Revert r173708's modifications to vm_object_page_remove().
Assume that a vnode is mapped shared and mlocked(), and then the vnode
is truncated, or truncated and then again extended past the mapping
point EOF. Truncation removes the pages past the truncation point,
and if pages are later created at this range, they are not properly
mapped into the mlocked region, and their wiring count is wrong.
The revert leaves the invalidated but wired pages on the object queue,
which means that the pages are found by vm_object_unwire() when the
mapped range is munlock()ed, and reused by the buffer cache when the
vnode is extended again.
The changes in r173708 were required since then vm_map_unwire() looked
at the page tables to find the page to unwire. This is no longer
needed with the vm_object_unwire() introduction, which follows the
objects shadow chain.
Also eliminate OBJPR_NOTWIRED flag for vm_object_page_remove(), which
is now redundand, we do not remove wired pages.
Reported by: trasz, Dmitry Sivachenko <trtrmitya@gmail.com>
Suggested and reviewed by: alc
Tested by: pho
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
Move including netinet/icmp6.h around to avoid a problem when including
netinet/icmp6.h and net/netmap.h. Both use ni_flags...
This allows to build multistack with SCTP support.
With the removal of b_saveaddr in the r285819, b_data must be reset to
b_kvabase when the buffer is reclaimed. Otherwise, if b_data for the
mapped buffer was adjusted with the page-offset portion of b_offset,
nothing would re-adjust the b_data, which breaks buffer management
code which expects page-aligned b_data (see e.g. bpman_qenter(), which
skips partial pages).
Fix a minor issue with the GB_KVAALLOC requests, which could result in
returning the mapped buffer if the reused buffer is mapped and have
the right amount of KVA reserved.
Improve assertion in the vfs_buf_check_mapped() to catch unmapped
buffers which have their b_data incorrectly adjusted with offset.
Reported and tested by: pho (previous version)
Reviewed by: jeff (previous version)
Sponsored by: The FreeBSD Foundation
Synchronize PIN input/output modes with gnu/dts/include/dt-bindings/pinctrl/am33xx.h
gpio driver requires exact value to match SoC pin mode with GPIO pin direction
OF_getencprop_alloc shouldn't be used to get string value. If string
length + 1 is not divisible by 4 this function returns NULL property
value. Otherwise - string with each 4 letters inverted
Document the fact that system(3) can easily be misused due to shell meta
characters are honored. While I'm there also mention posix_spawn in the
SEE ALSO section.
Add a comment discussing the appropriate use of the atomic_*() functions
with acquire and release semantics versus the *mb() functions on amd64
processors.
ar: add -U (unique) option to disable -D (deterministic) mode
This is required in order for us to support deterministic mode by
default. If multiple -D or -U options are specified on the command
line, the final one takes precedence. GNU ar also uses -U for this.
An equivalent change will be applied to ELF Tool Chain's version of ar.
PR: 196929
MFC after: 1 month
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D3175
marius [Fri, 24 Jul 2015 17:01:16 +0000 (17:01 +0000)]
- Since r253161, uart_intr() abuses FILTER_SCHEDULE_THREAD for signaling
uart_bus_attach() during its test that 20 iterations weren't sufficient
for clearing all pending interrupts, assuming this means that hardware
is broken and doesn't deassert interrupts. However, under pressure, 20
iterations also can be insufficient for clearing all pending interrupts,
leading to a panic as intr_event_handle() tries to schedule an interrupt
handler not registered. Solve this by introducing a flag that is set in
test mode and otherwise restores pre-r253161 behavior of uart_intr(). The
approach of additionally registering uart_intr() as handler as suggested
in PR 194979 is not taken as that in turn would abuse special pccard and
pccbb handling code of intr_event_handle(). [1]
- Const'ify uart_driver_name.
- Fix some minor style bugs.
marius [Fri, 24 Jul 2015 16:00:35 +0000 (16:00 +0000)]
- In mpt_send_handshake_cmd(), use bus_space_write_stream_4(9) for writing
raw data to the doorbell offset in order to clarify the intent and for
avoiding unnecessarily converting the endianess back and forth.
Unfortunately, the same can't be done in mpt_recv_handshake_reply() as
16-bit data needs to be read using 32-bit bus accessors.
- In mpt_recv_handshake_reply(), get rid of a redundant variable.
marius [Fri, 24 Jul 2015 15:13:21 +0000 (15:13 +0000)]
o Revert the other functional half of r239864, i. e. the merge of r134227
from x86 to use smp_ipi_mtx spin lock not only for smp_rendezvous_cpus()
but also for the MD cache invalidation, TLB demapping and remote register
reading IPIs due to the following reasons:
- The cross-IPI SMP deadlock x86 otherwise is subject to can't happen on
sparc64. That's because on sparc64, spin locks don't disable interrupts
completely but only raise the processor interrupt level to PIL_TICK. This
means that IPIs still get delivered and direct dispatch IPIs such as the
cache invalidation etc. IPIs in question are still executed.
- In smp_rendezvous_cpus(), smp_ipi_mtx is held not only while sending an
IPI_RENDEZVOUS, but until all CPUs have processed smp_rendezvous_action().
Consequently, smp_ipi_mtx may be locked for an extended amount of time as
queued IPIs (as opposed to the direct ones) such as IPI_RENDEZVOUS are
scheduled via a soft interrupt. Moreover, given that this soft interrupt
is only delivered at PIL_RENDEZVOUS, processing of smp_rendezvous_action()
on a target may be interrupted by f. e. a tick interrupt at PIL_TICK, in
turn leading to the target in question trying to send an IPI by itself
while IPI_RENDEZVOUS isn't fully handled, yet, and, thus, resulting in a
deadlock.
o As mentioned in the commit message of r245850, on least some sun4u platforms
concurrent sending of IPIs by different CPUs is fatal. Therefore, hold the
reintroduced MD ipi_mtx also while delivering cross-traps via MI helpers,
i. e. ipi_{all_but_self,cpu,selected}().
o Akin to x86, let the last CPU to process cpu_mp_bootstrap() set smp_started
instead of the BSP in cpu_mp_unleash(). This ensures that all APs actually
are started, when smp_started is no longer 0.
o In all MD and MI IPI helpers, check for smp_started == 1 rather than for
smp_cpus > 1 or nothing at all. This avoids races during boot causing IPIs
trying to be delivered to APs that in fact aren't up and running, yet.
While at it, move setting of the cpu_ipi_{selected,single}() pointers to
the appropriate delivery functions from mp_init() to cpu_mp_start() where
it's better suited and allows to get rid of the global isjbus variable.
o Given that now concurrent IPI delivery no longer is possible, also nuke
the delays before completely disabling interrupts again in the CPU-specific
cross-trap delivery functions, previously giving other CPUs a window for
sending IPIs on their part. Actually, we now should be able to entirely get
rid of completely disabling interrupts in these functions. Such a change
needs more testing, though.
o In {s,}tick_get_timecount_mp(), make the {s,}tick variable static. While not
necessary for correctness, this avoids page faults when accessing the stack
of a foreign CPU as {s,}tick now is locked into the TLBs as part of static
kernel data. Hence, {s,}tick_get_timecount_mp() always execute as fast as
possible, avoiding jitter.
Fix an issue with MAC OS locking and also optimize the case
where we are sending back a stream-reset and a sack timer is running, in
that case we should just send the SACK.
If apropos(1) and whatis(1) are not hardlinks to man(1) that means the system is
using mandocdb, then man -k should spawn apropos(1) and/or whatis(1) directly
ed [Fri, 24 Jul 2015 07:46:02 +0000 (07:46 +0000)]
Implement the basic system calls that operate on pathnames.
Summary:
Unlike FreeBSD, CloudABI does not use null terminated strings for its
pathnames. Introduce a function called copyin_path() that can be used by
all of the filesystem system calls that use pathnames. This change
already implements the system calls that don't depend on any additional
functionality (e.g., conversion of struct stat).
Also implement the socket system calls that operate on pathnames, namely
the ones used by the C library functions bindat() and connectat(). These
don't receive a 'struct sockaddr_un', but just the pathname, meaning
they could be implemented in such a way that they don't depend on the
size of sun_path. For now, just use the existing interfaces.
Add a missing #include to cloudabi_syscalldefs.h to get this code to
build, as one of its macros depends on UINT64_C().
Test Plan:
These implementations have already been tested in the CloudABI branch on
GitHub. They pass all of the tests.
jeff [Thu, 23 Jul 2015 19:13:41 +0000 (19:13 +0000)]
Refactor unmapped buffer address handling.
- Use pointer assignment rather than a combination of pointers and
flags to switch buffers between unmapped and mapped. This eliminates
multiple flags and generally simplifies the logic.
- Eliminate b_saveaddr since it is only used with pager bufs which have
their b_data re-initialized on each allocation.
- Gather up some convenience routines in the buffer cache for
manipulating buf space and buf malloc space.
- Add an inline, buf_mapped(), to standardize checks around unmapped
buffers.
ed [Thu, 23 Jul 2015 11:11:01 +0000 (11:11 +0000)]
Allow cap_rights_{set,clear,is_set} to be called with no arguments.
In the CloudABI code I sometimes call into cap_rights_* without
providing any arguments. Though one could argue that this doesn't make
sense, in this specific case it's hard to avoid, as the rights that
should be tested against are forwarded by a couple of wrapper macros.
jeff [Thu, 23 Jul 2015 02:20:41 +0000 (02:20 +0000)]
- Don't defeat the FIFO nature of the buffer cache by eliminating the
most recently used buffer when we are under paging pressure. This is
a perversion of the buffer and page replacement algorithms and recent
improvements to the page daemon have rendered it unnecessary. In the
event that low-memory deadlocks become an issue it would be possible
to make a daemon or event handler that performs a similar action on
the oldest buffers rather than the newest. Since the buf cache
is analogous to the page cache and some minimum working set is desired
another possibility is to simply shrink the minimum working set which
has less downside now that file pages are not directly mapped.
Sponsored by: EMC / Isilon
Reviewed by: alc, kib (with some minor objection)
Tested by: pho
sfxge: added fallbacks for pre 4.2.1 firmware support
Driver must be able to start against older firmware that is missing
recently added MCDI calls, otherwise firmware upgrade will not be
possible.
Submitted by: Richard Houldsworth <rhouldsworth at solarflare.com>
Sponsored by: Solarflare Communications, Inc.
MFC after: 2 days
Differential Revision: https://reviews.freebsd.org/D3145
Fix several problems with Stream Reset.
1) We were not handling (or sending) the IN_PROGRESS case if
the other side (or our side) was not able to reset (awaiting more data).
2) We would improperly send a stream-reset when we should not. Not
waiting until the TSN had been assigned when data was inqueue.
ed [Wed, 22 Jul 2015 07:32:49 +0000 (07:32 +0000)]
Add Makefiles for CloudABI kernel modules.
Place all of the machine/pointer size independent code in a kernel
module called 'cloudabi'. All of the 64-bit specific code goes in a
separate module called 'cloudabi64'. The latter is only enabled on
amd64, as it is the only architecture supported.
Various changes to the registers displayed in DDB for x86.
- Fix segment registers to only display the low 16 bits.
- Remove unused handlers and entries for the debug registers.
- Display xcr0 (if valid) in 'show sysregs'.
- Add '0x' prefix to MSR values to match other values in 'show sysregs'.
- MFamd64: Display various MSRs in 'show sysregs'.
- Add a 'show dbregs' to display the value of debug registers.
- Dynamically size the column width for register values to properly
align columns on 64-bit platforms.
- Display %gs for i386 in 'show registers'.
Fix counter reads on platforms where sizeof(uint64_t) != sizeof(uint64_t *).
In the kernel, structs such as tcpstat are manipulated as an array of
counter_u64_t (uint64_t *), but made visible to userland as an array of
uint64_t. kread_counters() was previously copying the counter array into
user space and sequentially overwriting each counter with its value. This
mostly affects IPsec counters, as other counters are exported via sysctl.
PR: 201700
Tested by: Jason Unovitch
MFC after: 1 week
Let the unwinder handle faults during function prologues or epilogues.
The i386 and amd64 DDB stack unwinders contain code to detect and handle
the case where the first frame is not completely set up or torn down. This
code was accidentally unused however, since db_backtrace() was never called
with a non-NULL trap frame. This change fixes that.
Also remove get_rsp() from the amd64 code. It appears to have come from
i386, which needs to take into account whether the exception triggered a
CPL switch, since SS:ESP is only pushed onto the stack if so. On amd64,
SS:RSP is pushed regardless, so get_rsp() was doing the wrong thing for
kernel-mode exceptions. As a result, we can also remove custom print
functions for these registers.
Improve stack unwinding on i386 and amd64 after an IP fault.
If we can't find a symbol corresponding to the faulting instruction, assume
that the previously-executed function is a call and attempt to find the
calling function using the return address on the stack. Otherwise we end
up associating the last stack frame with the current call, which is
incorrect and causes the unwinder to skip printing of the calling function,
resulting in a confusing backtrace.
Don't return undefined symbols to a DDB symbol lookup.
Undefined symbols have a value of zero, so it makes no sense to return
such a symbol when performing a lookup by value. This occurs for example
when unwinding the stack after calling a NULL function pointer, and we
confusingly report the faulting function as uart_sab82532_class() on
amd64.
Convert db_print_loc_and_inst() to only attempt disassembly if we managed
to find a symbol corresponding to the IP. Otherwise we may fault and
re-enter the debugger.
Remove some dead code from DDB's amd64 stack unwinder.
The amd64 port copied some code from i386 to fetch function arguments and
display them in backtraces. However, it was commented out and can't easily
be implemented since the function arguments are passed in
registers rather than on the stack in amd64. Remove it in preparation for
some bug fixes in this area.
The smp_rendezvous_cpus() function should ensure that all accesses
done by the functions called on other CPUs, are visible to the caller.
Pair otherwise useless acquire on smp_rv_waiters[3] with a release add
to ensure synchronized with relation, which guarantees visibility.
Reviewed by: alc
Sponsored by: The FreeBSD Foundation
MFC after: 3 weeks
- Add required MAC/VLAN filter when adding an LAA
- Fix bug where code did not check for I40E_SUCCESS from a successful
i40e_validate_mac_address() call in ixl_init_locked(), when setting
an LAA.
nvd: set d_delmaxsize to full capacity of NVMe namespace
The NVMe specification has no ability to specify a maximum delete size
that is less than the full capacity of the namespace - so just using the
namespace size is the correct value here.
This fixes reported issues where ZFS trim on init looked like it was
hanging the system - previously the default I/O max size (128KB on
Intel NVMe controllers) was used for delete operations which worked out
to only about 8MB/s. With this patch I can add an 800GB DC P3700
drive to a ZFS pool in about 15-20 seconds.
Reported by: Dylan Just <dylan@techtangents.com>
MFC after: 3 days
Sponsored by: Intel
cem [Tue, 21 Jul 2015 20:33:36 +0000 (20:33 +0000)]
vt: Draw logos per CPU core
This feature is inspired by another Unix-alike OS commonly found on
airplane headrests.
A number of beasties[0] are drawn at top of framebuffer during boot,
based on the number of active SMP CPUs[1]. Console buffer output
continues to scroll in the screen area below beastie(s)[2].
After some time[3] has passed, the beasties are erased leaving the
entire terminal for use.
Includes two 80x80 vga16 beastie graphics and an 80x80 vga16 orb
graphic. (The graphics are RLE compressed to save some space -- 3x 3200
bytes uncompressed, or 4208 compressed.)
[0]: The user may select the style of beastie with
kern.vt.splash_cpu_style=(0|1|2)
[1]: Or the number may be overridden with tunable kern.vt.splash_ncpu.
[2]: https://www.youtube.com/watch?v=UP2jizfr3_o
[3]: Configurable with kern.vt.splash_cpu_duration (seconds, def. 10).
Don't allow malloc() to wait for resource while holding a lock in ITS
malloc() should not go to sleep in case of lack of resource while
the kernel thread is holding a non-sleepable lock.
- change malloc() flags to M_NOWAIT in such cases implement
lpi_free_chunk() routine as it will be needed when ITT
allocation fails in its_device_alloc_locked()
- do not increase verbosity of this code since upper layers will
communicate an error if the interrupt setup fails
Obtained from: Semihalf
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D3121
ed [Tue, 21 Jul 2015 15:08:13 +0000 (15:08 +0000)]
Make clock_gettime() and clock_getres() work for CloudABI programs.
Though the standard C library uses a 'struct timespec' using a 64-bit
'time_t', there is no need to use such a type at the system call level.
CloudABI uses a simple 64-bit unsigned timestamp in nanoseconds. This is
sufficient to express any time value from 1970 to 2554.
The CloudABI low-level interface also supports fetching timestamp values
with a lower precision. Instead of overloading the clock ID argument for
this purpose, the system call provides a precision argument that may be
used to specify the maximum slack. The current system call
implementation does not use this information, but it's good to already
have this available.
Expose cloudabi_convert_timespec(), as we're going to need this for
fstat() as well.
Add support for vendor specific function for PCI devid acquisition in ITS
It is possible that some HW will use different PCI devids,
hence allow to replace the default domain:bus:slot:func schema
by implementing and registering custom function.
Obtained from: Semihalf
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D3118
The part of r285680 which removed release semantic for two stores to
it_need was wrong [*]. Restore the releases and add a comment
explaining why it is needed.
Noted by: alc [*]
Reviewed by: bde [*]
Sponsored by: The FreeBSD Foundation
ed [Tue, 21 Jul 2015 12:47:15 +0000 (12:47 +0000)]
Make thread creation work for CloudABI processes.
Summary:
Remove the stub system call that was put in place during the system call
import and replace it by a target-dependent version stored in sys/amd64.
Initialize the thread in a way similar to cpu_set_upcall_kse(). We
provide the entry point with two arguments: the thread ID and the
argument pointer.
Test Plan:
Thread creation still seems to work, both for FreeBSD and CloudABI
binaries.
ed [Tue, 21 Jul 2015 10:52:05 +0000 (10:52 +0000)]
Unbreak "last reboot".
According to the last(1) man page, the "reboot" pseudo-user should print
all system reboot entries. This got broken by the utmpx import, as
records are typed.
Re-add support for "last reboot" by specifically matching against
SHUTDOWN_TIME and BOOT_TIME records.
When a tunneling protocol is being used with UDP we must release the
lock on the INP before calling the tunnel protocol, else a LOR
may occur (it does with SCTP for sure). Instead we must acquire a
ref count and release the lock, taking care to allow for the case
where the UDP socket has gone away and *not* unlocking since the
refcnt decrement on the inp will do the unlock in that case.
hiren [Tue, 21 Jul 2015 06:48:36 +0000 (06:48 +0000)]
Remove a couple of TUNABLE_INT() calls which are unnecessary after r267961.
r267961 did remove them but they "reappeared" when ixgbe(4) rewrite happened in
r280182.
lseek() allows an offset to be set beyond the end of file. Using
it to check that partition has enough space to write bootcode doesn't
work. Use the known size of provider instead.
Fix some issues with the application of libxo to ls(1)
* Add whitespace trimming to some fields (username, group, size, inode, blocks) to avoid whitespace in JSON strings
* fix -m mode, was invalid JSON (repeated keys), and was missing outer array container
* in -n mode, numeric uids and gids were returned as strings
Fix an out-of-order execution issue regarding pkg(8):
- pkg(8) cannot be removed before subsequent reinvocations
- The PKG_CACHEDIR cannot be cleaned after the repo*.sqlite
has been removed
- pkg(8) cannot be removed as a precursor to any of the other
steps involved here
MFC after: 3 days
X-MFC-With: r285722
X-MFC-Before: 10.2-{BETA3,RC1} (whichever happens next)
Sponsored by: The FreeBSD Foundation