This implements all of the v8.1-Atomics instructions except
for compare-and-swap, which is decoded elsewhere.
Backports commit 74608ea45434c9b07055b21885e093528c5ed98c from qemu
The insns in the ARMv8.1-Atomics are added to the existing
load/store exclusive and load/store reg opcode spaces.
Rearrange the top-level decoders for these to accomodate.
The Atomics insns themselves still generate Unallocated.
Backports commit 68412d2ecedbab5a43b0d346cddb27e00d724aff from qemu
Given that this atomic operation will be used by both risc-v
and aarch64, let's not duplicate code across the two targets.
Backports commit 5507c2bf35aa6b4705939349184e71afd5e058b2 from qemu
These operations are re-invented by several targets so far.
Several supported hosts have insns for these, so place the
expanders out-of-line for a future introduction of tcg opcodes.
Backports commit b87fb8cd9f9a0ba599ff79e7bf03222da02e5724 from qemu
Some versions of gcc produce a spurious warning if the result of
__atomic_compare_echange_n() is not used and the type involved
is a signed 8 bit value:
error: value computed is not used [-Werror=unused-value]
This has been seen on at least
gcc (Ubuntu 5.4.0-6ubuntu1~16.04.9) 5.4.0 20160609
Work around this by using an explicit cast to void to indicate
that we don't care about the return value.
We don't currently use our atomic_cmpxchg() macro on any signed
8 bit types, but the upcoming support for the Arm v8.1-Atomics
will require it.
Backports commit cd95fc28fb6d8afced0d70ce52c294d0761a9daa from qemu
It is implementation defined whether a multiply-add of
(0,inf,qnan) or (inf,0,qnan) raises InvalidaOperation or
not, so we let the target-specific pickNaNMulAdd function
handle this. This means that we must do the "return the
default NaN in default NaN mode" check after the call,
not before. Correct the ordering, and restore the comment
from the old propagateFloat64MulAddNaN() that warned about
this corner case.
This fixes a regression from 2.11 for Arm guests where we would
incorrectly fail to set the Invalid flag for these cases.
Backports commit 1839189bbf89889076aadf0c793c1b57977b28d7 from qemu
Notes:
- DISAS_TOO_MANY replaces the former "break" in the translation loop.
However, care must be taken not to overwrite a previous condition
in is_jmp; that's why in translate_insn we first check is_jmp and
return if it's != DISAS_NEXT.
- Added an assert in translate_insn, before exiting due to an exception,
to make sure that is_jmp is set to DISAS_NORETURN (the exception
generation function always sets it.)
- Added an assert for the default case in is_jmp's switch.
Backports commit 18f440edfb974feaff8490d4861844b5a2b7a3b5 from qemu
No changes to the logic here; this is just to make the diff
that follows easier to read.
While at it, remove the unnecessary 'struct' in
'struct TranslationBlock'.
Note that checkpatch complains with a false positive:
ERROR: space prohibited after that '&' (ctx:WxW)
\#75: FILE: target/mips/translate.c:20220:
+ ctx->kscrexist = (env->CP0_Config4 >> CP0C4_KScrExist) & 0xff;
^
Backports commit 12be92588cf26a192f1b62846906983fc1e102a7 from qemu
Notes:
- BS_EXCP in generate_exception_err and after hen_helper_wait
becomes DISAS_NORETURN, because we do not return after
raising an exception.
- Some uses of BS_EXCP are misleading in that they're used
only as a "not BS_STOP" exit condition, i.e. they have nothing
to do with an actual exception. For those cases, define
and use DISAS_EXIT, which is clearer. With this and the
above change, BS_EXCP goes away completely.
- fix a comment typo (s/intetrupt/interrupt/).
Backports commit b28425babc2ad4b90cd87d07a1809d3322b9c065 from qemu
The TB after BS_STOP is not fixed (e.g. helper_mtc0_hwrena
changes hflags, which ends up changing the TB flags via
cpu_get_tb_cpu_state). This requires a full lookup (i.e.
with flags) via lookup_and_goto_ptr instead of gen_goto_tb,
since the latter only looks at the PC for in-page goto's. Fix it.
Backports commit cd314a7d0190a03122ca0606ecf71b4b873a22c6 from qemu.
Notes:
- Moved the cross-page check from the end of translate_insn to
init_disas_context.
Backports commit 6e61bc941025345ab01c48d116bef60bb8990406 from qemu
Notes:
- pc and npc are left unmodified, since they can point to out-of-TB
jump targets.
- Got rid of last_pc in gen_intermediate_code(), using base.pc_next
instead. Only update pc_next (1) on a breakpoint (so that tb->size
includes the insn), and (2) after reading the current instruction
from memory. This allows us to use base.pc_next in the BP check,
which is what the translator loop does.
Backports commit af00be490b30d7f576d12ac7b2bc5406ca6fda3f from qemu
While at it, use int for both num_insns and max_insns to make
sure we have same-type comparisons.
Backports commit b542683d77b4f56cef0221b267c341616d87bce9 from qemu
If the PC is in the last page of the address space, next_page_start
overflows to 0. Fix it.
Backports commit 6cd79443d33e6ba6b4c5b787eb713ca1cec56328 from qemu
If the PC is in the last page of the address space, next_page_start
overflows to 0. Fix it.
Backports commit bfe7ad5be77a6a8925a7ab1628452c8942222102 from qemu
In 6001f7729e12 we partially attempt to address the branch
displacement overflow caused by 15fa08f845.
However, gcc/testsuite/gcc.target/aarch64/advsimd-intrinsics/vqtbX.c
is a testcase that contains a TB so large as to overflow anyway.
The limit here of 8000 ops produces a maximum output TB size of
24112 bytes on a ppc64le host with that test case. This is still
much less than the maximum forward branch distance of 32764 bytes.
Backports commit abebf92597186be2bc48d487235da28b1127860f from qemu
The VPUNPCKLD* instructions are all "non-destructive source",
indicated by "NDS" in the encoding string in the x86 ISA manual.
This means that they take two source operands, one of which is
encoded in the VEX.vvvv field. We were incorrectly treating them
as if they were destructive-source and passing 0 as the 'v'
argument of tcg_out_vex_modrm(). This meant we were always
using %xmm0 as one of the source operands, causing incorrect
results if the register allocator happened to want to use
something else. For instance the input AArch64 insn:
DUP v26.16b, w21
which becomes TCG IR ops:
dup_vec v128,e8,tmp2,x21
st_vec v128,e8,tmp2,env,$0xa40
was assembled to:
0x607c568c: c4 c1 7a 7e 86 e8 00 00 vmovq 0xe8(%r14), %xmm0
0x607c5694: 00
0x607c5695: c5 f9 60 c8 vpunpcklbw %xmm0, %xmm0, %xmm1
0x607c5699: c5 f9 61 c9 vpunpcklwd %xmm1, %xmm0, %xmm1
0x607c569d: c5 f9 70 c9 00 vpshufd $0, %xmm1, %xmm1
0x607c56a2: c4 c1 7a 7f 8e 40 0a 00 vmovdqu %xmm1, 0xa40(%r14)
0x607c56aa: 00
when the vpunpcklwd insn should be "%xmm1, %xmm1, %xmm1".
This resulted in our incorrectly setting the output vector to
q26=0000320000003200:0000320000003200
when given an input of x21 == 0000000002803200
rather than the expected all-zeroes.
Pass the correct source register number to tcg_out_vex_modrm()
for these insns.
Backports commit 7eb30ef0ba2eb59e7430d4848ae8d4bf4e50f768 from qemu
The Werror stems from the compiler finding a path through the second
switch via a missing default case in which src1 is uninitialized, and
not being able to prove that the missing default case is unreachable
due to the first switch.
Simplify the second switch to merge default with OS_LONG,
which returns directly. This removes the unreachable path.
Backports commit 5cbc61110738accb16ff8ed1f08a32906d02790f from qemu.
For v8M the instructions VLLDM and VLSTM support lazy saving
and restoring of the secure floating-point registers. Even
if the floating point extension is not implemented, these
instructions must act as NOPs in Secure state, so they can
be used as part of the secure-to-nonsecure call sequence.
Fixes: https://bugs.launchpad.net/qemu/+bug/1768295
Backports commit b1e5336a9899016c53d59eba53ebf6abcc21995c from qemu
The duplication of id_tlbtr_reginfo was unintentionally added within
3281af8114c6b8ead02f08b58e3c36895c1ea047 which should have been
id_mpuir_reginfo.
The effect was that for OMAP and StrongARM CPUs we would
incorrectly UNDEF writes to MPUIR rather than NOPing them.
Backports commit 100061121c1f69a672ce7bb3e9e3781f8018f9f6 from qemu
Path analysis shows that size == 3 && !is_q has been eliminated.
Fixes: Coverity CID1385853
Backports commit a8766e3172c1671cab297c1ef4566a3c5d094822 from qemu
The (size > 3 && !is_q) condition is identical to the preceeding test
of bit 3 in immh; eliminate it. For the benefit of Coverity, assert
that size is within the bounds we expect.
Fixes: Coverity CID1385846
Fixes: Coverity CID1385849
Fixes: Coverity CID1385852
Fixes: Coverity CID1385857
Backports commit 8dae46970532afcf93470b00e83ca9921980efc3 from qemu
For convenience and clarity, make it possible to call qobject_ref() at
the time when the reference is associated with a variable, or
argument, by making qobject_ref() return the same pointer as given.
Use that to simplify the callers.
Backports commit f5a74a5a50387c6f980b2e2f94f062487a1826da from qemu
Now that we can safely call QOBJECT() on QObject * as well as its
subtypes, we can have macros qobject_ref() / qobject_unref() that work
everywhere instead of having to use QINCREF() / QDECREF() for QObject
and qobject_incref() / qobject_decref() for its subtypes.
The replacement is mechanical, except I broke a long line, and added a
cast in monitor_qmp_cleanup_req_queue_locked(). Unlike
qobject_decref(), qobject_unref() doesn't accept void *.
Note that the new macros evaluate their argument exactly once, thus no
need to shout them.
Backports commit cb3e7f08aeaab0ab13e629ce8496dca150a449ba from qemu
By moving the base fields to a QObjectBase_, QObject can be a type
which also has a 'base' field. This allows writing a generic QOBJECT()
macro that will work with any QObject type, including QObject
itself. The container_of() macro ensures that the object to cast has a
QObjectBase_ base field, giving some type safety guarantees. QObject
must have no members but QObjectBase_ base, or else QOBJECT() breaks.
QObjectBase_ is not a typedef and uses a trailing underscore to make
it obvious it is not for normal use and to avoid potential abuse.
Backports commit 3d3eacaeccaab718ea0e2ddaa578bfae9e311c59 from qemu
All QObject types have the base QObject as their first field. This
allows the simplification of qobject_to().
Backports commit 7ee9edfdb117da47c86c9764d90f0be11a648666 from qemu
ppc64 uses a BC instruction to call the tcg_out_qemu_ld/st
slow path. BC instruction uses a relative address encoded
on 14 bits.
The slow path functions are added at the end of the generated
instructions buffer, in the reverse order of the callers.
So more we have slow path functions more the distance between
the caller (BC) and the function increases.
This patch changes the behavior to generate the functions in
the same order of the callers.
Backports commit 6001f7729e12dd1d810291e4cbf83cee8e07441d from qemu
Drop TCGV_PTR_TO_NAT and TCGV_NAT_TO_PTR internal macros.
Add tcg_temp_local_new_ptr, tcg_gen_brcondi_ptr, tcg_gen_ext_i32_ptr,
tcg_gen_trunc_i64_ptr, tcg_gen_extu_ptr_i64, tcg_gen_trunc_ptr_i32.
Use inlines instead of macros where possible.
Backports commit 5bfa803448638a45542441fd6b7cc1241403ea72 from qemu
In db432672, we allow wide inputs for operations such as add.
However, in 212be173 and 3774030a we didn't do the same for
compare and multiply.
Backports commit 9a938d86b04025ac605db0ea9819e5896bf576ec from qemu
I found with qemu 2.11.x or newer that I would get an illegal instruction
error running some Intel binaries on my ARM chromebook. On investigation,
I found it was quitting on memory barriers.
qemu instruction:
mb $0x31
was translating as:
0x604050cc: 5bf07ff5 blpl #0x600250a8
After patch it gives:
0x604050cc: f57ff05b dmb ish
In short, I found INSN_DMB_ISH (memory barrier for ARMv7) appeared to be
correct based on online docs, but due to some endian-related shenanigans it
had to be byte-swapped to suit qemu; it appears INSN_DMB_MCR (memory
barrier for ARMv6) also should be byte swapped (and this patch does so).
I have not checked for correctness of aarch64's barrier instruction.
Backports commit 3f814b803797c007abfe5c4041de754e01723031 from qemu
floatx80_sin() and floatx80_cos() are derived from one
sincos() function. They have both unused code coming from
their common origin. Remove it.
Backports commit 6361d2984ce88912976a34e1797a5ad5139c649b from qemu
This patch fixes decrement of the pointers for subx mem, mem instructions.
Without the patch pointers are decremented by OS_* constant value instead of
retrieving the corresponding data size and using it as a decrement.
Backports commit 355d4d1c00e708907ff391c24ca708f1c9c06bf0 from qemu
This is a bug fix to ensure 64-bit reads of these registers don't read
adjacent data.
Backports commit e4e91a217c17fff4045dd4b423cdcb471b3d6a0e from qemu
Because the design of the PMU requires that the counter values be
converted between their delta and guest-visible forms for mode
filtering, an additional hook which occurs before the EL is changed is
necessary.
Backports commit b5c53d1b3886387874f8c8582b205aeb3e4c3df6 from qemu
This eliminates the need for fetching it from el_change_hook_opaque, and
allows for supporting multiple el_change_hooks without having to hack
something together to find the registered opaque belonging to GICv3.
Backports commit d5a5e4c93dae0dc3feb402cf7ee78d846da1a7e1 from qemu
In commit 95695effe8caa552b8f2 we changed the v7M/v8M stack
pop code to use a new v7m_stack_read() function that checks
whether the read should fail due to an MPU or bus abort.
We missed one call though, the one which reads the signature
word for the callee-saved register part of the frame.
Correct the omission.
Backports commit 4818bad98c8212fbbb0525d10761b6b65279ab92 from qemu
Remove a stale TODO comment -- we have now made the arm_ldl_ptw()
and arm_ldq_ptw() functions propagate physical memory read errors
out to their callers.
Backports commit 145772707fe80395b87c244ccf5699a756f1946b from qemu
Without bounding the increment, we can overflow exp either here
in scalbn_decomposed or when adding the bias in round_canonical.
This can result in e.g. underflowing to 0 instead of overflowing
to infinity.
The old softfloat code did bound the increment.
Backports commit ce8d4082054519f2eaac39958edde502860a7fc6 from qemu
The re-factoring of div_floats changed the order of checking meaning
an operation like -inf/0 erroneously raises the divbyzero flag.
IEEE-754 (2008) specifies this should only occur for operations on
finite operands.
We fix this by moving the check on the dividend being Inf/0 to before
the divisor is zero check.
Backports commit 9cb4e398c2f95c1e837fe9c570e124a55259f725 from qemu
