)]
libressl: add back missing header file
Add back /usr/include/priv/openssl/gost.h that was mistakenly removed
in commit
7dae3d51+.
Pointed out by: Sascha Wildner
Tomohiro Kusumi [Thu, 23 Jul 2020 22:26:25 +0000 (07:26 +0900)]
sys/vfs/ext2fs: Modify XXX comment on disabled htree dirents creation
Sascha Wildner [Thu, 23 Jul 2020 20:52:52 +0000 (22:52 +0200)]
Fix some cases where %zd was used wrongly across the tree.
François Tigeot [Thu, 23 Jul 2020 08:05:47 +0000 (10:05 +0200)]
drm/linux: Fix stop_machine()
It never released the big kernel lock.
Spotted-by: Sascha Wildner <saw@online.de>
Fixed and patch submitted-by: Matthew Dillon <dillon@apollo.backplane.com>
Daniel Fojt [Thu, 23 Jul 2020 04:55:15 +0000 (06:55 +0200)]
Fix 'make upgrade' for updated libressl.
Require -DREMOVE_DEPRECATED to remove old versions of libprivate_crypto
and libprivate_ssl, as pkg(8) may still be linked against them.
Sascha Wildner [Wed, 22 Jul 2020 20:46:19 +0000 (22:46 +0200)]
kernel/e1000: Add missing 'break'.
Tomohiro Kusumi [Wed, 22 Jul 2020 19:12:32 +0000 (04:12 +0900)]
sys/vfs/ext2fs: Add XXX comment on disabled htree dirents creation
Tomohiro Kusumi [Wed, 22 Jul 2020 16:10:17 +0000 (01:10 +0900)]
sys/vfs/ext2fs: Stop checking for failures from malloc(M_WAITOK)
from freebsd/freebsd@
193af7936290dc30509e922b19a32be2d421ff6f
François Tigeot [Wed, 22 Jul 2020 16:21:01 +0000 (18:21 +0200)]
drm/linux: Add get_task_pid()
Sascha Wildner [Wed, 22 Jul 2020 16:18:28 +0000 (18:18 +0200)]
Remove fuse.ko via 'make upgrade'.
So the ever older module doesn't stay around.
Daniel Fojt [Tue, 21 Jul 2020 11:07:52 +0000 (13:07 +0200)]
Upgrade libressl. 2/2
Update README.{DELETED,DRAGONFLY} and adapt Makefiles to vendor changes.
Daniel Fojt [Mon, 20 Jul 2020 08:59:26 +0000 (10:59 +0200)]
Upgrade libressl. 1/2
Merge branch 'vendor/LIBRESSL' into master.
Daniel Fojt [Sat, 18 Jul 2020 07:42:07 +0000 (09:42 +0200)]
vendor/libressl: upgrade from 2.9.1 to 3.1.3
Summary of notable changes:
- fixed CVE-2019-1547 and CVE-2019-1563
- various side channels in DSA and ECDSA were addressed
- fixed PVK format output issue with openssl(1) dsa and rsa subcommand
- added cms subcommand to openssl(1)
- added -addext option to openssl(1) req subcommand
- added -groups option to openssl(1) s_server subcommand
- improved compatibility by backporting functionality and documentation
from OpenSSL 1.1.1
- added RSA-PSS and RSA-OAEP methods from OpenSSL 1.1.1
- completed the port of RSA_METHOD accessors from the OpenSSL 1.1 API
- ported Cryptographic Message Syntax (CMS) implementation from OpenSSL
1.1.1 and enabled by default
- completed initial TLS 1.3 implementation, TLS 1.3 is now enabled
by default for the client side, with the server side to be enabled
in a future release
- provided TLSv1.3 cipher suite aliases to match the names used
in RFC 8446
For details see ChangeLog.
Pierre-Alain TORET [Wed, 22 Jul 2020 08:51:46 +0000 (10:51 +0200)]
libc/sys : display the table properly in sigaction(2) manual page
Taken-from: FreeBSD (r306496)
François Tigeot [Wed, 22 Jul 2020 06:16:12 +0000 (08:16 +0200)]
drm: Implement dma-buf operations, enable most drm_prime code
Inspired-by: OpenBSD
Tomohiro Kusumi [Tue, 21 Jul 2020 16:54:07 +0000 (01:54 +0900)]
sys/vfs/ext2fs: Remove ext2_{bread,brelse}() wrapper variants
Given
f9507b19dfcb670d0003163dcff7a6722eb4abda,
remove ext2_{bread,brelse}() wrapper variants which initially
emulated FreeBSD behavior, and use regular bread()/brelse() variants.
Sascha Wildner [Mon, 20 Jul 2020 22:53:13 +0000 (00:53 +0200)]
libdmsg: Fix wrong check in dmsg_crypto_negotiate()'s error path.
Tomohiro Kusumi [Mon, 20 Jul 2020 17:10:11 +0000 (02:10 +0900)]
sys/vfs/ext2fs: Add redundant ext2_brelse() on ext2_bread() error
Preparation to remove ext2_{bread,brelse}() wrappers that were
initially adopted, and use regular bread()/brelse() variants.
Although not interested in making cleanups that would cause
nonfunctional diff from its origin, this one is better without
FreeBSD compat wrappers.
Sascha Wildner [Mon, 20 Jul 2020 16:58:18 +0000 (18:58 +0200)]
<sys/objcache.h>: Mark two kernel functions as being like malloc().
Tomohiro Kusumi [Mon, 20 Jul 2020 16:16:15 +0000 (01:16 +0900)]
sys/vfs/ext2fs: Fix function name in ihash collision message
Matthew Dillon [Mon, 20 Jul 2020 06:36:46 +0000 (23:36 -0700)]
kernel - Detect fixed meltdown hardware on Intel (2)
* Adjust last commit to work as intended.
Reported-by: mjg
Tomohiro Kusumi [Sun, 19 Jul 2020 16:40:47 +0000 (01:40 +0900)]
sys/vfs/ext2fs: Drop i_devvp initialization in ext2_vget()
This line originally came from ext2_vget() in sys/gnu/vfs/ext2fs
is meaningless in sys/vfs/ext2fs.
Tomohiro Kusumi [Sun, 19 Jul 2020 16:08:44 +0000 (01:08 +0900)]
sys/vfs/ext2fs: Add ext2_alloc_vnode()
Tomohiro Kusumi [Sun, 19 Jul 2020 15:12:07 +0000 (00:12 +0900)]
sys/vfs/ext2fs: Cleanup ext2_vget()/ext2_valloc()
ext2_valloc() shouldn't use VFS_VGET() (whereas sys/gnu/vfs/ext2fs does),
but is similar.
No functional changes.
Sascha Wildner [Sun, 19 Jul 2020 10:46:18 +0000 (12:46 +0200)]
linux/kernel.h: Add some __printf() where appropriate.
Sascha Wildner [Sun, 19 Jul 2020 10:03:18 +0000 (12:03 +0200)]
<sys/stdint.h>: Use __GNUC_PREREQ__.
Sascha Wildner [Sun, 19 Jul 2020 08:41:29 +0000 (10:41 +0200)]
kernel: Prepare for -fno-common becoming default in future compilers.
Add an explicit -fcommon to the compilation of genassym.c and
vmx_genassym.c, just like we did for asmcontext.c in libc already.
genassym.o and vmx_genassym.o are used for the generation of the
respective *.h files (using genassym.sh) which fails without
-fcommon. They are not themselves linked into the kernel.
See
dadd01b6c5da8ff83652d6c3be89b16f373af69a.
Sascha Wildner [Sun, 19 Jul 2020 08:08:09 +0000 (10:08 +0200)]
Sync ACPICA with Intel's version
20200717.
* Do not increment OperationRegion reference counts for field units.
* Fix some issues in iasl(8).
For detailed list, please see sys/contrib/dev/acpica/changes.txt.
Matthew Dillon [Sun, 19 Jul 2020 06:19:45 +0000 (23:19 -0700)]
kernel - Detect fixed meltdown hardware on Intel
* Detect fixed meltdown hardware on Intel.
* Display standard cpu features2 and features3 info during boot.
Requested-by: mjg
Tomohiro Kusumi [Sat, 18 Jul 2020 17:19:54 +0000 (02:19 +0900)]
sys/vfs/ext2fs: Disable ext3 htree dirents for now
Seeing occasional failure with buildworld test on ext2,
but not when mkfs'd with ^dir_index.
Tomohiro Kusumi [Sat, 18 Jul 2020 16:10:06 +0000 (01:10 +0900)]
sys/vfs/fuse: Remove FUSE from Makefile for now
Disable FUSE build for now.
Not much progress since first committed in 2019.
Note that FUSE is difficult to implement in BSD's from the way
VFS is designed (can't properly implement "fh").
Tomohiro Kusumi [Sat, 18 Jul 2020 15:33:34 +0000 (00:33 +0900)]
sys/vfs/fuse: Fix fuse_vop_readdir()
Tomohiro Kusumi [Sat, 18 Jul 2020 15:33:10 +0000 (00:33 +0900)]
sys/vfs/autofs: Fix autofs_readdir()
Tomohiro Kusumi [Sat, 18 Jul 2020 17:15:45 +0000 (02:15 +0900)]
sys/vfs/ext2fs: Use bp->b_loffset in ext2_readdir()
Tomohiro Kusumi [Fri, 17 Jul 2020 16:12:07 +0000 (01:12 +0900)]
Add non-GPL ext2 filesystem (port from FreeBSD)
The new ext2 will replace the existing sys/gnu/vfs/ext2fs in future,
which is known to have lots of issues. This is already better than
sys/gnu/vfs/ext2fs in many ways, but will keep it as an option for now
as there seems to be htree dirents and/or vop_readdir related bugs.
The new implementation supports ext3 htree dirents. Read support for
ext4 extents which exists in FreeBSD ext2 isn't implemented yet (TODO).
ACL/xattr are unsupported, and likely never will be.
With exception of DragonFly VFS specific part that can't reuse FreeBSD
code (vfsops, many vnops, bread/getblk/bio i/f difference, etc),
the new ext2 is designed to be close to upstream as much as possible.
The code is there for a reason to make clear the *real* diff without
introducing nonfunctional diff noises.
Daniel Fojt [Fri, 17 Jul 2020 05:11:08 +0000 (07:11 +0200)]
Upgrade ldns and drill(1). 2/2
- regenerate config.h, net.h and util.h
- update README.{DELETED,DRAGONFLY}
- update manpage drill.1: remove empty section "BUGS", merge duplicate
sections "SEE ALSO" and remove a reference to dig(1)
Daniel Fojt [Fri, 17 Jul 2020 05:10:04 +0000 (07:10 +0200)]
Upgrade ldns and drill(1). 1/2
Merge branch 'vendor/LDNS' into master.
Daniel Fojt [Thu, 16 Jul 2020 06:54:34 +0000 (08:54 +0200)]
vendor/ldns: upgrade from 1.7.0 to 1.7.1
Notable bugfixes:
- drill -x crashes with malformed IPv4 address
- drill -DT fails for CNAME domain
- CDS & CDNSKEY RRsets should be signed with the KSK
- several potential NULL pointer dereferences
- memory leak with more EDNS sections
- time sensitive TSIG compare vulnerability
- only chase DS if signer is parent of owner
For details see commit history or changelog on github:
https://github.com/NLnetLabs/ldns/blob/release-1.7.1/Changelog
Matthew Dillon [Thu, 16 Jul 2020 23:30:36 +0000 (16:30 -0700)]
kernel - use built-in for ffsl()
* Use built-in for ffsl(), reduces generated code and improves
performance by using CMOV
* Remove old unused/dead code in ffs()
Submitted-by: mjg
Sascha Wildner [Thu, 16 Jul 2020 20:00:47 +0000 (22:00 +0200)]
Reference efibootmgr(8) in boot(8) manual page.
Sascha Wildner [Thu, 16 Jul 2020 17:38:22 +0000 (19:38 +0200)]
Update the pciconf(8) database.
June 24, 2020 snapshot from https://pci-ids.ucw.cz
Sascha Wildner [Thu, 16 Jul 2020 07:30:00 +0000 (09:30 +0200)]
Fix a few typos across the tree.
Sascha Wildner [Wed, 15 Jul 2020 18:41:16 +0000 (20:41 +0200)]
kernel: Remove some unused variables and a dead sysctl.
Reported-by: mjg
Sascha Wildner [Wed, 15 Jul 2020 09:24:34 +0000 (11:24 +0200)]
libc/asmcontext: Fix building with -fno-common.
asmcontext.h generation using genassym.sh isn't working properly with
-fno-common, so for now just specify -fcommon at the end of the flags.
Right now this is only needed for testing -fno-common because it is
not GCC's default yet.
This concludes the work to make buildworld succeed with -fno-common.
Thanks to Krzysztof Piecuch for helping with it.
Sascha Wildner [Mon, 13 Jul 2020 19:04:53 +0000 (21:04 +0200)]
libc/asmcontext: Remove unneeded ASSYM()s.
Sascha Wildner [Wed, 15 Jul 2020 09:23:53 +0000 (11:23 +0200)]
mail(1): Fix building with -fno-common.
Similar to how we "fixed" tip(1).
-fno-common will be default in future compilers (GCC 10, for example).
Sascha Wildner [Wed, 15 Jul 2020 09:23:19 +0000 (11:23 +0200)]
tip(1): Fix building with -fno-common.
We do so by adding 'EXTERN' to all offending variables and then defining
'EXTERN' to 'extern' for all .c files, except for one (tip.c). This is
the same "fix" that FreeBSD chose, to be cleaned up better at a later
point in time.
-fno-common will be default in future compilers (GCC 10, for example).
Sascha Wildner [Wed, 15 Jul 2020 09:22:49 +0000 (11:22 +0200)]
telnet(1)/telnetd(8): Fix building with -fno-common.
-fno-common will be default in future compilers (GCC 10, for example).
François Tigeot [Wed, 15 Jul 2020 09:23:35 +0000 (11:23 +0200)]
drm: restore pre-Linux 4.10 print functions
Newer versions tend to print garbage
François Tigeot [Mon, 13 Jul 2020 21:50:01 +0000 (23:50 +0200)]
drm: Update base driver to Linux 4.10.17
Thanks to Matthew Dillon <dillon@apollo.backplane.com>
for some VM fixes.
François Tigeot [Mon, 13 Jul 2020 05:57:03 +0000 (07:57 +0200)]
drm/linux: Do not sleep when not required in schedule_timeout()
Krzysztof Piecuch [Sat, 11 Jul 2020 17:40:22 +0000 (17:40 +0000)]
tftp(1): Fix building with -fno-common
-fno-common will be default in future compilers (GCC 10, for example).
Krzysztof Piecuch [Fri, 10 Jul 2020 22:58:31 +0000 (22:58 +0000)]
fsck(8), fsdb(8): Fix building with -fno-common
-fno-common will be default in future compilers (GCC 10, for example).
François Tigeot [Sun, 12 Jul 2020 07:57:59 +0000 (09:57 +0200)]
drm/linux: break from wait_event_xxx() loops sooner
François Tigeot [Sun, 12 Jul 2020 07:52:15 +0000 (09:52 +0200)]
drm/linux: Fix lost wakeup event issues with wait_event_xxx() routines
Submitted-by: Matthew Dillon <dillon@apollo.backplane.com>
François Tigeot [Sun, 12 Jul 2020 07:30:37 +0000 (09:30 +0200)]
drm/linux: Implement module_param() directives
* Properly define KBUILD_MODNAME where required
* Rewrite linux/moduleparam.h, Automatically generate suitable
TUNABLE_INT() directives
François Tigeot [Sun, 12 Jul 2020 05:20:52 +0000 (07:20 +0200)]
drm/linux: Import the radix-tree code from FreeBSD
Daniel Fojt [Sat, 11 Jul 2020 17:43:00 +0000 (19:43 +0200)]
libexec/dma: split body of mails not respecting RFC5322
When receiving mails with body not respecting RFC5322 [1] (may happen
with automatically generated mails, for example from cron(8)), try to
split by words finding the last space before 1000's character. If this
fails, then simply refuse to further receive and process such input.
Taken from: FreeBSD (revision 326641)
Discussed with: Sascha Wildner
[1] https://tools.ietf.org/html/rfc5322#section-2.1.1
François Tigeot [Sat, 11 Jul 2020 10:54:04 +0000 (12:54 +0200)]
drm/linux: Fix an unlock without lock issue in dma_fence_default_wait()
François Tigeot [Sat, 11 Jul 2020 09:58:18 +0000 (11:58 +0200)]
drm/linux: fix races in the ww mutex code that can cause lockups
Submitted-by: Matthew Dillon <dillon@apollo.backplane.com>
Sascha Wildner [Fri, 10 Jul 2020 20:19:05 +0000 (22:19 +0200)]
wsp(4): Clean up the Makefile.