reclamation synchronously from get_pv_entry() instead of
asynchronously as part of the page daemon. Additionally, limit the
reclamation to inactive pages unless allocation from the PV entry zone
or reclamation from the inactive queue fails. Previously, reclamation
destroyed mappings to both inactive and active pages. get_pv_entry()
still, however, wakes up the page daemon when reclamation occurs. The
reason being that the page daemon may move some pages from the active
queue to the inactive queue, making some new pages available to future
reclamations.
Print the "reclaiming PV entries" message at most once per minute, but
don't stop printing it after the fifth time. This way, we do not give
the impression that the problem has gone away.
Reviewed by: tegge
Previously, pvzone's initialization was split between pmap_init() and
pmap_init2(). This split initialization was the underlying cause of
some UMA panics during initialization. Specifically, if the UMA boot
pages was exhausted before the pvzone was fully initialized, then UMA,
through no fault of its own, would use an inappropriate back-end
allocator leading to a panic. (Previously, as a workaround, we have
increased the UMA boot pages.) Fortunately, there is no longer any
reason that pvzone's initialization cannot be completed in
pmap_init().
Eliminate a check for whether pv_entry_high_water has been initialized
or not from get_pv_entry(). Since pvzone's initialization is
completed in pmap_init(), this check is no longer needed.
Use cnt.v_page_count, the actual count of available physical pages,
instead of vm_page_array_size to compute the maximum number of pv
entries.
Introduce the vm.pmap.pv_entries tunable on alpha and ia64.
Eliminate some unnecessary white space.
Discussed with: tegge (item #1)
Tested by: marcel (ia64)
- Prefer '_' to ' ', as it results in more easily parsed results in
memory monitoring tools such as vmstat.
- Remove punctuation that is incompatible with using memory type names
as file names, such as '/' characters.
- Disambiguate some collisions by adding subsystem prefixes to some
memory types.
- Generally prefer lower case to upper case.
- If the same type is defined in multiple architecture directories,
attempt to use the same name in additional cases.
Not all instances were caught in this change, so more work is required to
finish this conversion. Similar changes are required for UMA zone names.
that caused a premature exit after calling a fast interrupt handler
and bypassing a much needed critical_exit() and the scheduling of
the interrupt thread for non-fast handlers. In short: unbreak :-)
and increase flexibility to allow various different approaches to be tried
in the future.
- Split struct ithd up into two pieces. struct intr_event holds the list
of interrupt handlers associated with interrupt sources.
struct intr_thread contains the data relative to an interrupt thread.
Currently we still provide a 1:1 relationship of events to threads
with the exception that events only have an associated thread if there
is at least one threaded interrupt handler attached to the event. This
means that on x86 we no longer have 4 bazillion interrupt threads with
no handlers. It also means that interrupt events with only INTR_FAST
handlers no longer have an associated thread either.
- Renamed struct intrhand to struct intr_handler to follow the struct
intr_foo naming convention. This did require renaming the powerpc
MD struct intr_handler to struct ppc_intr_handler.
- INTR_FAST no longer implies INTR_EXCL on all architectures except for
powerpc. This means that multiple INTR_FAST handlers can attach to the
same interrupt and that INTR_FAST and non-INTR_FAST handlers can attach
to the same interrupt. Sharing INTR_FAST handlers may not always be
desirable, but having sio(4) and uhci(4) fight over an IRQ isn't fun
either. Drivers can always still use INTR_EXCL to ask for an interrupt
exclusively. The way this sharing works is that when an interrupt
comes in, all the INTR_FAST handlers are executed first, and if any
threaded handlers exist, the interrupt thread is scheduled afterwards.
This type of layout also makes it possible to investigate using interrupt
filters ala OS X where the filter determines whether or not its companion
threaded handler should run.
- Aside from the INTR_FAST changes above, the impact on MD interrupt code
is mostly just 's/ithread/intr_event/'.
- A new MI ddb command 'show intrs' walks the list of interrupt events
dumping their state. It also has a '/v' verbose switch which dumps
info about all of the handlers attached to each event.
- We currently don't destroy an interrupt thread when the last threaded
handler is removed because it would suck for things like ppbus(8)'s
braindead behavior. The code is present, though, it is just under
#if 0 for now.
- Move the code to actually execute the threaded handlers for an interrrupt
event into a separate function so that ithread_loop() becomes more
readable. Previously this code was all in the middle of ithread_loop()
and indented halfway across the screen.
- Made struct intr_thread private to kern_intr.c and replaced td_ithd
with a thread private flag TDP_ITHREAD.
- In statclock, check curthread against idlethread directly rather than
curthread's proc against idlethread's proc. (Not really related to intr
changes)
Tested on: alpha, amd64, i386, sparc64
Tested on: arm, ia64 (older version of patch by cognet and marcel)
get a new pv under high system load where the available pv entries
have been exhausted before the pagedaemon has a chance to wake up
to reclaim some.
Prior to this, the NULL pointer dereference ended up causing
secondary panics with rather less than useful resulting tracebacks.
Reviewed by: alc, jhb
MFC after: 1 week
changes in MD code are trivial, before this change, trapsignal and
sendsig use discrete parameters, now they uses member fields of
ksiginfo_t structure. For sendsig, this change allows us to pass
POSIX realtime signal value to user code.
2. Remove cpu_thread_siginfo, it is no longer needed because we now always
generate ksiginfo_t data and feed it to libpthread.
3. Add p_sigqueue to proc structure to hold shared signals which were
blocked by all threads in the proc.
4. Add td_sigqueue to thread structure to hold all signals delivered to
thread.
5. i386 and amd64 now return POSIX standard si_code, other arches will
be fixed.
6. In this sigqueue implementation, pending signal set is kept as before,
an extra siginfo list holds additional siginfo_t data for signals.
kernel code uses psignal() still behavior as before, it won't be failed
even under memory pressure, only exception is when deleting a signal,
we should call sigqueue_delete to remove signal from sigqueue but
not SIGDELSET. Current there is no kernel code will deliver a signal
with additional data, so kernel should be as stable as before,
a ksiginfo can carry more information, for example, allow signal to
be delivered but throw away siginfo data if memory is not enough.
SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can
not be caught or masked.
The sigqueue() syscall allows user code to queue a signal to target
process, if resource is unavailable, EAGAIN will be returned as
specification said.
Just before thread exits, signal queue memory will be freed by
sigqueue_flush.
Current, all signals are allowed to be queued, not only realtime signals.
Earlier patch reviewed by: jhb, deischen
Tested on: i386, amd64
osf1_signal.c:1.41, amd64/amd64/trap.c:1.291, linux_socket.c:1.60,
svr4_fcntl.c:1.36, svr4_ioctl.c:1.23, svr4_ipc.c:1.18, svr4_misc.c:1.81,
svr4_signal.c:1.34, svr4_stat.c:1.21, svr4_stream.c:1.55,
svr4_termios.c:1.13, svr4_ttold.c:1.15, svr4_util.h:1.10,
ext2_alloc.c:1.43, i386/i386/trap.c:1.279, vm86.c:1.58,
unaligned.c:1.12, imgact_elf.c:1.164, ffs_alloc.c:1.133:
Now that Giant is acquired in uprintf() and tprintf(), the caller no
longer leads to acquire Giant unless it also holds another mutex that
would generate a lock order reversal when calling into these functions.
Specifically not backed out is the acquisition of Giant in nfs_socket.c
and rpcclnt.c, where local mutexes are held and would otherwise violate
the lock order with Giant.
This aligns this code more with the eventual locking of ttys.
Suggested by: bde
variable and returns the previous value of the variable.
Tested on: i386, alpha, sparc64, arm (cognet)
Reviewed by: arch@
Submitted by: cognet (arm)
MFC after: 1 week
as they both interact with the tty code (!MPSAFE) and may sleep if the
tty buffer is full (per comment).
Modify all consumers of uprintf() and tprintf() to hold Giant around
calls into these functions. In most cases, this means adding an
acquisition of Giant immediately around the function. In some cases
(nfs_timer()), it means acquiring Giant higher up in the callout.
With these changes, UFS no longer panics on SMP when either blocks are
exhausted or inodes are exhausted under load due to races in the tty
code when running without Giant.
NB: Some reduction in calls to uprintf() in the svr4 code is probably
desirable.
NB: In the case of nfs_timer(), calling uprintf() while holding a mutex,
or even in a callout at all, is a bad idea, and will generate warnings
and potential upset. This needs to be fixed, but was a problem before
this change.
NB: uprintf()/tprintf() sleeping is generally a bad ideas, as is having
non-MPSAFE tty code.
MFC after: 1 week
This kernel config briefly describes some of the major MAC policies
available on FreeBSD. The hope is that this will raise the awareness
about MAC and get more people interested.
Discussed with: scottl
take the lock from interrupt context, which causes an implicit
lock order reversal. We've been using the lock carefully enough
that making it a spin lock should not be harmful.
was not invalidated if the PTE was not actually being removed. In
an UP kernel this didn't cause problems, because the new mapping
would preempt the old one. In an SMP kernel this could lead to the
use of stale translations when processes move between CPUs at the
"right" moment. This fixes the last of the obvious SMP problems
and it should be safe to enable SMP by default now.
o In pmap_remove_pte: minor code refactoring to avoid duplication.
o Test all PTE pointers against NULL. Don't use implicit boolean
tests.
o s/vhpt_base/pmap_vhpt_base/g
o s/vhpt_bucket/pmap_vhpt_bucket/g
o Declare the above in <machine/pmap.h>
o Move the vm.stats.vhpt.* sysctls to machdep.vhpt.*
o Create a tunable machdep.vhpt.log2size, with corresponding sysctl.
The tunable allows the user to specify the VHPT size from the loader.
o Don't keep track of the number of PTEs in the VHPT. Calculate the
population when necessary by iterating the buckets and summing up
the length of the buckets.
o Don't perform the tpa instruction with a bucket lock held. The
instruction can (theoretically) fault and locking is not needed.
is never 0, so one cannot test for a NULL pointer after a physical
address is translated into a virtual pointer with said macro. Instead,
keep the physical address around and test it against 0. Note that
this obviously implies that a PTE can never be allocated at physical
address 0. This isn't exactly guaranteed, but hasn't been a problem
so far. We test the physical address against 0 for as long as the ia64
port exists...
it to __MINSIGSTKSZ. Define MINSIGSTKSZ in <sys/signal.h>.
This is done in order to use MINSIGSTKSZ for the macro PTHREAD_STACK_MIN
in <pthread.h> (soon <limits.h>) without having to include the whole
<sys/signal.h> header.
Discussed with: bde
high FP registers. It was not that the IPI got lost due to the
perceived unreliability of the IPI delivery, but rather that the
IPI was not assigned a vector (ugh). Sending a 0 vector to a CPU
results in a stray external interrupt.
Add a KASSERT to ipi_send() to catch this. The initialization of
the IPIs could be better, but it's not at all sure what the future
of the code is. Avoid wasting a lot of time on something that is
going to be rewritten anyway.
o Allocate a VHPT per CPU. The VHPT is a hash table that the CPU
uses to look up translations it can't find in the TLB. As such,
the VHPT serves as a level 1 cache (the TLB being a level 0 cache)
and best results are obtained when it's not shared between CPUs.
The collision chain (i.e. the hash bucket) is shared between CPUs,
as all buckets together constitute our collection of PTEs. To
achieve this, the collision chain does not point to the first PTE
in the list anymore, but to a hash bucket head structure. The
head structure contains the pointer to the first PTE in the list,
as well as a mutex to lock the bucket. Thus, each bucket is locked
independently of each other. With at least 1024 buckets in the VHPT,
this provides for sufficiently finei-grained locking to make the
ssolution scalable to large SMP machines.
o Add synchronisation to the lazy FP context switching. We do this
with a seperate per-thread lock. On SMP machines the lazy high FP
context switching without synchronisation caused inconsistent
state, which resulted in a panic. Since the use of the high FP
registers is not common, it's possible that races exist. The ia64
package build has proven to be a good stress test, so this will
get plenty of exercise in the near future.
o Don't use the local ID of the processor we want to send the IPI to
as the argument to ipi_send(). use the struct pcpu pointer instead.
The reason for this is that IPI delivery is unreliable. It has been
observed that sending an IPI to a CPU causes it to receive a stray
external interrupt. As such, we need a way to make the delivery
reliable. The intended solution is to queue requests in the target
CPU's per-CPU structure and use a single IPI to inform the CPU that
there's a new entry in the queue. If that IPI gets lost, the CPU
can check it's queue at any convenient time (such as for each
clock interrupt). This also allows us to send requests to a CPU
without interrupting it, if such would be beneficial.
With these changes SMP is almost working. There are still some random
process crashes and the machine can hang due to having the IPI lost
that deals with the high FP context switch.
The overhead of introducing the hash bucket head structure results
in a performance degradation of about 1% for UP (extra pointer
indirection). This is surprisingly small and is offset by gaining
reasonably/good scalable SMP support.
allocating a VHPT per CPU. Since we don't yet know how many CPUs
are actually in the system at the time we need to allocate the
VHPTs, we allocate for MAXCPU processors. This can result in a
lot of wasted space for 2-way machines. So, for now, limit MAXCPU
to something smaller until we have something more dynamic.
variables rather than void * variables. This makes it easier and simpler
to get asm constraints and volatile keywords correct.
MFC after: 3 days
Tested on: i386, alpha, sparc64
Compiled on: ia64, powerpc, amd64
Kernel toolchain busted on: arm
address, writting non-canonical address can cause kernel a panic,
by restricting base values to 0..VM_MAXUSER_ADDRESS, ensuring
only canonical values get written to the registers.
Reviewed by: peter, Josepha Koshy < joseph.koshy at gmail dot com >
Approved by: re (scottl)
This case is triggered with ptrace(2) and the PT_SETREGS function.
Change the return type of the function to int so that errors can be
passed on to the caller.
Approved by: re (scottl)
pointer doesn't point to the first instruction of that function, but
rather to a descriptor. The descriptor has the address of the first
instruction, as well as the value of the global pointer. The symbol
table doesn't know anything about descriptors, so if you lookup the
name of a function you get the address of the first instruction. The
cast from the address, which is the result of the symbol lookup, to a
function pointer as is done in db_fncall is therefore invalid.
Abstract this detail behind the DB_CALL macro. By default DB_CALL is
defined as db_fncall_generic, which yields the old behaviour. On ia64
the macro is defined as db_fncall_ia64, in which a descriptor is
constructed to yield a valid function pointer.
While here, introduce DB_MAXARGS. DB_MAXARGS replaces the existing
(local) MAXARGS. The DB_MAXARGS macro can be defined by platforms to
create a convenient maximum. By default this will be the legacy 10.
On ia64 we define this macro to be 8, for 8 is the maximum number of
arguments that can be passed in registers. This avoids having to
implement spilling of arguments on the memory stack.
Approved by: re (dwhite)
by amd64 and i386: For buffered writes we collect data and write it
out a ${DEV_BSIZE}-sized block at a time. The fragsz variable is used
to keep track of how much data we have collected in the buffer so far
and it's reset to zero immediately after writing a block to the dump
device.
When the last, possibly partially filled buffer is flushed, we didn't
reset fragsz to 0 and as such would stop reflecting reality. Since we
currently only need to do buffered writes once, this isn't a problem.
However, when kernel dumps are made by hand (say by callling doadump
from within DDB), the improperly cleared state from the first call to
dumpsys causes the next call to dumpsys to create an invalid code file.
This change resets fragsz after flushing the partially filled buffer so
that it fixes the two problems at once.
Approved by: re (scottl)
This is good enough to be able to run a RELENG_4 gdb binary against
a RELENG_4 application, along with various other tools (eg: 4.x gcore).
We use this at work.
ia32_reg.[ch]: handle the 32 bit register file format, used by ptrace,
procfs and core dumps.
procfs_*regs.c: vary the format of proc/XXX/*regs depending on the client
and target application.
procfs_map.c: Don't print a 64 bit value to 32 bit consumers, or their
sscanf fails. They expect an unsigned long.
imgact_elf.c: produce a valid 32 bit coredump for 32 bit apps.
sys_process.c: handle 32 bit consumers debugging 32 bit targets. Note
that 64 bit consumers can still debug 32 bit targets.
IA64 has got stubs for ia32_reg.c.
Known limitations: a 5.x/6.x gdb uses get/setcontext(), which isn't
implemented in the 32/64 wrapper yet. We also make a tiny patch to
gdb pacify it over conflicting formats of ld-elf.so.1.
Approved by: re
immediate is not saved by the architecture. Any of the break.{mifx}
instructions have their immediate saved in cr.iim on interruption.
Consequently, when we handle the break interrupt, we end up with a
break value of 0 when it was a break.b. The immediate is important
because it distinguishes between different uses of the break and
which are defined by the runtime specification.
The bottomline is that when the GNU debugger replaces a B-unit
instruction with a break instruction in the inferior, we would not
send the process a SIGTRAP when we encounter it, because the value
is not one we recognize as a debugger breakpoint.
This change adds logic to decode the bundle in which the break
instruction lives whenever the break value is 0. The assumption
being that it's a break.b and we fetch the immediate directly out
of the instruction. If the break instruction was not a break.b,
but any of break.{mifx} with an immediate of 0, we would be doing
unnecessary work. But since a break 0 is invalid, this is not a
problem and it will still result in a SIGILL being sent to the
process.
Approved by: re (scottl)
vm_page's machine-dependent fields. Use this function in
vm_pageq_add_new_page() so that the vm_page's machine-dependent and
machine-independent fields are initialized at the same time.
Remove code from pmap_init() for initializing the vm_page's
machine-dependent fields.
Remove stale comments from pmap_init().
Eliminate the Boolean variable pmap_initialized from the alpha, amd64,
i386, and ia64 pmap implementations. Its use is no longer required
because of the above changes and earlier changes that result in physical
memory that is being mapped at initialization time being mapped without
pv entries.
Tested by: cognet, kensmith, marcel
- Implement sampling modes and logging support in hwpmc(4).
- Separate MI and MD parts of hwpmc(4) and allow sharing of
PMC implementations across different architectures.
Add support for P4 (EMT64) style PMCs to the amd64 code.
- New pmcstat(8) options: -E (exit time counts) -W (counts
every context switch), -R (print log file).
- pmc(3) API changes, improve our ability to keep ABI compatibility
in the future. Add more 'alias' names for commonly used events.
- bug fixes & documentation.
