Now that the movcond TCG op is available, it's possible to replace
shl and shr helpers by TCG code. The code generated by TCG is slightly
longer than the code generated by GCC for the helper but is still worth
it as this avoid all the consequences of using an helper: globals saved
back to memory, no possible optimization, call overhead, etc.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Now that the setcond TCG op is available, it's possible to replace
add_cc and sub_cc helpers by TCG code. The code generated by TCG is
actually very close to the one generated by GCC for the helper, and
this avoid all the consequences of using an helper: globals saved back
to memory, no possible optimization, call overhead, etc.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Use globals for CC flags instead of loading/storing them each they are
accessed. This allows some optimizations to be performed by the TCG
optimization passes.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reinstate the display of VFP registers in cpu_dump_state(), if
the CPU has them (this code had been #if 0'd out a for a long time).
We drop the attempt ot display the values as floating point, since
this makes assumptions about the host 'float' and 'double' formats
and is not done by eg the i386 cpu_dump_state().
This display is gated on the CPU_DUMP_FPU flag, as for x86.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
For all targets that currently call tcg_gen_debug_insn_start,
add CPU_LOG_TB_OP_OPT to the condition that gates it.
This is useful for comparing optimization dumps, when the
pre-optimization dump is merely noise.
Signed-off-by: Richard Henderson <rth@twiddle.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Convert code load functions and switch to AREG0 free mode.
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Convert remaining helpers to AREG0 free mode: add an explicit
CPUState parameter instead of relying on AREG0.
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Add an explicit CPUState parameter instead of relying on AREG0.
For easier review, convert only op helpers which don't return any value.
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Report from smatch:
target-arm/helper.c:651 arm946_prbs_read(6) error:
buffer overflow 'env->cp15.c6_region' 8 <= 8
target-arm/helper.c:661 arm946_prbs_write(6) error:
buffer overflow 'env->cp15.c6_region' 8 <= 8
c7_region is an array with 8 elements, so the index must be less than 8.
Signed-off-by: Stefan Weil <sw@weilnetz.de>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Fix a variety of typos in comments in target-arm files.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Peter Crosthwaite <peter.crosthwaite@petalogix.com>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: Peter A. G. Crosthwaite <peter.crosthwaite@petalogix.com>
Reviewed-by: Andreas Färber <afaerber@suse.de>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Implement the changes to the TTBCR register required for LPAE:
* many fewer bits should be RAZ/WI
* since TTBCR changes can result in a change of ASID, we must
flush the TLB on writes to it
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Implement the different format of the PAR when long descriptor
translation tables are in use. Note that we assume that
get_phys_addr() returns a long-descriptor format DFSR value on
failure if long descriptors are in use; this added subtlety tips
the balance and makes it worth adding a comment documenting the
API to get_phys_addr().
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
In the implementation of get_phys_addr(), consistently use
target_phys_addr_t to hold the physical address rather than
uint32_t.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Under LPAE, the cp15 registers PAR, TTBR0 and TTBR1 are extended
to 64 bits, with a 64 bit (MRRC/MCRR) access path to read the
full width of the register. Add the state fields for the top
half and the 64 bit access path. Actual use of the top half of
the register will come with the addition of the long-descriptor
translation table format support.
For the PAR we also need to correct the masking applied for
32 bit writes (there are no bits reserved if LPAE is implemented)
and clear the high half when doing a 32 bit result VA-to-PA
lookup.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
LPAE extends the DBGDRAR and DBGDSAR debug registers to 64 bits; we
only implement these as dummy RAZ versions; provide dummies for
the 64 bit accesses as well.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Add implementations of the AMAIR0 and AMAIR1 LPAE
Auxiliary Memory Attribute Indirection Registers.
These are implementation defined and we choose to
implement them as RAZ/WI, matching the Cortex-A7
and Cortex-A15.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Extend feature flags to 64 bits, as we've just run out of space
in the 32 bit integer we were using for them.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Implement the privileged-execute-never (PXN) translation table bit.
It is implementation-defined whether this is implemented, so we give
it its own ARM_FEATURE_ flag. LPAE requires PXN, so add also an
LPAE feature flag and the implication logic, as a placeholder
for actually implementing LPAE at a later date.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Make target_phys_addr_t 64 bits for ARM targets, and set
TARGET_PHYS_ADDR_SPACE_BITS to 40. This should have no effect for ARM
boards where physical addresses really are 32 bits (except perhaps a
slight performance hit on 32 bit hosts for system emulation) but allows
us to implement the Large Physical Address Extensions for Cortex-A15,
which mean 40 bit physical addresses.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Fix errors in the TCG temp handling in the 64 bit coprocessor
write path: we were reusing a 32 bit temp after it had been
freed by store_reg(), and failing to free a 64 bit temp.
This bug has no visible effect at this point because there
aren't any non-NOP 64 bit registers yet; it needs to be fixed
as a prerequisite for the 64 bit registers in LPAE support.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Fix a couple of cases where cp register names were copy-and-pasted.
These are harmless since we don't use the name for anything (except
debugging convenience) but could be confusing.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Igor Mitsyanko <i.mitsyanko@samsung.com>
Fix a copy-and-paste error in the register description for TTBR1
that meant it was a duplicate of TTBR0 rather than affecting the
correct bit of CPU state.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
The coprocessor register rework broke cp15 based WFI instructions.
We incorrectly fall through the normal register write case, which
incorrectly adds a forced block termination. We've already done
a special version of this (DISAS_WFI), so return immediately.
Signed-off-by: Paul Brook <paul@codesourcery.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
All the uses of ARM_CPUID() to vary behaviour have now been
removed, so we can delete the ARM_CPUID_* macros now.
The one exception is the TI915T/925T, because of its odd behaviour
where the MIDR value can be changed at runtime.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Andreas Färber <afaerber@suse.de>
Move the v6 optional block cache ops to the new cp15 framework.
This includes only providing them on the CPUs which implemented
them, rather than the previous blunderbuss approach of making
all MCRR instructions on all CPUs act as NOPs.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Remove the no-longer-used CPUARMState c0_cachetype field.
Although this was a constant register we had it in our
migration state. Drop this (with resulting version bump)
because for ARM currently we prefer cleaner migration
code and have not stabilised migration format yet.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Convert the MPIDR to the new cp15 register scheme.
This includes giving it its own feature bit rather
than doing a CPUID value check.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Convert cp15 crn=9 registers (mostly cache lockdown) to the new scheme.
Note that this change makes OMAPCP cores RAZ/WI the whole c9 space. This is
a change from previous behaviour, but a return to the behaviour of commit
c3d2689d when OMAP1 support was first added -- subsequent commits have
clearly accidentally relegated the OMAPCP RAZ condition to only a subset of
the crn=9 space when adding support for other cores.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Convert the cp15 crn=6 registers to the new scheme.
Note that this includes some minor tidyup: drop an unnecessary
underdecoding of op2 on OMAPCP cores, and only implement the
pre-v6 c6,c0,0,1 IFAR on the 1026 and not on the other ARMv5
cores, which didn't have it.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Convert the cp15 crn=7 registers to the new scheme.
Note that to do this we have to distinguish some registers
used on the ARM9 and ARM10 from some which are ARM1176
only. This is because the old code returned a value of 0
but always set the Z flag (by clearing env->ZF, since we
store the Z flag in CPUState inverted). This is inconsistent
with actual ARM CPU behaviour, which only sets flags for
reads to r15 and sets them based on the top bits of the result.
However it happened to work for the two common use cases for
cp15 crn=7 reads:
* On ARM9 and ARM10 the cache clean-and-test operations are
typically done with a destination of r15 so that you can do
a "loop: mrc ... ; bne loop" to keep cleaning until the cache
is finally clean; always setting the Z flag means this loop
terminates immediately
* on ARM1176 the Cache Dirty Status Register reads as zero
if the cache is dirty; returning 0 means this is correctly
implemented for QEMU
Since the new coprocessor register framework does the right
thing of always setting flags based on the returned result
for reads to r15, we need to split these up so that we can
return (1<<30) for the ARM9/ARM10 registers but 0 for the
ARM1176 one.
This allows us to remove the nasty hack which always sets Z.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
We RAZ/WI the entire block of crn=10 registers. Note that this
actually covers not just the implementation-defined TLB
lockdown registers but also a number of v7 VMSA memory
attribute registers which we would need to implement to
support TEX remap. We retain the previous QEMU behaviour
in this conversion, though.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Convert the cp15 crn=13 registers (FCSEIDR, CONTEXTIDR,
and the ARM946 Trace Process Identifier Register).
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Convert the cp15 crn=2 registers (MMU page table control,
MPU cache control) to the new scheme.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Convert the cp15 c3 register (MMU domain access control
or MPU write buffer control). NB that this is horribly
underdecoded for modern cores (should be crn=3,crm=0,
opc1=0,opc2=0) but this change preserves the existing
QEMU behaviour.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>