Include the cluster number in the hash we use to look
up TBs. This is important because a TB that is valid
for one cluster at a given physical address and set
of CPU flags is not necessarily valid for another:
the two clusters may have different views of physical
memory, or may have different CPU features (eg FPU
present or absent).
We put the cluster number in the high 8 bits of the
TB cflags. This gives us up to 256 clusters, which should
be enough for anybody. If we ever need more, or need
more bits in cflags for other purposes, we could make
tb_hash_func() take more data (and expand qemu_xxhash7()
to qemu_xxhash8()).
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com>
Message-id: 20190121152218.9592-4-peter.maydell@linaro.org
We set up TLB entries in tlb_set_page_with_attrs(), where we have
some logic for determining whether the TLB entry is considered
to be RAM-backed, and thus has a valid addend field. When we
look at the TLB entry in get_page_addr_code(), we use different
logic for determining whether to treat the page as RAM-backed
and use the addend field. This is confusing, and in fact buggy,
because the code in tlb_set_page_with_attrs() correctly decides
that rom_device memory regions not in romd mode are not RAM-backed,
but the code in get_page_addr_code() thinks they are RAM-backed.
This typically results in "Bad ram pointer" assertion if the
guest tries to execute from such a memory region.
Fix this by making get_page_addr_code() just look at the
TLB_MMIO bit in the code_address field of the TLB, which
tlb_set_page_with_attrs() sets if and only if the addend
field is not valid for code execution.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Tested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-id: 20180713150945.12348-1-peter.maydell@linaro.org
There is no need for a stub, since tb_invalidate_phys_addr can be excised
altogether when TCG is disabled. This is a bit cleaner since it avoids
using code that is clearly specific to user-mode emulation (it calls
mmap_lock/unlock) for the !CONFIG_TCG case.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Fix the --disable-tcg breakage introduced by 8bca9a03ec:
$ configure --disable-tcg
[...]
$ make -C i386-softmmu exec.o
make: Entering directory 'i386-softmmu'
CC exec.o
In file included from source/qemu/exec.c:62:0:
source/qemu/include/exec/ram_addr.h:96:6: error: conflicting types for ‘tb_invalidate_phys_range’
void tb_invalidate_phys_range(ram_addr_t start, ram_addr_t end);
^~~~~~~~~~~~~~~~~~~~~~~~
In file included from source/qemu/exec.c:24:0:
source/qemu/include/exec/exec-all.h:309:6: note: previous declaration of ‘tb_invalidate_phys_range’ was here
void tb_invalidate_phys_range(target_ulong start, target_ulong end);
^~~~~~~~~~~~~~~~~~~~~~~~
source/qemu/exec.c:1043:6: error: conflicting types for ‘tb_invalidate_phys_addr’
void tb_invalidate_phys_addr(AddressSpace *as, hwaddr addr, MemTxAttrs attrs)
^~~~~~~~~~~~~~~~~~~~~~~
In file included from source/qemu/exec.c:24:0:
source/qemu/include/exec/exec-all.h:308:6: note: previous declaration of ‘tb_invalidate_phys_addr’ was here
void tb_invalidate_phys_addr(target_ulong addr);
^~~~~~~~~~~~~~~~~~~~~~~
make: *** [source/qemu/rules.mak:69: exec.o] Error 1
make: Leaving directory 'i386-softmmu'
Tested to build x86_64-softmmu and i386-softmmu targets.
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-id: 20180629200710.27626-1-f4bug@amsat.org
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Place them in exec.c, exec-all.h and ram_addr.h. This removes
knowledge of translate-all.h (which is an internal header) from
several files outside accel/tcg and removes knowledge of
AddressSpace from translate-all.c (as it only operates on ram_addr_t).
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Use mmap_lock in user-mode to protect TCG state and the page descriptors.
In !user-mode, each vCPU has its own TCG state, so no locks needed.
Per-page locks are used to protect the page descriptors.
Per-TB locks are used in both modes to protect TB jumps.
Some notes:
- tb_lock is removed from notdirty_mem_write by passing a
locked page_collection to tb_invalidate_phys_page_fast.
- tcg_tb_lookup/remove/insert/etc have their own internal lock(s),
so there is no need to further serialize access to them.
- do_tb_flush is run in a safe async context, meaning no other
vCPU threads are running. Therefore acquiring mmap_lock there
is just to please tools such as thread sanitizer.
- Not visible in the diff, but tb_invalidate_phys_page already
has an assert_memory_lock.
- cpu_io_recompile is !user-only, so no mmap_lock there.
- Added mmap_unlock()'s before all siglongjmp's that could
be called in user-mode while mmap_lock is held.
+ Added an assert for !have_mmap_lock() after returning from
the longjmp in cpu_exec, just like we do in cpu_exec_step_atomic.
Performance numbers before/after:
Host: AMD Opteron(tm) Processor 6376
ubuntu 17.04 ppc64 bootup+shutdown time
700 +-+--+----+------+------------+-----------+------------*--+-+
| + + + + + *B |
| before ***B*** ** * |
|tb lock removal ###D### *** |
600 +-+ *** +-+
| ** # |
| *B* #D |
| *** * ## |
500 +-+ *** ### +-+
| * *** ### |
| *B* # ## |
| ** * #D# |
400 +-+ ** ## +-+
| ** ### |
| ** ## |
| ** # ## |
300 +-+ * B* #D# +-+
| B *** ### |
| * ** #### |
| * *** ### |
200 +-+ B *B #D# +-+
| #B* * ## # |
| #* ## |
| + D##D# + + + + |
100 +-+--+----+------+------------+-----------+------------+--+-+
1 8 16 Guest CPUs 48 64
png: https://imgur.com/HwmBHXe
debian jessie aarch64 bootup+shutdown time
90 +-+--+-----+-----+------------+------------+------------+--+-+
| + + + + + + |
| before ***B*** B |
80 +tb lock removal ###D### **D +-+
| **### |
| **## |
70 +-+ ** # +-+
| ** ## |
| ** # |
60 +-+ *B ## +-+
| ** ## |
| *** #D |
50 +-+ *** ## +-+
| * ** ### |
| **B* ### |
40 +-+ **** # ## +-+
| **** #D# |
| ***B** ### |
30 +-+ B***B** #### +-+
| B * * # ### |
| B ###D# |
20 +-+ D ##D## +-+
| D# |
| + + + + + + |
10 +-+--+-----+-----+------------+------------+------------+--+-+
1 8 16 Guest CPUs 48 64
png: https://imgur.com/iGpGFtv
The gains are high for 4-8 CPUs. Beyond that point, however, unrelated
lock contention significantly hurts scalability.
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Emilio G. Cota <cota@braap.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
This applies to both user-mode and !user-mode emulation.
Instead of relying on a global lock, protect the list of incoming
jumps with tb->jmp_lock. This lock also protects tb->cflags,
so update all tb->cflags readers outside tb->jmp_lock to use
atomic reads via tb_cflags().
In order to find the destination TB (and therefore its jmp_lock)
from the origin TB, we introduce tb->jmp_dest[].
I considered not using a linked list of jumps, which simplifies
code and makes the struct smaller. However, it unnecessarily increases
memory usage, which results in a performance decrease. See for
instance these numbers booting+shutting down debian-arm:
Time (s) Rel. err (%) Abs. err (s) Rel. slowdown (%)
------------------------------------------------------------------------------
before 20.88 0.74 0.154512 0.
after 20.81 0.38 0.079078 -0.33524904
GTree 21.02 0.28 0.058856 0.67049808
GHashTable + xxhash 21.63 1.08 0.233604 3.5919540
Using a hash table or a binary tree to keep track of the jumps
doesn't really pay off, not only due to the increased memory usage,
but also because most TBs have only 0 or 1 jumps to them. The maximum
number of jumps when booting debian-arm that I measured is 35, but
as we can see in the histogram below a TB with that many incoming jumps
is extremely rare; the average TB has 0.80 incoming jumps.
n_jumps: 379208; avg jumps/tb: 0.801099
dist: [0.0,1.0)|▄█▁▁▁▁▁▁▁▁▁▁▁ ▁▁▁▁▁▁ ▁▁▁ ▁▁▁ ▁|[34.0,35.0]
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Emilio G. Cota <cota@braap.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
The appended adds assertions to make sure we do not longjmp with page
locks held. Note that user-mode has nothing to check, since page_locks
are !user-mode only.
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Emilio G. Cota <cota@braap.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Groundwork for supporting parallel TCG generation.
Instead of using a global lock (tb_lock) to protect changes
to pages, use fine-grained, per-page locks in !user-mode.
User-mode stays with mmap_lock.
Sometimes changes need to happen atomically on more than one
page (e.g. when a TB that spans across two pages is
added/invalidated, or when a range of pages is invalidated).
We therefore introduce struct page_collection, which helps
us keep track of a set of pages that have been locked in
the appropriate locking order (i.e. by ascending page index).
This commit first introduces the structs and the function helpers,
to then convert the calling code to use per-page locking. Note
that tb_lock is not removed yet.
While at it, rename tb_alloc_page to tb_page_add, which pairs with
tb_page_remove.
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Emilio G. Cota <cota@braap.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
This commit does several things, but to avoid churn I merged them all
into the same commit. To wit:
- Use uintptr_t instead of TranslationBlock * for the list of TBs in a page.
Just like we did in (c37e6d7e "tcg: Use uintptr_t type for
jmp_list_{next|first} fields of TB"), the rationale is the same: these
are tagged pointers, not pointers. So use a more appropriate type.
- Only check the least significant bit of the tagged pointers. Masking
with 3/~3 is unnecessary and confusing.
- Introduce the TB_FOR_EACH_TAGGED macro, and use it to define
PAGE_FOR_EACH_TB, which improves readability. Note that
TB_FOR_EACH_TAGGED will gain another user in a subsequent patch.
- Update tb_page_remove to use PAGE_FOR_EACH_TB. In case there
is a bug and we attempt to remove a TB that is not in the list, instead
of segfaulting (since the list is NULL-terminated) we will reach
g_assert_not_reached().
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Emilio G. Cota <cota@braap.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
This paves the way for enabling scalable parallel generation of TCG code.
Instead of tracking TBs with a single binary search tree (BST), use a
BST for each TCG region, protecting it with a lock. This is as scalable
as it gets, since each TCG thread operates on a separate region.
The core of this change is the introduction of struct tcg_region_tree,
which contains a pointer to a GTree and an associated lock to serialize
accesses to it. We then allocate an array of tcg_region_tree's, adding
the appropriate padding to avoid false sharing based on
qemu_dcache_linesize.
Given a tc_ptr, we first find the corresponding region_tree. This
is done by special-casing the first and last regions first, since they
might be of size != region.size; otherwise we just divide the offset
by region.stride. I was worried about this division (several dozen
cycles of latency), but profiling shows that this is not a fast path.
Note that region.stride is not required to be a power of two; it
is only required to be a multiple of the host's page size.
Note that with this design we can also provide consistent snapshots
about all region trees at once; for instance, tcg_tb_foreach
acquires/releases all region_tree locks before/after iterating over them.
For this reason we now drop tb_lock in dump_exec_info().
As an alternative I considered implementing a concurrent BST, but this
can be tricky to get right, offers no consistent snapshots of the BST,
and performance and scalability-wise I don't think it could ever beat
having separate GTrees, given that our workload is insert-mostly (all
concurrent BST designs I've seen focus, understandably, on making
lookups fast, which comes at the expense of convoluted, non-wait-free
insertions/removals).
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Emilio G. Cota <cota@braap.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Currently we don't support board configurations that put an IOMMU
in the path of the CPU's memory transactions, and instead just
assert() if the memory region fonud in address_space_translate_for_iotlb()
is an IOMMUMemoryRegion.
Remove this limitation by having the function handle IOMMUs.
This is mostly straightforward, but we must make sure we have
a notifier registered for every IOMMU that a transaction has
passed through, so that we can flush the TLB appropriately
when any of the IOMMUs change their mappings.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Message-id: 20180604152941.20374-5-peter.maydell@linaro.org
The API for cpu_transaction_failed() says that it takes the physical
address for the failed transaction. However we were actually passing
it the offset within the target MemoryRegion. We don't currently
have any target CPU implementations of this hook that require the
physical address; fix this bug so we don't get confused if we ever
do add one.
Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20180611125633.32755-3-peter.maydell@linaro.org
In icount mode, instructions that access io memory spaces in the middle
of the translation block invoke TB recompilation. After recompilation,
such instructions become last in the TB and are allowed to access io
memory spaces.
When the code includes instruction like i386 'xchg eax, 0xffffd080'
which accesses APIC, QEMU goes into an infinite loop of the recompilation.
This instruction includes two memory accesses - one read and one write.
After the first access, APIC calls cpu_report_tpr_access, which restores
the CPU state to get the current eip. But cpu_restore_state_from_tb
resets the cpu->can_do_io flag which makes the second memory access invalid.
Therefore the second memory access causes a recompilation of the block.
Then these operations repeat again and again.
This patch moves resetting cpu->can_do_io flag from
cpu_restore_state_from_tb to cpu_loop_exit* functions.
It also adds a parameter for cpu_restore_state which controls restoring
icount. There is no need to restore icount when we only query CPU state
without breaking the TB. Restoring it in such cases leads to the
incorrect flow of the virtual time.
In most cases new parameter is true (icount should be recalculated).
But there are two cases in i386 and openrisc when the CPU state is only
queried without the need to break the TB. This patch fixes both of
these cases.
Signed-off-by: Pavel Dovgalyuk <Pavel.Dovgaluk@ispras.ru>
Message-Id: <20180409091320.12504.35329.stgit@pasha-VirtualBox>
[rth: Make can_do_io setting unconditional; move from cpu_exec;
make cpu_loop_exit_{noexc,restore} call cpu_loop_exit.]
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
The MC68040 MMU provides the size of the access that
triggers the page fault.
This size is set in the Special Status Word which
is written in the stack frame of the access fault
exception.
So we need the size in m68k_cpu_unassigned_access() and
m68k_cpu_handle_mmu_fault().
To be able to do that, this patch modifies the prototype of
handle_mmu_fault handler, tlb_fill() and probe_write().
do_unassigned_access() already includes a size parameter.
This patch also updates handle_mmu_fault handlers and
tlb_fill() of all targets (only parameter, no code change).
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20180118193846.24953-2-laurent@vivier.eu>
Normally we create an address space for that CPU and pass that address
space into the function. Let's just do it inside to unify address space
creations. It'll simplify my next patch to rename those address spaces.
Signed-off-by: Peter Xu <peterx@redhat.com>
Message-Id: <20171123092333.16085-3-peterx@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
We are still seeing signals during translation time when we walk over
a page protection boundary. This expands the check to ensure the host
PC is inside the code generation buffer. The original suggestion was
to check versus tcg_ctx.code_gen_ptr but as we now segment the
translation buffer we have to settle for just a general check for
being inside.
I've also fixed up the declaration to make it clear it can deal with
invalid addresses. A later patch will fix up the call sites.
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Reported-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20171108153245.20740-2-alex.bennee@linaro.org
Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Cc: Richard Henderson <rth@twiddle.net>
Tested-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
We don't really free anything in this function anymore; we just remove
the TB from the binary search tree.
Suggested-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Emilio G. Cota <cota@braap.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
This is a prerequisite for supporting multiple TCG contexts, since
we will have threads generating code in separate regions of
code_gen_buffer.
For this we need a new field (.size) in struct tb_tc to keep
track of the size of the translated code. This field uses a size_t
to avoid adding a hole to the struct, although really an unsigned
int would have been enough.
The comparison function we use is optimized for the common case:
insertions. Profiling shows that upon booting debian-arm, 98%
of comparisons are between existing tb's (i.e. a->size and b->size
are both !0), which happens during insertions (and removals, but
those are rare). The remaining cases are lookups. From reading the glib
sources we see that the first key is always the lookup key. However,
the code does not assume this to always be the case because this
behaviour is not guaranteed in the glib docs. However, we embed
this knowledge in the code as a branch hint for the compiler.
Note that tb_free does not free space in the code_gen_buffer anymore,
since we cannot easily know whether the tb is the last one inserted
in code_gen_buffer. The next patch in this series renames tb_free
to tb_remove to reflect this.
Performance-wise, lookups in tb_find_pc are the same as before:
O(log n). However, insertions are O(log n) instead of O(1), which
results in a small slowdown when booting debian-arm:
Performance counter stats for 'build/arm-softmmu/qemu-system-arm \
-machine type=virt -nographic -smp 1 -m 4096 \
-netdev user,id=unet,hostfwd=tcp::2222-:22 \
-device virtio-net-device,netdev=unet \
-drive file=img/arm/jessie-arm32.qcow2,id=myblock,index=0,if=none \
-device virtio-blk-device,drive=myblock \
-kernel img/arm/aarch32-current-linux-kernel-only.img \
-append console=ttyAMA0 root=/dev/vda1 \
-name arm,debug-threads=on -smp 1' (10 runs):
- Before:
8048.598422 task-clock (msec) # 0.931 CPUs utilized ( +- 0.28% )
16,974 context-switches # 0.002 M/sec ( +- 0.12% )
0 cpu-migrations # 0.000 K/sec
10,125 page-faults # 0.001 M/sec ( +- 1.23% )
35,144,901,879 cycles # 4.367 GHz ( +- 0.14% )
<not supported> stalled-cycles-frontend
<not supported> stalled-cycles-backend
65,758,252,643 instructions # 1.87 insns per cycle ( +- 0.33% )
10,871,298,668 branches # 1350.707 M/sec ( +- 0.41% )
192,322,212 branch-misses # 1.77% of all branches ( +- 0.32% )
8.640869419 seconds time elapsed ( +- 0.57% )
- After:
8146.242027 task-clock (msec) # 0.923 CPUs utilized ( +- 1.23% )
17,016 context-switches # 0.002 M/sec ( +- 0.40% )
0 cpu-migrations # 0.000 K/sec
18,769 page-faults # 0.002 M/sec ( +- 0.45% )
35,660,956,120 cycles # 4.378 GHz ( +- 1.22% )
<not supported> stalled-cycles-frontend
<not supported> stalled-cycles-backend
65,095,366,607 instructions # 1.83 insns per cycle ( +- 1.73% )
10,803,480,261 branches # 1326.192 M/sec ( +- 1.95% )
195,601,289 branch-misses # 1.81% of all branches ( +- 0.39% )
8.828660235 seconds time elapsed ( +- 0.38% )
Reviewed-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Emilio G. Cota <cota@braap.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Now that we have curr_cflags, we can include CF_USE_ICOUNT
early and then remove it as necessary.
Reviewed-by: Emilio G. Cota <cota@braap.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
These flags are used by target/*/translate.c,
and affect code generation.
Reviewed-by: Emilio G. Cota <cota@braap.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
