falso
/
qemu-irix
mirror of https://github.com/n64decomp/qemu-irix.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

tcg/arm: Improve tlb load for armv7 

Use UBFX to avoid limitation on CPU_TLB_BITS.  Since we're dropping
the initial shift, we need to replace the page masking.  We can use
MOVW+BIC to do this without shifting.  The result is the same size
as the armv6 path with one less conditional instruction.

Signed-off-by: Richard Henderson <rth@twiddle.net>
This commit is contained in:
Richard Henderson 2017-07-27 13:16:16 -07:00
parent e9823b4c33
commit 647ab96aaf
1 changed files with 53 additions and 21 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

74
tcg/arm/tcg-target.inc.c
View File

@ -1173,18 +1173,33 @@ static TCGReg tcg_out_tlb_read(TCGContext *s, TCGReg addrlo, TCGReg addrhi,
unsigned s_bits = opc & MO_SIZE; unsigned s_bits = opc & MO_SIZE;
unsigned a_bits = get_alignment_bits(opc); unsigned a_bits = get_alignment_bits(opc);
/* Should generate something like the following: /* V7 generates the following:
* shr tmp, addrlo, #TARGET_PAGE_BITS (1) * ubfx r0, addrlo, #TARGET_PAGE_BITS, #CPU_TLB_BITS
* add r2, env, #high * add r2, env, #high
* and r0, tmp, #(CPU_TLB_SIZE - 1) (2) * add r2, r2, r0, lsl #CPU_TLB_ENTRY_BITS
* add r2, r2, r0, lsl #CPU_TLB_ENTRY_BITS (3) * ldr r0, [r2, #cmp]
* ldr r0, [r2, #cmp] (4) * ldr r2, [r2, #add]
* movw tmp, #page_align_mask
* bic tmp, addrlo, tmp
* cmp r0, tmp
*
* Otherwise we generate:
* shr tmp, addrlo, #TARGET_PAGE_BITS
* add r2, env, #high
* and r0, tmp, #(CPU_TLB_SIZE - 1)
* add r2, r2, r0, lsl #CPU_TLB_ENTRY_BITS
* ldr r0, [r2, #cmp]
* ldr r2, [r2, #add]
* tst addrlo, #s_mask * tst addrlo, #s_mask
* ldr r2, [r2, #add] (5)
* cmpeq r0, tmp, lsl #TARGET_PAGE_BITS * cmpeq r0, tmp, lsl #TARGET_PAGE_BITS
*/ */
tcg_out_dat_reg(s, COND_AL, ARITH_MOV, TCG_REG_TMP, if (use_armv7_instructions) {
0, addrlo, SHIFT_IMM_LSR(TARGET_PAGE_BITS)); tcg_out_extract(s, COND_AL, TCG_REG_R0, addrlo,
TARGET_PAGE_BITS, CPU_TLB_BITS);
} else {
tcg_out_dat_reg(s, COND_AL, ARITH_MOV, TCG_REG_TMP,
0, addrlo, SHIFT_IMM_LSR(TARGET_PAGE_BITS));
}
/* We checked that the offset is contained within 16 bits above. */ /* We checked that the offset is contained within 16 bits above. */
if (add_off > 0xfff || (use_armv6_instructions && cmp_off > 0xff)) { if (add_off > 0xfff || (use_armv6_instructions && cmp_off > 0xff)) {
@ -1194,9 +1209,10 @@ static TCGReg tcg_out_tlb_read(TCGContext *s, TCGReg addrlo, TCGReg addrhi,
add_off -= cmp_off & 0xff00; add_off -= cmp_off & 0xff00;
cmp_off &= 0xff; cmp_off &= 0xff;
} }
if (!use_armv7_instructions) {
tcg_out_dat_imm(s, COND_AL, ARITH_AND, tcg_out_dat_imm(s, COND_AL, ARITH_AND,
TCG_REG_R0, TCG_REG_TMP, CPU_TLB_SIZE - 1); TCG_REG_R0, TCG_REG_TMP, CPU_TLB_SIZE - 1);
}
tcg_out_dat_reg(s, COND_AL, ARITH_ADD, TCG_REG_R2, base, tcg_out_dat_reg(s, COND_AL, ARITH_ADD, TCG_REG_R2, base,
TCG_REG_R0, SHIFT_IMM_LSL(CPU_TLB_ENTRY_BITS)); TCG_REG_R0, SHIFT_IMM_LSL(CPU_TLB_ENTRY_BITS));
@ -1212,24 +1228,40 @@ static TCGReg tcg_out_tlb_read(TCGContext *s, TCGReg addrlo, TCGReg addrhi,
} }
} }
/* Load the tlb addend. */
tcg_out_ld32_12(s, COND_AL, TCG_REG_R2, TCG_REG_R2, add_off);
/* Check alignment. We don't support inline unaligned acceses, /* Check alignment. We don't support inline unaligned acceses,
but we can easily support overalignment checks. */ but we can easily support overalignment checks. */
if (a_bits < s_bits) { if (a_bits < s_bits) {
a_bits = s_bits; a_bits = s_bits;
} }
if (a_bits) {
tcg_out_dat_imm(s, COND_AL, ARITH_TST, 0, addrlo, (1 << a_bits) - 1); if (use_armv7_instructions) {
tcg_target_ulong mask = ~(TARGET_PAGE_MASK | ((1 << a_bits) - 1));
int rot = encode_imm(mask);
if (rot >= 0) {
tcg_out_dat_imm(s, COND_AL, ARITH_BIC, TCG_REG_TMP, addrlo,
rotl(mask, rot) | (rot << 7));
} else {
tcg_out_movi32(s, COND_AL, TCG_REG_TMP, mask);
tcg_out_dat_reg(s, COND_AL, ARITH_BIC, TCG_REG_TMP,
addrlo, TCG_REG_TMP, 0);
}
tcg_out_dat_reg(s, COND_AL, ARITH_CMP, 0, TCG_REG_R0, TCG_REG_TMP, 0);
} else {
if (a_bits) {
tcg_out_dat_imm(s, COND_AL, ARITH_TST, 0, addrlo,
(1 << a_bits) - 1);
}
tcg_out_dat_reg(s, (a_bits ? COND_EQ : COND_AL), ARITH_CMP,
0, TCG_REG_R0, TCG_REG_TMP,
SHIFT_IMM_LSL(TARGET_PAGE_BITS));
} }
/* Load the tlb addend. */
tcg_out_ld32_12(s, COND_AL, TCG_REG_R2, TCG_REG_R2, add_off);
tcg_out_dat_reg(s, (a_bits ? COND_EQ : COND_AL), ARITH_CMP, 0,
TCG_REG_R0, TCG_REG_TMP, SHIFT_IMM_LSL(TARGET_PAGE_BITS));
if (TARGET_LONG_BITS == 64) { if (TARGET_LONG_BITS == 64) {
tcg_out_dat_reg(s, COND_EQ, ARITH_CMP, 0, tcg_out_dat_reg(s, COND_EQ, ARITH_CMP, 0, TCG_REG_R1, addrhi, 0);
TCG_REG_R1, addrhi, SHIFT_IMM_LSL(0));
} }
return TCG_REG_R2; return TCG_REG_R2;