GitLab

When translating, the condexec bits for the TB are in the TB flags;
the CPUState condexec bits may be different.

This patch fixes https://bugs.launchpad.net/bugs/604872

 where we might
segfault if we took an exception in the middle of a TB with an IT
block, because when we came to retranslate in cpu_restore_state()
the CPUState condexec bits would have advanced compared to the start
of the TB and we would generate different (wrong) code.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

The Thumb/ARM state for the TB being translated should come from
the TB flags, not the CPUState.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

When translating, the VFP vector length and stride for this TB are encoded
in the TB flags; the CPUState copies may be different and must not be used.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

When translating code, whether the VFP unit is enabled for this TB
is stored in a bit in the TB flags. Use this rather than incorrectly
reading the FPEXC from the CPUState passed to translation.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

Add symbolic constants for the bitfields we use in the TB flags.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

When translating the SRS instruction, handle the "store registers
to stack of current mode" case in the helper function rather than
inline. This means the generated code does not make assumptions
about the current CPU mode which might not be valid when the TB
is executed later.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

VSQRTS always uses the standard FPSCR value as it is a Neon instruction.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

Add support to the ARM helper routines for a second fp_status value
which should be used for operations which the ARM ARM indicates use
"ARM standard floating-point arithmetic" rather than being controlled
by the rounding/flush/NaN settings in the FPSCR.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

The implementation of the ARM VRSQRTS instruction (which calculates
(3 - op1 * op2) / 2) was missing the division operation. It also
did not handle the special cases of (0,inf) and (inf,0).
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

Add a utility function to softfloat to test whether a float32
is zero or denormal.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

lsi_ram_read*() and lsi_ram_write*() are not consistent, one uses
leXX_to_cpu() the other uses nothing. As the comment above the RAM
declaration says: "Script ram is stored as 32-bit words in host
byteorder.", remove the leXX_to_cpu() calls.

This fixes the boot of an ARM versatile machine on MIPS and PowerPC
hosts.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

Board id can't be written with stl_phys() as it's read-only part of
memory. Use stl_p() on the memory buffer instead.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

When a TCG variable is anded with a value and the compared with the same
value, we can simply invert the comparison and compare it with 0. The
generated code is smaller.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

As exception is not the normal path, don't bother saving PC, before
raising one, instead rely on code retranslation to get the CPU state.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

Add the ftrv XMTRX,FVn instruction, which computes the 4-row x 4-column
matrix XMTRX by the 4-dimensional vector FVn.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

Add the fipr FVm,FVn instruction, which computes the inner products of
a 4-dimensional single precision floating-point vector.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

FPU exception support where not implemented on SH4. Implement them by
clearing the softfloat exceptions flags before an FP instruction (the
SH4 FPU also clear them before an instruction), and calling a function
to update the FPSCR register after an FP instruction. This function
update the corresponding FPSCR bits (both flags and cumulative flags)
and trigger exception if enabled.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

When the FPSCR.DN bit is set, the SH4 FPU treat denormalized numbers as
zero. Enable the corresponding softfloat option when this bit is set.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

Define FPSCR constants for all field and use them instead of hardcoded
values.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

SH4 FPU doesn't propagate NaN, and instead always regenerate new ones.
Enable the default-NaN mode by default.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

When the default-NaN mode is enabled, it should return the default NaN
value, but it should anyway raise the invalid operation flag if one of
the operand is an sNaN.

I have checked that this behavior matches the ARM and SH4 manuals, as
well as real SH4 hardware.
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

We need to be able to catch exceptions correctly and thus enable softfloat
on SH4.

As all machines except i386 and x86_64 are using softfloat, make it the
default and change the case to detect i386 and x86_64. Note that CRIS
doesn't have an FPU, so it can be configured with both softfloat-native
and softfloat.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

Since commit d1807a4f

 ./configure tries
to test files and directories with "test -f", which only test for regular
files. Test with "test -e", which looks for any kind of files.

This unbreak the configure script when not using a separate object
directory.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

These are not in any release, so I am just renaming them.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>

This will help getting config.guess and config.sub from the srcdir.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>

Not used in the submodules.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>

Not necessary since we use mkdir -p and from this patch test -f.

Also, dirname returns "." if a path has no directory component,
as is the case for "sh configure".
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>

"ln -sf" does not really do anything more than "ln -s" on Solaris.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>

This also allows overriding it with SDL_CONFIG, and warning in suspicious
cross-compilation scenarios.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>

This can still be requested with PKG_CONFIG=/path/to/pkg-config.
Just do not use it as a default, and print a warning.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>

Do not hardcode the list of 64-bit CPUs.  Use sizeof(void *) to
compute it.  Renaming it to HOST_LONG_BITS to HOST_POINTER_BITS
is left for later.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>

Non-existent -I paths are dropped silently by the compiler, but still
it is not polite to pass bogus options.  Configure-time tests do not
need any include files from the source path, so only include -I flags
at make time (when they're properly expanded).
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>

The "test the C compiler works ok" comes before a bunch of flags
are added for --cpu or just depending on the host.  It helps
debugging if the test is done after these flags are (unconditionally)
added.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>

I didn't test with sparse, but the old code using += before a variable
was set was wrong.  Sparse support should probably be ripped out or
redone, but this at least keeps some sanity.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Blue Swirl <blauwirbel@gmail.com>