The RISC-V vector instruction set utilizes the LMUL field to group
multiple registers, enabling variable-length vector registers. This
implementation uses only the first register number of each group while
reserving the other register numbers within the group.
In TCG, each VEC_IR can have 3 types (TCG_TYPE_V64/128/256), and the
host runtime needs to adjust LMUL based on the type to use different
register groups.
This presents challenges for TCG's register allocation. Currently, we
avoid modifying the register allocation part of TCG and only expose the
minimum number of vector registers.
For example, when the host vlen is 64 bits and type is TCG_TYPE_V256, with
LMUL equal to 4, we use 4 vector registers as one register group. We can
use a maximum of 8 register groups, but the V0 register number is reserved
as a mask register, so we can effectively use at most 7 register groups.
Moreover, when type is smaller than TCG_TYPE_V256, only 7 registers are
forced to be used. This is because TCG cannot yet dynamically constrain
registers with type; likewise, when the host vlen is 128 bits and
TCG_TYPE_V256, we can use at most 15 registers.
There is not much pressure on vector register allocation in TCG now, so
using 7 registers is feasible and will not have a major impact on code
generation.
This patch:
1. Reserves vector register 0 for use as a mask register.
2. When using register groups, reserves the additional registers within
each group.
Signed-off-by: Huang Shiyuan <swung0x48@outlook.com>
Co-authored-by: TANG Tiancheng <tangtiancheng.ttc@alibaba-inc.com>
Signed-off-by: TANG Tiancheng <tangtiancheng.ttc@alibaba-inc.com>
Reviewed-by: Liu Zhiwei <zhiwei_liu@linux.alibaba.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20241007025700.47259-3-zhiwei_liu@linux.alibaba.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
The DEBUG_PLUGIN_GEN_OPS ifdef is replaced with "-d op_plugin".
The second pr_ops call can be obtained with "-d op".
Reviewed-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Since we no longer emit plugin helpers during the initial code
translation phase, we don't need to specially mark plugin helpers.
Reviewed-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Introduce a new plugin_mem_cb op to hold the address temp
and meminfo computed by tcg-op-ldst.c. Because this now
has its own opcode, we no longer need PLUGIN_GEN_FROM_MEM.
Reviewed-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Introduce a new plugin_cb op and migrate one operation.
By using emit_before_op, we do not need to emit opcodes
early and modify them later -- we can simply emit the
final set of opcodes once.
Reviewed-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
For normal helpers, read the function pointer from the
structure earlier. For plugins, this will allow the
function pointer to come from elsewhere.
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Add the enumerators, adjust the helpers to match, and dump.
Not supported anywhere else just yet.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Move all of it into accel/tcg/monitor.c. This puts everything
about tcg that is only used by the monitor in the same place.
Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
The two concrete type functions already existed, merely needing
a bit of hardening to invalid inputs.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Begin disconnecting CONFIG_SOFTMMU from !CONFIG_USER_ONLY.
Introduce a variable which can be set at startup to select
one method or another for user-only.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
The tcg/tcg.h header is a big bucket, containing stuff related to
the translators and the JIT backend. The places that initialize
tcg or create new threads do not need all of that, so split out
these three functions to a new header.
Suggested-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Now that there is no padding between CPUNegativeOffsetState
and CPUArchState, this value is constant across all targets.
Reviewed-by: Anton Johansson <anjo@rev.ng>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
This finally paves the way for tcg/ to be built once per mode.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Create tcg/tcg-op-gvec-common.h, moving everything that does not
concern TARGET_LONG_BITS. Adjust tcg-op-gvec.c to use the new header.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
The replacement isn't ideal, as the raw count of bits
is not easily synced with exec/cpu-all.h, but it does
remove from tcg.h the target dependency on TARGET_PAGE_BITS_MIN
which is built into TLB_FLAGS_MASK.
Reviewed-by: Anton Johansson <anjo@rev.ng>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
This will enable replacement of TARGET_INSN_START_WORDS in tcg.c.
Split out "tcg/insn-start-words.h" and use it in target/.
Reviewed-by: Anton Johansson <anjo@rev.ng>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
