1; RUN: llc < %s -mcpu=cortex-a15 -verify-machineinstrs -arm-atomic-cfg-tidy=0 | FileCheck %s 2 3; Check a spill right after a function call with large struct byval is correctly 4; generated. 5; PR16393 6 7; CHECK: set_stored_macroblock_parameters 8; CHECK: str r{{.*}}, [sp, [[SLOT:#[0-9]+]]] @ 4-byte Spill 9; CHECK: bl RestoreMVBlock8x8 10; CHECK: bl RestoreMVBlock8x8 11; CHECK: bl RestoreMVBlock8x8 12; CHECK: ldr r{{.*}}, [sp, [[SLOT]]] @ 4-byte Reload 13 14target triple = "armv7l-unknown-linux-gnueabihf" 15 16%structA = type { double, [16 x [16 x i16]], [16 x [16 x i16]], [16 x [16 x i16]], i32****, i32***, i32, i16, [4 x i32], [4 x i32], i8**, [16 x i8], [16 x i8], i32, i64, i32, i16******, i16******, [2 x [4 x [4 x i8]]], i32, i32, i32, i32, i32, i32, i32, i32, i32 } 17%structB = type { i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, float, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i8**, i8**, i32, i32***, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, [9 x [16 x [16 x i16]]], [5 x [16 x [16 x i16]]], [9 x [8 x [8 x i16]]], [2 x [4 x [16 x [16 x i16]]]], [16 x [16 x i16]], [16 x [16 x i32]], i32****, i32***, i32***, i32***, i32****, i32****, %structC*, %structD*, %structK*, i32*, i32*, i32, i32, i32, i32, [4 x [4 x i32]], i32, i32, i32, i32, i32, double, i32, i32, i32, i32, i16******, i16******, i16******, i16******, [15 x i16], i32, i32, i32, i32, i32, i32, i32, i32, [6 x [32 x i32]], i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, [1 x i32], i32, i32, [2 x i32], i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, %structL*, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, double**, double***, i32***, double**, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, [3 x [2 x i32]], [2 x i32], i32, i32, i16, i32, i32, i32, i32, i32 } 18%structC = type { i32, i32, [100 x %structD*], i32, float, float, float } 19%structD = type { i32, i32, i32, i32, i32, i32, %structE*, %structH*, %structJ*, i32, i32*, i32*, i32*, i32, i32*, i32*, i32*, i32 (i32)*, [3 x [2 x i32]] } 20%structE = type { %structF*, %structG, %structG } 21%structF = type { i32, i32, i8, i32, i32, i8, i8, i32, i32, i8*, i32 } 22%structG = type { i32, i32, i32, i32, i32, i8*, i32*, i32, i32 } 23%structH = type { [3 x [11 x %structI]], [2 x [9 x %structI]], [2 x [10 x %structI]], [2 x [6 x %structI]], [4 x %structI], [4 x %structI], [3 x %structI] } 24%structI = type { i16, i8, i32 } 25%structJ = type { [2 x %structI], [4 x %structI], [3 x [4 x %structI]], [10 x [4 x %structI]], [10 x [15 x %structI]], [10 x [15 x %structI]], [10 x [5 x %structI]], [10 x [5 x %structI]], [10 x [15 x %structI]], [10 x [15 x %structI]] } 26%structK = type { i32, i32, i32, [2 x i32], i32, [8 x i32], %structK*, %structK*, i32, [2 x [4 x [4 x [2 x i32]]]], [16 x i8], [16 x i8], i32, i64, [4 x i32], [4 x i32], i64, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i16, double, i32, i32, i32, i32, i32, i32, i32, i32, i32 } 27%structL = type { i32, i32, i32, i32, i32, %structL* } 28%structM = type { i32, i32, i32, i32, i32, i32, [6 x [33 x i64]], [6 x [33 x i64]], [6 x [33 x i64]], [6 x [33 x i64]], i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i16**, i16****, i16****, i16*****, i16***, i8*, i8***, i64***, i64***, i16****, i8**, i8**, %structM*, %structM*, %structM*, i32, i32, i32, i32, i32, i32, i32 } 29%structN = type { i32, [16 x [16 x i32]], [16 x [16 x i32]], [16 x [16 x i32]], [3 x [16 x [16 x i32]]], [4 x i16], [4 x i8], [4 x i8], [4 x i8], [16 x [16 x i16]], [16 x [16 x i16]], [16 x [16 x i32]] } 30 31@cofAC = external global i32****, align 4 32@cofDC = external global i32***, align 4 33@rdopt = external global %structA*, align 4 34@img = external global %structB* 35@enc_picture = external global %structM* 36@si_frame_indicator = external global i32, align 4 37@sp2_frame_indicator = external global i32, align 4 38@lrec = external global i32**, align 4 39@tr8x8 = external global %structN, align 4 40@best_mode = external global i16, align 2 41@best_c_imode = external global i32, align 4 42@best_i16offset = external global i32, align 4 43@bi_pred_me = external global i16, align 2 44@b8mode = external global [4 x i32], align 4 45@b8pdir = external global [4 x i32], align 4 46@b4_intra_pred_modes = external global [16 x i8], align 1 47@b8_intra_pred_modes8x8 = external global [16 x i8], align 1 48@b4_ipredmode = external global [16 x i8], align 1 49@b8_ipredmode8x8 = external global [4 x [4 x i8]], align 1 50@rec_mbY = external global [16 x [16 x i16]], align 2 51@lrec_rec = external global [16 x [16 x i32]], align 4 52@rec_mbU = external global [16 x [16 x i16]], align 2 53@rec_mbV = external global [16 x [16 x i16]], align 2 54@lrec_rec_U = external global [16 x [16 x i32]], align 4 55@lrec_uv = external global i32***, align 4 56@lrec_rec_V = external global [16 x [16 x i32]], align 4 57@cbp = external global i32, align 4 58@cbp_blk = external global i64, align 8 59@luma_transform_size_8x8_flag = external global i32, align 4 60@frefframe = external global [4 x [4 x i8]], align 1 61@brefframe = external global [4 x [4 x i8]], align 1 62 63; Function Attrs: nounwind 64declare void @llvm.memcpy.p0i8.p0i8.i32(i8* nocapture, i8* nocapture, i32, i32, i1) #0 65 66; Function Attrs: nounwind 67declare void @llvm.memset.p0i8.i32(i8* nocapture, i8, i32, i32, i1) #0 68 69; Function Attrs: nounwind 70declare void @SetMotionVectorsMB(%structK* nocapture, i32) #1 71 72; Function Attrs: nounwind 73define void @set_stored_macroblock_parameters() #1 { 74entry: 75 %0 = load %structB*, %structB** @img, align 4 76 %1 = load i32, i32* undef, align 4 77 %mb_data = getelementptr inbounds %structB, %structB* %0, i32 0, i32 61 78 %2 = load %structK*, %structK** %mb_data, align 4 79 br label %for.body 80 81for.body: ; preds = %for.body, %entry 82 br i1 undef, label %for.end, label %for.body 83 84for.end: ; preds = %for.body 85 br i1 undef, label %for.body20, label %if.end 86 87for.body20: ; preds = %for.end 88 unreachable 89 90if.end: ; preds = %for.end 91 br i1 undef, label %if.end40, label %for.cond31.preheader 92 93for.cond31.preheader: ; preds = %if.end 94 unreachable 95 96if.end40: ; preds = %if.end 97 br i1 undef, label %if.end43, label %if.then42 98 99if.then42: ; preds = %if.end40 100 br label %if.end43 101 102if.end43: ; preds = %if.then42, %if.end40 103 br i1 undef, label %if.end164, label %for.cond47.preheader 104 105for.cond47.preheader: ; preds = %if.end43 106 br i1 undef, label %for.body119, label %if.end164 107 108for.body119: ; preds = %for.body119, %for.cond47.preheader 109 br i1 undef, label %for.body119, label %if.end164 110 111if.end164: ; preds = %for.body119, %for.cond47.preheader, %if.end43 112 store i32*** null, i32**** @cofDC, align 4 113 %mb_type = getelementptr inbounds %structK, %structK* %2, i32 %1, i32 8 114 br i1 undef, label %if.end230, label %if.then169 115 116if.then169: ; preds = %if.end164 117 br i1 undef, label %for.cond185.preheader, label %for.cond210.preheader 118 119for.cond185.preheader: ; preds = %if.then169 120 unreachable 121 122for.cond210.preheader: ; preds = %if.then169 123 unreachable 124 125if.end230: ; preds = %if.end164 126 tail call void @llvm.memcpy.p0i8.p0i8.i32(i8* undef, i8* bitcast ([4 x i32]* @b8mode to i8*), i32 16, i32 4, i1 false) 127 %b8pdir = getelementptr inbounds %structK, %structK* %2, i32 %1, i32 15 128 %3 = bitcast [4 x i32]* %b8pdir to i8* 129 tail call void @llvm.memcpy.p0i8.p0i8.i32(i8* %3, i8* bitcast ([4 x i32]* @b8pdir to i8*), i32 16, i32 4, i1 false) 130 br i1 undef, label %if.end236, label %if.then233 131 132if.then233: ; preds = %if.end230 133 unreachable 134 135if.end236: ; preds = %if.end230 136 %cmp242 = icmp ne i16 undef, 8 137 %4 = load i32, i32* @luma_transform_size_8x8_flag, align 4 138 %tobool245 = icmp ne i32 %4, 0 139 %or.cond812 = or i1 %cmp242, %tobool245 140 br i1 %or.cond812, label %if.end249, label %land.lhs.true246 141 142land.lhs.true246: ; preds = %if.end236 143 br i1 undef, label %if.end249, label %if.then248 144 145if.then248: ; preds = %land.lhs.true246 146 tail call void asm sideeffect "", "~{r1},~{r2},~{r3},~{r4},~{r5},~{r6},~{r7},~{r8},~{r9},~{r10},~{r11}"() nounwind 147 tail call void @RestoreMVBlock8x8(i32 1, i32 0, %structN* byval @tr8x8, i32 0) #0 148 tail call void @RestoreMVBlock8x8(i32 1, i32 2, %structN* byval @tr8x8, i32 0) #0 149 tail call void @RestoreMVBlock8x8(i32 1, i32 3, %structN* byval @tr8x8, i32 0) #0 150 br label %if.end249 151 152if.end249: ; preds = %if.then248, %land.lhs.true246, %if.end236 153 %5 = load i32, i32* @luma_transform_size_8x8_flag, align 4 154 %6 = load %structA*, %structA** @rdopt, align 4 155 %luma_transform_size_8x8_flag264 = getelementptr inbounds %structA, %structA* %6, i32 0, i32 21 156 store i32 %5, i32* %luma_transform_size_8x8_flag264, align 4 157 %7 = load i32, i32* undef, align 4 158 %add281 = add nsw i32 %7, 0 159 br label %for.body285 160 161for.body285: ; preds = %for.inc503, %if.end249 162 %8 = phi %structB* [ undef, %if.end249 ], [ %.pre1155, %for.inc503 ] 163 %i.21103 = phi i32 [ 0, %if.end249 ], [ %inc504, %for.inc503 ] 164 %block_x286 = getelementptr inbounds %structB, %structB* %8, i32 0, i32 37 165 %9 = load i32, i32* %block_x286, align 4 166 %add287 = add nsw i32 %9, %i.21103 167 %shr289 = ashr i32 %i.21103, 1 168 %add290 = add nsw i32 %shr289, 0 169 %arrayidx292 = getelementptr inbounds %structK, %structK* %2, i32 %1, i32 15, i32 %add290 170 %10 = load %structM*, %structM** @enc_picture, align 4 171 %ref_idx = getelementptr inbounds %structM, %structM* %10, i32 0, i32 35 172 %11 = load i8***, i8**** %ref_idx, align 4 173 %12 = load i8**, i8*** %11, align 4 174 %arrayidx313 = getelementptr inbounds i8*, i8** %12, i32 %add281 175 %13 = load i8*, i8** %arrayidx313, align 4 176 %arrayidx314 = getelementptr inbounds i8, i8* %13, i32 %add287 177 store i8 -1, i8* %arrayidx314, align 1 178 %14 = load %structB*, %structB** @img, align 4 179 %MbaffFrameFlag327 = getelementptr inbounds %structB, %structB* %14, i32 0, i32 100 180 %15 = load i32, i32* %MbaffFrameFlag327, align 4 181 %tobool328 = icmp eq i32 %15, 0 182 br i1 %tobool328, label %if.end454, label %if.then329 183 184if.then329: ; preds = %for.body285 185 %16 = load %structA*, %structA** @rdopt, align 4 186 br label %if.end454 187 188if.end454: ; preds = %if.then329, %for.body285 189 %17 = load i32, i32* %arrayidx292, align 4 190 %cmp457 = icmp eq i32 %17, 0 191 br i1 %cmp457, label %if.then475, label %lor.lhs.false459 192 193lor.lhs.false459: ; preds = %if.end454 194 %18 = load i32, i32* %mb_type, align 4 195 switch i32 %18, label %for.inc503 [ 196 i32 9, label %if.then475 197 i32 11, label %if.then475 198 i32 13, label %if.then475 199 i32 14, label %if.then475 200 ] 201 202if.then475: ; preds = %lor.lhs.false459, %lor.lhs.false459, %lor.lhs.false459, %lor.lhs.false459, %if.end454 203 store i16 0, i16* undef, align 2 204 br label %for.inc503 205 206for.inc503: ; preds = %if.then475, %lor.lhs.false459 207 %inc504 = add nsw i32 %i.21103, 1 208 %.pre1155 = load %structB*, %structB** @img, align 4 209 br label %for.body285 210} 211 212; Function Attrs: nounwind 213declare void @update_offset_params(i32, i32) #1 214 215; Function Attrs: nounwind 216declare void @RestoreMVBlock8x8(i32, i32, %structN* byval nocapture, i32) #1 217 218attributes #0 = { nounwind } 219attributes #1 = { nounwind "less-precise-fpmad"="false" "no-frame-pointer-elim"="false" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "unsafe-fp-math"="false" "use-soft-float"="false" } 220