computing-offload/generic_vdpa/qemu/target-hppa-Fix-deposit-assert-from-trans_shrpw_imm.patch
jiangdongxu 79c4324644 add generic_vdpa basecode
Change-Id: I2d302dda68298877c65c99147f5bf22186a59aac
2024-09-19 17:19:46 +08:00

89 lines
3.1 KiB
Diff

From 8964f3516cab9ed0183407136aa9f0dae87c49e3 Mon Sep 17 00:00:00 2001
From: tangbinzy <tangbin_yewu@cmss.chinamobile.com>
Date: Wed, 23 Nov 2022 15:38:38 -0800
Subject: [PATCH 12/29] target/hppa: Fix deposit assert from trans_shrpw_imm
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
mainline inclusion
commit 05bfd4db08608bc4c22de729780c1f74612fbc0e
category: bugfix
----------------------------------------------
Because sa may be 0,
tcg_gen_deposit_reg(dest, t0, cpu_gr[a->r1], 32 - sa, sa);
may attempt a zero-width deposit at bit 32, which will assert
for TARGET_REGISTER_BITS == 32.
Use the newer extract2 when possible, which itself includes the
rotri special case; otherwise mirror the code from trans_shrpw_sar,
using concat and shri.
Cc: qemu-stable@nongnu.org
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/635
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: tangbinzy <tangbin_yewu@cmss.chinamobile.com>
---
target/hppa/translate.c | 19 ++++++++++++-------
1 file changed, 12 insertions(+), 7 deletions(-)
diff --git a/target/hppa/translate.c b/target/hppa/translate.c
index 3b9744deb4..952027a28e 100644
--- a/target/hppa/translate.c
+++ b/target/hppa/translate.c
@@ -140,6 +140,7 @@
#define tcg_gen_deposit_z_reg tcg_gen_deposit_z_i64
#define tcg_gen_extract_reg tcg_gen_extract_i64
#define tcg_gen_sextract_reg tcg_gen_sextract_i64
+#define tcg_gen_extract2_reg tcg_gen_extract2_i64
#define tcg_const_reg tcg_const_i64
#define tcg_const_local_reg tcg_const_local_i64
#define tcg_constant_reg tcg_constant_i64
@@ -234,6 +235,7 @@
#define tcg_gen_deposit_z_reg tcg_gen_deposit_z_i32
#define tcg_gen_extract_reg tcg_gen_extract_i32
#define tcg_gen_sextract_reg tcg_gen_sextract_i32
+#define tcg_gen_extract2_reg tcg_gen_extract2_i32
#define tcg_const_reg tcg_const_i32
#define tcg_const_local_reg tcg_const_local_i32
#define tcg_constant_reg tcg_constant_i32
@@ -3204,19 +3206,22 @@ static bool trans_shrpw_imm(DisasContext *ctx, arg_shrpw_imm *a)
dest = dest_gpr(ctx, a->t);
t2 = load_gpr(ctx, a->r2);
- if (a->r1 == a->r2) {
+ if (a->r1 == 0) {
+ tcg_gen_extract_reg(dest, t2, sa, 32 - sa);
+ } else if (TARGET_REGISTER_BITS == 32) {
+ tcg_gen_extract2_reg(dest, t2, cpu_gr[a->r1], sa);
+ } else if (a->r1 == a->r2) {
TCGv_i32 t32 = tcg_temp_new_i32();
tcg_gen_trunc_reg_i32(t32, t2);
tcg_gen_rotri_i32(t32, t32, sa);
tcg_gen_extu_i32_reg(dest, t32);
tcg_temp_free_i32(t32);
- } else if (a->r1 == 0) {
- tcg_gen_extract_reg(dest, t2, sa, 32 - sa);
} else {
- TCGv_reg t0 = tcg_temp_new();
- tcg_gen_extract_reg(t0, t2, sa, 32 - sa);
- tcg_gen_deposit_reg(dest, t0, cpu_gr[a->r1], 32 - sa, sa);
- tcg_temp_free(t0);
+ TCGv_i64 t64 = tcg_temp_new_i64();
+ tcg_gen_concat_reg_i64(t64, t2, cpu_gr[a->r1]);
+ tcg_gen_shri_i64(t64, t64, sa);
+ tcg_gen_trunc_i64_reg(dest, t64);
+ tcg_temp_free_i64(t64);
}
save_gpr(ctx, a->t, dest);
--
2.27.0