o Disable ofw_console(4), sab(4) and zs(4).
sab(4) and zs(4) are disabled because the hardware controlled by
them is handled by uart(4)+puc(4) and the latter combination is
functionally complete and up to date.
ofw_console(4) is disabled because it doesn't claim the device it
controls (through OFW) and thus interferes with puc(4)+uart(4),
which has sufficient knowledge to extract the necessary information
from OFW to setup the console. Put differently, ofw_console(4) is
not a proper device driver and can only do harm. Its functionality
is completely handled by uart(4).
This commit makes uart(4) the default driver for serial ports.
MFC after: 2 weeks
bridge in the device tree which lacks the mandatory (also by the OFW PCI
bus binding spec) "reg" property. Change the code to just ignore nodes
missing the "reg" property instead of panicing when encountering such a
node. Also ignore nodes without a "name" property (guaranteed by the OFW
PCI bus binding spec). This brings the behaviour of the MD OFW PCI code
regarding such incomplete nodes in line with the EBus and the SBus code.
Tested by: Cyril Tikhomiroff <tikho@anor.net>
MFC after: 1 month
now use a pool mutex to manage the reference counts. This fixes races
resulting in use-after-free.
Tested by: kris, David Cornejo dave at dogwood dot com
Reported by: bmilekic's MemGuard
MFC after: 1 week
won't exist for EBus. Just fail the allocation by returning NULL.
Now drivers that are MI can try resources that the driver knows may
be used by the device.
on entry and it assumes the responsibility for releasing the page queues
lock if it must sleep.
Remove a bogus comment from pmap_enter_quick().
Using the first change, modify vm_map_pmap_enter() so that the page queues
lock is acquired and released once, rather than each time that a page
is mapped.
In such cases, the busying of the page and the unlocking of the
containing object by vm_map_pmap_enter() and vm_fault_prefault() is
unnecessary overhead. To eliminate this overhead, this change
modifies pmap_enter_quick() so that it expects the object to be locked
on entry and it assumes the responsibility for busying the page and
unlocking the object if it must sleep. Note: alpha, amd64, i386 and
ia64 are the only implementations optimized by this change; arm,
powerpc, and sparc64 still conservatively busy the page and unlock the
object within every pmap_enter_quick() call.
Additionally, this change is the first case where we synchronize
access to the page's PG_BUSY flag and busy field using the containing
object's lock rather than the global page queues lock. (Modifications
to the page's PG_BUSY flag and busy field have asserted both locks for
several weeks, enabling an incremental transition.)
These devices should be probed first because they are at fixed
locations and cannot be turned off. ISA PNP devices, on the other
hand, can be turned off and often can be flexible in the resources
they use. Probe them last, as always.
specified register, but a pointer to the in-memory representation of
that value. The reason for this is twofold:
1. Not all registers can be represented by a register_t. In particular
FP registers fall in that category. Passing the new register value
by reference instead of by value makes this point moot.
2. When we receive a G or P packet, both are for writing a register,
the packet will have the register value in target-byte order and
in the memory representation (modulo the fact that bytes are sent
as 2 printable hexadecimal numbers of course). We only need to
decode the packet to have a pointer to the register value.
This change fixes the bug of extracting the register value of the P
packet as a hexadecimal number instead of as a bit array. The quick
(and dirty) fix to bswap the register value in gdb_cpu_setreg() as
it has been added on i386 and amd64 can therefore be removed and has
in fact been that.
Tested on: alpha, amd64, i386, ia64, sparc64
clock found on the ISA bus (some USIIe, USIIi and USIIIi models) and
EBus (USIII models) instead of a MK48Txx clock.
Testet by: Matthew T. Lager" <freebsd@trinetworks.com> on Sun Fire V100,
Xavier Beaudouin <kiwi@oav.net> on Netra X1 (initial version)
- The claim in the commit log of rev. 1.11 of dev/uart/uart_cpu_sparc64.c
etc. that UARTs are the only relevant ISA devices on sparc64 turned out
to be false. While there are sparc64 models where UARTs are the only
devices on the ISA bus there are in fact also low-cost models where all
devices traditionally found on the EBus are hooked up to the ISA bus.
There are also models that use a mix between EBus and ISA devices with
things like an AT keyboard controller and other rather interesting
devices that we might want to support in the futute hook up to the ISA
bus.
In order to not need to add sparc64 specific device_identify methods to
all of the respective ISA drivers and also not add OFW specific code to
the common ISA code make the sparc64 ISA bus code fake up PnP devices so
most ISA drivers probe their devices without further changes.
Unfortunately Sun doesn't adhere to the ISA bindings defined in IEEE
1275-1994 for the properties of most of the ISA devices which would
allow to obtain the vendor and logical IDs from their properties. So we
we just use a simple table which maps the name properties to PnP IDs.
This could be done in a more sophisticated way but I courrently don't
see the need for this. [1]
- Add the children with fully mapped and specified resources (in the OFW
sense) similar to what is done in the EBus code for the IRQ resources
of the children as adjusting the resources and the resource list entries
respectively in isa_alloc_resource() as done perviously causes trouble
with drivers which use rman_get_start(), pass-through or allocate and
release resources multiple times, etc.
Adjusting the resources might be better off in a bus_activate_resource
method but the common ISA code currently doesn't allow for an
isa_activate_resource(). [2]
With this change:
- ppbus(4) and lpt(4) attach and work (modulo ECP mode, which requires
real ISADMA code but it currently only consists of stubs on sparc64).
- atkbdc(4) and atkbdc(4) attach, no further testing done.
- fdc(4) itself attaches but causes a hang while attaching fd0 also
when is DMA disabled, further work in fdc(4) is required here as e.g.
fd0 uses the address of fd1 on sparc64 (not sure if sparc64 supports
more than one floppy drive at all).
All of these drivers previously caused panics in the sparc64 ISA code.
- Minor changes, e.g. use __FBSDID, remove a dupe word in a comment and
declare one global variable which isn't used outside of isa.c static.
o dev/uart/uart_cpu_sparc64.c and modules/uart/Makefile:
- Remove the code for registering the UARTs on the ISA bus from the
sparc64 uart_cpu_identify() again and rely on probing them via PnP.
Original idea by: tmm [1]
No objections by: tmm [1], [2]
I have in mind for the genclock interface):
- Recognize the MK48T18 as well (differs from the MK48T08 only in
packaging options and voltages).
- Allow MD code to provide functions for reading/writing NVRAM/RTC
locations.
If passed NULL, the old behaviour using bus_space_{read,write}_1() is
used. Otherwise, all access to the chip goes via the MD functions.
This is necessary for mvmeppc boards where the mk48txx NVRAM/RTC is
not directly addressable.
- Cleanup MI mk48txx(4) todclock driver:
- Prepare mk48txxvar.h and leave only register definitions in
mk48txxreg.h.
- Define struct mk48txx_softc as usual devices and allocate necessary
members in it.
- Change mk48txx_attach() to only take a device_t.
o While converting the sparc64 eeprom driver to the above changes:
- Remove some dead code and stale comments.
- Use the NVRAM size provided by the mk48txx driver instead of hardcoding
it as suggested by a comment.
- Add a comment about why it doesn't make much sense to read the hostid
directly from the NVRAM except for displaying it when attaching.
- Don't print the hostid if it reads all zero because it's stored
elsewhere.
Change the spelling of the "catch" option to be consistent with the new
options. Implement the "no wait" option. An implementation of the "CPU
private" for i386 will be committed at a later date.
need for most calls to vm_page_busy(). Specifically, most calls to
vm_page_busy() occur immediately prior to a call to vm_page_remove().
In such cases, the containing vm object is locked across both calls.
Consequently, the setting of the vm page's PG_BUSY flag is not even
visible to other threads that are following the synchronization
protocol.
This change (1) eliminates the calls to vm_page_busy() that
immediately precede a call to vm_page_remove() or functions, such as
vm_page_free() and vm_page_rename(), that call it and (2) relaxes the
requirement in vm_page_remove() that the vm page's PG_BUSY flag is
set. Now, the vm page's PG_BUSY flag is set only when the vm object
lock is released while the vm page is still in transition. Typically,
this is when it is undergoing I/O.
control the number of lines per page rather than a constant. The variable
can be examined and changed in ddb as '$lines'. Setting the variable to
0 will effectively turn off paging.
- Change db_putchar() to force out pending whitespace before outputting
newlines and carriage returns so that one can rub out content on the
current line via '\r \r' type strings.
- Change the simple pager to rub out the --More-- prompt explicitly when
the routine exits.
- Add some aliases to the simple pager to make it more compatible with
more(1): 'e' and 'j' do a single line. 'd' does half a page, and
'f' does a full page.
MFC after: 1 month
Inspired by: kris
on UltraSPARC workstations. The driver is based on OpenBSD's SBus
cs4231 driver and heavily modified to incorporate into sound(4)
infrastructure. Due to the lack of APCDMA documentation, the DMA
code of SBus cs4231 came from OpenBSD's driver.
The driver runs without Giant lock and supports both SBus and EBus
based CS4231 audio controller. Special thanks to marius for providing
feedbacks during the driver writing. His feedback made it possible
to write hiccup free playback code under high system loads.
Approved by: jake (mentor)
Reviewed by: marius (initial version)
Tested by: marius, kwm, Julian C. Dunn(jdunn AT opentrend DOT net)
be used to announce various system activity.
The auxio device provides auxiliary I/O functions and is found on various
SBus/EBus UltraSPARC models. At present, only front panel LED is
controlled by this driver.
Approved by: jake (mentor)
Reviewed by: joerg
Tested by: joerg
MP machines (hopefully). CPU timers are OK on UP machines but we
don't keep the timers in sync on MP machines so if the CPU's timer
is chosen as the primary timecounter it's possible for time to
not be monotonically increasing because different CPU's counters
may be used at different times. But the CPU's counters are otherwise
one of the higher quality counters available. So, on UP machines
we'll use a relatively high quality value but on MP machines we'll
use a quality that should prevent the CPU's counters from being chosen.
Requested by: green (who did the first version of the patch)
Reviewed by: marius, green
MFC after: 1 week
the structure space had been obtained from malloc() its contents is
random garbage. The choice of value being set is part of a larger effort
to solve some timecounter issues on MP machines (while working on that
we noticed this problem).
Noticed by: marius
Reviewed by: marius, green
MFC after: 3 days
longer than 'normal'. The cause is still being tracked down but
in the meantime there are machines where raising IPI_RETRIES does
help - it's not just a case of the machine staying locked up longer
and then panic-ing anyway. Several helpful folks on sparc64@ tried
a patch that helped figure out what to raise this number to.
Discussed on: sparc64@
MFC after: 3 days
MAXWIN to the register window manipulation functions - rwindow_load()
calls rwindow_save() so this one addition should take care of both.
This should help find places that pcb_nsaved doesn't get initialized
properly.
Suggested by: jake
for the new thread. The rest of the fields in the pcb wind up being
written to before they're read as a normal part of the pcb usage but
this field may be read upon return to userland, having it be uninitialized
garbage is bad.
Submitted by: Andrew Belashov (bel at orel dot ru)
Reviewed by: jake
MFC after: 3 days
a stack trace from ddb, the output will pause with a '--More--' prompt
every 18 lines. If you hit Enter, it will print another line and prompt
again. If you hit space it will output another page and then prompt.
If you hit 'q' or 'x' it will abort the rest of the stack trace.
- Fix the sparc64 userland stack trace to honor the total count of lines
to print. This is useful if your trace happens to walk back onto
0xdeadc0de and gets stuck in an endless loop.
MFC after: 1 month
Tested on: i386, alpha, sparc64
example the maximum segment size is 64K while the boundary is set
to 8K due to controller limitations. It is impossible to NOT cross
the boundary for any segment size that's larger than the boundary.
So, once we inherited the boundary from the parent tag, make sure
to reduce the maximum segment size to the boundary if it was larger.
MT5 candidate.
and was propagated to nearly every platform. The boundary of the child needs
to consider the boundary of the parent and pick the minimum of the two, not
the maximum. However, if either is 0 then pick the appropriate one.
This bug was exposed by a recent change to ATA, which should now be fixed by
this change. The alignment and maxsegsz tag attributes likely also need
a similar review in the near future.
This is a MT5 candidate.
Reviewed by: marcel
Submitted by: sos (in part)
It can be switched back once 5.3 is tested and released. Also turn on
PREEMPTION as many of the stability problems with it have been fixed.
MT5: 3 days.
but with slightly cleaned up interfaces.
The KSE structure has become the same as the "per thread scheduler
private data" structure. In order to not make the diffs too great
one is #defined as the other at this time.
The KSE (or td_sched) structure is now allocated per thread and has no
allocation code of its own.
Concurrency for a KSEGRP is now kept track of via a simple pair of counters
rather than using KSE structures as tokens.
Since the KSE structure is different in each scheduler, kern_switch.c
is now included at the end of each scheduler. Nothing outside the
scheduler knows the contents of the KSE (aka td_sched) structure.
The fields in the ksegrp structure that are to do with the scheduler's
queueing mechanisms are now moved to the kg_sched structure.
(per ksegrp scheduler private data structure). In other words how the
scheduler queues and keeps track of threads is no-one's business except
the scheduler's. This should allow people to write experimental
schedulers with completely different internal structuring.
A scheduler call sched_set_concurrency(kg, N) has been added that
notifies teh scheduler that no more than N threads from that ksegrp
should be allowed to be on concurrently scheduled. This is also
used to enforce 'fainess' at this time so that a ksegrp with
10000 threads can not swamp a the run queue and force out a process
with 1 thread, since the current code will not set the concurrency above
NCPU, and both schedulers will not allow more than that many
onto the system run queue at a time. Each scheduler should eventualy develop
their own methods to do this now that they are effectively separated.
Rejig libthr's kernel interface to follow the same code paths as
linkse for scope system threads. This has slightly hurt libthr's performance
but I will work to recover as much of it as I can.
Thread exit code has been cleaned up greatly.
exit and exec code now transitions a process back to
'standard non-threaded mode' before taking the next step.
Reviewed by: scottl, peter
MFC after: 1 week
The removed argument could trivially be derived from the remaining one.
That in turn should be the same as curthread, but it is possible that curthread could be expensive to derive on some syste,s so leave it as an argument.
Having both proc and thread as an argumen tjust gives an opportunity for
them to get out sync.
MFC after: 3 days
in diagnostics. It has outlived its usefulness and has started
causing panics for people who turn on DIAGNOSTIC, in what is otherwise
good code.
MFC after: 2 days
these two reasons:
1. On ia64 a function pointer does not hold the address of the first
instruction of a functions implementation. It holds the address
of a function descriptor. Hence the user(), btrap(), eintr() and
bintr() prototypes are wrong for getting the actual code address.
2. The logic forces interrupt, trap and exception entry points to
be layed-out contiguously. This can not be achieved on ia64 and is
generally just bad programming.
The MCOUNT_FROMPC_USER macro is used to set the frompc argument to
some kernel address which represents any frompc that falls outside
the kernel text range. The macro can expand to ~0U to bail out in
that case.
The MCOUNT_FROMPC_INTR macro is used to set the frompc argument to
some kernel address to represent a call to a trap or interrupt
handler. This to avoid that the trap or interrupt handler appear to
be called from everywhere in the call graph. The macro can expand
to ~0U to prevent adjusting frompc. Note that the argument is selfpc,
not frompc.
This commit defines the macros on all architectures equivalently to
the original code in sys/libkern/mcount.c. People can take it from
here...
Compile-tested on: alpha, amd64, i386, ia64 and sparc64
Boot-tested on: i386
compile option. All FreeBSD packet filters now use the PFIL_HOOKS API and
thus it becomes a standard part of the network stack.
If no hooks are connected the entire packet filter hooks section and related
activities are jumped over. This removes any performance impact if no hooks
are active.
Both OpenBSD and DragonFlyBSD have integrated PFIL_HOOKS permanently as well.
- Add some commented out NICs from i386 GENERIC. Most of them look like they
would work but I'm not sure if they are endian-clean and can't test. There
was a report that sk(4) works on sparc64 but it doesn't look like it would
because it doesn't use busdma.
- Improve some of the descriptions of sparc64 specific devices.
There's no functional change, i.e. no added or deleted uncommented devices or
options, in this commit.
- Chase the split of pcm(4). This unbreaks LINT compiles.
- sc(4) basically works and a lot of its options should be supported.
- Add the creator and ofw_console drivers.
- vinum(4) should work, at least its module was turned on for sparc64 a while
ago.
- Don't build sio(4). Its EBus front-end was removed a while ago and the ISA
one hardly works. Use uart(4) instead, it's not perfect yet but works much
better.
variable. If set to "true" OF_getetheraddr() will now return the unique
MAC address stored in the "local-mac-address" property of the device's
OFW node if present and the host address/system default MAC address if
the node doesn't doesn't have such a property. If set to "false" the
host address will be returned for all devices like before this change.
This brings the behaviour of device drivers for NICs with OFW support/
FCode, i.e. dc(4) for on-board DM9102A on Sun machines, gem(4) and hme(4),
regarding "local-mac-address?" in line with NetBSD and Solaris.
The man pages of the respective drivers will be updated separately to
reflect this change.
- Remove OF_getetheraddr2() which was used as a stopgap in dc(4). Its
functionality is now part of OF_getetheraddr().
subset ("compatible", "device_type", "model" and "name") of the standard
properties in drivers for devices on Open Firmware supported busses. The
standard properties "reg", "interrupts" und "address" are not covered by
this interface because they are only of interest in the respective bridge
code. There's a remaining standard property "status" which is unclear how
to support properly but which also isn't used in FreeBSD at present.
This ofw_bus kobj-interface allows to replace the various (ebus_get_node(),
ofw_pci_get_node(), etc.) and partially inconsistent (central_get_type()
vs. sbus_get_device_type(), etc.) existing IVAR ones with a common one.
This in turn allows to simplify and remove code-duplication in drivers for
devices that can hang off of more than one OFW supported bus.
- Convert the sparc64 Central, EBus, FHC, PCI and SBus bus drivers and the
drivers for their children to use the ofw_bus kobj-interface. The IVAR-
interfaces of the Central, EBus and FHC are entirely replaced by this. The
PCI bus driver used its own kobj-interface and now also uses the ofw_bus
one. The IVARs special to the SBus, e.g. for retrieving the burst size,
remain.
Beware: this causes an ABI-breakage for modules of drivers which used the
IVAR-interfaces, i.e. esp(4), hme(4), isp(4) and uart(4), which need to be
recompiled.
The style-inconsistencies introduced in some of the bus drivers will be
fixed by tmm@ in a generic clean-up of the respective drivers later (he
requested to add the changes in the "new" style).
- Convert the powerpc MacIO bus driver and the drivers for its children to
use the ofw_bus kobj-interface. This invloves removing the IVARs related
to the "reg" property which were unused and a leftover from the NetBSD
origini of the code. There's no ABI-breakage caused by this because none
of these driver are currently built as modules.
There are other powerpc bus drivers which can be converted to the ofw_bus
kobj-interface, e.g. the PCI bus driver, which should be done together
with converting powerpc to use the OFW PCI code from sparc64.
- Make the SBus and FHC front-end of zs(4) and the sparc64 eeprom(4) take
advantage of the ofw_bus kobj-interface and simplify them a bit.
Reviewed by: grehan, tmm
Approved by: re (scottl)
Discussed with: tmm
Tested with: Sun AX1105, AXe, Ultra 2, Ultra 60; PPC cross-build on i386
to allow dumping per-thread machine specific notes. On ia64 we use this
function to flush the dirty registers onto the backingstore before we
write out the PRSTATUS notes.
Tested on: alpha, amd64, i386, ia64 & sparc64
Not tested on: arm, powerpc
Implement the protection check required by the pmap_extract_and_hold()
specification.
Remove the acquisition and release of Giant from pmap_extract_and_hold() and
pmap_protect().
Many thanks to Ken Smith for resolving a sparc64-specific initialization
problem in my original patch.
Tested by: kensmith@
being defined, define and use a new MD macro, cpu_spinwait(). It only
expands to something on i386 and amd64, so the compiled code should be
identical.
Name of the macro found by: jhb
Reviewed by: jhb
their own directory and module, leaving the MD parts in the MD
area (the MD parts _are_ part of the modules). /dev/mem and /dev/io
are now loadable modules, thus taking us one step further towards
a kernel created entirely out of modules. Of course, there is nothing
preventing the kernel from having these statically compiled.
those architectures without pmap locking.
- Eliminate the acquisition and release of Giant from vm_map_protect().
(Translation: mprotect(2) runs to completion without touching Giant on
alpha, amd64, i386 and ia64.)
dereference curthread. It is called only from critical_{enter,exit}(),
which already dereferences curthread. This doesn't seem to affect SMP
performance in my benchmarks, but improves MySQL transaction throughput
by about 1% on UP on my Xeon.
Head nodding: jhb, bmilekic
the thread ID and call db_trace_thread().
Since arm has all the logic in db_stack_trace_cmd(), rename the
new DB_COMMAND function to db_stack_trace to avoid conflicts on
arm.
While here, have db_stack_trace parse its own arguments so that
we can use a more natural radix for IDs. If the ID is not a thread
ID, or more precisely when no thread exists with the ID, try if
there's a process with that ID and return the first thread in it.
This makes it easier to print stack traces from the ps output.
requested by: rwatson@
tested on: amd64, i386, ia64
pmap_protect() and pmap_remove(). In general, they require the lock in
order to modify a page's pv list or flags. In some cases, however,
pmap_protect() can avoid acquiring the lock.
Most of the changes are a direct result of adding thread awareness.
Typically, DDB_REGS is gone. All registers are taken from the
trapframe and backtraces use the PCB based contexts. DDB_REGS was
defined to be a trapframe on all platforms anyway.
Thread awareness introduces the following new commands:
thread X switch to thread X (where X is the TID),
show threads list all threads.
The backtrace code has been made more flexible so that one can
create backtraces for any thread by giving the thread ID as an
argument to trace.
With this change, ia64 has support for breakpoints.
o Make debugging code conditional upon KDB instead of DDB.
o Call kdb_enter() instead of Debugger().
o Remove implementation of Debugger().
o Check kdb_active instead of db_active.
o Call kdb_trap() according to the new world order.
a PCB from a trapframe for purposes of unwinding the stack. The PCB
is used as the thread context and all but the thread that entered the
debugger has a valid PCB.
This function can also be used to create a context for the threads
running on the CPUs that have been stopped when the debugger got
entered. This however is not done at the time of this commit.
in which multiple (presumably different) debugger backends can be
configured and which provides basic services to those backends.
Besides providing services to backends, it also serves as the single
point of contact for any and all code that wants to make use of the
debugger functions, such as entering the debugger or handling of the
alternate break sequence. For this purpose, the frontend has been
made non-optional.
All debugger requests are forwarded or handed over to the current
backend, if applicable. Selection of the current backend is done by
the debug.kdb.current sysctl. A list of configured backends can be
obtained with the debug.kdb.available sysctl. One can enter the
debugger by writing to the debug.kdb.enter sysctl.
backend improves over the old GDB support in the following ways:
o Unified implementation with minimal MD code.
o A simple interface for devices to register themselves as debug
ports, ala consoles.
o Compression by using run-length encoding.
o Implements GDB threading support.
than as one-off hacks in various other parts of the kernel:
- Add a function maybe_preempt() that is called from sched_add() to
determine if a thread about to be added to a run queue should be
preempted to directly. If it is not safe to preempt or if the new
thread does not have a high enough priority, then the function returns
false and sched_add() adds the thread to the run queue. If the thread
should be preempted to but the current thread is in a nested critical
section, then the flag TDF_OWEPREEMPT is set and the thread is added
to the run queue. Otherwise, mi_switch() is called immediately and the
thread is never added to the run queue since it is switch to directly.
When exiting an outermost critical section, if TDF_OWEPREEMPT is set,
then clear it and call mi_switch() to perform the deferred preemption.
- Remove explicit preemption from ithread_schedule() as calling
setrunqueue() now does all the correct work. This also removes the
do_switch argument from ithread_schedule().
- Do not use the manual preemption code in mtx_unlock if the architecture
supports native preemption.
- Don't call mi_switch() in a loop during shutdown to give ithreads a
chance to run if the architecture supports native preemption since
the ithreads will just preempt DELAY().
- Don't call mi_switch() from the page zeroing idle thread for
architectures that support native preemption as it is unnecessary.
- Native preemption is enabled on the same archs that supported ithread
preemption, namely alpha, i386, and amd64.
This change should largely be a NOP for the default case as committed
except that we will do fewer context switches in a few cases and will
avoid the run queues completely when preempting.
Approved by: scottl (with his re@ hat)
than a a stack-limited list. This removes the artifical limit on s/g list
size.
cvs: ----------------------------------------------------------------------
Ultra2 users may want to set OFWCONS_POLL_HZ to a value of '20'.
I have left default value at '4' as higher values can consume a more
than is acceptable amount of CPU, and we don't have a consensus yet
what is an optimal value.
Submitted by: Pyun YongHyeon <yongari@kt-is.co.kr>
its primary use is for the FEPS/FAS366 SCSI found in Sun Ultra 1e and 2
machines. Once the pci front-end is ported, this driver can replace the
amd(4) driver.
The code as-is is fairly stable. I've disabled tagged-queueing until I can
figure out a corruption bug related to it. I'm importing it now so that
people with these machines can (finally) stop netbooting and report bugs
before 5.3.
mbuma is an Mbuf & Cluster allocator built on top of a number of
extensions to the UMA framework, all included herein.
Extensions to UMA worth noting:
- Better layering between slab <-> zone caches; introduce
Keg structure which splits off slab cache away from the
zone structure and allows multiple zones to be stacked
on top of a single Keg (single type of slab cache);
perhaps we should look into defining a subset API on
top of the Keg for special use by malloc(9),
for example.
- UMA_ZONE_REFCNT zones can now be added, and reference
counters automagically allocated for them within the end
of the associated slab structures. uma_find_refcnt()
does a kextract to fetch the slab struct reference from
the underlying page, and lookup the corresponding refcnt.
mbuma things worth noting:
- integrates mbuf & cluster allocations with extended UMA
and provides caches for commonly-allocated items; defines
several zones (two primary, one secondary) and two kegs.
- change up certain code paths that always used to do:
m_get() + m_clget() to instead just use m_getcl() and
try to take advantage of the newly defined secondary
Packet zone.
- netstat(1) and systat(1) quickly hacked up to do basic
stat reporting but additional stats work needs to be
done once some other details within UMA have been taken
care of and it becomes clearer to how stats will work
within the modified framework.
From the user perspective, one implication is that the
NMBCLUSTERS compile-time option is no longer used. The
maximum number of clusters is still capped off according
to maxusers, but it can be made unlimited by setting
the kern.ipc.nmbclusters boot-time tunable to zero.
Work should be done to write an appropriate sysctl
handler allowing dynamic tuning of kern.ipc.nmbclusters
at runtime.
Additional things worth noting/known issues (READ):
- One report of 'ips' (ServeRAID) driver acting really
slow in conjunction with mbuma. Need more data.
Latest report is that ips is equally sucking with
and without mbuma.
- Giant leak in NFS code sometimes occurs, can't
reproduce but currently analyzing; brueffer is
able to reproduce but THIS IS NOT an mbuma-specific
problem and currently occurs even WITHOUT mbuma.
- Issues in network locking: there is at least one
code path in the rip code where one or more locks
are acquired and we end up in m_prepend() with
M_WAITOK, which causes WITNESS to whine from within
UMA. Current temporary solution: force all UMA
allocations to be M_NOWAIT from within UMA for now
to avoid deadlocks unless WITNESS is defined and we
can determine with certainty that we're not holding
any locks when we're M_WAITOK.
- I've seen at least one weird socketbuffer empty-but-
mbuf-still-attached panic. I don't believe this
to be related to mbuma but please keep your eyes
open, turn on debugging, and capture crash dumps.
This change removes more code than it adds.
A paper is available detailing the change and considering
various performance issues, it was presented at BSDCan2004:
http://www.unixdaemons.com/~bmilekic/netbuf_bmilekic.pdf
Please read the paper for Future Work and implementation
details, as well as credits.
Testing and Debugging:
rwatson,
brueffer,
Ketrien I. Saihr-Kesenchedra,
...
Reviewed by: Lots of people (for different parts)
struct vmspace is freed from cpu_sched_exit() to pmap_release().
This has the advantage of being able to rely on MI code to decide
when a free should occur, instead of having to inspect the reference
count ourselves.
At the same time, turn the per-CPU vmspace pointer into a pmap pointer,
so that pmap_release() can deal with pmaps exclusively.
Reviewed (and embrassing bug spotted) by: jake
