If the CPU is a PMSA config with no MPU implemented, then the
SCTLR.M bit should be RAZ/WI, so that the guest can never
turn on the non-existent MPU.
Backports commit 06312febfb2d35367006ef23608ddd6a131214d4 from qemu
Fix the handling of QOM properties for PMSA CPUs with no MPU:
Allow no-MPU to be specified by either:
* has-mpu = false
* pmsav7_dregion = 0
and make setting one imply the other. Don't clear the PMSA
feature bit in this situation.
Backports commit f50cd31413d8bc9d1eef8edd1f878324543bf65d from qemu
ARM CPUs come in two flavours:
* proper MMU ("VMSA")
* only an MPU ("PMSA")
For PMSA, the MPU may be implemented, or not (in which case there
is default "always acts the same" behaviour, but it isn't guest
programmable).
QEMU is a bit confused about how we indicate this: we have an
ARM_FEATURE_MPU, but it's not clear whether this indicates
"PMSA, not VMSA" or "PMSA and MPU present" , and sometimes we
use it for one purpose and sometimes the other.
Currently trying to implement a PMSA-without-MPU core won't
work correctly because we turn off the ARM_FEATURE_MPU bit
and then a lot of things which should still exist get
turned off too.
As the first step in cleaning this up, rename the feature
bit to ARM_FEATURE_PMSA, which indicates a PMSA CPU (with
or without MPU).
Backports commit 452a095526a0537f16c271516a2200877a272ea8 from qemu
Make M profile use completely separate ARMMMUIdx values from
those that A profile CPUs use. This is a prelude to adding
support for the MPU and for v8M, which together will require
6 MMU indexes which don't map cleanly onto the A profile
uses:
non secure User
non secure Privileged
non secure Privileged, execution priority < 0
secure User
secure Privileged
secure Privileged, execution priority < 0
Backports commit e7b921c2d9efc249f99b9feb0e7dca82c96aa5c4 from qemu
The v7M exception architecture requires that if a synchronous
exception cannot be taken immediately (because it is disabled
or at too low a priority) then it should be escalated to
HardFault (and the HardFault exception is then taken).
Implement this escalation logic.
Backports commit a73c98e159d18155445d29b6044be6ad49fd802f from qemu
The M profile CPU's MPU has an awkward corner case which we
would like to implement with a different MMU index.
We can avoid having to bump the number of MMU modes ARM
uses, because some of our existing MMU indexes are only
used by non-M-profile CPUs, so we can borrow one.
To avoid that getting too confusing, clean up the code
to try to keep the two meanings of the index separate.
Instead of ARMMMUIdx enum values being identical to core QEMU
MMU index values, they are now the core index values with some
high bits set. Any particular CPU always uses the same high
bits (so eventually A profile cores and M profile cores will
use different bits). New functions arm_to_core_mmu_idx()
and core_to_arm_mmu_idx() convert between the two.
In general core index values are stored in 'int' types, and
ARM values are stored in ARMMMUIdx types.
Backports commit 8bd5c82030b2cb09d3eef6b444f1620911cc9fc5 from qemu
The PMUv3 driver of linux kernel (in arch/arm64/kernel/perf_event.c)
relies on the PMUVER field of id_aa64dfr0_el1 to decide if PMU support
is present or not. This patch clears the PMUVER field under TCG mode
when vPMU=off. Without it, PMUv3 will init insider guest VMs even
with vPMU=off. This patch also removes a redundant line inside the
if-statement.
Backports commit 2b3ffa929249b15a75d8bde3e8e57a744f52aff0 from qemu
When identifying the DFSR format for an alignment fault, use
the mmu index that we are passed, rather than calling cpu_mmu_index()
to get the mmu index for the current CPU state. This doesn't actually
make any difference since the only cases where the current MMU index
differs from the index used for the load are the "unprivileged
load/store" instructions, and in that case the mmu index may
differ but the translation regime is the same (apart from the
"use from Hyp mode" case which is UNPREDICTABLE).
However it's the more logical thing to do.
Backports commit e517d95b63427fae9f03958dbc005c36b4ebf2cf from qemu
This patch converts the old "is_write" bool into IOMMUAccessFlags. The
difference is that "is_write" can only express either read/write, but
sometimes what we really want is "none" here (neither read nor write).
Replay is an good example - during replay, we should not check any RW
permission bits since thats not an actual IO at all.
Backports commit bf55b7afce53718ef96f4e6616da62c0ccac37dd from qemu
This function is an abstraction helper for address_space_translate() and
address_space_get_iotlb_entry(). It does the lookup of address into
memory region section, then does proper IOMMU translation if necessary.
Refactor the two existing functions to use it.
This fixes vhost when IOMMU is disabled by guest.
Backports commit a764040cc831cfe5b8bf1c80e8341b9bf2de3ce8 from qemu
In case where the conditional write is the first write to the page,
TLB_NOTDIRTY will be set and stop_the_world is triggered. Handle this as
a special case and set the dirty bit. After that fall through to the
actual atomic instruction below.
Backports commit 7f9af1abdcc69fd1d3d8d2be68464329600616d6 from qemu
Users of tcg_gen_atomic_cmpxchg and do_atomic_op rightfully utilize
the output. Even though this code is dead, it gets translated, and
without the initialization we encounter a tcg_error.
Backports commit 79b1af906245558c30e0a5faf26cb52b63f83cce from qemu
We already require gcc 4.1 or newer (for the atomic
support), so the fallback codepaths for older gcc
versions than that are now dead code and we can
just delete them.
NB: clang reports itself as gcc 4.2 (regardless of
clang version), so clang won't be using the fallbacks
either.
Backports commit fa54abb8c298f892639ffc4bc2f61448ac3be4a1 from qemu
Now that we've rewritten M-profile exception return so that the magic
PC values are not visible to other parts of QEMU, we can delete the
special casing of them elsewhere.
Backports commit f4e8e4edda875cab9df91dc4ae9767f7cb1f50aa from qemu
On M profile, return from exceptions happen when code in Handler mode
executes one of the following function call return instructions:
* POP or LDM which loads the PC
* LDR to PC
* BX register
and the new PC value is 0xFFxxxxxx.
QEMU tries to implement this by not treating the instruction
specially but then catching the attempt to execute from the magic
address value. This is not ideal, because:
* there are guest visible differences from the architecturally
specified behaviour (for instance jumping to 0xFFxxxxxx via a
different instruction should not cause an exception return but it
will in the QEMU implementation)
* we have to account for it in various places (like refusing to take
an interrupt if the PC is at a magic value, and making sure that
the MPU doesn't deny execution at the magic value addresses)
Drop these hacks, and instead implement exception return the way the
architecture specifies -- by having the relevant instructions check
for the magic value and raise the 'do an exception return' QEMU
internal exception immediately.
The effect on the generated code is minor:
bx lr, old code (and new code for Thread mode):
TCG:
mov_i32 tmp5,r14
movi_i32 tmp6,$0xfffffffffffffffe
and_i32 pc,tmp5,tmp6
movi_i32 tmp6,$0x1
and_i32 tmp5,tmp5,tmp6
st_i32 tmp5,env,$0x218
exit_tb $0x0
set_label $L0
exit_tb $0x7f2aabd61993
x86_64 generated code:
0x7f2aabe87019: mov %ebx,%ebp
0x7f2aabe8701b: and $0xfffffffffffffffe,%ebp
0x7f2aabe8701e: mov %ebp,0x3c(%r14)
0x7f2aabe87022: and $0x1,%ebx
0x7f2aabe87025: mov %ebx,0x218(%r14)
0x7f2aabe8702c: xor %eax,%eax
0x7f2aabe8702e: jmpq 0x7f2aabe7c016
bx lr, new code when in Handler mode:
TCG:
mov_i32 tmp5,r14
movi_i32 tmp6,$0xfffffffffffffffe
and_i32 pc,tmp5,tmp6
movi_i32 tmp6,$0x1
and_i32 tmp5,tmp5,tmp6
st_i32 tmp5,env,$0x218
movi_i32 tmp5,$0xffffffffff000000
brcond_i32 pc,tmp5,geu,$L1
exit_tb $0x0
set_label $L1
movi_i32 tmp5,$0x8
call exception_internal,$0x0,$0,env,tmp5
x86_64 generated code:
0x7fe8fa1264e3: mov %ebp,%ebx
0x7fe8fa1264e5: and $0xfffffffffffffffe,%ebx
0x7fe8fa1264e8: mov %ebx,0x3c(%r14)
0x7fe8fa1264ec: and $0x1,%ebp
0x7fe8fa1264ef: mov %ebp,0x218(%r14)
0x7fe8fa1264f6: cmp $0xff000000,%ebx
0x7fe8fa1264fc: jae 0x7fe8fa126509
0x7fe8fa126502: xor %eax,%eax
0x7fe8fa126504: jmpq 0x7fe8fa122016
0x7fe8fa126509: mov %r14,%rdi
0x7fe8fa12650c: mov $0x8,%esi
0x7fe8fa126511: mov $0x56095dbeccf5,%r10
0x7fe8fa12651b: callq *%r10
which is a difference of one cmp/branch-not-taken. This will
be lost in the noise of having to exit generated code and
look up the next TB anyway.
Backports commit 3bb8a96f5348913ee130169504f3642f501b113e from qemu
For M profile exception-return handling we'd like to generate different
code for some instructions depending on whether we are in Handler
mode or Thread mode. This isn't the same as "are we privileged
or user", so we need an extra bit in the TB flags to distinguish.
Backports commit 064c379c99b835bdcc478d21a3849507ea07d53a from qemu
Move the code to generate the "condition failed" instruction
codepath out of the if (singlestepping) {} else {}. This
will allow adding support for handling a new is_jmp type
which can't be neatly split into "singlestepping case"
versus "not singlestepping case".
Backports commit f021b2c4627890d82fbcc300db3bd782b37b7f8a from qemu
arm: Abstract out "are we singlestepping" test to utility function
We now test for "are we singlestepping" in several places and
it's not a trivial check because we need to care about both
architectural singlestep and QEMU gdbstub singlestep. We're
also about to add another place that needs to make this check,
so pull the condition out into a function.
Backports commit b636649f5a2e108413dd171edaf320f781f57942 from qemu
Move the utility routines gen_set_condexec() and gen_set_pc_im()
up in the file, as we will want to use them from a function
placed earlier in the file than their current location.
Backports commit 4d5e8c969a74c86124fc2284ea603cc6dd3c5dfa from qemu
We currently have two places that do:
if (dc->ss_active) {
gen_step_complete_exception(dc);
} else {
gen_exception_internal(EXCP_DEBUG);
}
Factor this out into its own function, as we're about to add
a third place that needs the same logic.
Backports commit 5425415ebba5fa20558e1ef25e1997a6f5ea4c7c from qemu
In Thumb mode, the only instructions which can cause an interworking
branch by writing the PC are BLX, BX, BXJ, LDR, POP and LDM. Unlike
ARM mode, data processing instructions which target the PC do not
cause interworking branches.
When we added support for doing interworking branches on writes to
PC from data processing instructions in commit 21aeb3430ce7ba, we
accidentally changed a Thumb instruction to have interworking
branch behaviour for writes to PC. (MOV, MOVS register-shifted
register, encoding T2; this is the standard encoding for
LSL/LSR/ASR/ROR (register).)
For this encoding, behaviour with Rd == R15 is specified as
UNPREDICTABLE, so allowing an interworking branch is within
spec, but it's confusing and differs from our handling of this
class of UNPREDICTABLE for other Thumb ALU operations. Make
it perform a simple (non-interworking) branch like the others.
Backports commit bedb8a6b09c1754c3b9f155750c62dc087706698 from qemu
For M-profile CPUs, the BXJ instruction does not exist at all, and
the encoding should always UNDEF. We were accidentally implementing
it to behave like A-profile BXJ; correct the error.
Backports commit 9d7c59c84d4530d05e8702b1c3a31e6da00a397e from qemu
In tlb_fill() we construct a syndrome register value from a
fault status register value which is filled in by arm_tlb_fill().
arm_tlb_fill() returns FSR values which might be in the format
used with short-format page descriptors, or the format used
with long-format (LPAE) descriptors. The syndrome register
always uses LPAE-format FSR status codes.
It isn't actually possible to end up delivering a syndrome
register value to the guest for a fault which is reported
with a short-format FSR (that kind of stage 1 fault will only
happen for an AArch32 translation regime which doesn't have
a syndrome register, and can never be redirected to an AArch64
or Hyp exception level). Add an assertion which checks this,
and adjust the code so that we construct a syndrome with
an invalid status code, rather than allowing set bits in
the FSR input to randomly corrupt other fields in the syndrome.
Backports commit 65ed2ed90d9d81fd4b639029be850ea5651f919f from qemu
The excnames[] array is defined in internals.h because we used
to use it from two different source files for handling logging
of AArch32 and AArch64 exception entry. Refactoring means that
it's now used only in arm_log_exception() in helper.c, so move
the array into that function.
Backports commit 2c4a7cc5afb1bfc1728a39abd951ddd7714c476e from qemu
Recent changes have added new EXCP_ values to ARM but forgot
to update the excnames[] array which is used to provide
human-readable strings when printing information about the
exception for debug logging. Add the missing entries, and
add a comment to the list of #defines to help avoid the mistake
being repeated in future.
Backports commit 32b81e620ea562d56ab2733421b5da1082b237a2 from qemu
For "ldp x0, x1, [x0]", if the second load is on a second page and
the second page is unmapped, the exception would be raised with x0
already modified. This means the instruction couldn't be restarted.
Backports commit 2d1bbf51c2cb948da4b6fd5f91cf3ecc80b28156 from qemu
MemoryRegionCache did not know about virtio support for IOMMUs (because the
two features were developed at the same time). Revert MemoryRegionCache
to "normal" address_space_* operations for 2.9, as it is simpler than
undoing the virtio patches.
Backports commit 90c4fe5fc517a045e7a7cf2f23472e114042ca29 from qemu
The C store helper functions take the address argument as a
target_ulong type; if this is 32 bit but the host is 64 bit
then the SPARC calling convention requires that the caller
must zero extend the value. We weren't doing this, which
meant we could pass values to the caller with high bits set
and QEMU would crash if it was compiled with optimizations.
In particular, the i386 BIOS would not start.
Backports commit 5c32be5baf41aec4f4675d2bf24f9948756abf3c from qemu
The C store helper functions take the data argument as a uint8_t,
uint16_t, etc depending on the store size. The SPARC calling
convention requires that data types smaller than the register
size must be extended by the caller. We weren't doing this,
which meant that if QEMU was compiled with optimizations enabled
we could end up storing incorrect values to guest memory.
(In particular the i386 guest BIOS would crash on startup.)
Add code to the trampolines that call the store helpers to
do the zero extension as required.
Backports commit 709a340d679d95a0c6cbb9b5f654498f04345b50 from qemu
The existing code for "host" and "max" CPU models overrides every
single feature in the CPU object at realize time, even the ones
that were explicitly enabled or disabled by the user using
"feat=on" or "feat=off", while features set using +feat/-feat are
kept.
This means "-cpu host,+invtsc" works as expected, while
"-cpu host,invtsc=on" doesn't.
This was a known bug, already documented in a comment inside
x86_cpu_expand_features(). What makes this bug worse now is that
libvirt 3.0.0 and newer now use "feat=on|off" instead of
+feat/-feat when it detects a QEMU version that supports it (see
libvirt commit d47db7b16dd5422c7e487c8c8ee5b181a2f9cd66).
Change the feature property getter/setter to set a
env->user_features field, to keep track of features that were
explicitly changed using QOM properties. Then make the
max_features code not override user features when handling "-cpu
host" and "-cpu max".
This will also allow us to remove the plus_features/minus_features
hack in the future, but I plan to do that after 2.9.0 is
released.
Backports commit d4a606b38b5d4b3689b86cc1575908e82179ecfb from qemu
The Cygwin target is really compiling for native Win32 with -mno-cygwin.
Except, GCC 4.7.0 has finally removed the long deprecated -mno-cygwin
option, and that happened about five years ago.
Let it rest in peace.
Backports commit c8645752ce31cc044ecc5f969a986fdcb6aab590 from qemu
It is unnecessary to test R6 from delay/forbidden slot check
in gen_msa_branch().
https://bugs.launchpad.net/qemu/+bug/1663287
Backports commit 075a1fe788d36b271ec25507466c30b9a90b5d54 from qemu
this fixes many warnings like:
target/mips/translate.c:6253:13: warning: Value stored to 'rn' is never read
rn = "invalid sel";
^ ~~~~~~~~~~~~~
Backports commit 3570d7f6672836140f0a1ec9bf95dd5ea50a2aaa from qemu
static code analyzer complain:
target/mips/helper.c:453:5: warning: Function call argument is an uninitialized value
qemu_log_mask(CPU_LOG_MMU,
^~~~~~~~~~~~~~~~~~~~~~~~~~
'physical' and 'prot' are uninitialized if 'ret' is not TLBRET_MATCH.
Backports commit def74c0cf05722b2e502d4b4f1219966c5b0cbd3 from qemu
Our implementation of writes to the APSR for M-profile via the MSR
instruction was badly broken.
First and worst, we had the sense wrong on the test of bit 2 of the
SYSm field -- this is supposed to request an APSR write if bit 2 is 0
but we were doing it if bit 2 was 1. This bug was introduced in
commit 58117c9bb429cd, so hasn't been in a QEMU release.
Secondly, the choice of exactly which parts of APSR should be written
is defined by bits in the 'mask' field. We were not passing these
through from instruction decode, making it impossible to check them
in the helper.
Pass the mask bits through from the instruction decode to the helper
function and process them appropriately; fix the wrong sense of the
SYSm bit 2 check.
Invalid mask values and invalid combinations of mask and register
number are UNPREDICTABLE; we choose to treat them as if the mask
values were valid.
Backports commit b28b3377d7e9ba35611d454d5a63ef50cab1f8c5 from qemu
For M profile (unlike A profile) the reset value of R14 is specified
as 0xffffffff. (The rationale is that this is an illegal exception
return value, so if guest code tries to return to it it will result
in a helpful exception.)
Registers r0 to r12 and the flags are architecturally UNKNOWN on
reset, so we leave those at zero.
Backports commit 056f43df9168413f304500b69c33158d66efb7cf from qemu
For M profile CPUs, FAULTMASK should be 0 on reset, like PRIMASK.
QEMU stores FAULTMASK in the PSTATE F bit, so (as with PRIMASK in the
I bit) we have to clear these to undo the A profile default of 1.
Update the comment accordingly and move it so that it's closer to the
code it's referring to.
Backports commit dc7abe4d65ad39390b2db120f5ad18f8f6576f8b from qemu
For v7M attempts to access a nonexistent coprocessor are reported
differently from plain undefined instructions (as UsageFaults of type
NOCP rather than type UNDEFINSTR). Split them out into a new
EXCP_NOCP so we can report the FSR value correctly.
Backports commit 7517748e3f71a3099e57915fba95c4c308e6d842 from qemu
For v7M attempts to access a nonexistent coprocessor are reported
differently from plain undefined instructions (as UsageFaults of type
NOCP rather than type UNDEFINSTR). Split them out into a new
EXCP_NOCP so we can report the FSR value correctly.
Backports commit 7517748e3f71a3099e57915fba95c4c308e6d842 from qemu
When we take an exception for an undefined instruction, set the
appropriate CFSR bit.
Backports commit 81dd9648c69bb89afdd6f4bb3ed6f3efdac96524 from qemu
The CCR.STACKALIGN bit controls whether the CPU is supposed to force
8-alignment of the stack pointer on entry to the exception handler.
Backports commit dc858c6633a9af8b80c1509cf6f825e4390d3ad1 from qemu
Add the structure fields, VMState fields, reset code and macros for
the v7M system control registers CCR, CFSR, HFSR, DFSR, MMFAR and
BFAR.
Backports commit 2c4da50d9477fb830d778bb5d6a11215aa359b44 from qemu
Give an explicit error and abort when a load
from the vector table fails. Architecturally this
should HardFault (which will then immediately
fail to load the HardFault vector and go into Lockup).
Since we don't model Lockup, just report this guest
error via cpu_abort(). This is more helpful than the
previous behaviour of reading a zero, which is the
address of the reset stack pointer and not a sensible
location to jump to.
Backports commit 1b9ea408fca1ce8caae67b792355b023c69c5ac5 from qemu
For v7m we need to catch attempts to execute from special
addresses at 0xfffffff0 and above. Previously we did this
with the aid of a hacky special purpose lump of memory
in the address space and a check in translate.c for whether
we were translating code at those addresses.
We can implement this more cleanly using a CPU
unassigned access handler which throws the exception
if the unassigned access is for one of the special addresses.
Backports commit 542b3478a00cb7ef51c259255b3ab1e2a7daada2 from qemu
The MRS and MSR instruction handling has a number of flaws:
* unprivileged accesses should only be able to read
CONTROL and the xPSR subfields, and only write APSR
(others RAZ/WI)
* privileged access should not be able to write xPSR
subfields other than APSR
* accesses to unimplemented registers should log as
guest errors, not abort QEMU
Backports commit 58117c9bb429cd9552d998687aa99088eb1d8528 from qemu
The v7m CONTROL register bit 1 is SPSEL, which indicates
the stack being used. We were storing this information
not in v7m.control but in the separate v7m.other_sp
structure field. Unfortunately, the code handling reads
of the CONTROL register didn't take account of this, and
so if SPSEL was updated by an exception entry or exit then
a subsequent guest read of CONTROL would get the wrong value.
Using a separate structure field doesn't really gain us
anything in efficiency, so drop this unnecessary complexity
in favour of simply storing all the bits in v7m.control.
This is a migration compatibility break for M profile
CPUs only.
Backports commit abc24d86cc0364f402e438fae3acb14289b40734 from qemu
The MRS instruction requires that bits [19..16] are all 1s, and for
A/R profile also that bits [7..0] are all 0s. At this point in the
decode tree we have checked all of the rest of the instruction but
were allowing these to be any value. If these bits are not set then
the result is architecturally UNPREDICTABLE, but choosing to UNDEF is
more helpful to the user and avoids unexpected odd behaviour if the
encodings are used for some purpose in future architecture versions.
Backports commit 3d54026fb06d1aea7ebb4e9825970b06bebcacac from qemu
M profile doesn't have the MSR(banked) and MRS(banked) instructions
and uses the encodings for different kinds of M-profile MRS/MSR.
Guard the relevant bits of the decode logic to make sure we don't
accidentally fall into them by accident on M-profile.
(The bit being checked for this (bit 5) is part of the SYSm field on
M-profile, but since no currently allocated system registers have
encodings with bit 5 of SYSm set, this hasn't been a problem in
practice.)
Backports commit 43ac65742319ef5ac4461daf43316b189cd21e89 from qemu
M profile doesn't have the HVC or SMC encodings, so make them always
UNDEF rather than generating calls to helper functions that assume
A/R profile.
Backports commit 001b3cab51ebfcb13e8dd03ea25bfa3bd0c517a3 from qemu
The 'name' parameter to memory_region_init_* had been marked as debug
only, however vmstate_region_ram uses it as a parameter to
qemu_ram_set_idstr to set RAMBlock names and these form part of the
migration stream.
Backports commit e8f5fe2de125a0bfbefbaa6a69af81f4817cb7a0 from qemu
The power state spec section 5.1.5 AFFINITY_INFO defines the
affinity info return values as
0 ON
1 OFF
2 ON_PENDING
I grepped QEMU for power_state to ensure that no assumptions
of OFF=0 were being made.
Backports commit d5affb0d8677e1a8a8fe03fa25005b669e7cdc02 from qemu
In armv8, this register implements more than a single bit, with
fine-grained enables for read access to event counters, cycles
counters, and write access to the software increment. This change
implements those checks using custom access functions for the relevant
registers.
Backports commit 6ecd0b6ba0591ef280ed984103924d4bdca5ac32 from qemu
glibc blacklists TSX on Haswell CPUs with model==60 and
stepping < 4. To make the Haswell CPU model more useful, make
those guests actually use TSX by changing CPU stepping to 4.
References:
* glibc commit 2702856bf45c82cf8e69f2064f5aa15c0ceb6359
https://sourceware.org/git/?p=glibc.git;a=commit;h=2702856bf45c82cf8e69f2064f5aa15c0ceb6359
Backports commit ec56a4a7b07e2943f49da273a31e3195083b1f2e from qemu
Helper function for code that needs to check the host CPU
vendor/family/model/stepping values.
Backports commit 20271d484069f154fb262507e63adc3a37e885d2 from qemu
..just like the rest of the displayed ESR register. Otherwise people
might scratch their heads if a not obviously hex number is displayed
for the EC field.
Backports commit 6568da459b611845ef55526cd23afc9fa9f4647f from qemu
Paths through the softmmu code during code generation now need to be audited
to check for double locking of tb_lock. In particular, VMEXIT can take tb_lock
through cpu_vmexit -> cpu_x86_update_cr4 -> tlb_flush.
To avoid this, split VMEXIT delivery in two parts, similar to what is done with
exceptions. cpu_vmexit only records the VMEXIT exit code and information, and
cc->do_interrupt can then deliver it when it is safe to take the lock.
Backports commit 10cde894b63146139f981857e4eedf756fa53dcb from qemu
The translation code uses cpu_ld*_code which can trigger a tlb_fill
which if it fails will erroneously attempts a fault resolution. This
never works during translation as the TB being generated hasn't been
added yet. The target should have checked retaddr before calling
cpu_restore_state but for those that have yet to be fixed we do it
here to avoid a recursive tb_lock() under MTTCG's new locking regime
Backports commit d8b2239bcd8872a5c5f7534d1658fc2365caab2d from qemu
This suppresses the incorrect warning when forcing MTTCG for x86
guests on x86 hosts. A future patch will still warn when
TARGET_SUPPORT_MTTCG hasn't been defined for the guest (which is still
pending for x86).
Backports commit 72c1701f62e8d44eb24a0583a958edc280105455 from qemu
Fix the design flaw demonstrated in the previous commit: new method
check_list() lets input visitors report that unvisited input remains
for a list, exactly like check_struct() lets them report that
unvisited input remains for a struct or union.
Implement the method for the qobject input visitor (straightforward),
and the string input visitor (less so, due to the magic list syntax
there). The opts visitor's list magic is even more impenetrable, and
all I can do there today is a stub with a FIXME comment. No worse
than before.
Backports commit a4a1c70dc759e5b81627e96564f344ab43ea86eb from qemu
The split between tests/test-qobject-input-visitor.c and
tests/test-qobject-input-strict.c now makes less sense than ever. The
next commit will take care of that.
Backports commit 048abb7b20c9f822ad9d4b730bade73b3311a47a from qemu
Commit 240f64b made all qobject input visitors created outside tests
strict, except for the one in object_property_set_qobject(). That one
was left behind only because Eric couldn't spare the time to figure
out whether making it strict would break anything, with a TODO
comment. Time to resolve it.
Strict makes a difference only for otherwise successful visits of QAPI
structs or unions. Let's examine what the callers of
object_property_set_qobject() visit:
* object_property_set_str(), object_property_set_bool(),
object_property_set_int() visit a QString, QBool, QInt,
respectively. Strictness can't matter.
* qmp_qom_set visits its @value argument. Comes straight from QMP and
can be anything ('any' in the QAPI schema). Strictness matters when
the property's set() method visits a struct or union QAPI type.
No such methods exist, thus switching to strict can't break
anything.
If we acquire such methods in the future, we'll *want* the visitor
to be strict, so that unexpected members get rejected as they should
be.
Switch to strict.
Backports commit 05601ed2de60df0e344d6b783a6bc0c1ff2b5d1f from qemu
The string input visitor tries to cope with null input. Null input
isn't used anywhere, and isn't covered by tests. Unsurprisingly, it
doesn't fully work: start_list() crashes because it passes the input
via parse_str() to strtoll() unchecked.
Make string_input_visitor_new() assert its argument isn't null, and
drop the code trying to deal with null input.
The opts visitor crashes when you try to actually visit something with
null input. Make opts_visitor_new() assert its argument isn't null,
mostly for clarity.
qobject_input_visitor_new() already asserts its argument isn't null.
Backports commit f332e830e38b3ff3953ef02ac04e409ae53769c5 from qemu
visit_optional() is to be called only between visit_start_struct() and
visit_end_struct(). Visitors that don't support struct visits,
i.e. don't implement start_struct(), end_struct(), have no use for it.
Clarify documentation.
The string input visitor doesn't support struct visits. Its
parse_optional() is therefore useless. Drop it.
Backports commit a8aec6de2ac1a5e36989fdfba29067b361009b75 from qemu
Error messages refer to nodes of the QObject being visited by name.
Trouble is the names are sometimes less than helpful:
* The name of the root QObject is whatever @name argument got passed
to the visitor, except NULL gets mapped to "null". We commonly pass
NULL. Not good.
Avoiding errors "at the root" mitigates. For instance,
visit_start_struct() can only fail when the visited object is not a
dictionary, and we commonly ensure it is beforehand.
* The name of a QDict's member is the member key. Good enough only
when this happens to be unique.
* The name of a QList's member is "null". Not good.
Improve error messages by referring to nodes by path instead, as
follows:
* The path of the root QObject is whatever @name argument got passed
to the visitor, except NULL gets mapped to "<anonymous>".
* The path of a root QDict's member is the member key.
* The path of a root QList's member is "[%u]", where %u is the list
index, starting at zero.
* The path of a non-root QDict's member is the path of the QDict
concatenated with "." and the member key.
* The path of a non-root QList's member is the path of the QList
concatenated with "[%u]", where %u is the list index.
For example, the incorrect QMP command
{ "execute": "blockdev-add", "arguments": { "node-name": "foo", "driver": "raw", "file": {"driver": "file" } } }
now fails with
{"error": {"class": "GenericError", "desc": "Parameter 'file.filename' is missing"}}
instead of
{"error": {"class": "GenericError", "desc": "Parameter 'filename' is missing"}}
and
{ "execute": "input-send-event", "arguments": { "device": "bar", "events": [ [] ] } }
now fails with
{"error": {"class": "GenericError", "desc": "Invalid parameter type for 'events[0]', expected: object"}}
instead of
{"error": {"class": "GenericError", "desc": "Invalid parameter type for 'null', expected: QDict"}}
Aside: calling the thing "parameter" is suboptimal for QMP, because
the root object is "arguments" there.
The qobject output visitor doesn't have this problem because it should
not fail. Same for dealloc and clone visitors.
The string visitors don't have this problem because they visit just
one value, whose name needs to be passed to the visitor as @name. The
string output visitor shouldn't fail anyway.
The options visitor uses QemuOpts names. Their name space is flat, so
the use of QDict member keys as names is fine. NULL names used with
roots and lists could conceivably result in bad error messages. Left
for another day.
Backports commit a9fc37f6bc3f2ab90585cb16493da9f6dcfbfbcf from qemu
qobject_input_start_struct() sets *list, except when it fails because
qobject_input_get_object() fails, i.e. the input object doesn't exist.
All the other input visitor start_struct(), start_list(),
start_alternate() always set *obj / *list.
Change qobject_input_start_struct() to match.
Backports commit 58561c27669ddf1c6d39ff8ce25837c6f2d9d92c from qemu
The QObject input visitor has three error message formats:
* Parameter '%s' is missing
* "Invalid parameter type for '%s', expected: %s"
* "QMP input object member '%s' is unexpected"
The '%s' are member names (or "null", but I'll fix that later).
The last error message calls the thing "QMP input object member"
instead of "parameter". Misleading when the visitor is used on
QObjects that don't come from QMP. Change it to "Parameter '%s' is
unexpected".
Backports commit 910f738b851a263396fc85b2052e47f884ffead3 from qemu
The QERR_ macros are leftovers from the days of "rich" error objects.
QERR_QMP_BAD_INPUT_OBJECT, QERR_QMP_BAD_INPUT_OBJECT_MEMBER,
QERR_QMP_EXTRA_MEMBER are used in just one place now, except for one
use that has crept into qobject-input-visitor.c.
Drop these macros, to make the (bad) error messages more visible.
Backports commit 99fb0c53c038105bae68b02a3d9f1cbf7951ba10 from qemu
At the moment ram device's memory regions are DEVICE_NATIVE_ENDIAN. It's
incorrect. This memory region is backed by a MMIO area in host, so the
uint64_t data that MemoryRegionOps read from/write to this area should be
host-endian rather than target-endian. Hence, current code does not work
when target and host endianness are different which is the most common case
on PPC64. To fix it, this introduces DEVICE_HOST_ENDIAN for the ram device.
This has been tested on PPC64 BE/LE host/guest in all possible combinations
including TCG.
Backports commit c99a29e702528698c0ce2590f06ca7ff239f7c39 from qemu
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.
Backports commit 55ac0a9bf4e1b1adfc7d73586a7aa085f58c9851 from qemu
CPU runnability checks and CPU model expansion have slightly
different requirements. Document the steps involved in loading a
CPU model and realizing a CPU, so their requirements and purpose
are clearly defined.
This patch doesn't change any implementation. It just add
comments, rename the x86_cpu_load_features() function for clarity
(so it won't be confused with x86_cpu_load_def()), and move
x86_cpu_filter_features() closer to it.
Backports commit b8d834a00fa3ed4dad7d371e1a00938a126a54a0 from qemu
To fix the following warnings:
In file included from /users/pranith/qemu/tcg/tcg.c:255:
/users/pranith/qemu/tcg/aarch64/tcg-target.inc.c:879:24: warning: implicit conversion from enumeration type 'TCGMemOp' (aka 'enum TCGMemOp') to different enumeration type 'TCGType' (aka 'enum TCGType')
[-Wenum-conversion]
tcg_out_cmp(s, ext, a, b, b_const);
~~~~~~~~~~~ ^~~
/users/pranith/qemu/tcg/aarch64/tcg-target.inc.c:893:36: warning: implicit conversion from enumeration type 'TCGMemOp' (aka 'enum TCGMemOp') to different enumeration type 'TCGType' (aka 'enum TCGType')
[-Wenum-conversion]
tcg_out_insn(s, 3201, CBZ, ext, a, offset);
~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~
/users/pranith/qemu/tcg/aarch64/tcg-target.inc.c:389:65: note: expanded from macro 'tcg_out_insn'
glue(tcg_out_insn_,FMT)(S, glue(glue(glue(I,FMT),_),OP), ## __VA_ARGS__)
^
/users/pranith/qemu/tcg/aarch64/tcg-target.inc.c:895:37: warning: implicit conversion from enumeration type 'TCGMemOp' (aka 'enum TCGMemOp') to different enumeration type 'TCGType' (aka 'enum TCGType')
[-Wenum-conversion]
tcg_out_insn(s, 3201, CBNZ, ext, a, offset);
~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~
/users/pranith/qemu/tcg/aarch64/tcg-target.inc.c:389:65: note: expanded from macro 'tcg_out_insn'
glue(tcg_out_insn_,FMT)(S, glue(glue(glue(I,FMT),_),OP), ## __VA_ARGS__)
^
/users/pranith/qemu/tcg/aarch64/tcg-target.inc.c:1610:27: warning: implicit conversion from enumeration type 'TCGType' (aka 'enum TCGType') to different enumeration type 'TCGMemOp' (aka 'enum TCGMemOp')
[-Wenum-conversion]
tcg_out_brcond(s, ext, a2, a0, a1, const_args[1], arg_label(args[3]));
~~~~~~~~~~~~~~ ^~~
backports commit dc1eccd661ada3b746ca4438e444993c36a0f04f from qemu
In float64_to_uint64_round_to_zero() a typo meant that we were
taking the uint64_t return value from float64_to_uint64() and
putting it into an int64_t variable before returning it as
uint64_t again. Use uint64_t instead of pointlessly casting it
back and forth to int64_t.
Backports commit d000b477f2693dbca97cd8ea751c2e0b71890662 from qemu
In get_page_addr_code(), if the guest PC doesn't correspond to RAM
then we currently run the CPU's do_unassigned_access() hook if it has
one, and otherwise we give up and exit QEMU with a more-or-less
useful message. This code assumes that the do_unassigned_access hook
will never return, because if it does then we'll plough on attempting
to use a non-RAM TLB entry to get a RAM address and will abort() in
qemu_ram_addr_from_host_nofail(). Unfortunately some CPU
implementations of this hook do return: Microblaze, SPARC and the ARM
v7M.
Change the code to call report_bad_exec() if the hook returns, as
well as if it didn't have one. This means we can tidy it up to use
the cpu_unassigned_access() function which wraps the "get the CPU
class and call the hook if it has one" work, since we aren't trying
to distinguish "no hook" from "hook existed and returned" any more.
This brings the handling of this hook into line with the handling
used for data accesses, where "hook returned" is treated the
same as "no hook existed" and gets you the default behaviour.
Backports commit 44d7ce0ef39cb45e13d384574d79799eb3d39834 from qemu
The aarch64 crypto instructions for AES and SHA are missing the
check for if the FPU is enabled.
Backports commit a4f5c5b72380deeccd53a6890ea3782f10ca8054 from qemu
This enables the multi-threaded system emulation by default for ARMv7
and ARMv8 guests using the x86_64 TCG backend. This is because on the
guest side:
- The ARM translate.c/translate-64.c have been converted to
- use MTTCG safe atomic primitives
- emit the appropriate barrier ops
- The ARM machine has been updated to
- hold the BQL when modifying shared cross-vCPU state
- defer powerctl changes to async safe work
All the host backends support the barrier and atomic primitives but
need to provide same-or-better support for normal load/store
operations.
Backports commit ca759f9e387db87e1719911f019bc60c74be9ed8 from qemu
The WFE and YIELD instructions are really only hints and in TCG's case
they were useful to move the scheduling on from one vCPU to the next. In
the parallel context (MTTCG) this just causes an unnecessary cpu_exit
and contention of the BQL.
Backports commit c22edfebff29f63d793032e4fbd42a035bb73e27 from qemu
This moves the helper function closer to where it is called and updates
the error message to report via error_report instead of the deprecated
fprintf.
Backports commit 857baec1d9e80947f0c1007c3a3d2331d62b4b53 from qemu
While the vargs approach was flexible the original MTTCG ended up
having munge the bits to a bitmap so the data could be used in
deferred work helpers. Instead of hiding that in cputlb we push the
change to the API to make it take a bitmap of MMU indexes instead.
For ARM some the resulting flushes end up being quite long so to aid
readability I've tended to move the index shifting to a new line so
all the bits being or-ed together line up nicely, for example:
tlb_flush_page_by_mmuidx(other_cs, pageaddr,
(1 << ARMMMUIdx_S1SE1) |
(1 << ARMMMUIdx_S1SE0));
Backports commit 0336cbf8532935d8e23c2aabf3e2ce2c0697b6ac from qemu
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.
Backports relevant parts of commit 08e73c48b053566bfe0c994f154f73991cd0ff0e from qemu
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.
Backports relevant parts of commit 372579427a5040a26dfee78464b50e2bdf27ef26
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.
Backports commit e5143e30fb87fbf179029387f83f98a5a9b27f19 from qemu
..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.
Backports commit 791158d93b27f22a17c2ada06621831d54f09a2c from qemu
Also atomically sets the unicorn equivalents
We know there will be cases where MTTCG won't work until additional work
is done in the front/back ends to support. It will however be useful to
be able to turn it on.
As a result MTTCG will default to off unless the combination is
supported. However the user can turn it on for the sake of testing.
Backports commit 8d4e9146b3568022ea5730d92841345d41275d66 from qemu
We'll be using the memory ordering definitions to define values for
both the host and guest. To avoid fighting with circular header
dependencies just move these types into their own minimal header.
Backports commit 20937143145b8f5a4194e5c407731ba38797864e from qemu
Enable tcg lock debug asserts in a debug build by default instead of
relying on DEBUG_LOCKING. None of the other DEBUG_* macros have
asserts, so this patch removes DEBUG_LOCKING and enable these asserts
in a debug build.
Backports commit 6ac3d7e845549f08473f020c1c70f14b8911a67e from qemu
This makes qemu_strtosz(), qemu_strtosz_mebi() and
qemu_strtosz_metric() similar to qemu_strtoi64(), except negative
values are rejected.
Backports commit f17fd4fdf0df3d2f3444399d04c38d22b9a3e1b7 from qemu
Change the qemu_strtosz() & friends to return -EINVAL when @endptr is
null and the conversion doesn't consume the string completely.
Matches how qemu_strtol() & friends work.
Only test_qemu_strtosz_simple() passes a null @endptr. No functional
change there, because its conversion consumes the string.
Simplify callers that use @endptr only to fail when it doesn't point
to '\0' to pass a null @endptr instead.
Backports commit 4fcdf65ae2c00ae69f7625f26ed41f37d77b403c from qemu
Writing QEMU_STRTOSZ_DEFSUFFIX_* instead of '*' gains nothing. Get
rid of these eyesores.
Backports commit 17f942560e54f8ee72996bc3276c697503606d7b from qemu
