2021-02-08 08:46:16 +03:00
|
|
|
/*
|
2023-03-07 05:58:15 +03:00
|
|
|
* Copyright(c) 2019-2023 Qualcomm Innovation Center, Inc. All Rights Reserved.
|
2021-02-08 08:46:16 +03:00
|
|
|
*
|
|
|
|
* This program is free software; you can redistribute it and/or modify
|
|
|
|
* it under the terms of the GNU General Public License as published by
|
|
|
|
* the Free Software Foundation; either version 2 of the License, or
|
|
|
|
* (at your option) any later version.
|
|
|
|
*
|
|
|
|
* This program is distributed in the hope that it will be useful,
|
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
|
|
* GNU General Public License for more details.
|
|
|
|
*
|
|
|
|
* You should have received a copy of the GNU General Public License
|
|
|
|
* along with this program; if not, see <http://www.gnu.org/licenses/>.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include "qemu/osdep.h"
|
|
|
|
#include "cpu.h"
|
|
|
|
#include "internal.h"
|
|
|
|
#include "tcg/tcg-op.h"
|
2021-08-13 22:39:10 +03:00
|
|
|
#include "tcg/tcg-op-gvec.h"
|
2021-02-08 08:46:16 +03:00
|
|
|
#include "insn.h"
|
|
|
|
#include "opcodes.h"
|
|
|
|
#include "translate.h"
|
2021-04-09 04:07:43 +03:00
|
|
|
#define QEMU_GENERATE /* Used internally by macros.h */
|
2021-02-08 08:46:16 +03:00
|
|
|
#include "macros.h"
|
2021-08-13 22:39:10 +03:00
|
|
|
#include "mmvec/macros.h"
|
2021-04-09 04:07:43 +03:00
|
|
|
#undef QEMU_GENERATE
|
2021-02-08 08:46:16 +03:00
|
|
|
#include "gen_tcg.h"
|
2021-05-18 20:01:09 +03:00
|
|
|
#include "gen_tcg_hvx.h"
|
2022-09-23 20:38:24 +03:00
|
|
|
#include "genptr.h"
|
|
|
|
|
2022-09-23 20:38:25 +03:00
|
|
|
TCGv gen_read_reg(TCGv result, int num)
|
|
|
|
{
|
|
|
|
tcg_gen_mov_tl(result, hex_gpr[num]);
|
|
|
|
return result;
|
|
|
|
}
|
|
|
|
|
2022-09-23 20:38:24 +03:00
|
|
|
TCGv gen_read_preg(TCGv pred, uint8_t num)
|
|
|
|
{
|
|
|
|
tcg_gen_mov_tl(pred, hex_pred[num]);
|
|
|
|
return pred;
|
|
|
|
}
|
2021-02-08 08:46:16 +03:00
|
|
|
|
2023-01-05 13:23:49 +03:00
|
|
|
#define IMMUTABLE (~0)
|
|
|
|
|
Hexagon (target/hexagon) Short-circuit packet register writes
In certain cases, we can avoid the overhead of writing to hex_new_value
and write directly to hex_gpr. We add need_commit field to DisasContext
indicating if the end-of-packet commit is needed. If it is not needed,
get_result_gpr() and get_result_gpr_pair() can return hex_gpr.
We pass the ctx->need_commit to helpers when needed.
Finally, we can early-exit from gen_reg_writes during packet commit.
There are a few instructions whose semantics write to the result before
reading all the inputs. Therefore, the idef-parser generated code is
incompatible with short-circuit. We tell idef-parser to skip them.
For debugging purposes, we add a cpu property to turn off short-circuit.
When the short-circuit property is false, we skip the analysis and force
the end-of-packet commit.
Here's a simple example of the TCG generated for
0x004000b4: 0x7800c020 { R0 = #0x1 }
BEFORE:
---- 004000b4
movi_i32 new_r0,$0x1
mov_i32 r0,new_r0
AFTER:
---- 004000b4
movi_i32 r0,$0x1
This patch reintroduces a use of check_for_attrib, so we remove the
G_GNUC_UNUSED added earlier in this series.
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Brian Cain <bcain@quicinc.com>
Message-Id: <20230427230012.3800327-12-tsimpson@quicinc.com>
2023-04-28 02:00:02 +03:00
|
|
|
const target_ulong reg_immut_masks[TOTAL_PER_THREAD_REGS] = {
|
2023-01-05 13:23:49 +03:00
|
|
|
[HEX_REG_USR] = 0xc13000c0,
|
|
|
|
[HEX_REG_PC] = IMMUTABLE,
|
|
|
|
[HEX_REG_GP] = 0x3f,
|
|
|
|
[HEX_REG_UPCYCLELO] = IMMUTABLE,
|
|
|
|
[HEX_REG_UPCYCLEHI] = IMMUTABLE,
|
|
|
|
[HEX_REG_UTIMERLO] = IMMUTABLE,
|
|
|
|
[HEX_REG_UTIMERHI] = IMMUTABLE,
|
|
|
|
};
|
|
|
|
|
|
|
|
static inline void gen_masked_reg_write(TCGv new_val, TCGv cur_val,
|
|
|
|
target_ulong reg_mask)
|
|
|
|
{
|
|
|
|
if (reg_mask) {
|
|
|
|
TCGv tmp = tcg_temp_new();
|
|
|
|
|
|
|
|
/* new_val = (new_val & ~reg_mask) | (cur_val & reg_mask) */
|
|
|
|
tcg_gen_andi_tl(new_val, new_val, ~reg_mask);
|
|
|
|
tcg_gen_andi_tl(tmp, cur_val, reg_mask);
|
|
|
|
tcg_gen_or_tl(new_val, new_val, tmp);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-04-28 02:00:07 +03:00
|
|
|
TCGv get_result_gpr(DisasContext *ctx, int rnum)
|
2021-02-08 08:46:16 +03:00
|
|
|
{
|
Hexagon (target/hexagon) Short-circuit packet register writes
In certain cases, we can avoid the overhead of writing to hex_new_value
and write directly to hex_gpr. We add need_commit field to DisasContext
indicating if the end-of-packet commit is needed. If it is not needed,
get_result_gpr() and get_result_gpr_pair() can return hex_gpr.
We pass the ctx->need_commit to helpers when needed.
Finally, we can early-exit from gen_reg_writes during packet commit.
There are a few instructions whose semantics write to the result before
reading all the inputs. Therefore, the idef-parser generated code is
incompatible with short-circuit. We tell idef-parser to skip them.
For debugging purposes, we add a cpu property to turn off short-circuit.
When the short-circuit property is false, we skip the analysis and force
the end-of-packet commit.
Here's a simple example of the TCG generated for
0x004000b4: 0x7800c020 { R0 = #0x1 }
BEFORE:
---- 004000b4
movi_i32 new_r0,$0x1
mov_i32 r0,new_r0
AFTER:
---- 004000b4
movi_i32 r0,$0x1
This patch reintroduces a use of check_for_attrib, so we remove the
G_GNUC_UNUSED added earlier in this series.
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Brian Cain <bcain@quicinc.com>
Message-Id: <20230427230012.3800327-12-tsimpson@quicinc.com>
2023-04-28 02:00:02 +03:00
|
|
|
if (ctx->need_commit) {
|
2023-04-28 02:00:07 +03:00
|
|
|
if (rnum == HEX_REG_USR) {
|
|
|
|
return hex_new_value_usr;
|
|
|
|
} else {
|
2023-04-28 02:00:08 +03:00
|
|
|
if (ctx->new_value[rnum] == NULL) {
|
|
|
|
ctx->new_value[rnum] = tcg_temp_new();
|
|
|
|
tcg_gen_movi_tl(ctx->new_value[rnum], 0);
|
|
|
|
}
|
|
|
|
return ctx->new_value[rnum];
|
2023-04-28 02:00:07 +03:00
|
|
|
}
|
Hexagon (target/hexagon) Short-circuit packet register writes
In certain cases, we can avoid the overhead of writing to hex_new_value
and write directly to hex_gpr. We add need_commit field to DisasContext
indicating if the end-of-packet commit is needed. If it is not needed,
get_result_gpr() and get_result_gpr_pair() can return hex_gpr.
We pass the ctx->need_commit to helpers when needed.
Finally, we can early-exit from gen_reg_writes during packet commit.
There are a few instructions whose semantics write to the result before
reading all the inputs. Therefore, the idef-parser generated code is
incompatible with short-circuit. We tell idef-parser to skip them.
For debugging purposes, we add a cpu property to turn off short-circuit.
When the short-circuit property is false, we skip the analysis and force
the end-of-packet commit.
Here's a simple example of the TCG generated for
0x004000b4: 0x7800c020 { R0 = #0x1 }
BEFORE:
---- 004000b4
movi_i32 new_r0,$0x1
mov_i32 r0,new_r0
AFTER:
---- 004000b4
movi_i32 r0,$0x1
This patch reintroduces a use of check_for_attrib, so we remove the
G_GNUC_UNUSED added earlier in this series.
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Brian Cain <bcain@quicinc.com>
Message-Id: <20230427230012.3800327-12-tsimpson@quicinc.com>
2023-04-28 02:00:02 +03:00
|
|
|
} else {
|
|
|
|
return hex_gpr[rnum];
|
|
|
|
}
|
Hexagon (target/hexagon) Remove gen_log_predicated_reg_write[_pair]
We assign the instruction destination register to hex_new_value[num]
instead of a TCG temp that gets copied back to hex_new_value[num].
We introduce new functions get_result_gpr[_pair] to facilitate getting
the proper destination register.
Since we preload hex_new_value for predicated instructions, we don't
need the check for slot_cancelled. So, we call gen_log_reg_write instead.
We update the helper function generation and gen_tcg.h to maintain the
disable-hexagon-idef-parser configuration.
Here is a simple example of the differences in the TCG code generated:
IN:
0x00400094: 0xf900c102 { if (P0) R2 = and(R0,R1) }
BEFORE
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
mov_i32 loc2,$0x0
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 loc2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
and_i32 tmp0,slot_cancelled,$0x8
movcond_i32 new_r2,tmp0,$0x0,loc2,new_r2,eq
mov_i32 r2,new_r2
AFTER
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 new_r2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
mov_i32 r2,new_r2
We'll remove the unnecessary manipulation of slot_cancelled in a
subsequent patch.
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20230307025828.1612809-13-tsimpson@quicinc.com>
2023-03-07 05:58:26 +03:00
|
|
|
}
|
2021-02-08 08:46:16 +03:00
|
|
|
|
Hexagon (target/hexagon) Remove gen_log_predicated_reg_write[_pair]
We assign the instruction destination register to hex_new_value[num]
instead of a TCG temp that gets copied back to hex_new_value[num].
We introduce new functions get_result_gpr[_pair] to facilitate getting
the proper destination register.
Since we preload hex_new_value for predicated instructions, we don't
need the check for slot_cancelled. So, we call gen_log_reg_write instead.
We update the helper function generation and gen_tcg.h to maintain the
disable-hexagon-idef-parser configuration.
Here is a simple example of the differences in the TCG code generated:
IN:
0x00400094: 0xf900c102 { if (P0) R2 = and(R0,R1) }
BEFORE
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
mov_i32 loc2,$0x0
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 loc2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
and_i32 tmp0,slot_cancelled,$0x8
movcond_i32 new_r2,tmp0,$0x0,loc2,new_r2,eq
mov_i32 r2,new_r2
AFTER
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 new_r2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
mov_i32 r2,new_r2
We'll remove the unnecessary manipulation of slot_cancelled in a
subsequent patch.
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20230307025828.1612809-13-tsimpson@quicinc.com>
2023-03-07 05:58:26 +03:00
|
|
|
static TCGv_i64 get_result_gpr_pair(DisasContext *ctx, int rnum)
|
|
|
|
{
|
|
|
|
TCGv_i64 result = tcg_temp_new_i64();
|
Hexagon (target/hexagon) Short-circuit packet register writes
In certain cases, we can avoid the overhead of writing to hex_new_value
and write directly to hex_gpr. We add need_commit field to DisasContext
indicating if the end-of-packet commit is needed. If it is not needed,
get_result_gpr() and get_result_gpr_pair() can return hex_gpr.
We pass the ctx->need_commit to helpers when needed.
Finally, we can early-exit from gen_reg_writes during packet commit.
There are a few instructions whose semantics write to the result before
reading all the inputs. Therefore, the idef-parser generated code is
incompatible with short-circuit. We tell idef-parser to skip them.
For debugging purposes, we add a cpu property to turn off short-circuit.
When the short-circuit property is false, we skip the analysis and force
the end-of-packet commit.
Here's a simple example of the TCG generated for
0x004000b4: 0x7800c020 { R0 = #0x1 }
BEFORE:
---- 004000b4
movi_i32 new_r0,$0x1
mov_i32 r0,new_r0
AFTER:
---- 004000b4
movi_i32 r0,$0x1
This patch reintroduces a use of check_for_attrib, so we remove the
G_GNUC_UNUSED added earlier in this series.
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Brian Cain <bcain@quicinc.com>
Message-Id: <20230427230012.3800327-12-tsimpson@quicinc.com>
2023-04-28 02:00:02 +03:00
|
|
|
tcg_gen_concat_i32_i64(result, get_result_gpr(ctx, rnum),
|
|
|
|
get_result_gpr(ctx, rnum + 1));
|
Hexagon (target/hexagon) Remove gen_log_predicated_reg_write[_pair]
We assign the instruction destination register to hex_new_value[num]
instead of a TCG temp that gets copied back to hex_new_value[num].
We introduce new functions get_result_gpr[_pair] to facilitate getting
the proper destination register.
Since we preload hex_new_value for predicated instructions, we don't
need the check for slot_cancelled. So, we call gen_log_reg_write instead.
We update the helper function generation and gen_tcg.h to maintain the
disable-hexagon-idef-parser configuration.
Here is a simple example of the differences in the TCG code generated:
IN:
0x00400094: 0xf900c102 { if (P0) R2 = and(R0,R1) }
BEFORE
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
mov_i32 loc2,$0x0
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 loc2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
and_i32 tmp0,slot_cancelled,$0x8
movcond_i32 new_r2,tmp0,$0x0,loc2,new_r2,eq
mov_i32 r2,new_r2
AFTER
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 new_r2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
mov_i32 r2,new_r2
We'll remove the unnecessary manipulation of slot_cancelled in a
subsequent patch.
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20230307025828.1612809-13-tsimpson@quicinc.com>
2023-03-07 05:58:26 +03:00
|
|
|
return result;
|
2021-02-08 08:46:16 +03:00
|
|
|
}
|
|
|
|
|
2023-04-28 01:59:53 +03:00
|
|
|
void gen_log_reg_write(DisasContext *ctx, int rnum, TCGv val)
|
2021-02-08 08:46:16 +03:00
|
|
|
{
|
2023-01-05 13:23:49 +03:00
|
|
|
const target_ulong reg_mask = reg_immut_masks[rnum];
|
|
|
|
|
|
|
|
gen_masked_reg_write(val, hex_gpr[rnum], reg_mask);
|
Hexagon (target/hexagon) Short-circuit packet register writes
In certain cases, we can avoid the overhead of writing to hex_new_value
and write directly to hex_gpr. We add need_commit field to DisasContext
indicating if the end-of-packet commit is needed. If it is not needed,
get_result_gpr() and get_result_gpr_pair() can return hex_gpr.
We pass the ctx->need_commit to helpers when needed.
Finally, we can early-exit from gen_reg_writes during packet commit.
There are a few instructions whose semantics write to the result before
reading all the inputs. Therefore, the idef-parser generated code is
incompatible with short-circuit. We tell idef-parser to skip them.
For debugging purposes, we add a cpu property to turn off short-circuit.
When the short-circuit property is false, we skip the analysis and force
the end-of-packet commit.
Here's a simple example of the TCG generated for
0x004000b4: 0x7800c020 { R0 = #0x1 }
BEFORE:
---- 004000b4
movi_i32 new_r0,$0x1
mov_i32 r0,new_r0
AFTER:
---- 004000b4
movi_i32 r0,$0x1
This patch reintroduces a use of check_for_attrib, so we remove the
G_GNUC_UNUSED added earlier in this series.
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Brian Cain <bcain@quicinc.com>
Message-Id: <20230427230012.3800327-12-tsimpson@quicinc.com>
2023-04-28 02:00:02 +03:00
|
|
|
tcg_gen_mov_tl(get_result_gpr(ctx, rnum), val);
|
2021-04-09 04:07:44 +03:00
|
|
|
if (HEX_DEBUG) {
|
|
|
|
/* Do this so HELPER(debug_commit_end) will know */
|
|
|
|
tcg_gen_movi_tl(hex_reg_written[rnum], 1);
|
|
|
|
}
|
2021-02-08 08:46:16 +03:00
|
|
|
}
|
|
|
|
|
2023-04-28 01:59:53 +03:00
|
|
|
static void gen_log_reg_write_pair(DisasContext *ctx, int rnum, TCGv_i64 val)
|
2021-02-08 08:46:16 +03:00
|
|
|
{
|
2023-01-05 13:23:49 +03:00
|
|
|
TCGv val32 = tcg_temp_new();
|
|
|
|
|
2021-02-08 08:46:16 +03:00
|
|
|
/* Low word */
|
2023-01-05 13:23:49 +03:00
|
|
|
tcg_gen_extrl_i64_i32(val32, val);
|
Hexagon (target/hexagon) Short-circuit packet register writes
In certain cases, we can avoid the overhead of writing to hex_new_value
and write directly to hex_gpr. We add need_commit field to DisasContext
indicating if the end-of-packet commit is needed. If it is not needed,
get_result_gpr() and get_result_gpr_pair() can return hex_gpr.
We pass the ctx->need_commit to helpers when needed.
Finally, we can early-exit from gen_reg_writes during packet commit.
There are a few instructions whose semantics write to the result before
reading all the inputs. Therefore, the idef-parser generated code is
incompatible with short-circuit. We tell idef-parser to skip them.
For debugging purposes, we add a cpu property to turn off short-circuit.
When the short-circuit property is false, we skip the analysis and force
the end-of-packet commit.
Here's a simple example of the TCG generated for
0x004000b4: 0x7800c020 { R0 = #0x1 }
BEFORE:
---- 004000b4
movi_i32 new_r0,$0x1
mov_i32 r0,new_r0
AFTER:
---- 004000b4
movi_i32 r0,$0x1
This patch reintroduces a use of check_for_attrib, so we remove the
G_GNUC_UNUSED added earlier in this series.
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Brian Cain <bcain@quicinc.com>
Message-Id: <20230427230012.3800327-12-tsimpson@quicinc.com>
2023-04-28 02:00:02 +03:00
|
|
|
gen_log_reg_write(ctx, rnum, val32);
|
2021-02-08 08:46:16 +03:00
|
|
|
|
|
|
|
/* High word */
|
2023-01-05 13:23:49 +03:00
|
|
|
tcg_gen_extrh_i64_i32(val32, val);
|
Hexagon (target/hexagon) Short-circuit packet register writes
In certain cases, we can avoid the overhead of writing to hex_new_value
and write directly to hex_gpr. We add need_commit field to DisasContext
indicating if the end-of-packet commit is needed. If it is not needed,
get_result_gpr() and get_result_gpr_pair() can return hex_gpr.
We pass the ctx->need_commit to helpers when needed.
Finally, we can early-exit from gen_reg_writes during packet commit.
There are a few instructions whose semantics write to the result before
reading all the inputs. Therefore, the idef-parser generated code is
incompatible with short-circuit. We tell idef-parser to skip them.
For debugging purposes, we add a cpu property to turn off short-circuit.
When the short-circuit property is false, we skip the analysis and force
the end-of-packet commit.
Here's a simple example of the TCG generated for
0x004000b4: 0x7800c020 { R0 = #0x1 }
BEFORE:
---- 004000b4
movi_i32 new_r0,$0x1
mov_i32 r0,new_r0
AFTER:
---- 004000b4
movi_i32 r0,$0x1
This patch reintroduces a use of check_for_attrib, so we remove the
G_GNUC_UNUSED added earlier in this series.
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Brian Cain <bcain@quicinc.com>
Message-Id: <20230427230012.3800327-12-tsimpson@quicinc.com>
2023-04-28 02:00:02 +03:00
|
|
|
gen_log_reg_write(ctx, rnum + 1, val32);
|
2021-02-08 08:46:16 +03:00
|
|
|
}
|
|
|
|
|
2023-04-28 02:00:03 +03:00
|
|
|
TCGv get_result_pred(DisasContext *ctx, int pnum)
|
|
|
|
{
|
|
|
|
if (ctx->need_commit) {
|
2023-04-28 02:00:09 +03:00
|
|
|
if (ctx->new_pred_value[pnum] == NULL) {
|
|
|
|
ctx->new_pred_value[pnum] = tcg_temp_new();
|
|
|
|
tcg_gen_movi_tl(ctx->new_pred_value[pnum], 0);
|
|
|
|
}
|
|
|
|
return ctx->new_pred_value[pnum];
|
2023-04-28 02:00:03 +03:00
|
|
|
} else {
|
|
|
|
return hex_pred[pnum];
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2022-09-23 20:38:24 +03:00
|
|
|
void gen_log_pred_write(DisasContext *ctx, int pnum, TCGv val)
|
2021-02-08 08:46:16 +03:00
|
|
|
{
|
2023-04-28 02:00:03 +03:00
|
|
|
TCGv pred = get_result_pred(ctx, pnum);
|
2021-02-08 08:46:16 +03:00
|
|
|
TCGv base_val = tcg_temp_new();
|
|
|
|
|
|
|
|
tcg_gen_andi_tl(base_val, val, 0xff);
|
2021-04-09 04:07:34 +03:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Section 6.1.3 of the Hexagon V67 Programmer's Reference Manual
|
|
|
|
*
|
|
|
|
* Multiple writes to the same preg are and'ed together
|
|
|
|
* If this is the first predicate write in the packet, do a
|
|
|
|
* straight assignment. Otherwise, do an and.
|
|
|
|
*/
|
|
|
|
if (!test_bit(pnum, ctx->pregs_written)) {
|
2023-04-28 02:00:03 +03:00
|
|
|
tcg_gen_mov_tl(pred, base_val);
|
2021-04-09 04:07:34 +03:00
|
|
|
} else {
|
2023-04-28 02:00:03 +03:00
|
|
|
tcg_gen_and_tl(pred, pred, base_val);
|
2021-04-09 04:07:34 +03:00
|
|
|
}
|
2023-04-28 01:59:59 +03:00
|
|
|
if (HEX_DEBUG) {
|
2023-04-28 02:00:10 +03:00
|
|
|
tcg_gen_ori_tl(ctx->pred_written, ctx->pred_written, 1 << pnum);
|
2023-04-28 01:59:59 +03:00
|
|
|
}
|
2023-03-07 05:58:19 +03:00
|
|
|
set_bit(pnum, ctx->pregs_written);
|
2021-02-08 08:46:16 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
static inline void gen_read_p3_0(TCGv control_reg)
|
|
|
|
{
|
|
|
|
tcg_gen_movi_tl(control_reg, 0);
|
|
|
|
for (int i = 0; i < NUM_PREGS; i++) {
|
|
|
|
tcg_gen_deposit_tl(control_reg, control_reg, hex_pred[i], i * 8, 8);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Certain control registers require special handling on read
|
2022-12-29 12:20:06 +03:00
|
|
|
* HEX_REG_P3_0_ALIASED aliased to the predicate registers
|
2021-02-08 08:46:16 +03:00
|
|
|
* -> concat the 4 predicate registers together
|
|
|
|
* HEX_REG_PC actual value stored in DisasContext
|
|
|
|
* -> assign from ctx->base.pc_next
|
|
|
|
* HEX_REG_QEMU_*_CNT changes in current TB in DisasContext
|
|
|
|
* -> add current TB changes to existing reg value
|
|
|
|
*/
|
|
|
|
static inline void gen_read_ctrl_reg(DisasContext *ctx, const int reg_num,
|
|
|
|
TCGv dest)
|
|
|
|
{
|
2022-12-29 12:20:06 +03:00
|
|
|
if (reg_num == HEX_REG_P3_0_ALIASED) {
|
2021-02-08 08:46:16 +03:00
|
|
|
gen_read_p3_0(dest);
|
|
|
|
} else if (reg_num == HEX_REG_PC) {
|
|
|
|
tcg_gen_movi_tl(dest, ctx->base.pc_next);
|
|
|
|
} else if (reg_num == HEX_REG_QEMU_PKT_CNT) {
|
|
|
|
tcg_gen_addi_tl(dest, hex_gpr[HEX_REG_QEMU_PKT_CNT],
|
|
|
|
ctx->num_packets);
|
|
|
|
} else if (reg_num == HEX_REG_QEMU_INSN_CNT) {
|
|
|
|
tcg_gen_addi_tl(dest, hex_gpr[HEX_REG_QEMU_INSN_CNT],
|
|
|
|
ctx->num_insns);
|
2021-09-30 22:29:00 +03:00
|
|
|
} else if (reg_num == HEX_REG_QEMU_HVX_CNT) {
|
|
|
|
tcg_gen_addi_tl(dest, hex_gpr[HEX_REG_QEMU_HVX_CNT],
|
|
|
|
ctx->num_hvx_insns);
|
2021-02-08 08:46:16 +03:00
|
|
|
} else {
|
|
|
|
tcg_gen_mov_tl(dest, hex_gpr[reg_num]);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void gen_read_ctrl_reg_pair(DisasContext *ctx, const int reg_num,
|
|
|
|
TCGv_i64 dest)
|
|
|
|
{
|
2022-12-29 12:20:06 +03:00
|
|
|
if (reg_num == HEX_REG_P3_0_ALIASED) {
|
2021-02-08 08:46:16 +03:00
|
|
|
TCGv p3_0 = tcg_temp_new();
|
|
|
|
gen_read_p3_0(p3_0);
|
|
|
|
tcg_gen_concat_i32_i64(dest, p3_0, hex_gpr[reg_num + 1]);
|
|
|
|
} else if (reg_num == HEX_REG_PC - 1) {
|
2021-10-03 03:47:50 +03:00
|
|
|
TCGv pc = tcg_constant_tl(ctx->base.pc_next);
|
2021-02-08 08:46:16 +03:00
|
|
|
tcg_gen_concat_i32_i64(dest, hex_gpr[reg_num], pc);
|
|
|
|
} else if (reg_num == HEX_REG_QEMU_PKT_CNT) {
|
|
|
|
TCGv pkt_cnt = tcg_temp_new();
|
|
|
|
TCGv insn_cnt = tcg_temp_new();
|
|
|
|
tcg_gen_addi_tl(pkt_cnt, hex_gpr[HEX_REG_QEMU_PKT_CNT],
|
|
|
|
ctx->num_packets);
|
|
|
|
tcg_gen_addi_tl(insn_cnt, hex_gpr[HEX_REG_QEMU_INSN_CNT],
|
|
|
|
ctx->num_insns);
|
|
|
|
tcg_gen_concat_i32_i64(dest, pkt_cnt, insn_cnt);
|
2021-09-30 22:29:00 +03:00
|
|
|
} else if (reg_num == HEX_REG_QEMU_HVX_CNT) {
|
|
|
|
TCGv hvx_cnt = tcg_temp_new();
|
|
|
|
tcg_gen_addi_tl(hvx_cnt, hex_gpr[HEX_REG_QEMU_HVX_CNT],
|
|
|
|
ctx->num_hvx_insns);
|
|
|
|
tcg_gen_concat_i32_i64(dest, hvx_cnt, hex_gpr[reg_num + 1]);
|
2021-02-08 08:46:16 +03:00
|
|
|
} else {
|
|
|
|
tcg_gen_concat_i32_i64(dest,
|
|
|
|
hex_gpr[reg_num],
|
|
|
|
hex_gpr[reg_num + 1]);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2022-02-10 05:15:56 +03:00
|
|
|
static void gen_write_p3_0(DisasContext *ctx, TCGv control_reg)
|
2021-02-08 08:46:16 +03:00
|
|
|
{
|
2022-02-10 05:15:56 +03:00
|
|
|
TCGv hex_p8 = tcg_temp_new();
|
2021-02-08 08:46:16 +03:00
|
|
|
for (int i = 0; i < NUM_PREGS; i++) {
|
2022-02-10 05:15:56 +03:00
|
|
|
tcg_gen_extract_tl(hex_p8, control_reg, i * 8, 8);
|
|
|
|
gen_log_pred_write(ctx, i, hex_p8);
|
2021-02-08 08:46:16 +03:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Certain control registers require special handling on write
|
2022-12-29 12:20:06 +03:00
|
|
|
* HEX_REG_P3_0_ALIASED aliased to the predicate registers
|
2021-02-08 08:46:16 +03:00
|
|
|
* -> break the value across 4 predicate registers
|
|
|
|
* HEX_REG_QEMU_*_CNT changes in current TB in DisasContext
|
|
|
|
* -> clear the changes
|
|
|
|
*/
|
|
|
|
static inline void gen_write_ctrl_reg(DisasContext *ctx, int reg_num,
|
|
|
|
TCGv val)
|
|
|
|
{
|
2022-12-29 12:20:06 +03:00
|
|
|
if (reg_num == HEX_REG_P3_0_ALIASED) {
|
2022-02-10 05:15:56 +03:00
|
|
|
gen_write_p3_0(ctx, val);
|
2021-02-08 08:46:16 +03:00
|
|
|
} else {
|
2023-04-28 01:59:53 +03:00
|
|
|
gen_log_reg_write(ctx, reg_num, val);
|
2021-02-08 08:46:16 +03:00
|
|
|
if (reg_num == HEX_REG_QEMU_PKT_CNT) {
|
|
|
|
ctx->num_packets = 0;
|
|
|
|
}
|
|
|
|
if (reg_num == HEX_REG_QEMU_INSN_CNT) {
|
|
|
|
ctx->num_insns = 0;
|
|
|
|
}
|
2021-09-30 22:29:00 +03:00
|
|
|
if (reg_num == HEX_REG_QEMU_HVX_CNT) {
|
|
|
|
ctx->num_hvx_insns = 0;
|
|
|
|
}
|
2021-02-08 08:46:16 +03:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void gen_write_ctrl_reg_pair(DisasContext *ctx, int reg_num,
|
|
|
|
TCGv_i64 val)
|
|
|
|
{
|
2022-12-29 12:20:06 +03:00
|
|
|
if (reg_num == HEX_REG_P3_0_ALIASED) {
|
Hexagon (target/hexagon) Remove gen_log_predicated_reg_write[_pair]
We assign the instruction destination register to hex_new_value[num]
instead of a TCG temp that gets copied back to hex_new_value[num].
We introduce new functions get_result_gpr[_pair] to facilitate getting
the proper destination register.
Since we preload hex_new_value for predicated instructions, we don't
need the check for slot_cancelled. So, we call gen_log_reg_write instead.
We update the helper function generation and gen_tcg.h to maintain the
disable-hexagon-idef-parser configuration.
Here is a simple example of the differences in the TCG code generated:
IN:
0x00400094: 0xf900c102 { if (P0) R2 = and(R0,R1) }
BEFORE
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
mov_i32 loc2,$0x0
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 loc2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
and_i32 tmp0,slot_cancelled,$0x8
movcond_i32 new_r2,tmp0,$0x0,loc2,new_r2,eq
mov_i32 r2,new_r2
AFTER
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 new_r2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
mov_i32 r2,new_r2
We'll remove the unnecessary manipulation of slot_cancelled in a
subsequent patch.
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20230307025828.1612809-13-tsimpson@quicinc.com>
2023-03-07 05:58:26 +03:00
|
|
|
TCGv result = get_result_gpr(ctx, reg_num + 1);
|
2021-02-08 08:46:16 +03:00
|
|
|
TCGv val32 = tcg_temp_new();
|
|
|
|
tcg_gen_extrl_i64_i32(val32, val);
|
2022-02-10 05:15:56 +03:00
|
|
|
gen_write_p3_0(ctx, val32);
|
2021-02-08 08:46:16 +03:00
|
|
|
tcg_gen_extrh_i64_i32(val32, val);
|
Hexagon (target/hexagon) Remove gen_log_predicated_reg_write[_pair]
We assign the instruction destination register to hex_new_value[num]
instead of a TCG temp that gets copied back to hex_new_value[num].
We introduce new functions get_result_gpr[_pair] to facilitate getting
the proper destination register.
Since we preload hex_new_value for predicated instructions, we don't
need the check for slot_cancelled. So, we call gen_log_reg_write instead.
We update the helper function generation and gen_tcg.h to maintain the
disable-hexagon-idef-parser configuration.
Here is a simple example of the differences in the TCG code generated:
IN:
0x00400094: 0xf900c102 { if (P0) R2 = and(R0,R1) }
BEFORE
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
mov_i32 loc2,$0x0
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 loc2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
and_i32 tmp0,slot_cancelled,$0x8
movcond_i32 new_r2,tmp0,$0x0,loc2,new_r2,eq
mov_i32 r2,new_r2
AFTER
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 new_r2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
mov_i32 r2,new_r2
We'll remove the unnecessary manipulation of slot_cancelled in a
subsequent patch.
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20230307025828.1612809-13-tsimpson@quicinc.com>
2023-03-07 05:58:26 +03:00
|
|
|
tcg_gen_mov_tl(result, val32);
|
2021-02-08 08:46:16 +03:00
|
|
|
} else {
|
2023-04-28 01:59:53 +03:00
|
|
|
gen_log_reg_write_pair(ctx, reg_num, val);
|
2021-02-08 08:46:16 +03:00
|
|
|
if (reg_num == HEX_REG_QEMU_PKT_CNT) {
|
|
|
|
ctx->num_packets = 0;
|
|
|
|
ctx->num_insns = 0;
|
|
|
|
}
|
2021-09-30 22:29:00 +03:00
|
|
|
if (reg_num == HEX_REG_QEMU_HVX_CNT) {
|
|
|
|
ctx->num_hvx_insns = 0;
|
|
|
|
}
|
2021-02-08 08:46:16 +03:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2022-09-23 20:38:24 +03:00
|
|
|
TCGv gen_get_byte(TCGv result, int N, TCGv src, bool sign)
|
2021-04-09 04:07:50 +03:00
|
|
|
{
|
|
|
|
if (sign) {
|
|
|
|
tcg_gen_sextract_tl(result, src, N * 8, 8);
|
|
|
|
} else {
|
|
|
|
tcg_gen_extract_tl(result, src, N * 8, 8);
|
|
|
|
}
|
|
|
|
return result;
|
|
|
|
}
|
|
|
|
|
2022-09-23 20:38:24 +03:00
|
|
|
TCGv gen_get_byte_i64(TCGv result, int N, TCGv_i64 src, bool sign)
|
2021-04-09 04:07:48 +03:00
|
|
|
{
|
|
|
|
TCGv_i64 res64 = tcg_temp_new_i64();
|
|
|
|
if (sign) {
|
|
|
|
tcg_gen_sextract_i64(res64, src, N * 8, 8);
|
|
|
|
} else {
|
|
|
|
tcg_gen_extract_i64(res64, src, N * 8, 8);
|
|
|
|
}
|
|
|
|
tcg_gen_extrl_i64_i32(result, res64);
|
|
|
|
|
|
|
|
return result;
|
|
|
|
}
|
|
|
|
|
2022-09-23 20:38:24 +03:00
|
|
|
TCGv gen_get_half(TCGv result, int N, TCGv src, bool sign)
|
2021-04-09 04:07:50 +03:00
|
|
|
{
|
|
|
|
if (sign) {
|
|
|
|
tcg_gen_sextract_tl(result, src, N * 16, 16);
|
|
|
|
} else {
|
|
|
|
tcg_gen_extract_tl(result, src, N * 16, 16);
|
|
|
|
}
|
|
|
|
return result;
|
|
|
|
}
|
|
|
|
|
2022-09-23 20:38:24 +03:00
|
|
|
void gen_set_half(int N, TCGv result, TCGv src)
|
2021-04-09 04:07:52 +03:00
|
|
|
{
|
|
|
|
tcg_gen_deposit_tl(result, result, src, N * 16, 16);
|
|
|
|
}
|
|
|
|
|
2022-09-23 20:38:24 +03:00
|
|
|
void gen_set_half_i64(int N, TCGv_i64 result, TCGv src)
|
2021-04-09 04:07:52 +03:00
|
|
|
{
|
|
|
|
TCGv_i64 src64 = tcg_temp_new_i64();
|
|
|
|
tcg_gen_extu_i32_i64(src64, src);
|
|
|
|
tcg_gen_deposit_i64(result, result, src64, N * 16, 16);
|
|
|
|
}
|
|
|
|
|
2022-09-23 20:38:24 +03:00
|
|
|
void gen_set_byte_i64(int N, TCGv_i64 result, TCGv src)
|
2021-04-09 04:07:48 +03:00
|
|
|
{
|
|
|
|
TCGv_i64 src64 = tcg_temp_new_i64();
|
|
|
|
tcg_gen_extu_i32_i64(src64, src);
|
|
|
|
tcg_gen_deposit_i64(result, result, src64, N * 8, 8);
|
|
|
|
}
|
|
|
|
|
2021-02-08 08:46:16 +03:00
|
|
|
static inline void gen_load_locked4u(TCGv dest, TCGv vaddr, int mem_index)
|
|
|
|
{
|
2023-05-02 16:57:35 +03:00
|
|
|
tcg_gen_qemu_ld_tl(dest, vaddr, mem_index, MO_TEUL);
|
2021-02-08 08:46:16 +03:00
|
|
|
tcg_gen_mov_tl(hex_llsc_addr, vaddr);
|
|
|
|
tcg_gen_mov_tl(hex_llsc_val, dest);
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void gen_load_locked8u(TCGv_i64 dest, TCGv vaddr, int mem_index)
|
|
|
|
{
|
2023-05-02 16:57:35 +03:00
|
|
|
tcg_gen_qemu_ld_i64(dest, vaddr, mem_index, MO_TEUQ);
|
2021-02-08 08:46:16 +03:00
|
|
|
tcg_gen_mov_tl(hex_llsc_addr, vaddr);
|
|
|
|
tcg_gen_mov_i64(hex_llsc_val_i64, dest);
|
|
|
|
}
|
|
|
|
|
2021-06-02 02:19:43 +03:00
|
|
|
static inline void gen_store_conditional4(DisasContext *ctx,
|
2021-02-08 08:46:16 +03:00
|
|
|
TCGv pred, TCGv vaddr, TCGv src)
|
|
|
|
{
|
|
|
|
TCGLabel *fail = gen_new_label();
|
|
|
|
TCGLabel *done = gen_new_label();
|
|
|
|
TCGv one, zero, tmp;
|
|
|
|
|
|
|
|
tcg_gen_brcond_tl(TCG_COND_NE, vaddr, hex_llsc_addr, fail);
|
|
|
|
|
2021-10-03 03:47:50 +03:00
|
|
|
one = tcg_constant_tl(0xff);
|
|
|
|
zero = tcg_constant_tl(0);
|
2021-02-08 08:46:16 +03:00
|
|
|
tmp = tcg_temp_new();
|
|
|
|
tcg_gen_atomic_cmpxchg_tl(tmp, hex_llsc_addr, hex_llsc_val, src,
|
|
|
|
ctx->mem_idx, MO_32);
|
2021-06-02 02:19:43 +03:00
|
|
|
tcg_gen_movcond_tl(TCG_COND_EQ, pred, tmp, hex_llsc_val,
|
2021-02-08 08:46:16 +03:00
|
|
|
one, zero);
|
|
|
|
tcg_gen_br(done);
|
|
|
|
|
|
|
|
gen_set_label(fail);
|
|
|
|
tcg_gen_movi_tl(pred, 0);
|
|
|
|
|
|
|
|
gen_set_label(done);
|
|
|
|
tcg_gen_movi_tl(hex_llsc_addr, ~0);
|
|
|
|
}
|
|
|
|
|
2021-06-02 02:19:43 +03:00
|
|
|
static inline void gen_store_conditional8(DisasContext *ctx,
|
2021-02-08 08:46:16 +03:00
|
|
|
TCGv pred, TCGv vaddr, TCGv_i64 src)
|
|
|
|
{
|
|
|
|
TCGLabel *fail = gen_new_label();
|
|
|
|
TCGLabel *done = gen_new_label();
|
|
|
|
TCGv_i64 one, zero, tmp;
|
|
|
|
|
|
|
|
tcg_gen_brcond_tl(TCG_COND_NE, vaddr, hex_llsc_addr, fail);
|
|
|
|
|
2021-10-03 03:47:50 +03:00
|
|
|
one = tcg_constant_i64(0xff);
|
|
|
|
zero = tcg_constant_i64(0);
|
2021-02-08 08:46:16 +03:00
|
|
|
tmp = tcg_temp_new_i64();
|
|
|
|
tcg_gen_atomic_cmpxchg_i64(tmp, hex_llsc_addr, hex_llsc_val_i64, src,
|
|
|
|
ctx->mem_idx, MO_64);
|
|
|
|
tcg_gen_movcond_i64(TCG_COND_EQ, tmp, tmp, hex_llsc_val_i64,
|
|
|
|
one, zero);
|
2021-06-02 02:19:43 +03:00
|
|
|
tcg_gen_extrl_i64_i32(pred, tmp);
|
2021-02-08 08:46:16 +03:00
|
|
|
tcg_gen_br(done);
|
|
|
|
|
|
|
|
gen_set_label(fail);
|
|
|
|
tcg_gen_movi_tl(pred, 0);
|
|
|
|
|
|
|
|
gen_set_label(done);
|
|
|
|
tcg_gen_movi_tl(hex_llsc_addr, ~0);
|
|
|
|
}
|
|
|
|
|
2023-04-28 02:00:11 +03:00
|
|
|
#ifndef CONFIG_HEXAGON_IDEF_PARSER
|
|
|
|
static TCGv gen_slotval(DisasContext *ctx)
|
|
|
|
{
|
|
|
|
int slotval = (ctx->pkt->pkt_has_store_s1 & 1) | (ctx->insn->slot << 1);
|
|
|
|
return tcg_constant_tl(slotval);
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
2022-09-23 20:38:24 +03:00
|
|
|
void gen_store32(TCGv vaddr, TCGv src, int width, uint32_t slot)
|
2021-04-09 04:07:50 +03:00
|
|
|
{
|
|
|
|
tcg_gen_mov_tl(hex_store_addr[slot], vaddr);
|
|
|
|
tcg_gen_movi_tl(hex_store_width[slot], width);
|
|
|
|
tcg_gen_mov_tl(hex_store_val32[slot], src);
|
|
|
|
}
|
|
|
|
|
2022-09-23 20:38:24 +03:00
|
|
|
void gen_store1(TCGv_env cpu_env, TCGv vaddr, TCGv src, uint32_t slot)
|
2021-04-09 04:07:50 +03:00
|
|
|
{
|
|
|
|
gen_store32(vaddr, src, 1, slot);
|
|
|
|
}
|
|
|
|
|
2022-09-23 20:38:24 +03:00
|
|
|
void gen_store1i(TCGv_env cpu_env, TCGv vaddr, int32_t src, uint32_t slot)
|
2021-04-09 04:07:50 +03:00
|
|
|
{
|
2021-10-03 03:47:50 +03:00
|
|
|
TCGv tmp = tcg_constant_tl(src);
|
2022-09-20 11:07:45 +03:00
|
|
|
gen_store1(cpu_env, vaddr, tmp, slot);
|
2021-04-09 04:07:50 +03:00
|
|
|
}
|
|
|
|
|
2022-09-23 20:38:24 +03:00
|
|
|
void gen_store2(TCGv_env cpu_env, TCGv vaddr, TCGv src, uint32_t slot)
|
2021-04-09 04:07:50 +03:00
|
|
|
{
|
|
|
|
gen_store32(vaddr, src, 2, slot);
|
|
|
|
}
|
|
|
|
|
2022-09-23 20:38:24 +03:00
|
|
|
void gen_store2i(TCGv_env cpu_env, TCGv vaddr, int32_t src, uint32_t slot)
|
2021-04-09 04:07:50 +03:00
|
|
|
{
|
2021-10-03 03:47:50 +03:00
|
|
|
TCGv tmp = tcg_constant_tl(src);
|
2022-09-20 11:07:45 +03:00
|
|
|
gen_store2(cpu_env, vaddr, tmp, slot);
|
2021-04-09 04:07:50 +03:00
|
|
|
}
|
|
|
|
|
2022-09-23 20:38:24 +03:00
|
|
|
void gen_store4(TCGv_env cpu_env, TCGv vaddr, TCGv src, uint32_t slot)
|
2021-04-09 04:07:50 +03:00
|
|
|
{
|
|
|
|
gen_store32(vaddr, src, 4, slot);
|
|
|
|
}
|
|
|
|
|
2022-09-23 20:38:24 +03:00
|
|
|
void gen_store4i(TCGv_env cpu_env, TCGv vaddr, int32_t src, uint32_t slot)
|
2021-04-09 04:07:50 +03:00
|
|
|
{
|
2021-10-03 03:47:50 +03:00
|
|
|
TCGv tmp = tcg_constant_tl(src);
|
2022-09-20 11:07:45 +03:00
|
|
|
gen_store4(cpu_env, vaddr, tmp, slot);
|
2021-04-09 04:07:50 +03:00
|
|
|
}
|
|
|
|
|
2022-09-23 20:38:24 +03:00
|
|
|
void gen_store8(TCGv_env cpu_env, TCGv vaddr, TCGv_i64 src, uint32_t slot)
|
2021-04-09 04:07:50 +03:00
|
|
|
{
|
|
|
|
tcg_gen_mov_tl(hex_store_addr[slot], vaddr);
|
|
|
|
tcg_gen_movi_tl(hex_store_width[slot], 8);
|
|
|
|
tcg_gen_mov_i64(hex_store_val64[slot], src);
|
|
|
|
}
|
|
|
|
|
2022-09-23 20:38:24 +03:00
|
|
|
void gen_store8i(TCGv_env cpu_env, TCGv vaddr, int64_t src, uint32_t slot)
|
2021-04-09 04:07:50 +03:00
|
|
|
{
|
2021-10-03 03:47:50 +03:00
|
|
|
TCGv_i64 tmp = tcg_constant_i64(src);
|
2022-09-20 11:07:45 +03:00
|
|
|
gen_store8(cpu_env, vaddr, tmp, slot);
|
2021-04-09 04:07:50 +03:00
|
|
|
}
|
|
|
|
|
2022-09-23 20:38:24 +03:00
|
|
|
TCGv gen_8bitsof(TCGv result, TCGv value)
|
2021-04-09 04:07:49 +03:00
|
|
|
{
|
2021-10-03 03:47:50 +03:00
|
|
|
TCGv zero = tcg_constant_tl(0);
|
|
|
|
TCGv ones = tcg_constant_tl(0xff);
|
2021-04-09 04:07:49 +03:00
|
|
|
tcg_gen_movcond_tl(TCG_COND_NE, result, value, zero, ones, zero);
|
|
|
|
|
|
|
|
return result;
|
|
|
|
}
|
|
|
|
|
2022-11-08 19:29:02 +03:00
|
|
|
static void gen_write_new_pc_addr(DisasContext *ctx, TCGv addr,
|
|
|
|
TCGCond cond, TCGv pred)
|
|
|
|
{
|
|
|
|
TCGLabel *pred_false = NULL;
|
|
|
|
if (cond != TCG_COND_ALWAYS) {
|
|
|
|
pred_false = gen_new_label();
|
|
|
|
tcg_gen_brcondi_tl(cond, pred, 0, pred_false);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (ctx->pkt->pkt_has_multi_cof) {
|
|
|
|
/* If there are multiple branches in a packet, ignore the second one */
|
|
|
|
tcg_gen_movcond_tl(TCG_COND_NE, hex_gpr[HEX_REG_PC],
|
2023-04-28 02:00:12 +03:00
|
|
|
ctx->branch_taken, tcg_constant_tl(0),
|
2022-11-08 19:29:02 +03:00
|
|
|
hex_gpr[HEX_REG_PC], addr);
|
2023-04-28 02:00:12 +03:00
|
|
|
tcg_gen_movi_tl(ctx->branch_taken, 1);
|
2022-11-08 19:29:02 +03:00
|
|
|
} else {
|
|
|
|
tcg_gen_mov_tl(hex_gpr[HEX_REG_PC], addr);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (cond != TCG_COND_ALWAYS) {
|
|
|
|
gen_set_label(pred_false);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void gen_write_new_pc_pcrel(DisasContext *ctx, int pc_off,
|
|
|
|
TCGCond cond, TCGv pred)
|
|
|
|
{
|
|
|
|
target_ulong dest = ctx->pkt->pc + pc_off;
|
2022-11-08 19:29:05 +03:00
|
|
|
if (ctx->pkt->pkt_has_multi_cof) {
|
|
|
|
gen_write_new_pc_addr(ctx, tcg_constant_tl(dest), cond, pred);
|
|
|
|
} else {
|
|
|
|
/* Defer this jump to the end of the TB */
|
|
|
|
ctx->branch_cond = TCG_COND_ALWAYS;
|
|
|
|
if (pred != NULL) {
|
|
|
|
ctx->branch_cond = cond;
|
2023-04-28 02:00:12 +03:00
|
|
|
tcg_gen_mov_tl(ctx->branch_taken, pred);
|
2022-11-08 19:29:05 +03:00
|
|
|
}
|
|
|
|
ctx->branch_dest = dest;
|
|
|
|
}
|
2022-11-08 19:29:02 +03:00
|
|
|
}
|
|
|
|
|
2023-04-05 19:42:11 +03:00
|
|
|
void gen_set_usr_field(DisasContext *ctx, int field, TCGv val)
|
2022-11-10 20:49:35 +03:00
|
|
|
{
|
2023-04-05 19:42:11 +03:00
|
|
|
TCGv usr = get_result_gpr(ctx, HEX_REG_USR);
|
|
|
|
tcg_gen_deposit_tl(usr, usr, val,
|
2022-11-10 20:49:35 +03:00
|
|
|
reg_field_info[field].offset,
|
|
|
|
reg_field_info[field].width);
|
|
|
|
}
|
|
|
|
|
2023-04-05 19:42:11 +03:00
|
|
|
void gen_set_usr_fieldi(DisasContext *ctx, int field, int x)
|
2022-11-10 20:49:35 +03:00
|
|
|
{
|
|
|
|
if (reg_field_info[field].width == 1) {
|
2023-04-05 19:42:11 +03:00
|
|
|
TCGv usr = get_result_gpr(ctx, HEX_REG_USR);
|
2022-11-10 20:49:35 +03:00
|
|
|
target_ulong bit = 1 << reg_field_info[field].offset;
|
|
|
|
if ((x & 1) == 1) {
|
2023-04-05 19:42:11 +03:00
|
|
|
tcg_gen_ori_tl(usr, usr, bit);
|
2022-11-10 20:49:35 +03:00
|
|
|
} else {
|
2023-04-05 19:42:11 +03:00
|
|
|
tcg_gen_andi_tl(usr, usr, ~bit);
|
2022-11-10 20:49:35 +03:00
|
|
|
}
|
|
|
|
} else {
|
|
|
|
TCGv val = tcg_constant_tl(x);
|
2023-04-05 19:42:11 +03:00
|
|
|
gen_set_usr_field(ctx, field, val);
|
2022-11-10 20:49:35 +03:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2022-11-08 19:29:03 +03:00
|
|
|
static void gen_compare(TCGCond cond, TCGv res, TCGv arg1, TCGv arg2)
|
|
|
|
{
|
|
|
|
TCGv one = tcg_constant_tl(0xff);
|
|
|
|
TCGv zero = tcg_constant_tl(0);
|
|
|
|
|
|
|
|
tcg_gen_movcond_tl(cond, res, arg1, arg2, one, zero);
|
|
|
|
}
|
|
|
|
|
2023-04-28 01:59:54 +03:00
|
|
|
#ifndef CONFIG_HEXAGON_IDEF_PARSER
|
|
|
|
static inline void gen_loop0r(DisasContext *ctx, TCGv RsV, int riV)
|
|
|
|
{
|
|
|
|
fIMMEXT(riV);
|
|
|
|
fPCALIGN(riV);
|
|
|
|
gen_log_reg_write(ctx, HEX_REG_LC0, RsV);
|
|
|
|
gen_log_reg_write(ctx, HEX_REG_SA0, tcg_constant_tl(ctx->pkt->pc + riV));
|
|
|
|
gen_set_usr_fieldi(ctx, USR_LPCFG, 0);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void gen_loop0i(DisasContext *ctx, int count, int riV)
|
|
|
|
{
|
|
|
|
gen_loop0r(ctx, tcg_constant_tl(count), riV);
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void gen_loop1r(DisasContext *ctx, TCGv RsV, int riV)
|
|
|
|
{
|
|
|
|
fIMMEXT(riV);
|
|
|
|
fPCALIGN(riV);
|
|
|
|
gen_log_reg_write(ctx, HEX_REG_LC1, RsV);
|
|
|
|
gen_log_reg_write(ctx, HEX_REG_SA1, tcg_constant_tl(ctx->pkt->pc + riV));
|
|
|
|
}
|
|
|
|
|
|
|
|
static void gen_loop1i(DisasContext *ctx, int count, int riV)
|
|
|
|
{
|
|
|
|
gen_loop1r(ctx, tcg_constant_tl(count), riV);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void gen_ploopNsr(DisasContext *ctx, int N, TCGv RsV, int riV)
|
|
|
|
{
|
|
|
|
fIMMEXT(riV);
|
|
|
|
fPCALIGN(riV);
|
|
|
|
gen_log_reg_write(ctx, HEX_REG_LC0, RsV);
|
|
|
|
gen_log_reg_write(ctx, HEX_REG_SA0, tcg_constant_tl(ctx->pkt->pc + riV));
|
|
|
|
gen_set_usr_fieldi(ctx, USR_LPCFG, N);
|
|
|
|
gen_log_pred_write(ctx, 3, tcg_constant_tl(0));
|
|
|
|
}
|
|
|
|
|
|
|
|
static void gen_ploopNsi(DisasContext *ctx, int N, int count, int riV)
|
|
|
|
{
|
|
|
|
gen_ploopNsr(ctx, N, tcg_constant_tl(count), riV);
|
|
|
|
}
|
2023-04-28 01:59:58 +03:00
|
|
|
|
|
|
|
static inline void gen_comparei(TCGCond cond, TCGv res, TCGv arg1, int arg2)
|
|
|
|
{
|
|
|
|
gen_compare(cond, res, arg1, tcg_constant_tl(arg2));
|
|
|
|
}
|
2023-04-28 01:59:54 +03:00
|
|
|
#endif
|
|
|
|
|
2022-11-08 19:29:04 +03:00
|
|
|
static void gen_cond_jumpr(DisasContext *ctx, TCGv dst_pc,
|
|
|
|
TCGCond cond, TCGv pred)
|
|
|
|
{
|
|
|
|
gen_write_new_pc_addr(ctx, dst_pc, cond, pred);
|
|
|
|
}
|
|
|
|
|
2023-03-07 05:58:15 +03:00
|
|
|
static void gen_cond_jumpr31(DisasContext *ctx, TCGCond cond, TCGv pred)
|
|
|
|
{
|
|
|
|
TCGv LSB = tcg_temp_new();
|
|
|
|
tcg_gen_andi_tl(LSB, pred, 1);
|
|
|
|
gen_cond_jumpr(ctx, hex_gpr[HEX_REG_LR], cond, LSB);
|
|
|
|
}
|
|
|
|
|
2022-11-08 19:29:03 +03:00
|
|
|
static void gen_cond_jump(DisasContext *ctx, TCGCond cond, TCGv pred,
|
|
|
|
int pc_off)
|
|
|
|
{
|
|
|
|
gen_write_new_pc_pcrel(ctx, pc_off, cond, pred);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void gen_cmpnd_cmp_jmp(DisasContext *ctx,
|
|
|
|
int pnum, TCGCond cond1, TCGv arg1, TCGv arg2,
|
|
|
|
TCGCond cond2, int pc_off)
|
|
|
|
{
|
|
|
|
if (ctx->insn->part1) {
|
|
|
|
TCGv pred = tcg_temp_new();
|
|
|
|
gen_compare(cond1, pred, arg1, arg2);
|
|
|
|
gen_log_pred_write(ctx, pnum, pred);
|
|
|
|
} else {
|
|
|
|
TCGv pred = tcg_temp_new();
|
2023-04-28 02:00:09 +03:00
|
|
|
tcg_gen_mov_tl(pred, ctx->new_pred_value[pnum]);
|
2022-11-08 19:29:03 +03:00
|
|
|
gen_cond_jump(ctx, cond2, pred, pc_off);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void gen_cmpnd_cmp_jmp_t(DisasContext *ctx,
|
|
|
|
int pnum, TCGCond cond, TCGv arg1, TCGv arg2,
|
|
|
|
int pc_off)
|
|
|
|
{
|
|
|
|
gen_cmpnd_cmp_jmp(ctx, pnum, cond, arg1, arg2, TCG_COND_EQ, pc_off);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void gen_cmpnd_cmp_jmp_f(DisasContext *ctx,
|
|
|
|
int pnum, TCGCond cond, TCGv arg1, TCGv arg2,
|
|
|
|
int pc_off)
|
|
|
|
{
|
|
|
|
gen_cmpnd_cmp_jmp(ctx, pnum, cond, arg1, arg2, TCG_COND_NE, pc_off);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void gen_cmpnd_cmpi_jmp_t(DisasContext *ctx,
|
|
|
|
int pnum, TCGCond cond, TCGv arg1, int arg2,
|
|
|
|
int pc_off)
|
|
|
|
{
|
|
|
|
TCGv tmp = tcg_constant_tl(arg2);
|
|
|
|
gen_cmpnd_cmp_jmp(ctx, pnum, cond, arg1, tmp, TCG_COND_EQ, pc_off);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void gen_cmpnd_cmpi_jmp_f(DisasContext *ctx,
|
|
|
|
int pnum, TCGCond cond, TCGv arg1, int arg2,
|
|
|
|
int pc_off)
|
|
|
|
{
|
|
|
|
TCGv tmp = tcg_constant_tl(arg2);
|
|
|
|
gen_cmpnd_cmp_jmp(ctx, pnum, cond, arg1, tmp, TCG_COND_NE, pc_off);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void gen_cmpnd_cmp_n1_jmp_t(DisasContext *ctx, int pnum, TCGCond cond,
|
|
|
|
TCGv arg, int pc_off)
|
|
|
|
{
|
|
|
|
gen_cmpnd_cmpi_jmp_t(ctx, pnum, cond, arg, -1, pc_off);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void gen_cmpnd_cmp_n1_jmp_f(DisasContext *ctx, int pnum, TCGCond cond,
|
|
|
|
TCGv arg, int pc_off)
|
|
|
|
{
|
|
|
|
gen_cmpnd_cmpi_jmp_f(ctx, pnum, cond, arg, -1, pc_off);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void gen_cmpnd_tstbit0_jmp(DisasContext *ctx,
|
|
|
|
int pnum, TCGv arg, TCGCond cond, int pc_off)
|
|
|
|
{
|
|
|
|
if (ctx->insn->part1) {
|
|
|
|
TCGv pred = tcg_temp_new();
|
|
|
|
tcg_gen_andi_tl(pred, arg, 1);
|
|
|
|
gen_8bitsof(pred, pred);
|
|
|
|
gen_log_pred_write(ctx, pnum, pred);
|
|
|
|
} else {
|
|
|
|
TCGv pred = tcg_temp_new();
|
2023-04-28 02:00:09 +03:00
|
|
|
tcg_gen_mov_tl(pred, ctx->new_pred_value[pnum]);
|
2022-11-08 19:29:03 +03:00
|
|
|
gen_cond_jump(ctx, cond, pred, pc_off);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2022-11-08 19:29:04 +03:00
|
|
|
static void gen_testbit0_jumpnv(DisasContext *ctx,
|
|
|
|
TCGv arg, TCGCond cond, int pc_off)
|
|
|
|
{
|
|
|
|
TCGv pred = tcg_temp_new();
|
|
|
|
tcg_gen_andi_tl(pred, arg, 1);
|
|
|
|
gen_cond_jump(ctx, cond, pred, pc_off);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void gen_jump(DisasContext *ctx, int pc_off)
|
|
|
|
{
|
|
|
|
gen_write_new_pc_pcrel(ctx, pc_off, TCG_COND_ALWAYS, NULL);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void gen_jumpr(DisasContext *ctx, TCGv new_pc)
|
|
|
|
{
|
|
|
|
gen_write_new_pc_addr(ctx, new_pc, TCG_COND_ALWAYS, NULL);
|
|
|
|
}
|
|
|
|
|
2022-11-08 19:29:02 +03:00
|
|
|
static void gen_call(DisasContext *ctx, int pc_off)
|
|
|
|
{
|
Hexagon (target/hexagon) Remove gen_log_predicated_reg_write[_pair]
We assign the instruction destination register to hex_new_value[num]
instead of a TCG temp that gets copied back to hex_new_value[num].
We introduce new functions get_result_gpr[_pair] to facilitate getting
the proper destination register.
Since we preload hex_new_value for predicated instructions, we don't
need the check for slot_cancelled. So, we call gen_log_reg_write instead.
We update the helper function generation and gen_tcg.h to maintain the
disable-hexagon-idef-parser configuration.
Here is a simple example of the differences in the TCG code generated:
IN:
0x00400094: 0xf900c102 { if (P0) R2 = and(R0,R1) }
BEFORE
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
mov_i32 loc2,$0x0
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 loc2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
and_i32 tmp0,slot_cancelled,$0x8
movcond_i32 new_r2,tmp0,$0x0,loc2,new_r2,eq
mov_i32 r2,new_r2
AFTER
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 new_r2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
mov_i32 r2,new_r2
We'll remove the unnecessary manipulation of slot_cancelled in a
subsequent patch.
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20230307025828.1612809-13-tsimpson@quicinc.com>
2023-03-07 05:58:26 +03:00
|
|
|
TCGv lr = get_result_gpr(ctx, HEX_REG_LR);
|
|
|
|
tcg_gen_movi_tl(lr, ctx->next_PC);
|
2022-11-08 19:29:02 +03:00
|
|
|
gen_write_new_pc_pcrel(ctx, pc_off, TCG_COND_ALWAYS, NULL);
|
|
|
|
}
|
|
|
|
|
2023-03-07 05:58:16 +03:00
|
|
|
static void gen_callr(DisasContext *ctx, TCGv new_pc)
|
|
|
|
{
|
Hexagon (target/hexagon) Remove gen_log_predicated_reg_write[_pair]
We assign the instruction destination register to hex_new_value[num]
instead of a TCG temp that gets copied back to hex_new_value[num].
We introduce new functions get_result_gpr[_pair] to facilitate getting
the proper destination register.
Since we preload hex_new_value for predicated instructions, we don't
need the check for slot_cancelled. So, we call gen_log_reg_write instead.
We update the helper function generation and gen_tcg.h to maintain the
disable-hexagon-idef-parser configuration.
Here is a simple example of the differences in the TCG code generated:
IN:
0x00400094: 0xf900c102 { if (P0) R2 = and(R0,R1) }
BEFORE
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
mov_i32 loc2,$0x0
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 loc2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
and_i32 tmp0,slot_cancelled,$0x8
movcond_i32 new_r2,tmp0,$0x0,loc2,new_r2,eq
mov_i32 r2,new_r2
AFTER
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 new_r2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
mov_i32 r2,new_r2
We'll remove the unnecessary manipulation of slot_cancelled in a
subsequent patch.
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20230307025828.1612809-13-tsimpson@quicinc.com>
2023-03-07 05:58:26 +03:00
|
|
|
TCGv lr = get_result_gpr(ctx, HEX_REG_LR);
|
|
|
|
tcg_gen_movi_tl(lr, ctx->next_PC);
|
2023-03-07 05:58:16 +03:00
|
|
|
gen_write_new_pc_addr(ctx, new_pc, TCG_COND_ALWAYS, NULL);
|
|
|
|
}
|
|
|
|
|
2022-11-08 19:29:02 +03:00
|
|
|
static void gen_cond_call(DisasContext *ctx, TCGv pred,
|
|
|
|
TCGCond cond, int pc_off)
|
|
|
|
{
|
Hexagon (target/hexagon) Remove gen_log_predicated_reg_write[_pair]
We assign the instruction destination register to hex_new_value[num]
instead of a TCG temp that gets copied back to hex_new_value[num].
We introduce new functions get_result_gpr[_pair] to facilitate getting
the proper destination register.
Since we preload hex_new_value for predicated instructions, we don't
need the check for slot_cancelled. So, we call gen_log_reg_write instead.
We update the helper function generation and gen_tcg.h to maintain the
disable-hexagon-idef-parser configuration.
Here is a simple example of the differences in the TCG code generated:
IN:
0x00400094: 0xf900c102 { if (P0) R2 = and(R0,R1) }
BEFORE
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
mov_i32 loc2,$0x0
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 loc2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
and_i32 tmp0,slot_cancelled,$0x8
movcond_i32 new_r2,tmp0,$0x0,loc2,new_r2,eq
mov_i32 r2,new_r2
AFTER
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 new_r2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
mov_i32 r2,new_r2
We'll remove the unnecessary manipulation of slot_cancelled in a
subsequent patch.
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20230307025828.1612809-13-tsimpson@quicinc.com>
2023-03-07 05:58:26 +03:00
|
|
|
TCGv lr = get_result_gpr(ctx, HEX_REG_LR);
|
2023-01-30 03:41:33 +03:00
|
|
|
TCGv lsb = tcg_temp_new();
|
2022-11-08 19:29:02 +03:00
|
|
|
TCGLabel *skip = gen_new_label();
|
|
|
|
tcg_gen_andi_tl(lsb, pred, 1);
|
|
|
|
gen_write_new_pc_pcrel(ctx, pc_off, cond, lsb);
|
|
|
|
tcg_gen_brcondi_tl(cond, lsb, 0, skip);
|
Hexagon (target/hexagon) Remove gen_log_predicated_reg_write[_pair]
We assign the instruction destination register to hex_new_value[num]
instead of a TCG temp that gets copied back to hex_new_value[num].
We introduce new functions get_result_gpr[_pair] to facilitate getting
the proper destination register.
Since we preload hex_new_value for predicated instructions, we don't
need the check for slot_cancelled. So, we call gen_log_reg_write instead.
We update the helper function generation and gen_tcg.h to maintain the
disable-hexagon-idef-parser configuration.
Here is a simple example of the differences in the TCG code generated:
IN:
0x00400094: 0xf900c102 { if (P0) R2 = and(R0,R1) }
BEFORE
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
mov_i32 loc2,$0x0
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 loc2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
and_i32 tmp0,slot_cancelled,$0x8
movcond_i32 new_r2,tmp0,$0x0,loc2,new_r2,eq
mov_i32 r2,new_r2
AFTER
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 new_r2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
mov_i32 r2,new_r2
We'll remove the unnecessary manipulation of slot_cancelled in a
subsequent patch.
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20230307025828.1612809-13-tsimpson@quicinc.com>
2023-03-07 05:58:26 +03:00
|
|
|
tcg_gen_movi_tl(lr, ctx->next_PC);
|
2022-11-08 19:29:02 +03:00
|
|
|
gen_set_label(skip);
|
|
|
|
}
|
|
|
|
|
2023-03-07 05:58:16 +03:00
|
|
|
static void gen_cond_callr(DisasContext *ctx,
|
|
|
|
TCGCond cond, TCGv pred, TCGv new_pc)
|
|
|
|
{
|
|
|
|
TCGv lsb = tcg_temp_new();
|
|
|
|
TCGLabel *skip = gen_new_label();
|
|
|
|
tcg_gen_andi_tl(lsb, pred, 1);
|
|
|
|
tcg_gen_brcondi_tl(cond, lsb, 0, skip);
|
|
|
|
gen_callr(ctx, new_pc);
|
|
|
|
gen_set_label(skip);
|
|
|
|
}
|
|
|
|
|
2023-04-28 01:59:55 +03:00
|
|
|
#ifndef CONFIG_HEXAGON_IDEF_PARSER
|
|
|
|
/* frame = ((LR << 32) | FP) ^ (FRAMEKEY << 32)) */
|
|
|
|
static TCGv_i64 gen_frame_scramble(void)
|
|
|
|
{
|
|
|
|
TCGv_i64 frame = tcg_temp_new_i64();
|
|
|
|
TCGv tmp = tcg_temp_new();
|
|
|
|
tcg_gen_xor_tl(tmp, hex_gpr[HEX_REG_LR], hex_gpr[HEX_REG_FRAMEKEY]);
|
|
|
|
tcg_gen_concat_i32_i64(frame, hex_gpr[HEX_REG_FP], tmp);
|
|
|
|
return frame;
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
2023-03-07 05:58:18 +03:00
|
|
|
/* frame ^= (int64_t)FRAMEKEY << 32 */
|
|
|
|
static void gen_frame_unscramble(TCGv_i64 frame)
|
|
|
|
{
|
|
|
|
TCGv_i64 framekey = tcg_temp_new_i64();
|
|
|
|
tcg_gen_extu_i32_i64(framekey, hex_gpr[HEX_REG_FRAMEKEY]);
|
|
|
|
tcg_gen_shli_i64(framekey, framekey, 32);
|
|
|
|
tcg_gen_xor_i64(frame, frame, framekey);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void gen_load_frame(DisasContext *ctx, TCGv_i64 frame, TCGv EA)
|
|
|
|
{
|
|
|
|
Insn *insn = ctx->insn; /* Needed for CHECK_NOSHUF */
|
|
|
|
CHECK_NOSHUF(EA, 8);
|
2023-05-02 16:57:35 +03:00
|
|
|
tcg_gen_qemu_ld_i64(frame, EA, ctx->mem_idx, MO_TEUQ);
|
2023-03-07 05:58:18 +03:00
|
|
|
}
|
|
|
|
|
2023-04-28 01:59:55 +03:00
|
|
|
#ifndef CONFIG_HEXAGON_IDEF_PARSER
|
|
|
|
/* Stack overflow check */
|
|
|
|
static void gen_framecheck(TCGv EA, int framesize)
|
|
|
|
{
|
|
|
|
/* Not modelled in linux-user mode */
|
|
|
|
/* Placeholder for system mode */
|
|
|
|
#ifndef CONFIG_USER_ONLY
|
|
|
|
g_assert_not_reached();
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
|
|
|
static void gen_allocframe(DisasContext *ctx, TCGv r29, int framesize)
|
|
|
|
{
|
|
|
|
TCGv r30 = tcg_temp_new();
|
|
|
|
TCGv_i64 frame;
|
|
|
|
tcg_gen_addi_tl(r30, r29, -8);
|
|
|
|
frame = gen_frame_scramble();
|
|
|
|
gen_store8(cpu_env, r30, frame, ctx->insn->slot);
|
|
|
|
gen_log_reg_write(ctx, HEX_REG_FP, r30);
|
|
|
|
gen_framecheck(r30, framesize);
|
|
|
|
tcg_gen_subi_tl(r29, r30, framesize);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void gen_deallocframe(DisasContext *ctx, TCGv_i64 r31_30, TCGv r30)
|
|
|
|
{
|
|
|
|
TCGv r29 = tcg_temp_new();
|
|
|
|
TCGv_i64 frame = tcg_temp_new_i64();
|
|
|
|
gen_load_frame(ctx, frame, r30);
|
|
|
|
gen_frame_unscramble(frame);
|
|
|
|
tcg_gen_mov_i64(r31_30, frame);
|
|
|
|
tcg_gen_addi_tl(r29, r30, 8);
|
|
|
|
gen_log_reg_write(ctx, HEX_REG_SP, r29);
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
Hexagon (target/hexagon) Remove gen_log_predicated_reg_write[_pair]
We assign the instruction destination register to hex_new_value[num]
instead of a TCG temp that gets copied back to hex_new_value[num].
We introduce new functions get_result_gpr[_pair] to facilitate getting
the proper destination register.
Since we preload hex_new_value for predicated instructions, we don't
need the check for slot_cancelled. So, we call gen_log_reg_write instead.
We update the helper function generation and gen_tcg.h to maintain the
disable-hexagon-idef-parser configuration.
Here is a simple example of the differences in the TCG code generated:
IN:
0x00400094: 0xf900c102 { if (P0) R2 = and(R0,R1) }
BEFORE
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
mov_i32 loc2,$0x0
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 loc2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
and_i32 tmp0,slot_cancelled,$0x8
movcond_i32 new_r2,tmp0,$0x0,loc2,new_r2,eq
mov_i32 r2,new_r2
AFTER
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 new_r2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
mov_i32 r2,new_r2
We'll remove the unnecessary manipulation of slot_cancelled in a
subsequent patch.
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20230307025828.1612809-13-tsimpson@quicinc.com>
2023-03-07 05:58:26 +03:00
|
|
|
static void gen_return(DisasContext *ctx, TCGv_i64 dst, TCGv src)
|
2023-03-07 05:58:18 +03:00
|
|
|
{
|
|
|
|
/*
|
|
|
|
* frame = *src
|
|
|
|
* dst = frame_unscramble(frame)
|
|
|
|
* SP = src + 8
|
|
|
|
* PC = dst.w[1]
|
|
|
|
*/
|
|
|
|
TCGv_i64 frame = tcg_temp_new_i64();
|
|
|
|
TCGv r31 = tcg_temp_new();
|
Hexagon (target/hexagon) Remove gen_log_predicated_reg_write[_pair]
We assign the instruction destination register to hex_new_value[num]
instead of a TCG temp that gets copied back to hex_new_value[num].
We introduce new functions get_result_gpr[_pair] to facilitate getting
the proper destination register.
Since we preload hex_new_value for predicated instructions, we don't
need the check for slot_cancelled. So, we call gen_log_reg_write instead.
We update the helper function generation and gen_tcg.h to maintain the
disable-hexagon-idef-parser configuration.
Here is a simple example of the differences in the TCG code generated:
IN:
0x00400094: 0xf900c102 { if (P0) R2 = and(R0,R1) }
BEFORE
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
mov_i32 loc2,$0x0
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 loc2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
and_i32 tmp0,slot_cancelled,$0x8
movcond_i32 new_r2,tmp0,$0x0,loc2,new_r2,eq
mov_i32 r2,new_r2
AFTER
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 new_r2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
mov_i32 r2,new_r2
We'll remove the unnecessary manipulation of slot_cancelled in a
subsequent patch.
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20230307025828.1612809-13-tsimpson@quicinc.com>
2023-03-07 05:58:26 +03:00
|
|
|
TCGv r29 = get_result_gpr(ctx, HEX_REG_SP);
|
2023-03-07 05:58:18 +03:00
|
|
|
|
|
|
|
gen_load_frame(ctx, frame, src);
|
|
|
|
gen_frame_unscramble(frame);
|
|
|
|
tcg_gen_mov_i64(dst, frame);
|
|
|
|
tcg_gen_addi_tl(r29, src, 8);
|
|
|
|
tcg_gen_extrh_i64_i32(r31, dst);
|
|
|
|
gen_jumpr(ctx, r31);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* if (pred) dst = dealloc_return(src):raw */
|
|
|
|
static void gen_cond_return(DisasContext *ctx, TCGv_i64 dst, TCGv src,
|
|
|
|
TCGv pred, TCGCond cond)
|
|
|
|
{
|
|
|
|
TCGv LSB = tcg_temp_new();
|
|
|
|
TCGLabel *skip = gen_new_label();
|
|
|
|
tcg_gen_andi_tl(LSB, pred, 1);
|
|
|
|
|
|
|
|
tcg_gen_brcondi_tl(cond, LSB, 0, skip);
|
Hexagon (target/hexagon) Remove gen_log_predicated_reg_write[_pair]
We assign the instruction destination register to hex_new_value[num]
instead of a TCG temp that gets copied back to hex_new_value[num].
We introduce new functions get_result_gpr[_pair] to facilitate getting
the proper destination register.
Since we preload hex_new_value for predicated instructions, we don't
need the check for slot_cancelled. So, we call gen_log_reg_write instead.
We update the helper function generation and gen_tcg.h to maintain the
disable-hexagon-idef-parser configuration.
Here is a simple example of the differences in the TCG code generated:
IN:
0x00400094: 0xf900c102 { if (P0) R2 = and(R0,R1) }
BEFORE
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
mov_i32 loc2,$0x0
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 loc2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
and_i32 tmp0,slot_cancelled,$0x8
movcond_i32 new_r2,tmp0,$0x0,loc2,new_r2,eq
mov_i32 r2,new_r2
AFTER
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 new_r2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
mov_i32 r2,new_r2
We'll remove the unnecessary manipulation of slot_cancelled in a
subsequent patch.
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20230307025828.1612809-13-tsimpson@quicinc.com>
2023-03-07 05:58:26 +03:00
|
|
|
gen_return(ctx, dst, src);
|
2023-03-07 05:58:18 +03:00
|
|
|
gen_set_label(skip);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* sub-instruction version (no RddV, so handle it manually) */
|
|
|
|
static void gen_cond_return_subinsn(DisasContext *ctx, TCGCond cond, TCGv pred)
|
|
|
|
{
|
Hexagon (target/hexagon) Remove gen_log_predicated_reg_write[_pair]
We assign the instruction destination register to hex_new_value[num]
instead of a TCG temp that gets copied back to hex_new_value[num].
We introduce new functions get_result_gpr[_pair] to facilitate getting
the proper destination register.
Since we preload hex_new_value for predicated instructions, we don't
need the check for slot_cancelled. So, we call gen_log_reg_write instead.
We update the helper function generation and gen_tcg.h to maintain the
disable-hexagon-idef-parser configuration.
Here is a simple example of the differences in the TCG code generated:
IN:
0x00400094: 0xf900c102 { if (P0) R2 = and(R0,R1) }
BEFORE
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
mov_i32 loc2,$0x0
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 loc2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
and_i32 tmp0,slot_cancelled,$0x8
movcond_i32 new_r2,tmp0,$0x0,loc2,new_r2,eq
mov_i32 r2,new_r2
AFTER
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 new_r2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
mov_i32 r2,new_r2
We'll remove the unnecessary manipulation of slot_cancelled in a
subsequent patch.
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20230307025828.1612809-13-tsimpson@quicinc.com>
2023-03-07 05:58:26 +03:00
|
|
|
TCGv_i64 RddV = get_result_gpr_pair(ctx, HEX_REG_FP);
|
2023-03-07 05:58:18 +03:00
|
|
|
gen_cond_return(ctx, RddV, hex_gpr[HEX_REG_FP], pred, cond);
|
2023-04-28 01:59:53 +03:00
|
|
|
gen_log_reg_write_pair(ctx, HEX_REG_FP, RddV);
|
2023-03-07 05:58:18 +03:00
|
|
|
}
|
|
|
|
|
2022-11-10 20:49:35 +03:00
|
|
|
static void gen_endloop0(DisasContext *ctx)
|
|
|
|
{
|
2023-01-30 03:41:33 +03:00
|
|
|
TCGv lpcfg = tcg_temp_new();
|
2022-11-10 20:49:35 +03:00
|
|
|
|
|
|
|
GET_USR_FIELD(USR_LPCFG, lpcfg);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* if (lpcfg == 1) {
|
2023-04-28 01:59:59 +03:00
|
|
|
* p3 = 0xff;
|
2022-11-10 20:49:35 +03:00
|
|
|
* }
|
|
|
|
*/
|
|
|
|
TCGLabel *label1 = gen_new_label();
|
|
|
|
tcg_gen_brcondi_tl(TCG_COND_NE, lpcfg, 1, label1);
|
|
|
|
{
|
2023-04-28 01:59:59 +03:00
|
|
|
gen_log_pred_write(ctx, 3, tcg_constant_tl(0xff));
|
2022-11-10 20:49:35 +03:00
|
|
|
}
|
|
|
|
gen_set_label(label1);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* if (lpcfg) {
|
|
|
|
* SET_USR_FIELD(USR_LPCFG, lpcfg - 1);
|
|
|
|
* }
|
|
|
|
*/
|
|
|
|
TCGLabel *label2 = gen_new_label();
|
|
|
|
tcg_gen_brcondi_tl(TCG_COND_EQ, lpcfg, 0, label2);
|
|
|
|
{
|
|
|
|
tcg_gen_subi_tl(lpcfg, lpcfg, 1);
|
2023-04-05 19:42:11 +03:00
|
|
|
gen_set_usr_field(ctx, USR_LPCFG, lpcfg);
|
2022-11-10 20:49:35 +03:00
|
|
|
}
|
|
|
|
gen_set_label(label2);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* If we're in a tight loop, we'll do this at the end of the TB to take
|
|
|
|
* advantage of direct block chaining.
|
|
|
|
*/
|
|
|
|
if (!ctx->is_tight_loop) {
|
|
|
|
/*
|
2023-05-24 17:41:48 +03:00
|
|
|
* if (LC0 > 1) {
|
|
|
|
* PC = SA0;
|
|
|
|
* LC0--;
|
2022-11-10 20:49:35 +03:00
|
|
|
* }
|
|
|
|
*/
|
|
|
|
TCGLabel *label3 = gen_new_label();
|
|
|
|
tcg_gen_brcondi_tl(TCG_COND_LEU, hex_gpr[HEX_REG_LC0], 1, label3);
|
|
|
|
{
|
Hexagon (target/hexagon) Remove gen_log_predicated_reg_write[_pair]
We assign the instruction destination register to hex_new_value[num]
instead of a TCG temp that gets copied back to hex_new_value[num].
We introduce new functions get_result_gpr[_pair] to facilitate getting
the proper destination register.
Since we preload hex_new_value for predicated instructions, we don't
need the check for slot_cancelled. So, we call gen_log_reg_write instead.
We update the helper function generation and gen_tcg.h to maintain the
disable-hexagon-idef-parser configuration.
Here is a simple example of the differences in the TCG code generated:
IN:
0x00400094: 0xf900c102 { if (P0) R2 = and(R0,R1) }
BEFORE
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
mov_i32 loc2,$0x0
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 loc2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
and_i32 tmp0,slot_cancelled,$0x8
movcond_i32 new_r2,tmp0,$0x0,loc2,new_r2,eq
mov_i32 r2,new_r2
AFTER
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 new_r2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
mov_i32 r2,new_r2
We'll remove the unnecessary manipulation of slot_cancelled in a
subsequent patch.
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20230307025828.1612809-13-tsimpson@quicinc.com>
2023-03-07 05:58:26 +03:00
|
|
|
TCGv lc0 = get_result_gpr(ctx, HEX_REG_LC0);
|
2022-11-10 20:49:35 +03:00
|
|
|
gen_jumpr(ctx, hex_gpr[HEX_REG_SA0]);
|
Hexagon (target/hexagon) Remove gen_log_predicated_reg_write[_pair]
We assign the instruction destination register to hex_new_value[num]
instead of a TCG temp that gets copied back to hex_new_value[num].
We introduce new functions get_result_gpr[_pair] to facilitate getting
the proper destination register.
Since we preload hex_new_value for predicated instructions, we don't
need the check for slot_cancelled. So, we call gen_log_reg_write instead.
We update the helper function generation and gen_tcg.h to maintain the
disable-hexagon-idef-parser configuration.
Here is a simple example of the differences in the TCG code generated:
IN:
0x00400094: 0xf900c102 { if (P0) R2 = and(R0,R1) }
BEFORE
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
mov_i32 loc2,$0x0
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 loc2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
and_i32 tmp0,slot_cancelled,$0x8
movcond_i32 new_r2,tmp0,$0x0,loc2,new_r2,eq
mov_i32 r2,new_r2
AFTER
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 new_r2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
mov_i32 r2,new_r2
We'll remove the unnecessary manipulation of slot_cancelled in a
subsequent patch.
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20230307025828.1612809-13-tsimpson@quicinc.com>
2023-03-07 05:58:26 +03:00
|
|
|
tcg_gen_subi_tl(lc0, hex_gpr[HEX_REG_LC0], 1);
|
2022-11-10 20:49:35 +03:00
|
|
|
}
|
|
|
|
gen_set_label(label3);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-03-07 05:58:17 +03:00
|
|
|
static void gen_endloop1(DisasContext *ctx)
|
|
|
|
{
|
|
|
|
/*
|
2023-05-24 17:41:48 +03:00
|
|
|
* if (LC1 > 1) {
|
|
|
|
* PC = SA1;
|
|
|
|
* LC1--;
|
2023-03-07 05:58:17 +03:00
|
|
|
* }
|
|
|
|
*/
|
|
|
|
TCGLabel *label = gen_new_label();
|
|
|
|
tcg_gen_brcondi_tl(TCG_COND_LEU, hex_gpr[HEX_REG_LC1], 1, label);
|
|
|
|
{
|
Hexagon (target/hexagon) Remove gen_log_predicated_reg_write[_pair]
We assign the instruction destination register to hex_new_value[num]
instead of a TCG temp that gets copied back to hex_new_value[num].
We introduce new functions get_result_gpr[_pair] to facilitate getting
the proper destination register.
Since we preload hex_new_value for predicated instructions, we don't
need the check for slot_cancelled. So, we call gen_log_reg_write instead.
We update the helper function generation and gen_tcg.h to maintain the
disable-hexagon-idef-parser configuration.
Here is a simple example of the differences in the TCG code generated:
IN:
0x00400094: 0xf900c102 { if (P0) R2 = and(R0,R1) }
BEFORE
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
mov_i32 loc2,$0x0
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 loc2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
and_i32 tmp0,slot_cancelled,$0x8
movcond_i32 new_r2,tmp0,$0x0,loc2,new_r2,eq
mov_i32 r2,new_r2
AFTER
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 new_r2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
mov_i32 r2,new_r2
We'll remove the unnecessary manipulation of slot_cancelled in a
subsequent patch.
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20230307025828.1612809-13-tsimpson@quicinc.com>
2023-03-07 05:58:26 +03:00
|
|
|
TCGv lc1 = get_result_gpr(ctx, HEX_REG_LC1);
|
2023-03-07 05:58:17 +03:00
|
|
|
gen_jumpr(ctx, hex_gpr[HEX_REG_SA1]);
|
Hexagon (target/hexagon) Remove gen_log_predicated_reg_write[_pair]
We assign the instruction destination register to hex_new_value[num]
instead of a TCG temp that gets copied back to hex_new_value[num].
We introduce new functions get_result_gpr[_pair] to facilitate getting
the proper destination register.
Since we preload hex_new_value for predicated instructions, we don't
need the check for slot_cancelled. So, we call gen_log_reg_write instead.
We update the helper function generation and gen_tcg.h to maintain the
disable-hexagon-idef-parser configuration.
Here is a simple example of the differences in the TCG code generated:
IN:
0x00400094: 0xf900c102 { if (P0) R2 = and(R0,R1) }
BEFORE
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
mov_i32 loc2,$0x0
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 loc2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
and_i32 tmp0,slot_cancelled,$0x8
movcond_i32 new_r2,tmp0,$0x0,loc2,new_r2,eq
mov_i32 r2,new_r2
AFTER
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 new_r2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
mov_i32 r2,new_r2
We'll remove the unnecessary manipulation of slot_cancelled in a
subsequent patch.
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20230307025828.1612809-13-tsimpson@quicinc.com>
2023-03-07 05:58:26 +03:00
|
|
|
tcg_gen_subi_tl(lc1, hex_gpr[HEX_REG_LC1], 1);
|
2023-03-07 05:58:17 +03:00
|
|
|
}
|
|
|
|
gen_set_label(label);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void gen_endloop01(DisasContext *ctx)
|
|
|
|
{
|
|
|
|
TCGv lpcfg = tcg_temp_new();
|
|
|
|
TCGLabel *label1 = gen_new_label();
|
|
|
|
TCGLabel *label2 = gen_new_label();
|
|
|
|
TCGLabel *label3 = gen_new_label();
|
|
|
|
TCGLabel *done = gen_new_label();
|
|
|
|
|
|
|
|
GET_USR_FIELD(USR_LPCFG, lpcfg);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* if (lpcfg == 1) {
|
2023-04-28 01:59:59 +03:00
|
|
|
* p3 = 0xff;
|
2023-03-07 05:58:17 +03:00
|
|
|
* }
|
|
|
|
*/
|
|
|
|
tcg_gen_brcondi_tl(TCG_COND_NE, lpcfg, 1, label1);
|
|
|
|
{
|
2023-04-28 01:59:59 +03:00
|
|
|
gen_log_pred_write(ctx, 3, tcg_constant_tl(0xff));
|
2023-03-07 05:58:17 +03:00
|
|
|
}
|
|
|
|
gen_set_label(label1);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* if (lpcfg) {
|
|
|
|
* SET_USR_FIELD(USR_LPCFG, lpcfg - 1);
|
|
|
|
* }
|
|
|
|
*/
|
|
|
|
tcg_gen_brcondi_tl(TCG_COND_EQ, lpcfg, 0, label2);
|
|
|
|
{
|
|
|
|
tcg_gen_subi_tl(lpcfg, lpcfg, 1);
|
2023-04-05 19:42:11 +03:00
|
|
|
gen_set_usr_field(ctx, USR_LPCFG, lpcfg);
|
2023-03-07 05:58:17 +03:00
|
|
|
}
|
|
|
|
gen_set_label(label2);
|
|
|
|
|
|
|
|
/*
|
2023-05-24 17:41:48 +03:00
|
|
|
* if (LC0 > 1) {
|
|
|
|
* PC = SA0;
|
|
|
|
* LC0--;
|
|
|
|
* } else if (LC1 > 1) {
|
|
|
|
* PC = SA1;
|
|
|
|
* LC1--;
|
2023-03-07 05:58:17 +03:00
|
|
|
* }
|
|
|
|
*/
|
|
|
|
tcg_gen_brcondi_tl(TCG_COND_LEU, hex_gpr[HEX_REG_LC0], 1, label3);
|
|
|
|
{
|
Hexagon (target/hexagon) Remove gen_log_predicated_reg_write[_pair]
We assign the instruction destination register to hex_new_value[num]
instead of a TCG temp that gets copied back to hex_new_value[num].
We introduce new functions get_result_gpr[_pair] to facilitate getting
the proper destination register.
Since we preload hex_new_value for predicated instructions, we don't
need the check for slot_cancelled. So, we call gen_log_reg_write instead.
We update the helper function generation and gen_tcg.h to maintain the
disable-hexagon-idef-parser configuration.
Here is a simple example of the differences in the TCG code generated:
IN:
0x00400094: 0xf900c102 { if (P0) R2 = and(R0,R1) }
BEFORE
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
mov_i32 loc2,$0x0
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 loc2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
and_i32 tmp0,slot_cancelled,$0x8
movcond_i32 new_r2,tmp0,$0x0,loc2,new_r2,eq
mov_i32 r2,new_r2
AFTER
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 new_r2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
mov_i32 r2,new_r2
We'll remove the unnecessary manipulation of slot_cancelled in a
subsequent patch.
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20230307025828.1612809-13-tsimpson@quicinc.com>
2023-03-07 05:58:26 +03:00
|
|
|
TCGv lc0 = get_result_gpr(ctx, HEX_REG_LC0);
|
2023-03-07 05:58:17 +03:00
|
|
|
gen_jumpr(ctx, hex_gpr[HEX_REG_SA0]);
|
Hexagon (target/hexagon) Remove gen_log_predicated_reg_write[_pair]
We assign the instruction destination register to hex_new_value[num]
instead of a TCG temp that gets copied back to hex_new_value[num].
We introduce new functions get_result_gpr[_pair] to facilitate getting
the proper destination register.
Since we preload hex_new_value for predicated instructions, we don't
need the check for slot_cancelled. So, we call gen_log_reg_write instead.
We update the helper function generation and gen_tcg.h to maintain the
disable-hexagon-idef-parser configuration.
Here is a simple example of the differences in the TCG code generated:
IN:
0x00400094: 0xf900c102 { if (P0) R2 = and(R0,R1) }
BEFORE
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
mov_i32 loc2,$0x0
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 loc2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
and_i32 tmp0,slot_cancelled,$0x8
movcond_i32 new_r2,tmp0,$0x0,loc2,new_r2,eq
mov_i32 r2,new_r2
AFTER
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 new_r2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
mov_i32 r2,new_r2
We'll remove the unnecessary manipulation of slot_cancelled in a
subsequent patch.
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20230307025828.1612809-13-tsimpson@quicinc.com>
2023-03-07 05:58:26 +03:00
|
|
|
tcg_gen_subi_tl(lc0, hex_gpr[HEX_REG_LC0], 1);
|
2023-03-07 05:58:17 +03:00
|
|
|
tcg_gen_br(done);
|
|
|
|
}
|
|
|
|
gen_set_label(label3);
|
|
|
|
tcg_gen_brcondi_tl(TCG_COND_LEU, hex_gpr[HEX_REG_LC1], 1, done);
|
|
|
|
{
|
Hexagon (target/hexagon) Remove gen_log_predicated_reg_write[_pair]
We assign the instruction destination register to hex_new_value[num]
instead of a TCG temp that gets copied back to hex_new_value[num].
We introduce new functions get_result_gpr[_pair] to facilitate getting
the proper destination register.
Since we preload hex_new_value for predicated instructions, we don't
need the check for slot_cancelled. So, we call gen_log_reg_write instead.
We update the helper function generation and gen_tcg.h to maintain the
disable-hexagon-idef-parser configuration.
Here is a simple example of the differences in the TCG code generated:
IN:
0x00400094: 0xf900c102 { if (P0) R2 = and(R0,R1) }
BEFORE
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
mov_i32 loc2,$0x0
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 loc2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
and_i32 tmp0,slot_cancelled,$0x8
movcond_i32 new_r2,tmp0,$0x0,loc2,new_r2,eq
mov_i32 r2,new_r2
AFTER
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 new_r2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
mov_i32 r2,new_r2
We'll remove the unnecessary manipulation of slot_cancelled in a
subsequent patch.
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20230307025828.1612809-13-tsimpson@quicinc.com>
2023-03-07 05:58:26 +03:00
|
|
|
TCGv lc1 = get_result_gpr(ctx, HEX_REG_LC1);
|
2023-03-07 05:58:17 +03:00
|
|
|
gen_jumpr(ctx, hex_gpr[HEX_REG_SA1]);
|
Hexagon (target/hexagon) Remove gen_log_predicated_reg_write[_pair]
We assign the instruction destination register to hex_new_value[num]
instead of a TCG temp that gets copied back to hex_new_value[num].
We introduce new functions get_result_gpr[_pair] to facilitate getting
the proper destination register.
Since we preload hex_new_value for predicated instructions, we don't
need the check for slot_cancelled. So, we call gen_log_reg_write instead.
We update the helper function generation and gen_tcg.h to maintain the
disable-hexagon-idef-parser configuration.
Here is a simple example of the differences in the TCG code generated:
IN:
0x00400094: 0xf900c102 { if (P0) R2 = and(R0,R1) }
BEFORE
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
mov_i32 loc2,$0x0
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 loc2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
and_i32 tmp0,slot_cancelled,$0x8
movcond_i32 new_r2,tmp0,$0x0,loc2,new_r2,eq
mov_i32 r2,new_r2
AFTER
---- 00400094
mov_i32 slot_cancelled,$0x0
mov_i32 new_r2,r2
and_i32 tmp0,p0,$0x1
brcond_i32 tmp0,$0x0,eq,$L1
and_i32 tmp0,r0,r1
mov_i32 new_r2,tmp0
br $L2
set_label $L1
or_i32 slot_cancelled,slot_cancelled,$0x8
set_label $L2
mov_i32 r2,new_r2
We'll remove the unnecessary manipulation of slot_cancelled in a
subsequent patch.
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20230307025828.1612809-13-tsimpson@quicinc.com>
2023-03-07 05:58:26 +03:00
|
|
|
tcg_gen_subi_tl(lc1, hex_gpr[HEX_REG_LC1], 1);
|
2023-03-07 05:58:17 +03:00
|
|
|
}
|
|
|
|
gen_set_label(done);
|
|
|
|
}
|
|
|
|
|
2022-11-08 19:29:04 +03:00
|
|
|
static void gen_cmp_jumpnv(DisasContext *ctx,
|
|
|
|
TCGCond cond, TCGv val, TCGv src, int pc_off)
|
|
|
|
{
|
|
|
|
TCGv pred = tcg_temp_new();
|
|
|
|
tcg_gen_setcond_tl(cond, pred, val, src);
|
|
|
|
gen_cond_jump(ctx, TCG_COND_EQ, pred, pc_off);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void gen_cmpi_jumpnv(DisasContext *ctx,
|
|
|
|
TCGCond cond, TCGv val, int src, int pc_off)
|
|
|
|
{
|
|
|
|
TCGv pred = tcg_temp_new();
|
|
|
|
tcg_gen_setcondi_tl(cond, pred, val, src);
|
|
|
|
gen_cond_jump(ctx, TCG_COND_EQ, pred, pc_off);
|
|
|
|
}
|
|
|
|
|
2022-11-08 19:28:58 +03:00
|
|
|
/* Shift left with saturation */
|
2023-04-05 19:42:11 +03:00
|
|
|
static void gen_shl_sat(DisasContext *ctx, TCGv dst, TCGv src, TCGv shift_amt)
|
2022-11-08 19:28:58 +03:00
|
|
|
{
|
2023-04-28 02:00:00 +03:00
|
|
|
TCGv tmp = tcg_temp_new(); /* In case dst == src */
|
2023-04-05 19:42:11 +03:00
|
|
|
TCGv usr = get_result_gpr(ctx, HEX_REG_USR);
|
2022-11-08 19:28:58 +03:00
|
|
|
TCGv sh32 = tcg_temp_new();
|
|
|
|
TCGv dst_sar = tcg_temp_new();
|
|
|
|
TCGv ovf = tcg_temp_new();
|
|
|
|
TCGv satval = tcg_temp_new();
|
|
|
|
TCGv min = tcg_constant_tl(0x80000000);
|
|
|
|
TCGv max = tcg_constant_tl(0x7fffffff);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Possible values for shift_amt are 0 .. 64
|
|
|
|
* We need special handling for values above 31
|
|
|
|
*
|
|
|
|
* sh32 = shift & 31;
|
|
|
|
* dst = sh32 == shift ? src : 0;
|
|
|
|
* dst <<= sh32;
|
|
|
|
* dst_sar = dst >> sh32;
|
|
|
|
* satval = src < 0 ? min : max;
|
|
|
|
* if (dst_asr != src) {
|
|
|
|
* usr.OVF |= 1;
|
|
|
|
* dst = satval;
|
|
|
|
* }
|
|
|
|
*/
|
|
|
|
|
|
|
|
tcg_gen_andi_tl(sh32, shift_amt, 31);
|
2023-04-28 02:00:00 +03:00
|
|
|
tcg_gen_movcond_tl(TCG_COND_EQ, tmp, sh32, shift_amt,
|
2022-11-08 19:28:58 +03:00
|
|
|
src, tcg_constant_tl(0));
|
2023-04-28 02:00:00 +03:00
|
|
|
tcg_gen_shl_tl(tmp, tmp, sh32);
|
|
|
|
tcg_gen_sar_tl(dst_sar, tmp, sh32);
|
2022-11-08 19:28:58 +03:00
|
|
|
tcg_gen_movcond_tl(TCG_COND_LT, satval, src, tcg_constant_tl(0), min, max);
|
|
|
|
|
|
|
|
tcg_gen_setcond_tl(TCG_COND_NE, ovf, dst_sar, src);
|
|
|
|
tcg_gen_shli_tl(ovf, ovf, reg_field_info[USR_OVF].offset);
|
2023-04-05 19:42:11 +03:00
|
|
|
tcg_gen_or_tl(usr, usr, ovf);
|
2022-11-08 19:28:58 +03:00
|
|
|
|
2023-04-28 02:00:00 +03:00
|
|
|
tcg_gen_movcond_tl(TCG_COND_EQ, dst, dst_sar, src, tmp, satval);
|
2022-11-08 19:28:58 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
static void gen_sar(TCGv dst, TCGv src, TCGv shift_amt)
|
|
|
|
{
|
|
|
|
/*
|
|
|
|
* Shift arithmetic right
|
|
|
|
* Robust when shift_amt is >31 bits
|
|
|
|
*/
|
|
|
|
TCGv tmp = tcg_temp_new();
|
|
|
|
tcg_gen_umin_tl(tmp, shift_amt, tcg_constant_tl(31));
|
|
|
|
tcg_gen_sar_tl(dst, src, tmp);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Bidirectional shift right with saturation */
|
2023-04-05 19:42:11 +03:00
|
|
|
static void gen_asr_r_r_sat(DisasContext *ctx, TCGv RdV, TCGv RsV, TCGv RtV)
|
2022-11-08 19:28:58 +03:00
|
|
|
{
|
2023-01-30 03:41:33 +03:00
|
|
|
TCGv shift_amt = tcg_temp_new();
|
2022-11-08 19:28:58 +03:00
|
|
|
TCGLabel *positive = gen_new_label();
|
|
|
|
TCGLabel *done = gen_new_label();
|
|
|
|
|
|
|
|
tcg_gen_sextract_i32(shift_amt, RtV, 0, 7);
|
|
|
|
tcg_gen_brcondi_tl(TCG_COND_GE, shift_amt, 0, positive);
|
|
|
|
|
|
|
|
/* Negative shift amount => shift left */
|
|
|
|
tcg_gen_neg_tl(shift_amt, shift_amt);
|
2023-04-05 19:42:11 +03:00
|
|
|
gen_shl_sat(ctx, RdV, RsV, shift_amt);
|
2022-11-08 19:28:58 +03:00
|
|
|
tcg_gen_br(done);
|
|
|
|
|
|
|
|
gen_set_label(positive);
|
|
|
|
/* Positive shift amount => shift right */
|
|
|
|
gen_sar(RdV, RsV, shift_amt);
|
|
|
|
|
|
|
|
gen_set_label(done);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Bidirectional shift left with saturation */
|
2023-04-05 19:42:11 +03:00
|
|
|
static void gen_asl_r_r_sat(DisasContext *ctx, TCGv RdV, TCGv RsV, TCGv RtV)
|
2022-11-08 19:28:58 +03:00
|
|
|
{
|
2023-01-30 03:41:33 +03:00
|
|
|
TCGv shift_amt = tcg_temp_new();
|
2022-11-08 19:28:58 +03:00
|
|
|
TCGLabel *positive = gen_new_label();
|
|
|
|
TCGLabel *done = gen_new_label();
|
|
|
|
|
|
|
|
tcg_gen_sextract_i32(shift_amt, RtV, 0, 7);
|
|
|
|
tcg_gen_brcondi_tl(TCG_COND_GE, shift_amt, 0, positive);
|
|
|
|
|
|
|
|
/* Negative shift amount => shift right */
|
|
|
|
tcg_gen_neg_tl(shift_amt, shift_amt);
|
|
|
|
gen_sar(RdV, RsV, shift_amt);
|
|
|
|
tcg_gen_br(done);
|
|
|
|
|
|
|
|
gen_set_label(positive);
|
|
|
|
/* Positive shift amount => shift left */
|
2023-04-05 19:42:11 +03:00
|
|
|
gen_shl_sat(ctx, RdV, RsV, shift_amt);
|
2022-11-08 19:28:58 +03:00
|
|
|
|
|
|
|
gen_set_label(done);
|
|
|
|
}
|
|
|
|
|
2023-04-28 02:00:06 +03:00
|
|
|
static void gen_insert_rp(DisasContext *ctx, TCGv RxV, TCGv RsV, TCGv_i64 RttV)
|
|
|
|
{
|
|
|
|
/*
|
|
|
|
* int width = fZXTN(6, 32, (fGETWORD(1, RttV)));
|
|
|
|
* int offset = fSXTN(7, 32, (fGETWORD(0, RttV)));
|
|
|
|
* size8u_t mask = ((fCONSTLL(1) << width) - 1);
|
|
|
|
* if (offset < 0) {
|
|
|
|
* RxV = 0;
|
|
|
|
* } else {
|
|
|
|
* RxV &= ~(mask << offset);
|
|
|
|
* RxV |= ((RsV & mask) << offset);
|
|
|
|
* }
|
|
|
|
*/
|
|
|
|
|
|
|
|
TCGv width = tcg_temp_new();
|
|
|
|
TCGv offset = tcg_temp_new();
|
|
|
|
TCGv_i64 mask = tcg_temp_new_i64();
|
|
|
|
TCGv_i64 result = tcg_temp_new_i64();
|
|
|
|
TCGv_i64 tmp = tcg_temp_new_i64();
|
|
|
|
TCGv_i64 offset64 = tcg_temp_new_i64();
|
|
|
|
TCGLabel *label = gen_new_label();
|
|
|
|
TCGLabel *done = gen_new_label();
|
|
|
|
|
|
|
|
tcg_gen_extrh_i64_i32(width, RttV);
|
|
|
|
tcg_gen_extract_tl(width, width, 0, 6);
|
|
|
|
tcg_gen_extrl_i64_i32(offset, RttV);
|
|
|
|
tcg_gen_sextract_tl(offset, offset, 0, 7);
|
|
|
|
/* Possible values for offset are -64 .. 63 */
|
|
|
|
tcg_gen_brcondi_tl(TCG_COND_GE, offset, 0, label);
|
|
|
|
/* For negative offsets, zero out the result */
|
|
|
|
tcg_gen_movi_tl(RxV, 0);
|
|
|
|
tcg_gen_br(done);
|
|
|
|
gen_set_label(label);
|
|
|
|
/* At this point, possible values of offset are 0 .. 63 */
|
|
|
|
tcg_gen_ext_i32_i64(mask, width);
|
|
|
|
tcg_gen_shl_i64(mask, tcg_constant_i64(1), mask);
|
|
|
|
tcg_gen_subi_i64(mask, mask, 1);
|
|
|
|
tcg_gen_extu_i32_i64(result, RxV);
|
|
|
|
tcg_gen_ext_i32_i64(tmp, offset);
|
|
|
|
tcg_gen_shl_i64(tmp, mask, tmp);
|
|
|
|
tcg_gen_andc_i64(result, result, tmp);
|
|
|
|
tcg_gen_extu_i32_i64(tmp, RsV);
|
|
|
|
tcg_gen_and_i64(tmp, tmp, mask);
|
|
|
|
tcg_gen_extu_i32_i64(offset64, offset);
|
|
|
|
tcg_gen_shl_i64(tmp, tmp, offset64);
|
|
|
|
tcg_gen_or_i64(result, result, tmp);
|
|
|
|
tcg_gen_extrl_i64_i32(RxV, result);
|
|
|
|
gen_set_label(done);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void gen_asr_r_svw_trun(DisasContext *ctx, TCGv RdV,
|
|
|
|
TCGv_i64 RssV, TCGv RtV)
|
|
|
|
{
|
|
|
|
/*
|
|
|
|
* for (int i = 0; i < 2; i++) {
|
|
|
|
* fSETHALF(i, RdV, fGETHALF(0, ((fSXTN(7, 32, RtV) > 0) ?
|
|
|
|
* (fCAST4_8s(fGETWORD(i, RssV)) >> fSXTN(7, 32, RtV)) :
|
|
|
|
* (fCAST4_8s(fGETWORD(i, RssV)) << -fSXTN(7, 32, RtV)))));
|
|
|
|
* }
|
|
|
|
*/
|
|
|
|
TCGv shift_amt32 = tcg_temp_new();
|
|
|
|
TCGv_i64 shift_amt64 = tcg_temp_new_i64();
|
|
|
|
TCGv_i64 tmp64 = tcg_temp_new_i64();
|
|
|
|
TCGv tmp32 = tcg_temp_new();
|
|
|
|
TCGLabel *label = gen_new_label();
|
|
|
|
TCGLabel *zero = gen_new_label();
|
|
|
|
TCGLabel *done = gen_new_label();
|
|
|
|
|
|
|
|
tcg_gen_sextract_tl(shift_amt32, RtV, 0, 7);
|
|
|
|
/* Possible values of shift_amt32 are -64 .. 63 */
|
|
|
|
tcg_gen_brcondi_tl(TCG_COND_LE, shift_amt32, 0, label);
|
|
|
|
/* After branch, possible values of shift_amt32 are 1 .. 63 */
|
|
|
|
tcg_gen_ext_i32_i64(shift_amt64, shift_amt32);
|
|
|
|
for (int i = 0; i < 2; i++) {
|
|
|
|
tcg_gen_sextract_i64(tmp64, RssV, i * 32, 32);
|
|
|
|
tcg_gen_sar_i64(tmp64, tmp64, shift_amt64);
|
|
|
|
tcg_gen_extrl_i64_i32(tmp32, tmp64);
|
|
|
|
tcg_gen_deposit_tl(RdV, RdV, tmp32, i * 16, 16);
|
|
|
|
}
|
|
|
|
tcg_gen_br(done);
|
|
|
|
gen_set_label(label);
|
|
|
|
tcg_gen_neg_tl(shift_amt32, shift_amt32);
|
|
|
|
/*At this point, possible values of shift_amt32 are 0 .. 64 */
|
|
|
|
tcg_gen_brcondi_tl(TCG_COND_GT, shift_amt32, 63, zero);
|
|
|
|
/*At this point, possible values of shift_amt32 are 0 .. 63 */
|
|
|
|
tcg_gen_ext_i32_i64(shift_amt64, shift_amt32);
|
|
|
|
for (int i = 0; i < 2; i++) {
|
|
|
|
tcg_gen_sextract_i64(tmp64, RssV, i * 32, 32);
|
|
|
|
tcg_gen_shl_i64(tmp64, tmp64, shift_amt64);
|
|
|
|
tcg_gen_extrl_i64_i32(tmp32, tmp64);
|
|
|
|
tcg_gen_deposit_tl(RdV, RdV, tmp32, i * 16, 16);
|
|
|
|
}
|
|
|
|
tcg_gen_br(done);
|
|
|
|
gen_set_label(zero);
|
|
|
|
/* When the shift_amt is 64, zero out the result */
|
|
|
|
tcg_gen_movi_tl(RdV, 0);
|
|
|
|
gen_set_label(done);
|
|
|
|
}
|
|
|
|
|
2021-08-13 22:39:10 +03:00
|
|
|
static intptr_t vreg_src_off(DisasContext *ctx, int num)
|
|
|
|
{
|
|
|
|
intptr_t offset = offsetof(CPUHexagonState, VRegs[num]);
|
|
|
|
|
|
|
|
if (test_bit(num, ctx->vregs_select)) {
|
|
|
|
offset = ctx_future_vreg_off(ctx, num, 1, false);
|
|
|
|
}
|
|
|
|
if (test_bit(num, ctx->vregs_updated_tmp)) {
|
|
|
|
offset = ctx_tmp_vreg_off(ctx, num, 1, false);
|
|
|
|
}
|
|
|
|
return offset;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void gen_log_vreg_write(DisasContext *ctx, intptr_t srcoff, int num,
|
2023-03-07 05:58:28 +03:00
|
|
|
VRegWriteType type)
|
2021-08-13 22:39:10 +03:00
|
|
|
{
|
|
|
|
intptr_t dstoff;
|
|
|
|
|
|
|
|
if (type != EXT_TMP) {
|
|
|
|
dstoff = ctx_future_vreg_off(ctx, num, 1, true);
|
|
|
|
tcg_gen_gvec_mov(MO_64, dstoff, srcoff,
|
|
|
|
sizeof(MMVector), sizeof(MMVector));
|
|
|
|
} else {
|
|
|
|
dstoff = ctx_tmp_vreg_off(ctx, num, 1, false);
|
|
|
|
tcg_gen_gvec_mov(MO_64, dstoff, srcoff,
|
|
|
|
sizeof(MMVector), sizeof(MMVector));
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void gen_log_vreg_write_pair(DisasContext *ctx, intptr_t srcoff, int num,
|
2023-03-07 05:58:28 +03:00
|
|
|
VRegWriteType type)
|
2021-08-13 22:39:10 +03:00
|
|
|
{
|
2023-03-07 05:58:28 +03:00
|
|
|
gen_log_vreg_write(ctx, srcoff, num ^ 0, type);
|
2021-08-13 22:39:10 +03:00
|
|
|
srcoff += sizeof(MMVector);
|
2023-03-07 05:58:28 +03:00
|
|
|
gen_log_vreg_write(ctx, srcoff, num ^ 1, type);
|
2021-08-13 22:39:10 +03:00
|
|
|
}
|
|
|
|
|
2023-03-07 05:58:28 +03:00
|
|
|
static intptr_t get_result_qreg(DisasContext *ctx, int qnum)
|
2021-08-13 22:39:10 +03:00
|
|
|
{
|
2023-04-28 02:00:04 +03:00
|
|
|
if (ctx->need_commit) {
|
|
|
|
return offsetof(CPUHexagonState, future_QRegs[qnum]);
|
|
|
|
} else {
|
|
|
|
return offsetof(CPUHexagonState, QRegs[qnum]);
|
|
|
|
}
|
2021-08-13 22:39:10 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
static void gen_vreg_load(DisasContext *ctx, intptr_t dstoff, TCGv src,
|
|
|
|
bool aligned)
|
|
|
|
{
|
|
|
|
TCGv_i64 tmp = tcg_temp_new_i64();
|
|
|
|
if (aligned) {
|
|
|
|
tcg_gen_andi_tl(src, src, ~((int32_t)sizeof(MMVector) - 1));
|
|
|
|
}
|
|
|
|
for (int i = 0; i < sizeof(MMVector) / 8; i++) {
|
2023-05-02 16:57:35 +03:00
|
|
|
tcg_gen_qemu_ld_i64(tmp, src, ctx->mem_idx, MO_TEUQ);
|
2021-08-13 22:39:10 +03:00
|
|
|
tcg_gen_addi_tl(src, src, 8);
|
|
|
|
tcg_gen_st_i64(tmp, cpu_env, dstoff + i * 8);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2022-11-08 19:28:56 +03:00
|
|
|
static void gen_vreg_store(DisasContext *ctx, TCGv EA, intptr_t srcoff,
|
|
|
|
int slot, bool aligned)
|
2021-08-13 22:39:10 +03:00
|
|
|
{
|
|
|
|
intptr_t dstoff = offsetof(CPUHexagonState, vstore[slot].data);
|
|
|
|
intptr_t maskoff = offsetof(CPUHexagonState, vstore[slot].mask);
|
|
|
|
|
2022-11-08 19:28:56 +03:00
|
|
|
if (is_gather_store_insn(ctx)) {
|
2021-08-13 22:39:10 +03:00
|
|
|
TCGv sl = tcg_constant_tl(slot);
|
|
|
|
gen_helper_gather_store(cpu_env, EA, sl);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
tcg_gen_movi_tl(hex_vstore_pending[slot], 1);
|
|
|
|
if (aligned) {
|
|
|
|
tcg_gen_andi_tl(hex_vstore_addr[slot], EA,
|
|
|
|
~((int32_t)sizeof(MMVector) - 1));
|
|
|
|
} else {
|
|
|
|
tcg_gen_mov_tl(hex_vstore_addr[slot], EA);
|
|
|
|
}
|
|
|
|
tcg_gen_movi_tl(hex_vstore_size[slot], sizeof(MMVector));
|
|
|
|
|
|
|
|
/* Copy the data to the vstore buffer */
|
|
|
|
tcg_gen_gvec_mov(MO_64, dstoff, srcoff, sizeof(MMVector), sizeof(MMVector));
|
|
|
|
/* Set the mask to all 1's */
|
|
|
|
tcg_gen_gvec_dup_imm(MO_64, maskoff, sizeof(MMQReg), sizeof(MMQReg), ~0LL);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void gen_vreg_masked_store(DisasContext *ctx, TCGv EA, intptr_t srcoff,
|
|
|
|
intptr_t bitsoff, int slot, bool invert)
|
|
|
|
{
|
|
|
|
intptr_t dstoff = offsetof(CPUHexagonState, vstore[slot].data);
|
|
|
|
intptr_t maskoff = offsetof(CPUHexagonState, vstore[slot].mask);
|
|
|
|
|
|
|
|
tcg_gen_movi_tl(hex_vstore_pending[slot], 1);
|
|
|
|
tcg_gen_andi_tl(hex_vstore_addr[slot], EA,
|
|
|
|
~((int32_t)sizeof(MMVector) - 1));
|
|
|
|
tcg_gen_movi_tl(hex_vstore_size[slot], sizeof(MMVector));
|
|
|
|
|
|
|
|
/* Copy the data to the vstore buffer */
|
|
|
|
tcg_gen_gvec_mov(MO_64, dstoff, srcoff, sizeof(MMVector), sizeof(MMVector));
|
|
|
|
/* Copy the mask */
|
|
|
|
tcg_gen_gvec_mov(MO_64, maskoff, bitsoff, sizeof(MMQReg), sizeof(MMQReg));
|
|
|
|
if (invert) {
|
|
|
|
tcg_gen_gvec_not(MO_64, maskoff, maskoff,
|
|
|
|
sizeof(MMQReg), sizeof(MMQReg));
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void vec_to_qvec(size_t size, intptr_t dstoff, intptr_t srcoff)
|
|
|
|
{
|
|
|
|
TCGv_i64 tmp = tcg_temp_new_i64();
|
|
|
|
TCGv_i64 word = tcg_temp_new_i64();
|
|
|
|
TCGv_i64 bits = tcg_temp_new_i64();
|
|
|
|
TCGv_i64 mask = tcg_temp_new_i64();
|
|
|
|
TCGv_i64 zero = tcg_constant_i64(0);
|
|
|
|
TCGv_i64 ones = tcg_constant_i64(~0);
|
|
|
|
|
|
|
|
for (int i = 0; i < sizeof(MMVector) / 8; i++) {
|
|
|
|
tcg_gen_ld_i64(tmp, cpu_env, srcoff + i * 8);
|
|
|
|
tcg_gen_movi_i64(mask, 0);
|
|
|
|
|
|
|
|
for (int j = 0; j < 8; j += size) {
|
|
|
|
tcg_gen_extract_i64(word, tmp, j * 8, size * 8);
|
|
|
|
tcg_gen_movcond_i64(TCG_COND_NE, bits, word, zero, ones, zero);
|
|
|
|
tcg_gen_deposit_i64(mask, mask, bits, j, size);
|
|
|
|
}
|
|
|
|
|
|
|
|
tcg_gen_st8_i64(mask, cpu_env, dstoff + i);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2022-09-23 20:38:24 +03:00
|
|
|
void probe_noshuf_load(TCGv va, int s, int mi)
|
2022-07-08 00:05:46 +03:00
|
|
|
{
|
|
|
|
TCGv size = tcg_constant_tl(s);
|
|
|
|
TCGv mem_idx = tcg_constant_tl(mi);
|
|
|
|
gen_helper_probe_noshuf_load(cpu_env, va, size, mem_idx);
|
|
|
|
}
|
|
|
|
|
2022-09-23 20:38:25 +03:00
|
|
|
/*
|
|
|
|
* Note: Since this function might branch, `val` is
|
|
|
|
* required to be a `tcg_temp_local`.
|
|
|
|
*/
|
2023-04-05 19:42:11 +03:00
|
|
|
void gen_set_usr_field_if(DisasContext *ctx, int field, TCGv val)
|
2022-09-23 20:38:25 +03:00
|
|
|
{
|
|
|
|
/* Sets the USR field if `val` is non-zero */
|
|
|
|
if (reg_field_info[field].width == 1) {
|
2023-04-05 19:42:11 +03:00
|
|
|
TCGv usr = get_result_gpr(ctx, HEX_REG_USR);
|
2022-09-23 20:38:25 +03:00
|
|
|
TCGv tmp = tcg_temp_new();
|
|
|
|
tcg_gen_extract_tl(tmp, val, 0, reg_field_info[field].width);
|
|
|
|
tcg_gen_shli_tl(tmp, tmp, reg_field_info[field].offset);
|
2023-04-05 19:42:11 +03:00
|
|
|
tcg_gen_or_tl(usr, usr, tmp);
|
2022-09-23 20:38:25 +03:00
|
|
|
} else {
|
|
|
|
TCGLabel *skip_label = gen_new_label();
|
|
|
|
tcg_gen_brcondi_tl(TCG_COND_EQ, val, 0, skip_label);
|
2023-04-05 19:42:11 +03:00
|
|
|
gen_set_usr_field(ctx, field, val);
|
2022-09-23 20:38:25 +03:00
|
|
|
gen_set_label(skip_label);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
void gen_sat_i32(TCGv dest, TCGv source, int width)
|
|
|
|
{
|
|
|
|
TCGv max_val = tcg_constant_tl((1 << (width - 1)) - 1);
|
|
|
|
TCGv min_val = tcg_constant_tl(-(1 << (width - 1)));
|
|
|
|
tcg_gen_smin_tl(dest, source, max_val);
|
|
|
|
tcg_gen_smax_tl(dest, dest, min_val);
|
|
|
|
}
|
|
|
|
|
|
|
|
void gen_sat_i32_ovfl(TCGv ovfl, TCGv dest, TCGv source, int width)
|
|
|
|
{
|
2023-04-28 02:00:00 +03:00
|
|
|
TCGv tmp = tcg_temp_new(); /* In case dest == source */
|
|
|
|
gen_sat_i32(tmp, source, width);
|
|
|
|
tcg_gen_setcond_tl(TCG_COND_NE, ovfl, source, tmp);
|
|
|
|
tcg_gen_mov_tl(dest, tmp);
|
2022-09-23 20:38:25 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
void gen_satu_i32(TCGv dest, TCGv source, int width)
|
|
|
|
{
|
2023-04-28 02:00:00 +03:00
|
|
|
TCGv tmp = tcg_temp_new(); /* In case dest == source */
|
2022-09-23 20:38:25 +03:00
|
|
|
TCGv max_val = tcg_constant_tl((1 << width) - 1);
|
|
|
|
TCGv zero = tcg_constant_tl(0);
|
2023-04-28 02:00:00 +03:00
|
|
|
tcg_gen_movcond_tl(TCG_COND_GTU, tmp, source, max_val, max_val, source);
|
|
|
|
tcg_gen_movcond_tl(TCG_COND_LT, tmp, source, zero, zero, tmp);
|
|
|
|
tcg_gen_mov_tl(dest, tmp);
|
2022-09-23 20:38:25 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
void gen_satu_i32_ovfl(TCGv ovfl, TCGv dest, TCGv source, int width)
|
|
|
|
{
|
2023-04-28 02:00:00 +03:00
|
|
|
TCGv tmp = tcg_temp_new(); /* In case dest == source */
|
|
|
|
gen_satu_i32(tmp, source, width);
|
|
|
|
tcg_gen_setcond_tl(TCG_COND_NE, ovfl, source, tmp);
|
|
|
|
tcg_gen_mov_tl(dest, tmp);
|
2022-09-23 20:38:25 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
void gen_sat_i64(TCGv_i64 dest, TCGv_i64 source, int width)
|
|
|
|
{
|
|
|
|
TCGv_i64 max_val = tcg_constant_i64((1LL << (width - 1)) - 1LL);
|
|
|
|
TCGv_i64 min_val = tcg_constant_i64(-(1LL << (width - 1)));
|
|
|
|
tcg_gen_smin_i64(dest, source, max_val);
|
|
|
|
tcg_gen_smax_i64(dest, dest, min_val);
|
|
|
|
}
|
|
|
|
|
|
|
|
void gen_sat_i64_ovfl(TCGv ovfl, TCGv_i64 dest, TCGv_i64 source, int width)
|
|
|
|
{
|
2023-04-28 02:00:00 +03:00
|
|
|
TCGv_i64 tmp = tcg_temp_new_i64(); /* In case dest == source */
|
2022-09-23 20:38:25 +03:00
|
|
|
TCGv_i64 ovfl_64;
|
2023-04-28 02:00:00 +03:00
|
|
|
gen_sat_i64(tmp, source, width);
|
2022-09-23 20:38:25 +03:00
|
|
|
ovfl_64 = tcg_temp_new_i64();
|
2023-04-28 02:00:00 +03:00
|
|
|
tcg_gen_setcond_i64(TCG_COND_NE, ovfl_64, tmp, source);
|
|
|
|
tcg_gen_mov_i64(dest, tmp);
|
2022-09-23 20:38:25 +03:00
|
|
|
tcg_gen_trunc_i64_tl(ovfl, ovfl_64);
|
|
|
|
}
|
|
|
|
|
|
|
|
void gen_satu_i64(TCGv_i64 dest, TCGv_i64 source, int width)
|
|
|
|
{
|
2023-04-28 02:00:00 +03:00
|
|
|
TCGv_i64 tmp = tcg_temp_new_i64(); /* In case dest == source */
|
2022-09-23 20:38:25 +03:00
|
|
|
TCGv_i64 max_val = tcg_constant_i64((1LL << width) - 1LL);
|
|
|
|
TCGv_i64 zero = tcg_constant_i64(0);
|
2023-04-28 02:00:00 +03:00
|
|
|
tcg_gen_movcond_i64(TCG_COND_GTU, tmp, source, max_val, max_val, source);
|
|
|
|
tcg_gen_movcond_i64(TCG_COND_LT, tmp, source, zero, zero, tmp);
|
|
|
|
tcg_gen_mov_i64(dest, tmp);
|
2022-09-23 20:38:25 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
void gen_satu_i64_ovfl(TCGv ovfl, TCGv_i64 dest, TCGv_i64 source, int width)
|
|
|
|
{
|
2023-04-28 02:00:00 +03:00
|
|
|
TCGv_i64 tmp = tcg_temp_new_i64(); /* In case dest == source */
|
2022-09-23 20:38:25 +03:00
|
|
|
TCGv_i64 ovfl_64;
|
2023-04-28 02:00:00 +03:00
|
|
|
gen_satu_i64(tmp, source, width);
|
2022-09-23 20:38:25 +03:00
|
|
|
ovfl_64 = tcg_temp_new_i64();
|
2023-04-28 02:00:00 +03:00
|
|
|
tcg_gen_setcond_i64(TCG_COND_NE, ovfl_64, tmp, source);
|
|
|
|
tcg_gen_mov_i64(dest, tmp);
|
2022-09-23 20:38:25 +03:00
|
|
|
tcg_gen_trunc_i64_tl(ovfl, ovfl_64);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Implements the fADDSAT64 macro in TCG */
|
2023-04-05 19:42:11 +03:00
|
|
|
void gen_add_sat_i64(DisasContext *ctx, TCGv_i64 ret, TCGv_i64 a, TCGv_i64 b)
|
2022-09-23 20:38:25 +03:00
|
|
|
{
|
2023-01-30 03:41:33 +03:00
|
|
|
TCGv_i64 sum = tcg_temp_new_i64();
|
2022-09-23 20:38:25 +03:00
|
|
|
TCGv_i64 xor = tcg_temp_new_i64();
|
|
|
|
TCGv_i64 cond1 = tcg_temp_new_i64();
|
2023-01-30 03:41:33 +03:00
|
|
|
TCGv_i64 cond2 = tcg_temp_new_i64();
|
2022-09-23 20:38:25 +03:00
|
|
|
TCGv_i64 cond3 = tcg_temp_new_i64();
|
|
|
|
TCGv_i64 mask = tcg_constant_i64(0x8000000000000000ULL);
|
|
|
|
TCGv_i64 max_pos = tcg_constant_i64(0x7FFFFFFFFFFFFFFFLL);
|
|
|
|
TCGv_i64 max_neg = tcg_constant_i64(0x8000000000000000LL);
|
|
|
|
TCGv_i64 zero = tcg_constant_i64(0);
|
|
|
|
TCGLabel *no_ovfl_label = gen_new_label();
|
|
|
|
TCGLabel *ovfl_label = gen_new_label();
|
|
|
|
TCGLabel *ret_label = gen_new_label();
|
|
|
|
|
|
|
|
tcg_gen_add_i64(sum, a, b);
|
|
|
|
tcg_gen_xor_i64(xor, a, b);
|
|
|
|
|
|
|
|
/* if (xor & mask) */
|
|
|
|
tcg_gen_and_i64(cond1, xor, mask);
|
|
|
|
tcg_gen_brcondi_i64(TCG_COND_NE, cond1, 0, no_ovfl_label);
|
|
|
|
|
|
|
|
/* else if ((a ^ sum) & mask) */
|
|
|
|
tcg_gen_xor_i64(cond2, a, sum);
|
|
|
|
tcg_gen_and_i64(cond2, cond2, mask);
|
|
|
|
tcg_gen_brcondi_i64(TCG_COND_NE, cond2, 0, ovfl_label);
|
|
|
|
/* fallthrough to no_ovfl_label branch */
|
|
|
|
|
|
|
|
/* if branch */
|
|
|
|
gen_set_label(no_ovfl_label);
|
|
|
|
tcg_gen_mov_i64(ret, sum);
|
|
|
|
tcg_gen_br(ret_label);
|
|
|
|
|
|
|
|
/* else if branch */
|
|
|
|
gen_set_label(ovfl_label);
|
|
|
|
tcg_gen_and_i64(cond3, sum, mask);
|
|
|
|
tcg_gen_movcond_i64(TCG_COND_NE, ret, cond3, zero, max_pos, max_neg);
|
2023-04-05 19:42:11 +03:00
|
|
|
gen_set_usr_fieldi(ctx, USR_OVF, 1);
|
2022-09-23 20:38:25 +03:00
|
|
|
|
|
|
|
gen_set_label(ret_label);
|
|
|
|
}
|
|
|
|
|
2021-02-08 08:46:16 +03:00
|
|
|
#include "tcg_funcs_generated.c.inc"
|
|
|
|
#include "tcg_func_table_generated.c.inc"
|