qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Taylor Simpson <tsimpson@quicinc.com>
To: qemu-devel@nongnu.org
Cc: ale@rev.ng, bcain@quicinc.com, tsimpson@quicinc.com,
	richard.henderson@linaro.org, f4bug@amsat.org
Subject: [PATCH v4 23/30] Hexagon HVX (target/hexagon) helper overrides - vector stores
Date: Tue, 12 Oct 2021 05:11:01 -0500	[thread overview]
Message-ID: <1634033468-23566-24-git-send-email-tsimpson@quicinc.com> (raw)
In-Reply-To: <1634033468-23566-1-git-send-email-tsimpson@quicinc.com>

Acked-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Taylor Simpson <tsimpson@quicinc.com>
---
 target/hexagon/gen_tcg_hvx.h | 218 +++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 218 insertions(+)

diff --git a/target/hexagon/gen_tcg_hvx.h b/target/hexagon/gen_tcg_hvx.h
index 2d1d778..cdcc938 100644
--- a/target/hexagon/gen_tcg_hvx.h
+++ b/target/hexagon/gen_tcg_hvx.h
@@ -682,4 +682,222 @@ static inline void assert_vhist_tmp(DisasContext *ctx)
 #define fGEN_TCG_V6_vL32b_nt_tmp_npred_ppu(SHORTCODE) \
     fGEN_TCG_PRED_VEC_LOAD_npred_ppu
 
+/* Vector stores */
+#define fGEN_TCG_V6_vS32b_pi(SHORTCODE)                    SHORTCODE
+#define fGEN_TCG_V6_vS32Ub_pi(SHORTCODE)                   SHORTCODE
+#define fGEN_TCG_V6_vS32b_nt_pi(SHORTCODE)                 SHORTCODE
+#define fGEN_TCG_V6_vS32b_ai(SHORTCODE)                    SHORTCODE
+#define fGEN_TCG_V6_vS32Ub_ai(SHORTCODE)                   SHORTCODE
+#define fGEN_TCG_V6_vS32b_nt_ai(SHORTCODE)                 SHORTCODE
+#define fGEN_TCG_V6_vS32b_ppu(SHORTCODE)                   SHORTCODE
+#define fGEN_TCG_V6_vS32Ub_ppu(SHORTCODE)                  SHORTCODE
+#define fGEN_TCG_V6_vS32b_nt_ppu(SHORTCODE)                SHORTCODE
+
+/* New value vector stores */
+#define fGEN_TCG_NEWVAL_VEC_STORE(GET_EA, INC) \
+    do { \
+        GET_EA; \
+        gen_vreg_store(ctx, insn, pkt, EA, OsN_off, insn->slot, true); \
+        INC; \
+    } while (0)
+
+#define fGEN_TCG_NEWVAL_VEC_STORE_pi \
+    fGEN_TCG_NEWVAL_VEC_STORE(fEA_REG(RxV), fPM_I(RxV, siV * sizeof(MMVector)))
+
+#define fGEN_TCG_V6_vS32b_new_pi(SHORTCODE) \
+    fGEN_TCG_NEWVAL_VEC_STORE_pi
+#define fGEN_TCG_V6_vS32b_nt_new_pi(SHORTCODE) \
+    fGEN_TCG_NEWVAL_VEC_STORE_pi
+
+#define fGEN_TCG_NEWVAL_VEC_STORE_ai \
+    fGEN_TCG_NEWVAL_VEC_STORE(fEA_RI(RtV, siV * sizeof(MMVector)), \
+                              do { } while (0))
+
+#define fGEN_TCG_V6_vS32b_new_ai(SHORTCODE) \
+    fGEN_TCG_NEWVAL_VEC_STORE_ai
+#define fGEN_TCG_V6_vS32b_nt_new_ai(SHORTCODE) \
+    fGEN_TCG_NEWVAL_VEC_STORE_ai
+
+#define fGEN_TCG_NEWVAL_VEC_STORE_ppu \
+    fGEN_TCG_NEWVAL_VEC_STORE(fEA_REG(RxV), fPM_M(RxV, MuV))
+
+#define fGEN_TCG_V6_vS32b_new_ppu(SHORTCODE) \
+    fGEN_TCG_NEWVAL_VEC_STORE_ppu
+#define fGEN_TCG_V6_vS32b_nt_new_ppu(SHORTCODE) \
+    fGEN_TCG_NEWVAL_VEC_STORE_ppu
+
+/* Predicated vector stores */
+#define fGEN_TCG_PRED_VEC_STORE(GET_EA, PRED, SRCOFF, ALIGN, INC) \
+    do { \
+        TCGv LSB = tcg_temp_new(); \
+        TCGLabel *false_label = gen_new_label(); \
+        TCGLabel *end_label = gen_new_label(); \
+        GET_EA; \
+        PRED; \
+        tcg_gen_brcondi_tl(TCG_COND_EQ, LSB, 0, false_label); \
+        tcg_temp_free(LSB); \
+        gen_vreg_store(ctx, insn, pkt, EA, SRCOFF, insn->slot, ALIGN); \
+        INC; \
+        tcg_gen_br(end_label); \
+        gen_set_label(false_label); \
+        tcg_gen_ori_tl(hex_slot_cancelled, hex_slot_cancelled, \
+                       1 << insn->slot); \
+        gen_set_label(end_label); \
+    } while (0)
+
+#define fGEN_TCG_PRED_VEC_STORE_pred_pi(ALIGN) \
+    fGEN_TCG_PRED_VEC_STORE(fLSBOLD(PvV), \
+                            fEA_REG(RxV), \
+                            VsV_off, ALIGN, \
+                            fPM_I(RxV, siV * sizeof(MMVector)))
+#define fGEN_TCG_PRED_VEC_STORE_npred_pi(ALIGN) \
+    fGEN_TCG_PRED_VEC_STORE(fLSBOLDNOT(PvV), \
+                            fEA_REG(RxV), \
+                            VsV_off, ALIGN, \
+                            fPM_I(RxV, siV * sizeof(MMVector)))
+#define fGEN_TCG_PRED_VEC_STORE_new_pred_pi \
+    fGEN_TCG_PRED_VEC_STORE(fLSBOLD(PvV), \
+                            fEA_REG(RxV), \
+                            OsN_off, true, \
+                            fPM_I(RxV, siV * sizeof(MMVector)))
+#define fGEN_TCG_PRED_VEC_STORE_new_npred_pi \
+    fGEN_TCG_PRED_VEC_STORE(fLSBOLDNOT(PvV), \
+                            fEA_REG(RxV), \
+                            OsN_off, true, \
+                            fPM_I(RxV, siV * sizeof(MMVector)))
+
+#define fGEN_TCG_V6_vS32b_pred_pi(SHORTCODE) \
+    fGEN_TCG_PRED_VEC_STORE_pred_pi(true)
+#define fGEN_TCG_V6_vS32b_npred_pi(SHORTCODE) \
+    fGEN_TCG_PRED_VEC_STORE_npred_pi(true)
+#define fGEN_TCG_V6_vS32Ub_pred_pi(SHORTCODE) \
+    fGEN_TCG_PRED_VEC_STORE_pred_pi(false)
+#define fGEN_TCG_V6_vS32Ub_npred_pi(SHORTCODE) \
+    fGEN_TCG_PRED_VEC_STORE_npred_pi(false)
+#define fGEN_TCG_V6_vS32b_nt_pred_pi(SHORTCODE) \
+    fGEN_TCG_PRED_VEC_STORE_pred_pi(true)
+#define fGEN_TCG_V6_vS32b_nt_npred_pi(SHORTCODE) \
+    fGEN_TCG_PRED_VEC_STORE_npred_pi(true)
+#define fGEN_TCG_V6_vS32b_new_pred_pi(SHORTCODE) \
+    fGEN_TCG_PRED_VEC_STORE_new_pred_pi
+#define fGEN_TCG_V6_vS32b_new_npred_pi(SHORTCODE) \
+    fGEN_TCG_PRED_VEC_STORE_new_npred_pi
+#define fGEN_TCG_V6_vS32b_nt_new_pred_pi(SHORTCODE) \
+    fGEN_TCG_PRED_VEC_STORE_new_pred_pi
+#define fGEN_TCG_V6_vS32b_nt_new_npred_pi(SHORTCODE) \
+    fGEN_TCG_PRED_VEC_STORE_new_npred_pi
+
+#define fGEN_TCG_PRED_VEC_STORE_pred_ai(ALIGN) \
+    fGEN_TCG_PRED_VEC_STORE(fLSBOLD(PvV), \
+                            fEA_RI(RtV, siV * sizeof(MMVector)), \
+                            VsV_off, ALIGN, \
+                            do { } while (0))
+#define fGEN_TCG_PRED_VEC_STORE_npred_ai(ALIGN) \
+    fGEN_TCG_PRED_VEC_STORE(fLSBOLDNOT(PvV), \
+                            fEA_RI(RtV, siV * sizeof(MMVector)), \
+                            VsV_off, ALIGN, \
+                            do { } while (0))
+#define fGEN_TCG_PRED_VEC_STORE_new_pred_ai \
+    fGEN_TCG_PRED_VEC_STORE(fLSBOLD(PvV), \
+                            fEA_RI(RtV, siV * sizeof(MMVector)), \
+                            OsN_off, true, \
+                            do { } while (0))
+#define fGEN_TCG_PRED_VEC_STORE_new_npred_ai \
+    fGEN_TCG_PRED_VEC_STORE(fLSBOLDNOT(PvV), \
+                            fEA_RI(RtV, siV * sizeof(MMVector)), \
+                            OsN_off, true, \
+                            do { } while (0))
+
+#define fGEN_TCG_V6_vS32b_pred_ai(SHORTCODE) \
+    fGEN_TCG_PRED_VEC_STORE_pred_ai(true)
+#define fGEN_TCG_V6_vS32b_npred_ai(SHORTCODE) \
+    fGEN_TCG_PRED_VEC_STORE_npred_ai(true)
+#define fGEN_TCG_V6_vS32Ub_pred_ai(SHORTCODE) \
+    fGEN_TCG_PRED_VEC_STORE_pred_ai(false)
+#define fGEN_TCG_V6_vS32Ub_npred_ai(SHORTCODE) \
+    fGEN_TCG_PRED_VEC_STORE_npred_ai(false)
+#define fGEN_TCG_V6_vS32b_nt_pred_ai(SHORTCODE) \
+    fGEN_TCG_PRED_VEC_STORE_pred_ai(true)
+#define fGEN_TCG_V6_vS32b_nt_npred_ai(SHORTCODE) \
+    fGEN_TCG_PRED_VEC_STORE_npred_ai(true)
+#define fGEN_TCG_V6_vS32b_new_pred_ai(SHORTCODE) \
+    fGEN_TCG_PRED_VEC_STORE_new_pred_ai
+#define fGEN_TCG_V6_vS32b_new_npred_ai(SHORTCODE) \
+    fGEN_TCG_PRED_VEC_STORE_new_npred_ai
+#define fGEN_TCG_V6_vS32b_nt_new_pred_ai(SHORTCODE) \
+    fGEN_TCG_PRED_VEC_STORE_new_pred_ai
+#define fGEN_TCG_V6_vS32b_nt_new_npred_ai(SHORTCODE) \
+    fGEN_TCG_PRED_VEC_STORE_new_npred_ai
+
+#define fGEN_TCG_PRED_VEC_STORE_pred_ppu(ALIGN) \
+    fGEN_TCG_PRED_VEC_STORE(fLSBOLD(PvV), \
+                            fEA_REG(RxV), \
+                            VsV_off, ALIGN, \
+                            fPM_M(RxV, MuV))
+#define fGEN_TCG_PRED_VEC_STORE_npred_ppu(ALIGN) \
+    fGEN_TCG_PRED_VEC_STORE(fLSBOLDNOT(PvV), \
+                            fEA_REG(RxV), \
+                            VsV_off, ALIGN, \
+                            fPM_M(RxV, MuV))
+#define fGEN_TCG_PRED_VEC_STORE_new_pred_ppu \
+    fGEN_TCG_PRED_VEC_STORE(fLSBOLD(PvV), \
+                            fEA_REG(RxV), \
+                            OsN_off, true, \
+                            fPM_M(RxV, MuV))
+#define fGEN_TCG_PRED_VEC_STORE_new_npred_ppu \
+    fGEN_TCG_PRED_VEC_STORE(fLSBOLDNOT(PvV), \
+                            fEA_REG(RxV), \
+                            OsN_off, true, \
+                            fPM_M(RxV, MuV))
+
+#define fGEN_TCG_V6_vS32b_pred_ppu(SHORTCODE) \
+    fGEN_TCG_PRED_VEC_STORE_pred_ppu(true)
+#define fGEN_TCG_V6_vS32b_npred_ppu(SHORTCODE) \
+    fGEN_TCG_PRED_VEC_STORE_npred_ppu(true)
+#define fGEN_TCG_V6_vS32Ub_pred_ppu(SHORTCODE) \
+    fGEN_TCG_PRED_VEC_STORE_pred_ppu(false)
+#define fGEN_TCG_V6_vS32Ub_npred_ppu(SHORTCODE) \
+    fGEN_TCG_PRED_VEC_STORE_npred_ppu(false)
+#define fGEN_TCG_V6_vS32b_nt_pred_ppu(SHORTCODE) \
+    fGEN_TCG_PRED_VEC_STORE_pred_ppu(true)
+#define fGEN_TCG_V6_vS32b_nt_npred_ppu(SHORTCODE) \
+    fGEN_TCG_PRED_VEC_STORE_npred_ppu(true)
+#define fGEN_TCG_V6_vS32b_new_pred_ppu(SHORTCODE) \
+    fGEN_TCG_PRED_VEC_STORE_new_pred_ppu
+#define fGEN_TCG_V6_vS32b_new_npred_ppu(SHORTCODE) \
+    fGEN_TCG_PRED_VEC_STORE_new_npred_ppu
+#define fGEN_TCG_V6_vS32b_nt_new_pred_ppu(SHORTCODE) \
+    fGEN_TCG_PRED_VEC_STORE_new_pred_ppu
+#define fGEN_TCG_V6_vS32b_nt_new_npred_ppu(SHORTCODE) \
+    fGEN_TCG_PRED_VEC_STORE_new_npred_ppu
+
+/* Masked vector stores */
+#define fGEN_TCG_V6_vS32b_qpred_pi(SHORTCODE)              SHORTCODE
+#define fGEN_TCG_V6_vS32b_nt_qpred_pi(SHORTCODE)           SHORTCODE
+#define fGEN_TCG_V6_vS32b_qpred_ai(SHORTCODE)              SHORTCODE
+#define fGEN_TCG_V6_vS32b_nt_qpred_ai(SHORTCODE)           SHORTCODE
+#define fGEN_TCG_V6_vS32b_qpred_ppu(SHORTCODE)             SHORTCODE
+#define fGEN_TCG_V6_vS32b_nt_qpred_ppu(SHORTCODE)          SHORTCODE
+#define fGEN_TCG_V6_vS32b_nqpred_pi(SHORTCODE)             SHORTCODE
+#define fGEN_TCG_V6_vS32b_nt_nqpred_pi(SHORTCODE)          SHORTCODE
+#define fGEN_TCG_V6_vS32b_nqpred_ai(SHORTCODE)             SHORTCODE
+#define fGEN_TCG_V6_vS32b_nt_nqpred_ai(SHORTCODE)          SHORTCODE
+#define fGEN_TCG_V6_vS32b_nqpred_ppu(SHORTCODE)            SHORTCODE
+#define fGEN_TCG_V6_vS32b_nt_nqpred_ppu(SHORTCODE)         SHORTCODE
+
+/* Store release not modelled in qemu, but need to suppress compiler warnings */
+#define fGEN_TCG_V6_vS32b_srls_pi(SHORTCODE) \
+    do { \
+        siV = siV; \
+    } while (0)
+#define fGEN_TCG_V6_vS32b_srls_ai(SHORTCODE) \
+    do { \
+        RtV = RtV; \
+        siV = siV; \
+    } while (0)
+#define fGEN_TCG_V6_vS32b_srls_ppu(SHORTCODE) \
+    do { \
+        MuV = MuV; \
+    } while (0)
+
 #endif
-- 
2.7.4


  parent reply	other threads:[~2021-10-12 10:53 UTC|newest]

Thread overview: 45+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-10-12 10:10 [PATCH v4 00/30] Hexagon HVX (target/hexagon) patch series Taylor Simpson
2021-10-12 10:10 ` [PATCH v4 01/30] Hexagon HVX (target/hexagon) README Taylor Simpson
2021-10-12 10:10 ` [PATCH v4 02/30] Hexagon HVX (target/hexagon) add Hexagon Vector eXtensions (HVX) to core Taylor Simpson
2021-10-12 10:10 ` [PATCH v4 03/30] Hexagon HVX (target/hexagon) register names Taylor Simpson
2021-10-12 10:10 ` [PATCH v4 04/30] Hexagon HVX (target/hexagon) instruction attributes Taylor Simpson
2021-10-12 10:10 ` [PATCH v4 05/30] Hexagon HVX (target/hexagon) macros Taylor Simpson
2021-10-12 10:10 ` [PATCH v4 06/30] Hexagon HVX (target/hexagon) import macro definitions Taylor Simpson
2021-10-12 10:10 ` [PATCH v4 07/30] Hexagon HVX (target/hexagon) semantics generator Taylor Simpson
2021-10-12 10:10 ` [PATCH v4 08/30] Hexagon HVX (target/hexagon) semantics generator - part 2 Taylor Simpson
2021-10-12 10:10 ` [PATCH v4 09/30] Hexagon HVX (target/hexagon) C preprocessor for decode tree Taylor Simpson
2021-10-12 10:10 ` [PATCH v4 10/30] Hexagon HVX (target/hexagon) instruction utility functions Taylor Simpson
2021-10-29 18:53   ` Richard Henderson
2021-10-29 23:37     ` Taylor Simpson
2021-10-12 10:10 ` [PATCH v4 11/30] Hexagon HVX (target/hexagon) helper functions Taylor Simpson
2021-10-29 18:58   ` Richard Henderson
2021-10-12 10:10 ` [PATCH v4 12/30] Hexagon HVX (target/hexagon) TCG generation Taylor Simpson
2021-10-29 18:59   ` Richard Henderson
2021-10-12 10:10 ` [PATCH v4 13/30] Hexagon HVX (target/hexagon) helper overrides infrastructure Taylor Simpson
2021-10-29 16:48   ` Philippe Mathieu-Daudé
2021-10-29 19:00   ` Richard Henderson
2021-10-12 10:10 ` [PATCH v4 14/30] Hexagon HVX (target/hexagon) helper overrides for histogram instructions Taylor Simpson
2021-10-29 19:04   ` Richard Henderson
2021-10-12 10:10 ` [PATCH v4 15/30] Hexagon HVX (target/hexagon) helper overrides - vector assign & cmov Taylor Simpson
2021-10-12 10:10 ` [PATCH v4 16/30] Hexagon HVX (target/hexagon) helper overrides - vector add & sub Taylor Simpson
2021-10-12 10:10 ` [PATCH v4 17/30] Hexagon HVX (target/hexagon) helper overrides - vector shifts Taylor Simpson
2021-10-12 10:10 ` [PATCH v4 18/30] Hexagon HVX (target/hexagon) helper overrides - vector max/min Taylor Simpson
2021-10-12 10:10 ` [PATCH v4 19/30] Hexagon HVX (target/hexagon) helper overrides - vector logical ops Taylor Simpson
2021-10-29 19:06   ` Richard Henderson
2021-10-12 10:10 ` [PATCH v4 20/30] Hexagon HVX (target/hexagon) helper overrides - vector compares Taylor Simpson
2021-10-12 10:10 ` [PATCH v4 21/30] Hexagon HVX (target/hexagon) helper overrides - vector splat and abs Taylor Simpson
2021-10-12 10:11 ` [PATCH v4 22/30] Hexagon HVX (target/hexagon) helper overrides - vector loads Taylor Simpson
2021-10-12 10:11 ` Taylor Simpson [this message]
2021-10-12 10:11 ` [PATCH v4 24/30] Hexagon HVX (target/hexagon) import semantics Taylor Simpson
2021-10-12 10:11 ` [PATCH v4 25/30] Hexagon HVX (target/hexagon) instruction decoding Taylor Simpson
2021-10-12 10:11 ` [PATCH v4 26/30] Hexagon HVX (target/hexagon) import instruction encodings Taylor Simpson
2021-10-29 19:08   ` Richard Henderson
2021-10-12 10:11 ` [PATCH v4 27/30] Hexagon HVX (tests/tcg/hexagon) vector_add_int test Taylor Simpson
2021-10-29 19:10   ` Richard Henderson
2021-10-12 10:11 ` [PATCH v4 28/30] Hexagon HVX (tests/tcg/hexagon) hvx_misc test Taylor Simpson
2021-10-29 19:11   ` Richard Henderson
2021-10-12 10:11 ` [PATCH v4 29/30] Hexagon HVX (tests/tcg/hexagon) scatter_gather test Taylor Simpson
2021-10-29 19:13   ` Richard Henderson
2021-10-12 10:11 ` [PATCH v4 30/30] Hexagon HVX (tests/tcg/hexagon) histogram test Taylor Simpson
2021-10-29 19:15   ` Richard Henderson
2021-10-29 19:18     ` Taylor Simpson

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1634033468-23566-24-git-send-email-tsimpson@quicinc.com \
    --to=tsimpson@quicinc.com \
    --cc=ale@rev.ng \
    --cc=bcain@quicinc.com \
    --cc=f4bug@amsat.org \
    --cc=qemu-devel@nongnu.org \
    --cc=richard.henderson@linaro.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).