Hi,
this should be the last major bit of the TriCore integer instructions. The
floating point ones are a whole
different story. These patches depend on my other TriCore patches
(https://patchwork.ozlabs.org/patch/438866/)
and add the promised mac instructions for subtract.
Note here that msub.q behaves a bit strange, e.g. a - (((b * c) << n) >> size),
with size = size of b and c.
So the result is only using the upper half of (a * b) << n. However if the lower
half contains a result > 0 this
is rounded up to 1 and added to the upper half, since everything is in q31
format.
Also this patchset adds the remaining instructions of the RRRR, RRRW and SYS
format. Note here that I only implemented
non trap instructions, since I'm planing to do trap handling in another patch.
Cheers,
Bastian
Bastian Koppelmann (6):
target-tricore: Add instructions of RRR1 opcode format, which have
0xa3 as first opcode
target-tricore: Add instructions of RRR1 opcode format, which have
0x63 as first opcode
target-tricore: Add instructions of RRR1 opcode format, which have
0xe3 as first opcode
target-tricore: Add instructions of RRRR opcode format
target-tricore: Add instructions of RRRW opcode format
target-tricore: Add instructions of SYS opcode format
target-tricore/cpu.h | 7 +
target-tricore/helper.h | 12 +
target-tricore/op_helper.c | 436 +++
target-tricore/translate.c | 6765 +++++++++++++++++++++++---------------
target-tricore/tricore-opcodes.h | 56 +-
5 files changed, 4525 insertions(+), 2751 deletions(-)