Arm CPUs support some subset of the granule (page) sizes 4K, 16K and
64K. The guest selects the one it wants using bits in the TCR_ELx
registers. If it tries to program these registers with a value that
is either reserved or which requests a size that the CPU does not
implement, the architecture requires that the CPU behaves as if the
field was programmed to some size that has been implemented.
Currently we don't implement this, and instead let the guest use any
granule size, even if the CPU ID register fields say it isn't
present.
Make aa64_va_parameters() check against the supported granule size
and force use of a different one if it is not implemented.
(A subsequent commit will make ARMVAParameters use the new enum
rather than the current pair of using16k/using64k bools.)
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20221003162315.2833797-2-peter.maydell@linaro.org
For a-profile aarch64, which does not bank system registers, it takes
quite a lot of code to switch between security states. In the process,
registers such as TCR_EL{1,2} must be swapped, which in itself requires
the flushing of softmmu tlbs. Therefore it doesn't buy us anything to
separate tlbs by security state.
Retain the distinction between Stage2 and Stage2_S.
This will be important as we implement FEAT_RME, and do not wish to
add a third set of mmu indexes for Realm state.
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20221001162318.153420-11-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Updates write_scr() to allow setting SCR_EL3.EnTP2 when FEAT_SME is
implemented. SCR_EL3 being a 64-bit register, valid_mask is changed
to uint64_t and the SCR_* constants in target/arm/cpu.h are extended
to 64-bit so that masking and bitwise not (~) behave as expected.
This enables booting Linux with Trusted Firmware-A at EL3 with
"-M virt,secure=on -cpu max".
Cc: qemu-stable@nongnu.org
Fixes: 78cb977666 ("target/arm: Enable SME for -cpu max")
Signed-off-by: Jerome Forissier <jerome.forissier@linaro.org>
Reviewed-by: Andre Przywara <andre.przywara@arm.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20221004072354.27037-1-jerome.forissier@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Our SDCR_VALID_MASK doesn't include all of the bits which are defined
by the current architecture. In particular in commit 0b42f4fab9 we
forgot to add SCCD, which meant that an AArch32 guest couldn't
actually use the SCCD bit to disable counting in Secure state.
Add all the currently defined bits; we don't implement all of them,
but this makes them be reads-as-written, which is architecturally
valid and matches how we currently handle most of the others in the
mask.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220923123412.1214041-4-peter.maydell@linaro.org
With FEAT_PMUv3p5, the event counters are now 64 bit, rather than 32
bit. (Previously, only the cycle counter could be 64 bit, and other
event counters were always 32 bits). For any given event counter,
whether the overflow event is noted for overflow from bit 31 or from
bit 63 is controlled by a combination of PMCR.LP, MDCR_EL2.HLP and
MDCR_EL2.HPMN.
Implement the 64-bit event counter handling. We choose to make our
counters always 64 bits, and mask out the top 32 bits on read or
write of PMXEVCNTR for CPUs which don't have FEAT_PMUv3p5.
(Note that the changes to pmenvcntr_op_start() and
pmenvcntr_op_finish() bring their logic closer into line with that of
pmccntr_op_start() and pmccntr_op_finish(), which already had to cope
with the overflow being either at 32 or 64 bits.)
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220822132358.3524971-10-peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
FEAT_PMUv3p5 introduces new bits which disable the cycle
counter from counting:
* MDCR_EL2.HCCD disables the counter when in EL2
* MDCR_EL3.SCCD disables the counter when Secure
Add the code to support these bits.
(Note that there is a third documented counter-disable
bit, MDCR_EL3.MCCD, which disables the counter when in
EL3. This is not present until FEAT_PMUv3p7, so is
out of scope for now.)
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220822132358.3524971-9-peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Our feature test functions that check the PMU version are named
isar_feature_{aa32,aa64,any}_pmu_8_{1,4}. This doesn't match the
current Arm ARM official feature names, which are FEAT_PMUv3p1 and
FEAT_PMUv3p4. Rename these functions to _pmuv3p1 and _pmuv3p4.
This commit was created with:
sed -i -e 's/pmu_8_/pmuv3p/g' target/arm/*.[ch]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220822132358.3524971-8-peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
In Armv8.6, a new AArch32 ID register ID_DFR1 is defined; implement
it. We don't have any CPUs with features that they need to advertise
here yet, but plumbing in the ID register gives it the right name
when debugging and will help in future when we do add a CPU that
has non-zero ID_DFR1 fields.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220819110052.2942289-5-peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
In commit 7390e0e9ab, we added support for SME loads and stores.
Unlike SVE loads and stores, these include handling of 128-bit
elements. The SME load/store functions call down into the existing
sve_cont_ldst_elements() function, which uses the element size MO_*
value as an index into the pred_esz_masks[] array. Because this code
path now has to handle MO_128, we need to add an extra element to the
array.
This bug was spotted by Coverity because it meant we were reading off
the end of the array.
Resolves: Coverity CID 1490539, 1490541, 1490543, 1490544, 1490545,
1490546, 1490548, 1490549, 1490550, 1490551, 1490555, 1490557,
1490558, 1490560, 1490561, 1490563
Fixes: 7390e0e9ab ("target/arm: Implement SME LD1, ST1")
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220718100144.3248052-1-peter.maydell@linaro.org
In regime_tcr() we return the appropriate TCR register for the
translation regime. For Secure EL2, we return the VSTCR_EL2 value,
but in this translation regime some fields that control behaviour are
in VTCR_EL2. When this code was originally written (as the comment
notes), QEMU didn't care about any of those fields, but we have since
added support for features such as LPA2 which do need the values from
those fields.
Synthesize a TCR value by merging in the relevant VTCR_EL2 fields to
the VSTCR_EL2 value.
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1103
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220714132303.1287193-8-peter.maydell@linaro.org
Change the representation of the TCR_EL* registers in the CPU state
struct from struct TCR to uint64_t. This allows us to drop the
custom vmsa_ttbcr_raw_write() function, moving the "enforce RES0"
checks to their more usual location in the writefn
vmsa_ttbcr_write(). We also don't need the resetfn any more.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220714132303.1287193-7-peter.maydell@linaro.org
The documentation for PROT_MTE says that it cannot be cleared
by mprotect. Further, the implementation of the VM_ARCH_CLEAR bit,
contains PROT_BTI confiming that bit should be cleared.
Introduce PAGE_TARGET_STICKY to allow target/arch/cpu.h to control
which bits may be reset during page_set_flags. This is sort of the
opposite of VM_ARCH_CLEAR, but works better with qemu's PAGE_* bits
that are separate from PROT_* bits.
Reported-by: Vitaly Buka <vitalybuka@google.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220711031420.17820-1-richard.henderson@linaro.org
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
This new behaviour is in the ARM pseudocode function
AArch64.CheckFPAdvSIMDEnabled, which applies to AArch32
via AArch32.CheckAdvSIMDOrFPEnabled when the EL to which
the trap would be delivered is in AArch64 mode.
Given that ARMv9 drops support for AArch32 outside EL0, the trap EL
detection ought to be trivially true, but the pseudocode still contains
a number of conditions, and QEMU has not yet committed to dropping A32
support for EL[12] when v9 features are present.
Since the computation of SME_TRAP_NONSTREAMING is necessarily different
for the two modes, we might as well preserve bits within TBFLAG_ANY and
allocate separate bits within TBFLAG_A32 and TBFLAG_A64 instead.
Note that DDI0616A.a has typos for bits [22:21] of LD1RO in the table
of instructions illegal in streaming mode.
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220708151540.18136-4-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
The architecture defines the OS DoubleLock as a register which
(similarly to the OS Lock) suppresses debug events for use in CPU
powerdown sequences. This functionality is required in Arm v7 and
v8.0; from v8.2 it becomes optional and in v9 it must not be
implemented.
Currently in QEMU we implement the OSDLR_EL1 register as a NOP. This
is wrong both for the "feature implemented" and the "feature not
implemented" cases: if the feature is implemented then the DLK bit
should read as written and cause suppression of debug exceptions, and
if it is not implemented then the bit must be RAZ/WI.
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Starting with v7 of the debug architecture, there are three extra
ID registers that add information on top of that provided in
DBGDIDR. These are DBGDEVID, DBGDEVID1 and DBGDEVID2. In the
v7 debug architecture, DBGDEVID is optional, present only of
DBGDIDR.DEVID_imp is set. In v7.1 all three must be present.
Implement the missing registers. Note that we only need to set the
values in the ARMISARegisters struct for the CPUs Cortex-A7, A15,
A53, A57 and A72 (plus the 32-bit 'max' which uses the Cortex-A53
values): earlier CPUs didn't implement v7 of the architecture, and
our other 64-bit CPUs (Cortex-A76, Neoverse-N1 and A64fx) don't have
AArch32 support at EL1.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220630194116.3438513-5-peter.maydell@linaro.org
When Streaming SVE mode is enabled, the size is taken from
SMCR_ELx instead of ZCR_ELx. The format is shared, but the
set of vector lengths is not. Further, Streaming SVE does
not require any particular length to be supported.
Adjust sve_vqm1_for_el to pass the current value of PSTATE.SM
to the new function.
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220620175235.60881-19-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Mirror the properties for SVE. The main difference is
that any arbitrary set of powers of 2 may be supported,
and not the stricter constraints that apply to SVE.
Include a property to control FEAT_SME_FA64, as failing
to restrict the runtime to the proper subset of insns
could be a major point for bugs.
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20220620175235.60881-18-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
This will be used for both Normal and Streaming SVE, and the value
does not necessarily come from ZCR_ELx. While we're at it, emphasize
the units in which the value is returned.
Patch produced by
git grep -l sve_zcr_len_for_el | \
xargs -n1 sed -i 's/sve_zcr_len_for_el/sve_vqm1_for_el/g'
and then adding a function comment.
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220607203306.657998-13-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
The bitmap need only hold 15 bits; bitmap is over-complicated.
We can simplify operations quite a bit with plain logical ops.
The introduction of SVE_VQ_POW2_MAP eliminates the need for
looping in order to search for powers of two. Simply perform
the logical ops and use count leading or trailing zeros as
required to find the result.
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220607203306.657998-12-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
The FEAT_DoubleFault extension adds the following:
* All external aborts on instruction fetches and translation table
walks for instruction fetches must be synchronous. For QEMU this
is already true.
* SCR_EL3 has a new bit NMEA which disables the masking of SError
interrupts by PSTATE.A when the SError interrupt is taken to EL3.
For QEMU we only need to make the bit writable, because we have no
sources of SError interrupts.
* SCR_EL3 has a new bit EASE which causes synchronous external
aborts taken to EL3 to be taken at the same entry point as SError.
(Note that this does not mean that they are SErrors for purposes
of PSTATE.A masking or that the syndrome register reports them as
SErrors: it just means that the vector offset is different.)
* The existing SCTLR_EL3.IESB has an effective value of 1 when
SCR_EL3.NMEA is 1. For QEMU this is a no-op because we don't need
different behaviour based on IESB (we don't need to do anything to
ensure that error exceptions are synchronized).
So for QEMU the things we need to change are:
* Make SCR_EL3.{NMEA,EASE} writable
* When taking a synchronous external abort at EL3, adjust the
vector entry point if SCR_EL3.EASE is set
* Advertise the feature in the ID registers
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220531151431.949322-1-peter.maydell@linaro.org
