Merge the original development branch due to breakage caused by the
MTTCG merge.
Conflicts:
cpu-exec.c
translate-common.c
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The cpu->exit_request check in cpu_loop_exec_tb is unnecessary,
because cpu->tcg_exit_req is always set after cpu->exit_request.
So let the TB exit and we will pick up the exit request later
in cpu_handle_interrupt.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
This patch adds check to break cpu loop when icount expires without
setting the TB_EXIT_ICOUNT_EXPIRED flag. It happens when there is no
available translated blocks and all instructions were executed.
In icount replay mode unnecessary tb_find will be called (which may
cause an exception) and execution will be non-deterministic.
Because cpu_loop_exec_tb cannot longjmp anymore, we can remove
the anticipated call to align_clocks in cpu_loop_exec_tb, as
well as the SyncClocks *sc argument.
Signed-off-by: Pavel Dovgalyuk <pavel.dovgaluk@ispras.ru>
Message-Id: <002801d2810f$18809c20$4981d460$@ru>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Pavel Dovgalyuk <dovgaluk@ispras.ru>
The patch enables handling atomic code in the guest. This should be
preferably done in cpu_handle_exception(), but the current assumptions
regarding when we can execute atomic sections cause a deadlock.
The current mechanism discards the flags which were set in atomic
execution. We ensure they are properly saved by calling the
cc->cpu_exec_enter/leave() functions around the loop.
As we are running cpu_exec_step_atomic() from the outermost loop we
need to avoid an abort() when single stepping over atomic code since
debug exception longjmp will point to the the setlongjmp in
cpu_exec(). We do this by setting a new jmp_env so that it jumps back
here on an exception.
Signed-off-by: Pranith Kumar <bobby.prani@gmail.com>
[AJB: tweak title, merge with new patches, add mmap_lock]
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
CC: Paolo Bonzini <pbonzini@redhat.com>
There are a couple of changes that occur at the same time here:
- introduce a single vCPU qemu_tcg_cpu_thread_fn
One of these is spawned per vCPU with its own Thread and Condition
variables. qemu_tcg_rr_cpu_thread_fn is the new name for the old
single threaded function.
- the TLS current_cpu variable is now live for the lifetime of MTTCG
vCPU threads. This is for future work where async jobs need to know
the vCPU context they are operating in.
The user to switch on multi-thread behaviour and spawn a thread
per-vCPU. For a simple test kvm-unit-test like:
./arm/run ./arm/locking-test.flat -smp 4 -accel tcg,thread=multi
Will now use 4 vCPU threads and have an expected FAIL (instead of the
unexpected PASS) as the default mode of the test has no protection when
incrementing a shared variable.
We enable the parallel_cpus flag to ensure we generate correct barrier
and atomic code if supported by the front and backends. This doesn't
automatically enable MTTCG until default_mttcg_enabled() is updated to
check the configuration is supported.
Signed-off-by: KONRAD Frederic <fred.konrad@greensocs.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
[AJB: Some fixes, conditionally, commit rewording]
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
There are now only two uses of the global exit_request left.
The first ensures we exit the run_loop when we first start to process
pending work and in the kick handler. This is just as easily done by
setting the first_cpu->exit_request flag.
The second use is in the round robin kick routine. The global
exit_request ensured every vCPU would set its local exit_request and
cause a full exit of the loop. Now the iothread isn't being held while
running we can just rely on the kick handler to push us out as intended.
We lightly re-factor the main vCPU thread to ensure cpu->exit_requests
cause us to exit the main loop and process any IO requests that might
come along. As an cpu->exit_request may legitimately get squashed
while processing the EXCP_INTERRUPT exception we also check
cpu->queued_work_first to ensure queued work is expedited as soon as
possible.
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
This finally allows TCG to benefit from the iothread introduction: Drop
the global mutex while running pure TCG CPU code. Reacquire the lock
when entering MMIO or PIO emulation, or when leaving the TCG loop.
We have to revert a few optimization for the current TCG threading
model, namely kicking the TCG thread in qemu_mutex_lock_iothread and not
kicking it in qemu_cpu_kick. We also need to disable RAM block
reordering until we have a more efficient locking mechanism at hand.
Still, a Linux x86 UP guest and my Musicpal ARM model boot fine here.
These numbers demonstrate where we gain something:
20338 jan 20 0 331m 75m 6904 R 99 0.9 0:50.95 qemu-system-arm
20337 jan 20 0 331m 75m 6904 S 20 0.9 0:26.50 qemu-system-arm
The guest CPU was fully loaded, but the iothread could still run mostly
independent on a second core. Without the patch we don't get beyond
32206 jan 20 0 330m 73m 7036 R 82 0.9 1:06.00 qemu-system-arm
32204 jan 20 0 330m 73m 7036 S 21 0.9 0:17.03 qemu-system-arm
We don't benefit significantly, though, when the guest is not fully
loading a host CPU.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Message-Id: <1439220437-23957-10-git-send-email-fred.konrad@greensocs.com>
[FK: Rebase, fix qemu_devices_reset deadlock, rm address_space_* mutex]
Signed-off-by: KONRAD Frederic <fred.konrad@greensocs.com>
[EGC: fixed iothread lock for cpu-exec IRQ handling]
Signed-off-by: Emilio G. Cota <cota@braap.org>
[AJB: -smp single-threaded fix, clean commit msg, BQL fixes]
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Reviewed-by: Pranith Kumar <bobby.prani@gmail.com>
[PM: target-arm changes]
Acked-by: Peter Maydell <peter.maydell@linaro.org>
..and make the definition local to cpus. In preparation for MTTCG the
concept of a global tcg_current_cpu will no longer make sense. However
we still need to keep track of it in the single-threaded case to be able
to exit quickly when required.
qemu_cpu_kick_no_halt() moves and becomes qemu_cpu_kick_rr_cpu() to
emphasise its use-case. qemu_cpu_kick now kicks the relevant cpu as
well as qemu_kick_rr_cpu() which will become a no-op in MTTCG.
For the time being the setting of the global exit_request remains.
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Reviewed-by: Pranith Kumar <bobby.prani@gmail.com>
The recent patch enabling lock assertions uncovered the missing lock
acquisition in cpu_exec_step(). This patch adds them.
Signed-off-by: Pranith Kumar <bobby.prani@gmail.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
The icount interrupt flag and tcg_exit_req serve almost the same
purpose, let's make them completely the same.
The former TB_EXIT_REQUESTED and TB_EXIT_ICOUNT_EXPIRED cases are
unified, since we can distinguish them from the value of the
interrupt flag.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Commit 2afbdf8 ("target-i386: exception handling for memory helpers",
2015-09-15) changed tlb_fill's cpu_restore_state+raise_exception_err
to raise_exception_err_ra. After this change, the cpu_restore_state
and raise_exception_err's cpu_loop_exit are merged into
raise_exception_err_ra's cpu_loop_exit_restore.
This actually fixed some bugs, but when SVM is enabled there is a
second path from raise_exception_err_ra to cpu_loop_exit. This is
the VMEXIT path, and now cpu_vmexit is called without a
cpu_restore_state before.
The fix is to pass the retaddr to cpu_vmexit (via
cpu_svm_check_intercept_param). All helpers can now use GETPC() to pass
the correct retaddr, too.
Cc: qemu-stable@nongnu.org
Fixes: 2afbdf8480
Reported-by: Alexander Boettcher <alexander.boettcher@genode-labs.com>
Tested-by: Alexander Boettcher <alexander.boettcher@genode-labs.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reorganize the sigsetjmp so that the restart case falls through
to cpu_handle_exception and the execution loop.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The sigsetjmp only needs to be prepared once for the whole execution
of cpu_exec. This patch takes care of the "== 0" side, using a
nested loop so that cpu_handle_interrupt goes straight back to
cpu_handle_exception without doing another sigsetjmp.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The siglongjmp goes straight back to the beginning of cpu_exec's
outermost loop. We do not need a siglongjmp, we can simply
leave the inner TB execution loop.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
This seems to have worked just fine so far on weakly-ordered
architectures, but I don't see anything that prevents the
reordering from:
store 1 to exit_request
store 1 to tcg_exit_req
load tcg_exit_req
store 0 to tcg_exit_req
load exit_request
store 0 to exit_request
store 1 to exit_request
store 1 to tcg_exit_req
to this:
store 1 to exit_request
store 1 to tcg_exit_req
load tcg_exit_req
load exit_request
store 1 to exit_request
store 1 to tcg_exit_req
store 0 to tcg_exit_req
store 0 to exit_request
therefore losing a request. It's possible that other memory barriers
(e.g. in rcu_read_unlock) are hiding it, but better safe than
sorry.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
When icount is active, tb_add_jump is surprisingly called with an
out of bounds basic block index. I have no idea how that can work,
but it does not seem like a good idea. Clear *last_tb for all
TB_EXIT_ICOUNT_EXPIRED cases, even when all you have to do is
refill icount_extra.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Introduce rules in the top level Makefile that are able to generate
trace.[ch] files in every subdirectory which has a trace-events file.
The top level directory is handled specially, so instead of creating
trace.h, it creates trace-root.h. This allows sub-directories to
include the top level trace-root.h file, without ambiguity wrt to
the trace.g file in the current sub-dir.
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Message-id: 20170125161417.31949-7-berrange@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
This patch improves interrupt handling in record/replay mode.
Now "interrupt" event is saved only when cc->cpu_exec_interrupt returns true.
This patch also adds missing return to cpu_exec_interrupt function.
Signed-off-by: Pavel Dovgalyuk <pavel.dovgaluk@ispras.ru>
Message-Id: <20170124071708.4572.64023.stgit@PASHA-ISP>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reuse the existing locking provided by stdio to keep in_asm, cpu,
op, op_opt, op_ind, and out_asm as contiguous blocks.
While it isn't possible to interleave e.g. in_asm or op_opt logs
because of the TB lock protecting all code generation, it is
possible to interleave cpu logs, or to interleave a cpu dump with
an out_asm dump.
For mingw32, we appear to have no viable solution for this. The locking
functions are not properly exported from the system runtime library.
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Richard Henderson <rth@twiddle.net>
This protects all translation related work with tb_lock() too ensure
thread safety. This effectively serialises all code generation. In
addition to the code generation we also take the lock for TB
invalidation. This has a knock on effect of meaning tb_lock() is held
for modification of the SoftMMU TLB by non-self threads which will be
used in later patches.
Signed-off-by: KONRAD Frederic <fred.konrad@greensocs.com>
Message-Id: <1439220437-23957-8-git-send-email-fred.konrad@greensocs.com>
Signed-off-by: Emilio G. Cota <cota@braap.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
[AJB: moved into tree, clean-up history]
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Message-Id: <20161027151030.20863-10-alex.bennee@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Even more important when debugging MTTCG is seeing which vCPU is
currently executing.
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Message-Id: <20161027151030.20863-5-alex.bennee@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
When we cannot emulate an atomic operation within a parallel
context, this exception allows us to stop the world and try
again in a serial context.
Reviewed-by: Emilio G. Cota <cota@braap.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <rth@twiddle.net>
ThreadSanitizer picks up potential races although we already use
barriers to ensure things are in the correct order when processing exit
requests. For true C11 defined behaviour across threads we need to use
relaxed atomic_set/atomic_read semantics to reassure tsan.
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20160930213106.20186-9-alex.bennee@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Use async_safe_run_on_cpu() to make tb_flush() thread safe. This is
possible now that code generation does not happen in the middle of
execution.
It can happen that multiple threads schedule a safe work to flush the
translation buffer. To keep statistics and debugging output sane, always
check if the translation buffer has already been flushed.
Signed-off-by: Sergey Fedorov <serge.fdrv@gmail.com>
Signed-off-by: Sergey Fedorov <sergey.fedorov@linaro.org>
[AJB: minor re-base fixes]
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <1470158864-17651-13-git-send-email-alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
In fact, this function does not exactly perform a lookup by physical
address as it is descibed for comment on get_page_addr_code(). Thus
it may be a bit confusing to have "physical" in it's name. So rename it
to tb_htable_lookup() to better reflect its actual functionality.
Signed-off-by: Sergey Fedorov <serge.fdrv@gmail.com>
Signed-off-by: Sergey Fedorov <sergey.fedorov@linaro.org>
Message-Id: <20160715175852.30749-13-sergey.fedorov@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
These functions are not too big and can be merged together. This makes
locking scheme more clear and easier to follow.
Signed-off-by: Sergey Fedorov <serge.fdrv@gmail.com>
Signed-off-by: Sergey Fedorov <sergey.fedorov@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20160715175852.30749-12-sergey.fedorov@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Lock contention in the hot path of moving between existing patched
TranslationBlocks is the main drag in multithreaded performance. This
patch pushes the tb_lock() usage down to the two places that really need
it:
- code generation (tb_gen_code)
- jump patching (tb_add_jump)
The rest of the code doesn't really need to hold a lock as it is either
using per-CPU structures, atomically updated or designed to be used in
concurrent read situations (qht_lookup).
To keep things simple I removed the #ifdef CONFIG_USER_ONLY stuff as the
locks become NOPs anyway until the MTTCG work is completed.
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Reviewed-by: Sergey Fedorov <sergey.fedorov@linaro.org>
Signed-off-by: Sergey Fedorov <sergey.fedorov@linaro.org>
Message-Id: <20160715175852.30749-10-sergey.fedorov@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
When invalidating a translation block, set an invalid flag into the
TranslationBlock structure first. It is also necessary to check whether
the target TB is still valid after acquiring 'tb_lock' but before calling
tb_add_jump() since TB lookup is to be performed out of 'tb_lock' in
future. Note that we don't have to check 'last_tb'; an already invalidated
TB will not be executed anyway and it is thus safe to patch it.
Suggested-by: Sergey Fedorov <serge.fdrv@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Ensure atomicity and ordering of CPU's 'tb_flushed' access for future
translation block lookup out of 'tb_lock'.
This field can only be touched from another thread by tb_flush() in user
mode emulation. So the only access to be sequential atomic is:
* a single write in tb_flush();
* reads/writes out of 'tb_lock'.
In future, before enabling MTTCG in system mode, tb_flush() must be safe
and this field becomes unnecessary.
Signed-off-by: Sergey Fedorov <serge.fdrv@gmail.com>
Signed-off-by: Sergey Fedorov <sergey.fedorov@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20160715175852.30749-5-sergey.fedorov@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Ensure atomicity of CPU's 'tb_jmp_cache' access for future translation
block lookup out of 'tb_lock'.
Note that this patch does *not* make CPU's TLB invalidation safe if it
is done from some other thread while the CPU is in its execution loop.
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Sergey Fedorov <serge.fdrv@gmail.com>
Signed-off-by: Sergey Fedorov <sergey.fedorov@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20160715175852.30749-4-sergey.fedorov@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
This is a small clean up. tb_find_fast() is a final consumer of this
variable so no need to pass it by reference. 'last_tb' is always updated
by subsequent cpu_loop_exec_tb() in cpu_exec().
This change also simplifies calling cpu_exec_nocache() in
cpu_handle_exception().
Signed-off-by: Sergey Fedorov <serge.fdrv@gmail.com>
Signed-off-by: Sergey Fedorov <sergey.fedorov@linaro.org>
Message-Id: <20160715175852.30749-3-sergey.fedorov@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
This will fix a compiler warning with -Wclobbered:
http://lists.nongnu.org/archive/html/qemu-devel/2016-07/msg03347.html
Reported-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Sergey Fedorov <serge.fdrv@gmail.com>
Signed-off-by: Sergey Fedorov <sergey.fedorov@linaro.org>
Message-Id: <20160715193123.28113-1-sergey.fedorov@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Having a fixed-size hash table for keeping track of all translation blocks
is suboptimal: some workloads are just too big or too small to get maximum
performance from the hash table. The MRU promotion policy helps improve
performance when the hash table is a little undersized, but it cannot
make up for severely undersized hash tables.
Furthermore, frequent MRU promotions result in writes that are a scalability
bottleneck. For scalability, lookups should only perform reads, not writes.
This is not a big deal for now, but it will become one once MTTCG matures.
The appended fixes these issues by using qht as the implementation of
the TB hash table. This solution is superior to other alternatives considered,
namely:
- master: implementation in QEMU before this patchset
- xxhash: before this patch, i.e. fixed buckets + xxhash hashing + MRU.
- xxhash-rcu: fixed buckets + xxhash + RCU list + MRU.
MRU is implemented here by adding an intermediate struct
that contains the u32 hash and a pointer to the TB; this
allows us, on an MRU promotion, to copy said struct (that is not
at the head), and put this new copy at the head. After a grace
period, the original non-head struct can be eliminated, and
after another grace period, freed.
- qht-fixed-nomru: fixed buckets + xxhash + qht without auto-resize +
no MRU for lookups; MRU for inserts.
The appended solution is the following:
- qht-dyn-nomru: dynamic number of buckets + xxhash + qht w/ auto-resize +
no MRU for lookups; MRU for inserts.
The plots below compare the considered solutions. The Y axis shows the
boot time (in seconds) of a debian jessie image with arm-softmmu; the X axis
sweeps the number of buckets (or initial number of buckets for qht-autoresize).
The plots in PNG format (and with errorbars) can be seen here:
http://imgur.com/a/Awgnq
Each test runs 5 times, and the entire QEMU process is pinned to a
single core for repeatability of results.
Host: Intel Xeon E5-2690
28 ++------------+-------------+-------------+-------------+------------++
A***** + + + master **A*** +
27 ++ * xxhash ##B###++
| A******A****** xxhash-rcu $$C$$$ |
26 C$$ A******A****** qht-fixed-nomru*%%D%%%++
D%%$$ A******A******A*qht-dyn-mru A*E****A
25 ++ %%$$ qht-dyn-nomru &&F&&&++
B#####% |
24 ++ #C$$$$$ ++
| B### $ |
| ## C$$$$$$ |
23 ++ # C$$$$$$ ++
| B###### C$$$$$$ %%%D
22 ++ %B###### C$$$$$$C$$$$$$C$$$$$$C$$$$$$C$$$$$$C
| D%%%%%%B###### @E@@@@@@ %%%D%%%@@@E@@@@@@E
21 E@@@@@@E@@@@@@F&&&@@@E@@@&&&D%%%%%%B######B######B######B######B######B
+ E@@@ F&&& + E@ + F&&& + +
20 ++------------+-------------+-------------+-------------+------------++
14 16 18 20 22 24
log2 number of buckets
Host: Intel i7-4790K
14.5 ++------------+------------+-------------+------------+------------++
A** + + + master **A*** +
14 ++ ** xxhash ##B###++
13.5 ++ ** xxhash-rcu $$C$$$++
| qht-fixed-nomru %%D%%% |
13 ++ A****** qht-dyn-mru @@E@@@++
| A*****A******A****** qht-dyn-nomru &&F&&& |
12.5 C$$ A******A******A*****A****** ***A
12 ++ $$ A*** ++
D%%% $$ |
11.5 ++ %% ++
B### %C$$$$$$ |
11 ++ ## D%%%%% C$$$$$ ++
| # % C$$$$$$ |
10.5 F&&&&&&B######D%%%%% C$$$$$$C$$$$$$C$$$$$$C$$$$$C$$$$$$ $$$C
10 E@@@@@@E@@@@@@B#####B######B######E@@@@@@E@@@%%%D%%%%%D%%%###B######B
+ F&& D%%%%%%B######B######B#####B###@@@D%%% +
9.5 ++------------+------------+-------------+------------+------------++
14 16 18 20 22 24
log2 number of buckets
Note that the original point before this patch series is X=15 for "master";
the little sensitivity to the increased number of buckets is due to the
poor hashing function in master.
xxhash-rcu has significant overhead due to the constant churn of allocating
and deallocating intermediate structs for implementing MRU. An alternative
would be do consider failed lookups as "maybe not there", and then
acquire the external lock (tb_lock in this case) to really confirm that
there was indeed a failed lookup. This, however, would not be enough
to implement dynamic resizing--this is more complex: see
"Resizable, Scalable, Concurrent Hash Tables via Relativistic
Programming" by Triplett, McKenney and Walpole. This solution was
discarded due to the very coarse RCU read critical sections that we have
in MTTCG; resizing requires waiting for readers after every pointer update,
and resizes require many pointer updates, so this would quickly become
prohibitive.
qht-fixed-nomru shows that MRU promotion is advisable for undersized
hash tables.
However, qht-dyn-mru shows that MRU promotion is not important if the
hash table is properly sized: there is virtually no difference in
performance between qht-dyn-nomru and qht-dyn-mru.
Before this patch, we're at X=15 on "xxhash"; after this patch, we're at
X=15 @ qht-dyn-nomru. This patch thus matches the best performance that we
can achieve with optimum sizing of the hash table, while keeping the hash
table scalable for readers.
The improvement we get before and after this patch for booting debian jessie
with arm-softmmu is:
- Intel Xeon E5-2690: 10.5% less time
- Intel i7-4790K: 5.2% less time
We could get this same improvement _for this particular workload_ by
statically increasing the size of the hash table. But this would hurt
workloads that do not need a large hash table. The dynamic (upward)
resizing allows us to start small and enlarge the hash table as needed.
A quick note on downsizing: the table is resized back to 2**15 buckets
on every tb_flush; this makes sense because it is not guaranteed that the
table will reach the same number of TBs later on (e.g. most bootup code is
thrown away after boot); it makes sense to grow the hash table as
more code blocks are translated. This also avoids the complication of
having to build downsizing hysteresis logic into qht.
Reviewed-by: Sergey Fedorov <serge.fedorov@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Emilio G. Cota <cota@braap.org>
Message-Id: <1465412133-3029-15-git-send-email-cota@braap.org>
Signed-off-by: Richard Henderson <rth@twiddle.net>
For some workloads such as arm bootup, tb_phys_hash is performance-critical.
The is due to the high frequency of accesses to the hash table, originated
by (frequent) TLB flushes that wipe out the cpu-private tb_jmp_cache's.
More info:
https://lists.nongnu.org/archive/html/qemu-devel/2016-03/msg05098.html
To dig further into this I modified an arm image booting debian jessie to
immediately shut down after boot. Analysis revealed that quite a bit of time
is unnecessarily spent in tb_phys_hash: the cause is poor hashing that
results in very uneven loading of chains in the hash table's buckets;
the longest observed chain had ~550 elements.
The appended addresses this with two changes:
1) Use xxhash as the hash table's hash function. xxhash is a fast,
high-quality hashing function.
2) Feed the hashing function with not just tb_phys, but also pc and flags.
This improves performance over using just tb_phys for hashing, since that
resulted in some hash buckets having many TB's, while others getting very few;
with these changes, the longest observed chain on a single hash bucket is
brought down from ~550 to ~40.
Tests show that the other element checked for in tb_find_physical,
cs_base, is always a match when tb_phys+pc+flags are a match,
so hashing cs_base is wasteful. It could be that this is an ARM-only
thing, though. UPDATE:
On Tue, Apr 05, 2016 at 08:41:43 -0700, Richard Henderson wrote:
> The cs_base field is only used by i386 (in 16-bit modes), and sparc (for a TB
> consisting of only a delay slot).
> It may well still turn out to be reasonable to ignore cs_base for hashing.
BTW, after this change the hash table should not be called "tb_hash_phys"
anymore; this is addressed later in this series.
This change gives consistent bootup time improvements. I tested two
host machines:
- Intel Xeon E5-2690: 11.6% less time
- Intel i7-4790K: 19.2% less time
Increasing the number of hash buckets yields further improvements. However,
using a larger, fixed number of buckets can degrade performance for other
workloads that do not translate as many blocks (600K+ for debian-jessie arm
bootup). This is dealt with later in this series.
Reviewed-by: Sergey Fedorov <sergey.fedorov@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Emilio G. Cota <cota@braap.org>
Message-Id: <1465412133-3029-8-git-send-email-cota@braap.org>
Signed-off-by: Richard Henderson <rth@twiddle.net>
It is not safe to make a direct jump to a TB spanning two pages in
system emulation because the mapping for the second page can get changed
but we don't take care of direct jumps in this case.
However in user mode emulation, this is not the case because there's
only static address translation and TBs are always invalidated properly.
Fixes: 5b053a4a28 ("tcg: Clean up direct block chaining safety checks")
Reported-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Sergey Fedorov <serge.fdrv@gmail.com>
Signed-off-by: Sergey Fedorov <sergey.fedorov@linaro.org>
Tested-by: Max Filippov <jcmvbkbc@gmail.com>
Message-id: 1463404380-29302-1-git-send-email-sergey.fedorov@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
exec-all.h contains TCG-specific definitions. It is not needed outside
TCG-specific files such as translate.c, exec.c or *helper.c.
One generic function had snuck into include/exec/exec-all.h; move it to
include/qom/cpu.h.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Simplify cpu_exec() by extracting TB execution code outside of
cpu_exec() into a new static inline function cpu_loop_exec_tb().
Signed-off-by: Sergey Fedorov <serge.fdrv@gmail.com>
Signed-off-by: Sergey Fedorov <sergey.fedorov@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Message-Id: <1462962111-32237-5-git-send-email-sergey.fedorov@linaro.org>
Signed-off-by: Richard Henderson <rth@twiddle.net>
Simplify cpu_exec() by extracting interrupt handling code outside of
cpu_exec() into a new static inline function cpu_handle_interrupt().
Signed-off-by: Sergey Fedorov <serge.fdrv@gmail.com>
Signed-off-by: Sergey Fedorov <sergey.fedorov@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Message-Id: <1462962111-32237-4-git-send-email-sergey.fedorov@linaro.org>
Signed-off-by: Richard Henderson <rth@twiddle.net>
Simplify cpu_exec() by extracting exception handling code out of
cpu_exec() into a new static inline function cpu_handle_exception().
Also make cpu_handle_debug_exception() inline as it is used only once.
Signed-off-by: Sergey Fedorov <serge.fdrv@gmail.com>
Signed-off-by: Sergey Fedorov <sergey.fedorov@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Message-Id: <1462962111-32237-3-git-send-email-sergey.fedorov@linaro.org>
Signed-off-by: Richard Henderson <rth@twiddle.net>
Simplify cpu_exec() by extracting CPU halt state handling code out of
cpu_exec() into a new static inline function cpu_handle_halt().
Signed-off-by: Sergey Fedorov <serge.fdrv@gmail.com>
Signed-off-by: Sergey Fedorov <sergey.fedorov@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Message-Id: <1462962111-32237-2-git-send-email-sergey.fedorov@linaro.org>
Signed-off-by: Richard Henderson <rth@twiddle.net>
This comment should have been deleted by commit 0ac087f1f3 ("removed
unused code") but somehow it is still here. There's no point to keep it.
Signed-off-by: Sergey Fedorov <serge.fdrv@gmail.com>
Signed-off-by: Sergey Fedorov <sergey.fedorov@linaro.org>
Message-Id: <1462286050-21778-1-git-send-email-sergey.fedorov@linaro.org>
Signed-off-by: Richard Henderson <rth@twiddle.net>
This field was used for telling cpu_interrupt() to unlink a chain of TBs
being executed when it worked that way. Now, cpu_interrupt() don't do
this anymore. So we don't need this field anymore.
Signed-off-by: Sergey Fedorov <serge.fdrv@gmail.com>
Signed-off-by: Sergey Fedorov <sergey.fedorov@linaro.org>
Message-Id: <1462273462-14036-1-git-send-email-sergey.fedorov@linaro.org>
Signed-off-by: Richard Henderson <rth@twiddle.net>
Move tb_add_jump() call and surrounding code from cpu_exec() into
tb_find_fast(). That simplifies cpu_exec() a little by hiding the direct
chaining optimization details into tb_find_fast(). It also allows to
move tb_lock()/tb_unlock() pair into tb_find_fast(), putting it closer
to tb_find_slow() which also manipulates the lock.
Suggested-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Sergey Fedorov <serge.fdrv@gmail.com>
Signed-off-by: Sergey Fedorov <sergey.fedorov@linaro.org>
Signed-off-by: Richard Henderson <rth@twiddle.net>
[rth: Fixed rebase typo in nochain test.]
