Add support for pmap_enter(..., psind=1) to the amd64 pmap. In other words,
add support for explicitly requesting that pmap_enter() create a 2MB page
mapping. (Essentially, this feature allows the machine-independent layer to
create superpage mappings preemptively, and not wait for automatic promotion
to occur.)
Export pmap_ps_enabled() to the machine-independent layer.
Add a flag to pmap_pv_insert_pde() that specifies whether it should fail or
reclaim a PV entry when one is not available.
Refactor pmap_enter_pde() into two functions, one by the same name, that is
a general-purpose function for creating PDE PG_PS mappings, and another,
pmap_enter_2mpage(), that is used to prefault 2MB read- and/or execute-only
mappings for execve(2), mmap(2), and shmat(2).
In vm_page_ps_test(), always check that the base pages within the specified
superpage all belong to the same object. To date, that check has not been
needed, but upcoming changes require it. (See the Differential Revision.)
linuxkpi compiler.h: avoid gcc -Wunused-value in dummy expressions
It looks like the __acquire and __release macros are for the consumption
of static analysis tools and have no semantic effect. Transform the
definitions from constant expressions to empty statements in order to
avoid -Wunused-value from gcc.
Likewise avoid future warnings for __chk_{user,io}_ptr, but with a cast
to void, because it looks like some linux kernel code may use those in
expression contexts.
Only filter out the PF ioctls if we're building without pf support.
Until now those were always filtered out, so truss did not show symbolic
names for pf ioctls.
dim [Sat, 22 Jul 2017 11:08:25 +0000 (11:08 +0000)]
Upgrade our copies of clang, llvm, lld, lldb, compiler-rt and libc++ to
5.0.0 (trunk r308421). Upstream has branched for the 5.0.0 release,
which should be in about a month. Please report bugs and regressions,
so we can get them into the release.
Please note that from 3.5.0 onwards, clang, llvm and lldb require C++11
support to build; see UPDATING for more information.
Improve publication of the newly allocated snapdata.
For freshly allocated snapdata, Lock sn_lock in advance, so
si_snapdata readers see the locked snapdata and not race.
For existing snapdata, if the thread was put to sleep waiting for
sn_lock, re-read si_snapdata. This either closes the race or makes
the reliance on LK_DRAIN less important.
Reported and tested by: pho
Sponsored by: The FreeBSD Foundation
MFC after: 2 weeks
It is possible for ffs_snapblkfree() to race and lock snaplock while
the devvp snapdata is instantiated, but no snapshots exist. In this
case the loop over snapshots in ffs_snapblkfree() is not executed, and
the local variable vp is left initialized to NULL.
Unlock using &sn->sn_lock and not vp->v_vnlock. For the inodes on the
snapshot list, the locks are same.
Reported and tested by: pho
Sponsored by: The FreeBSD Foundation
MFC after: 2 weeks
Account for lock recursion when transfering snaplock to the vnode lock
in ffs_snapremove().
Apparently ffs_snapremove() may be called with the snap lock recursed,
at least one trace demonstrated this when snapshot vnode was unlinked
while synced. It was inactivated from the syncer thread.
Reported and tested by: pho
Sponsored by: The FreeBSD Foundation
MFC after: 2 weeks
dim [Fri, 21 Jul 2017 17:59:54 +0000 (17:59 +0000)]
Pull in r295886 from upstream clang trunk (by Richard Smith):
PR32034: Evaluate _Atomic(T) in-place when T is a class or array type.
This is necessary in order for the evaluation of an _Atomic
initializer for those types to have an associated object, which an
initializer for class or array type needs.
This fixes an assertion when building recent versions of LinuxCNC.
Pollution from counter.h made __pcpu visible in amd64/pmap.c. Delete
the existing extern decl of __pcpu in amd64/pmap.c and avoid referring
to that symbol, instead accessing the pcpu region via PCPU_SET macros.
Also delete an unused extern decl of __pcpu from mp_x86.c.
Fix getsockopt() for listening sockets when using SO_SNDBUF, SO_RCVBUF,
SO_SNDLOWAT, SO_RCVLOWAT. Since r31972 it only worked for non-listening
sockets.
P1022 and MPC8536 include a 'jog' feature for clock control
(jog being a slower form of run mode). This is done by changing the
PLL multiplier, and cannot be done if any core is in doze or sleep mode.
BREs recently became prematurely sensitive to the branching operator, which
outright broke expressions that used it instead of failing silently. Test
that \| is matching a literal | for the time being.
r320062 introduced a bug when doing NFSv4.1 mounts against some non-FreeBSD servers.
r320062 used nm_rsize, nm_wsize to set the maximum request/response sizes for
the NFSv4.1 session. If rsize,wsize are not specified as options, the
value of nm_rsize, nm_wsize is 0 at session creation, resulting in
values for request/response that are too small.
This patch fixes the problem. A workaround is to specify rsize=N,wsize=N
mount options explicitly, so they are set before session creation.
This bug only affects NFSv4.1 mounts against some non-FreeBSD servers.
r320062 introduced a bug when doing NFSv4.1 mounts against some non-FreeBSD servers.
r320062 used nm_rsize, nm_wsize to set the maximum request/response sizes for
the NFSv4.1 session. If rsize,wsize are not specified as options, the
value of nm_rsize, nm_wsize is 0 at session creation, resulting in
values for request/response that are too small.
This patch fixes the problem. A workaround is to specify rsize=N,wsize=N
mount options explicitly, so they are set before session creation.
This bug only affects NFSv4.1 mounts against some non-FreeBSD servers.
dim [Thu, 20 Jul 2017 20:28:31 +0000 (20:28 +0000)]
Fix printf format warning in iflib.c
Clang 5.0.0 got better warnings about printf format strings using %zd,
and this leads to the following -Werror warning on e.g. arm:
sys/net/iflib.c:1517:8: error: format specifies type 'ssize_t' (aka 'int') but the argument has type 'bus_size_t' (aka 'unsigned long') [-Werror,-Wformat]
sctx->isc_tx_maxsize, nsegments, sctx->isc_tx_maxsegsize);
^~~~~~~~~~~~~~~~~~~~
sys/net/iflib.c:1517:41: error: format specifies type 'ssize_t' (aka 'int') but the argument has type 'bus_size_t' (aka 'unsigned long') [-Werror,-Wformat]
sctx->isc_tx_maxsize, nsegments, sctx->isc_tx_maxsegsize);
^~~~~~~~~~~~~~~~~~~~~~~
Fix this by casting bus_size_t arguments to uintmax_t, and using %ju
instead.
Reviewed by: emaste
MFC after: 3 days
Differential Revision: https://reviews.freebsd.org/D11679
dim [Thu, 20 Jul 2017 20:27:19 +0000 (20:27 +0000)]
Fix printf format warning in zfs_module.c
Clang 5.0.0 got better warnings about print format strings using %zd,
and this leads to the following -Werror warning on e.g. arm:
sys/boot/efi/boot1/zfs_module.c:186:18: error: format specifies type 'ssize_t' (aka 'int') but the argument has type 'off_t' (aka 'long long') [-Werror,-Wformat]
"(%lu)\n", st.st_size, spa->spa_name, filepath, EFI_ERROR_CODE(status));
^~~~~~~~~~
Fix this by casting off_t arguments to intmax_t, and using %jd instead.
Reviewed by: tsoome
MFC after: 3 days
Differential Revision: https://reviews.freebsd.org/D11678
Private functions like __aio_read and _aio_read were exposed in
FBSDprivate_1.0 by r169090, even though they've never been used outside of
librt. Also, remove some weak references from r156136 that have never
resolved.
efi: restrict visibility of EFIABI_ATTR-declared functions
In-tree gcc (4.2) doesn't understand __attribute__((ms_abi))
(EFIABI_ATTR). Avoid declaring functions with that attribute when the
compiler is detected to be gcc < 4.4.
- Visualize mutually exclusive options and their corresponding
arguments.
- Try to make the subtleties that are expressed in the code, and
potentially in the manpages, more apparent.
- Delete trailing whitespace.
- Fix leading indentation (convert single column spaces to tabs).
- Convert "[Ff]all through" to "FALLTHROUGH", per implicit project
style/spelling.
Restore igb(4) code dropped during iflib conversion
- restore newer code for vf, i350, i210, i211
- restore dmac init code for i354 and i350
- restore WUC/WUFC update
- check for igb mac type before attempting trying to assert
a media changed event.
- handle link events for igb(4) and em(4) devices differently
and appropriately for their respective model types.
Submitted by: Matt Macy <mmacy@mattmacy.io>
Sponsored by: Limelight Networks
Update the nfsv4 man page to reflect recent changes to support
the newer RFCs (5661 and 7530). The main man changes are for the
case of "numbers in strings" for user/groups that RFC7530 allows
and avoids use of nfsuserd(8).
- Document /etc/cron.d and /usr/local/etc/cron.d under FILES.
- Reword documentation for -n: add appropriate soft-stop and remove
contraction to appease igor.
Raise the watchdog timer interval to 2 ticks, there by guaranteeing
that it fires between 1ms and 2ms. `
Treat two consecutive occurrences of Heartbeat failures as a legitimate
Heartbeat failure
Disable stack growth when accessed by AIO daemons.
Commit message for r321173 incorrectly stated that the change disables
automatic stack growth from the AIO daemons contexts, with explanation
that this is currently prevents applying wrong resource limits. Fix
this by actually disabling the growth.
Noted by: alc
Reviewed by: alc, jhb
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
Long objdir paths make it easy to hit the version string length limit in
kernel dump headers. The build number and timestamp are unlikely to be
truncated and ought to be sufficient to protect against false positives.
bsdinstall: improve checksum mismatch error for snapshots
The usual case of a mismatched checksum for installer snapshots
(e.g., -CURRENT, -ALPHA*) is that a newer snapshot has been built and
the old install sets have been replaced. Provide a specific error
message for checksum mismatches there that suggests looking for a newer
snapshot.
Submitted by: Guangyuan Yang <yzgyyang@outlook.com>
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D11641
- Use "atf_check -x 'cmd1 | cmd2'" instead of "cmd1 | atf_check cmd2". The
two forms are idiomatically similar, but subtly different in the sense of
what program invokes the other, and there could be unwanted side effects
of the latter idiom dealing with forking, pipes, etc.
- Remove chmod and instead source coredump_phnum_restore_state.sh directly.
This avoids the need to check the result of the chmod call.
- Fix indentation in an if-block (4 column space indentation -> hard tab).
The testcase no longer fails on ^/head because readelf has established parity
with binutils' copy of readelf.
This issue is not seen on Jenkins because
`test_suites.FreeBSD.allow_sysctl_side_effects` isn't set in kyua.conf on
the CI host, i.e., the test is skipped.
ken [Wed, 19 Jul 2017 15:39:01 +0000 (15:39 +0000)]
Fix spurious timeouts on commands sent to mps(4) and mpr(4) controllers.
mps_wait_command() and mpr_wait_command() were using getmicrotime() to
determine elapsed time when checking for a timeout in polled mode.
getmicrotime() isn't guaranteed to monotonically increase, and that
caused spurious timeouts occasionally.
Switch to using getmicrouptime(), which does increase monotonically.
This fixes the spurious timeouts in my test case.
Rename vfs.nfsd.enable_uidtostring to vfs.nfs.enable_uidtostring.
It applies to both NFS client and NFS server, and is useful for both.
This is different from vfs.nfsd.enable_stringtouid, which is specific
to server side.
This was originally added so that only one of diskmbr.h or diskpc98.h
was chosen and is no longer needed after PC98's removal. However, the
special handling was also broken as it effectively prevented the decoding
of ioctls declared in other headers such as <sys/disk.h> or
<sys/disklabel.h>.
Convert assertion that only vmspace owner grows the stack, into a
check blocking grow from other processes accesses.
Debugger may access stack grow area with ptrace(2). In this case,
real state of the process is to not have the stack grown, which
provides more accurate inspection. Technical reason to avoid the grow
is to avoid applying wrong process (debugger) stack limit.
This change also has a consequence of making aio workers accesses past
the bottom of stacks into EFAULT, arguably the situation is a
programmers mistake.
Reported by: jhb
Discussed with: alc, jhb
Sponsored by: The FreeBSD Foundation
MFC after: 3 days
manu [Tue, 18 Jul 2017 19:50:02 +0000 (19:50 +0000)]
Allwinner EHCI: Do not fail if we cannot get a phy
If we cannot get a phy, do not detach the driver, some boards have phy
always enabled and not exposed.
While here do not release the clocks if we fails as we release them
in a10_ehci_detach.
sh: Remove broken #ifdef NOHACK code (related to sh -c).
Apart from the fact that subtle syntactic changes make a poor compile-time
option, the NOHACK case has been obviously broken since it was added,
because it uses q uninitialized if (*p != '\0').
I overlooked the fact that vdev_op_io_done hook is called even if the
actual I/O is skipped, for example, in the case of a missing vdev.
Arguably, this could be considered an issue in the zio pipeline engine,
but for now I am adding defensive code to check for io_bp being NULL
along with assertions that that happens only when it can be really
expected.
Propagate warning flags from kern.opts.mk and then fix minor -Werror
issues when building with gcc from -Wredundant-decls, -Wnested-externs,
-Wuninitialized.
ZPL_VERSION is unsigned long long, not an int. With this change, a zpool can be
created on a 32-bit system (tested on powerpcspe) and mounted correctly.