Dimitry Andric [Thu, 26 Mar 2020 17:28:54 +0000 (17:28 +0000)]
Merge commit 459e8e948 from llvm git (by Justin Hibbits):
[PowerPC]: Don't allow r0 as a target for LD_GOT_TPREL_L/32
Summary:
The linker is free to relax this (relocation R_PPC_GOT_TPREL16)
against R_PPC_TLS, if it sees fit (initial exec to local exec). If r0
is used, this can generate execution-invalid code (converts to 'addi
%rX, %r0, FOO, which translates in PPC-lingo to li %rX, FOO). Forbid
this instead.
This fixes static binaries using locales on FreeBSD/powerpc (tested
on FreeBSD/powerpcspe).
Dimitry Andric [Thu, 26 Mar 2020 17:27:41 +0000 (17:27 +0000)]
Merge commit f0990e104 from llvm git (by Justin Hibbits):
[PowerPC]: e500 target can't use lwsync, use msync instead
The e500 core has a silicon bug that triggers an illegal instruction
program trap on any sync other than msync. Other cores will typically
ignore illegal sync types, and the documentation even implies that
the 'illegal' bits are ignored.
Address this hardware deficiency by only using msync, like the PPC440.
Leandro Lupori [Thu, 26 Mar 2020 12:17:49 +0000 (12:17 +0000)]
[PowerPC64] Fix OPAL IPMI driver
This change fixes a couple of issues with OPAL IPMI driver and
implements a mechanism to detect timeouts and discard old messages left
in receive queue, to avoid old messages from being confused with the
reply of new ones.
Reviewed by: jhibbits
Sponsored by: Eldorado Research Institute (eldorado.org.br)
Differential Revision: https://reviews.freebsd.org/D24185
Property-based filters allow substring and regular expressions
(see re_format(7)) matching against various message attributes.
Filter specification starts with '#:' or ':' followed by three
comma-separated fields property, operator, "value". Value must be
double-quoted. A double quote and backslash must be escaped by a
blackslash.
Following properties are supported as test value:
o msg - body of the message received;
o programname - program name sent the message;
o hostname - hostname of message's originator;
o source - an alias for hostname.
Supported operators:
o contains - true if filter value is found as a substring of property;
o isequal - true if filter value is equal to property;
o startswith - true if property starts with filter value;
o regex - true if property matches basic regular expression defined
in filter value;
o ereregex - true if property matches extended regular expression
defined in filter value;
Operator may be prefixed by '!' to invert compare logic or by
'icase_' to make comparison function case insensitive.
Submitted by: Boris N. Lytochkin <lytboris at gmail com>
MFC after: 2 weeks
Relnotes: yes
Differential Revision: https://reviews.freebsd.org/D23468
Conrad Meyer [Wed, 25 Mar 2020 23:12:43 +0000 (23:12 +0000)]
Expand generic subword atomic primitives
The goal of this change is to make the atomic_load_acq_{8,16},
atomic_testandset{,_acq}_long, and atomic_testandclear_long primitives
available in MI-namespace.
The second goal is to get this draft out of my local tree, as anything that
requires a full tinderbox is a big burden out of tree. MD specifics can be
refined individually afterwards.
The generic implementations may not be ideal for your architecture; feel
free to implement better versions. If no subword_atomic definitions are
needed, the include can be removed from your arch's machine/atomic.h.
Generic definitions are guarded by defined macros of the same name. To
avoid picking up conflicting generic definitions, some macro defines are
added to various MD machine/atomic.h to register an existing implementation.
Include _atomic_subword.h in arm and arm64 machine/atomic.h.
For some odd reason, KCSAN only generates some versions of primitives.
Generate the _acq variants of atomic_load.*_8, atomic_load.*_16, and
atomic_testandset.*_long. There are other questionably disabled primitives,
but I didn't run into them, so I left them alone. KCSAN is only built for
amd64 in tinderbox for now.
Add atomic_subword implementations of atomic_load_acq_{8,16} implemented
using masking and atomic_load_acq_32.
Add generic atomic_subword implementations of atomic_testandset_long(),
atomic_testandclear_long(), and atomic_testandset_acq_long(), using
atomic_fcmpset_long() and atomic_fcmpset_acq_long().
On x86, add atomic_testandset_acq_long as an alias for
atomic_testandset_long.
Simon J. Gerraty [Wed, 25 Mar 2020 19:12:19 +0000 (19:12 +0000)]
Fix pkgfs stat so it satisfies libsecureboot
We need a valid st_dev, st_ino and st_mtime
to correctly track which files have been verified
and to update our notion of time.
ve_utc_set(): ignore utc if it would jump our current time
by more than VE_UTC_MAX_JUMP (20 years).
Allow testing of install command via userboot.
Need to fix its stat implementation too.
bhyveload also needs stat fixed - due to change to userboot.h
Call ve_error_get() from vectx_close() when hash is wrong.
Track the names of files we have hashed into pcr
For the purposes of measured boot, it is important
to be able to reproduce the hash reflected in
loader.ve.pcr
so loader.ve.hashed provides a list of names in the order they
were added.
Michael Tuexen [Wed, 25 Mar 2020 18:20:37 +0000 (18:20 +0000)]
Remove an optimization, which was incorrect a couple of times and
therefore doesn't seem worth to be there.
In this case COOKIE where not retransmitted anymore, when the
socket was already closed.
Ryan Moeller [Wed, 25 Mar 2020 15:56:18 +0000 (15:56 +0000)]
MFOpenZFS: ZVOLs should not be allowed to have children
zfs create, receive and rename can bypass this hierarchy rule. Update
both userland and kernel module to prevent this issue and use pyzfs
unit tests to exercise the ioctls directly.
Note: this commit slightly changes zfs_ioc_create() ABI. This allow to
differentiate a generic error (EINVAL) from the specific case where we
tried to create a dataset below a ZVOL (ZFS_ERR_WRONG_PARENT).
Reviewed-by: Paul Dagnelie <pcd@delphix.com> Reviewed-by: Matt Ahrens <mahrens@delphix.com> Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov> Reviewed-by: Tom Caputi <tcaputi@datto.com> Signed-off-by: loli10K <ezomori.nozomu@gmail.com>
Approved by: mav (mentor)
MFC after: 2 weeks
Sponsored by: iXsystems, Inc.
openzfs/zfs@d8d418ff0cc90776182534bce10b01e9487b63e4
This introduces a regression reported by koobs@ when running a pyhton
test suite on a loaded system.
This patch resulted in a failing accept() call, when the association
was setup and gracefully shutdown by the peer before accept was called.
So the following packetdrill script would fail:
Emmanuel Vadot [Wed, 25 Mar 2020 01:36:23 +0000 (01:36 +0000)]
pkgbase: Stop threating rc files as config files
rc scripts arent configuration files so use FILES instead of CONFS.
While here put rc scripts into related package (sendmail in the FreeBSD-sendmail
package, wpa_supplicant in the FreeBSD-wpa etc ...)
Emmanuel Vadot [Wed, 25 Mar 2020 01:35:13 +0000 (01:35 +0000)]
Re-apply r359267 now that tools are using the proper include path
Original commit message:
bsd.lib.mk: Do not include bsd.incs.mk for INTERNALLIB
f we're building an internal lib do not bother including bsd.incs.mk so we
will not install the headers.
This also "solves" a problem with pkgbase where a libXXX-development package
is created and due to how packages are created we add a dependency to a
libXXX package that doesn't exists.
Mitchell Horne [Tue, 24 Mar 2020 23:25:54 +0000 (23:25 +0000)]
Makefile.inc1: override MACHINE for native-xtools
For the final step of the native-xtools target, "everything" is built
with TARGET and TARGET_ARCH set to the architecture we wish to
cross-build for. However, CROSSENV overwrites the values of MACHINE and
MACHINE_ARCH, setting them to be identical to TARGET and TARGET_ARCH.
For native-xtools this is undesirable since we are building binaries to
run on the host architecture, that can cross compile for the target
architecture.
When building native-xtools for RISC-V, this issue manifests as an invalid
argument for "-march". The compiler is invoked with the target triple of the
host architecture, but the CFLAGS inherited from bsd.cpu.mk are that of the
target architecture.
Michael Tuexen [Tue, 24 Mar 2020 22:44:36 +0000 (22:44 +0000)]
Another cleanup of the timer code. Also be more pedantic about the
parameters of the timer start and stop routines. Several inconsistencies
have been fixed in earlier commits. Now they will be catched when running
an INVARIANTS system.
Ed Maste [Tue, 24 Mar 2020 21:37:34 +0000 (21:37 +0000)]
Makefile.inc1: remove old (2018) stale dependency hacks
We handle a number of stale dependency issues on an ad-hoc basis, in
order to support ongoing NO_CLEAN builds. These hacks do not need to
be maintained indefinitely; now remove those for issues that are more
than a year old.
Warner Losh [Tue, 24 Mar 2020 19:33:21 +0000 (19:33 +0000)]
exec{l,v}{e,p} arrived in 7th Edition research Unix to support the Bourne Shell
which introduced environment variables. Document that here. Verified by
consulting the TUHS archive.
Conrad Meyer [Tue, 24 Mar 2020 19:20:10 +0000 (19:20 +0000)]
Fix PNP matching for iflib NIC drivers
The previous descriptor string specified that all fields were significant for
match. However, the only significant fields for in-tree drivers are
vendor:devid, and the fictitious zero values constructed by PVID() did not
match real subvendor, subdevice, revision, and/or class values, resulting in no
automatic probe.
If a future iflib driver needs to match on other criteria, the descriptor
string can be updated accordingly. (E.g., "V32" and ~0 for unspecified values
in PVID().)
Mark Johnston [Tue, 24 Mar 2020 18:43:23 +0000 (18:43 +0000)]
Remove the secondary_stacks array in arm64 and riscv kernels.
Instead, dynamically allocate a page for the boot stack of each AP when
starting them up, like we do on x86. This shrinks the bss by
MAXCPU*KSTACK_PAGES pages, which corresponds to 4MB on arm64 and 256KB
on riscv.
Duplicate the logic used on x86 to free the bootstacks, by using a
sysinit to wait for each AP to switch to a thread before freeing its
stack.
While here, mark some static MD variables as such.
Mark Johnston [Tue, 24 Mar 2020 18:16:56 +0000 (18:16 +0000)]
Add regression tests for newsyslog.conf's p flag.
While here do a bit of cleanup:
- declare local variables as such,
- make tmpdir_create() clean up logfile directories, to handle a
previously interrupt test run more gracefully.
MFC after: 2 weeks
Sponsored by: The FreeBSD Foundation
Mark Johnston [Tue, 24 Mar 2020 18:16:36 +0000 (18:16 +0000)]
newsyslog: Fix stack corruption when initializing a zipwork structure.
This happens when compressing a previously uncompressed already-rotated
file, as happens when handling the 'p' flag in newsyslog.conf. The file
name is stored in a flexible array member, so these structures cannot be
stack allocated.
Also make sure that we call change_attrs() and do_zipwork() in dry-run
mode; they handle this properly, contrary to the commit log message for
r327451.
John Baldwin [Tue, 24 Mar 2020 18:16:02 +0000 (18:16 +0000)]
Disable rarely used architecture variants in make universe by default.
If EXTRA_TARGETS is defined, build all supported architecture
variants. By default, build architecture variants needed to provide
code coverage or that are commonly used.
Use this to disable building of all the hard-float and little-endian
MIPS architecture variants along with n32 by default.
Ed Maste [Tue, 24 Mar 2020 17:25:56 +0000 (17:25 +0000)]
iflib: split compound assertion
ThunderX cluster systems are panicking on boot with a failed assertion
MPASS(gtask != NULL && gtask->gt_taskqueue != NULL). Split the
assertion so that it's clear which part is failing.
Use IP_FW_NAT44_DESTROY opcode for IP_FW3 socket option to destroy
NAT instance.
The NAT44 group of opcodes for IP_FW3 socket option is modern way
to control NAT instances and this method can be used in future to
switch from numeric to named NAT instances, like was done for ipfw
tables.
The IP_FW_NAT_DEL opcode is the last remnant of old ipfw_ctl control
plane that doesn't support versioned operations. This interface will
be retired soon.
Emmanuel Vadot [Tue, 24 Mar 2020 01:09:04 +0000 (01:09 +0000)]
bsd.lib.mk: Do not include bsd.incs.mk for INTERNALLIB
If we're building an internal lib do not bother including bsd.incs.mk so we
will not install the headers.
This also "solves" a problem with pkgbase where a libXXX-development package
is created and due to how packages are created we add a dependency to a
libXXX package that doesn't exists.
Emmanuel Vadot [Tue, 24 Mar 2020 01:07:01 +0000 (01:07 +0000)]
pkgbase: Move device.hints from the runtime to the bootloader package
Also mark it as config file so if a user changes this file pkg will attempt
to merge the new file upon an update.
device.hints is neither related to runtime or loader but it make more sense
to have it in loader in case some user delete /boot/ and wants to recreate it,
now only two packages are required FreeBSD-bootloader and the kernel package.
While here change where we override the package for files installed in /boot,
this allow us to keep other tags (such as config).
Conrad Meyer [Mon, 23 Mar 2020 23:00:13 +0000 (23:00 +0000)]
arm: Fix atomic long APIs to correct 'u_long' signedness
As defined in atomic(9) and implemented on other architectures, the
atomic(9) functions all act on unsigned pointers and types. Prior to this
revision, arm implemented some atomic(9) 'long' sized routines with correct
unsigned type, but others were incorrectly signed.
John Baldwin [Mon, 23 Mar 2020 21:26:32 +0000 (21:26 +0000)]
Use the newer EINTEGRITY error when authentication fails.
GELI used to fail with EINVAL when a read request spanned a disk
sector whose contents did not match the sector's authentication tag.
The recently-added EINTEGRITY more closely matches to the error in
this case.
Ed Maste [Mon, 23 Mar 2020 18:17:56 +0000 (18:17 +0000)]
arch.7: remove Default Tool Chain footnote about xtoolchain
MIPS was the last arch to use external toolchain by default but uses
in-tree Clang and lld as of r359233, and now no table entries reference
the footnote.
Ed Maste [Mon, 23 Mar 2020 18:14:04 +0000 (18:14 +0000)]
arch.7: update Default Tool Chain intro text
All FreeBSD archs now use an in-tree toolchain - Clang and ELF Tool
Chain everywhere, and lld everywhere but 32-bit PowerPC (which still
uses ld.bfd). No archs use external toolchain by default.
Alex Richardson [Mon, 23 Mar 2020 17:51:44 +0000 (17:51 +0000)]
Fix newvers.sh on macOS 10.15
It appears that the macOS /bin/sh echo now defaults to -e and therefore the
`#define VERSTR` included newline characters instead of \n. This caused compiler
errors due to unterminated strings. Fix by using printf instead of echo.
A less fragile solution might be to bootstrap the in-tree /bin/sh but that
requires more changes.
Warner Losh [Mon, 23 Mar 2020 16:02:01 +0000 (16:02 +0000)]
Document when execl and execp entered unix.
For this, things are complicated. The first mention in the manual was in the 4th
edition manual (as an add on to exec(II)). The 2nd and 3rd editions didn't have
these in the manual (either as a separate page, or as an add-on to exec(II)). We
don't have good 1st, 2nd or 3rd edition distributions to look in. However,
there's a tape labeled 'last1120c' that we do have. This tape contains the last
version of the V2 edition of the C compiler on it (just after C got struct). On
this tape there was a libc.sa archive that contains source for execl and
execp. This source is sufficiently different from the V5 sources (which are the
next ones we have sources for) and have a slightly different calling convention
than later sources, suggesting that the early date for the last1120c tape is
correct (in that era, the epoch changed every year, leading to a one or two year
ambiguity on when the files could have been modified) and it should be though of
as V2. Since this was also a time of compiler development, and the calling
convetions are known to be under evolution, and since the rest of the sources in
libc.sa are consistent, that's further evidence that V2 is likely. Finally, 2nd
edition was the last version to fully support the 11/20 because it lacked many
basic features and bell labs moved off it to the 11/45 as soon as they could
afford to buy one, around this time era. The unix manuals make it sound like V3
might have supported the 11/20, but the same intro could also be read to mean it
didn't, at all, and that V3 was the first rewrite for the 11/45 ahead of the
rewrite in C that came with V4.
Taken together, the evidence leans most heavily to V2 (90% IMHO), and slightly
to V3 (8%) or possibly V4 (2%). I've not put all this in the man page, but have
left it here in case someone notices in the future that V4 is the first manual
page for it.
Alex Richardson [Mon, 23 Mar 2020 10:36:32 +0000 (10:36 +0000)]
Use Clang and LLD as the default toolchain for MIPS
Now that we have updated the in-tree version of LLVM to 10.0, we have all the
necessary LLVM changes to use Clang+LLD as the default toolchain for MIPS.
When deleting a user, if its home directory does not belong to it, it should
not be removed. This is the promise that the manpage makes, the tool should
ensure that it respects that promise.
Add a regression test about it
PR: 244967
Submitted by: Eric Hanneken <eric@erichanneken.com>
MFC after: 3 days
Alex Richardson [Sun, 22 Mar 2020 22:18:00 +0000 (22:18 +0000)]
Fix linking OCTEON1 kernel with LLD
LLD complains that the type of .dynamic was changed. Fix this by copying
the approach used in the mips64 ldscript.
I do not have hardware to test this change so I only verified that the
kernel links and the section layout looks sensible.
There is an example in tail(1) manual page explaining how to use tail(1) to
track the contents of /var/log/messages. The example uses the -f flag to
follow the file. The problem with the -f flag is that it cannot handle the
situation where /var/log/messages is rotated. Hence, use -F instead in the
example.
Reviewed by: asomers
MFC after: 3 days
Differential Revision: https://reviews.freebsd.org/D24157
Rick Macklem [Sun, 22 Mar 2020 18:18:30 +0000 (18:18 +0000)]
Fix an NFS mount attempt where VFS_STATFS() fails.
r353150 added mnt_rootvnode and this seems to have broken NFS mounts when the
VFS_STATFS() called just after VFS_MOUNT() returns an error.
Then the code calls VFS_UNMOUNT(), which calls vflush(), which returns EBUSY.
Then the thread get stuck sleeping on "mntref" in vfs_mount_destroy().
This patch fixes this problem.
projects / FreeBSD / FreeBSD.git / log