wasmtime / issue #4761 cranelift-fuzzgen fuzzbug: "Segv o... · git-wasmtime

Stream: git-wasmtime

Topic: wasmtime / issue #4761 cranelift-fuzzgen fuzzbug: "Segv o...

Wasmtime GitHub notifications bot (Aug 23 2022 at 15:58):

https://oss-fuzz.com/testcase-detail/4896860279537664

ERROR: AddressSanitizer: SEGV on unknown address (pc 0x6250000014d5 bp 0x7fff30133db0 sp 0x7fff30133ad0 T0)
--
  | ==743==The signal is caused by a READ memory access.
  | ==743==Hint: this fault was caused by a dereference of a high value address (see register values below).  Disassemble the provided pc to learn which register was used.
  | SCARINESS: 20 (wild-addr-read)
  | #0 0x6250000014d5  (<unknown module>)
  | #1 0x62500000175d  (<unknown module>)
  | #0 0x5579a6d0c6c4 in cranelift_filetests::function_runner::CompiledFunction::call::ha6b9162cd2e21784 wasmtime/cranelift/filetests/src/function_runner.rs:183:9

input:

IFWoAAAAAE0AAABSSUl9LCB7eAx6ZcYNR9T0RvCdMjI0MjaWMgAAAAA7ACypAH19f319fX0AAAA7
JAABqQB9fX19fX19fX0IAAAAAAAAAQAAwyvDw8PDAADJ/zw8PEHDw8MAAGUBb/PGANpGAAAAAAAj
OwAsqW9vb29vb28AfX19fX19fQAAADskAAGpAH19fX19fX19fQ7/////////CAAAAAAAAAEAAMMr
w8PDwwAAyf88PDxBw8PDAABlAW/zxgDaRgAAAAAAAAAAAAAAAACTgoL/////////MAAAAAAAAAUA
AAAACIKCgoKCgkEAgoKCgoJNAAAAgoKCgoKCgoKCgoKCwoKCgoKCgoKSioIEAAcAgoKCgoKCgoKC
goKCgoKCGIKC//8wAAAAAAAABQAAAAAIgoKCgoKCQQCCgoKCgk0AAACCgoKCgoKCgoKCgoLCeoKC
eYKCgpKKggQAAACCgoKCgoKCgoKCgoKCgoIYgoKCgoKCgoKCgoKCFACCgjOCgoKCgoL/////goKC
goKCgoKCgoKCgoJBAUFB

cc @afonso360

Wasmtime GitHub notifications bot (Aug 23 2022 at 16:27):

afonso360 commented on issue #4761:

<details>
<summary> Formatted </summary>

ubuntu@instance-20220805-0848:~/git/wasmtime/fuzz$ cargo fuzz fmt cranelift-fuzzgen ./4761.in --no-default-features

Output of `std::fmt::Debug`:

;; Fuzzgen test case

test interpret
test run
set enable_llvm_abi_extensions
target aarch64
target s390x
target x86_64

function u0:1(i128, b1, b1, i128, b1, i16 uext, i8 sext, i64 sext, i32, i64 sext, i128, i128 sext, i64 uext, f32, i128 sext) -> i64 sext, f32, i16, b1, b1 sext, b1 sext, i8, i128 sext, f64 sext, i128 sext, i128, b1, i32, b1 uext, i8, i128 sext system_v {
    ss0 = explicit_slot 59
    ss1 = explicit_slot 0
    ss2 = explicit_slot 44
    ss3 = explicit_slot 40
    ss4 = explicit_slot 111
    ss5 = explicit_slot 111
    ss6 = explicit_slot 111
    ss7 = explicit_slot 111
    jt0 = jump_table [block3, block2, block2, block2, block2, block2, block2, block2, block2]

block0(v0: i128, v1: b1, v2: b1, v3: i128, v4: b1, v5: i16, v6: i8, v7: i64, v8: i32, v9: i64, v10: i128, v11: i128, v12: i64, v13: f32, v14: i128):
    v39 = f64const 0x1.d7d7d7d7d006fp984
    v40 = iconst.i64 0xa901_0024_3b00_0000
    v41 = iconst.i64 0x7d7d_7d7d_7d7d_7d00
    v42 = iconcat v41, v40  ; v41 = 0x7d7d_7d7d_7d7d_7d00, v40 = 0xa901_0024_3b00_0000
    v43 = iconst.i64 0xffff_ffff_ffff_0e7d
    v44 = iconst.i64 2303
    v45 = iconcat v44, v43  ; v44 = 2303, v43 = 0xffff_ffff_ffff_0e7d
    v46 = iconst.i8 0
    v47 = bconst.b1 true
    v48 = iconst.i32 0xffff_ffff_c3c3_c3c3
    v49 = bconst.b1 false
    v50 = iconst.i8 -1
    v51 = iconst.i64 0xc3c3_c341_3c3c
    v52 = iconst.i128 0
    v53 = iconst.i64 0
    v54 = iconst.i32 0
    v55 = iconst.i16 0
    v56 = iconst.i8 0
    stack_store v52, ss0  ; v52 = 0
    stack_store v52, ss0+16  ; v52 = 0
    stack_store v52, ss0+32  ; v52 = 0
    stack_store v53, ss0+48  ; v53 = 0
    stack_store v55, ss0+56  ; v55 = 0
    stack_store v56, ss0+58  ; v56 = 0
    stack_store v52, ss2  ; v52 = 0
    stack_store v52, ss2+16  ; v52 = 0
    stack_store v53, ss2+32  ; v53 = 0
    stack_store v54, ss2+40  ; v54 = 0
    stack_store v52, ss3  ; v52 = 0
    stack_store v52, ss3+16  ; v52 = 0
    stack_store v53, ss3+32  ; v53 = 0
    stack_store v52, ss4  ; v52 = 0
    stack_store v52, ss4+16  ; v52 = 0
    stack_store v52, ss4+32  ; v52 = 0
    stack_store v52, ss4+48  ; v52 = 0
    stack_store v52, ss4+64  ; v52 = 0
    stack_store v52, ss4+80  ; v52 = 0
    stack_store v53, ss4+96  ; v53 = 0
    stack_store v54, ss4+104  ; v54 = 0
    stack_store v55, ss4+108  ; v55 = 0
    stack_store v56, ss4+110  ; v56 = 0
    stack_store v52, ss5  ; v52 = 0
    stack_store v52, ss5+16  ; v52 = 0
    stack_store v52, ss5+32  ; v52 = 0
    stack_store v52, ss5+48  ; v52 = 0
    stack_store v52, ss5+64  ; v52 = 0
    stack_store v52, ss5+80  ; v52 = 0
    stack_store v53, ss5+96  ; v53 = 0
    stack_store v54, ss5+104  ; v54 = 0
    stack_store v55, ss5+108  ; v55 = 0
    stack_store v56, ss5+110  ; v56 = 0
    stack_store v52, ss6  ; v52 = 0
    stack_store v52, ss6+16  ; v52 = 0
    stack_store v52, ss6+32  ; v52 = 0
    stack_store v52, ss6+48  ; v52 = 0
    stack_store v52, ss6+64  ; v52 = 0
    stack_store v52, ss6+80  ; v52 = 0
    stack_store v53, ss6+96  ; v53 = 0
    stack_store v54, ss6+104  ; v54 = 0
    stack_store v55, ss6+108  ; v55 = 0
    stack_store v56, ss6+110  ; v56 = 0
    stack_store v52, ss7  ; v52 = 0
    stack_store v52, ss7+16  ; v52 = 0
    stack_store v52, ss7+32  ; v52 = 0
    stack_store v52, ss7+48  ; v52 = 0
    stack_store v52, ss7+64  ; v52 = 0
    stack_store v52, ss7+80  ; v52 = 0
    stack_store v53, ss7+96  ; v53 = 0
    stack_store v54, ss7+104  ; v54 = 0
    stack_store v55, ss7+108  ; v55 = 0
    stack_store v56, ss7+110  ; v56 = 0
    v57 = iadd v6, v6
    nop
    stack_store v5, ss0
    nop
    nop
    nop
    nop
    nop
    nop
    nop
    nop
    nop
    nop
    nop
    nop
    v58 = uextend.i64 v46  ; v46 = 0
    v59 = udiv v51, v51  ; v51 = 0xc3c3_c341_3c3c, v51 = 0xc3c3_c341_3c3c
    v60 = udiv v59, v59
    nop
    nop
    nop
    nop
    nop
    nop
    v61 = iadd v0, v0
    nop
    v62 = isub v8, v8
    v63 = ushr v62, v46  ; v46 = 0
    nop
    v64 = ushr v62, v46  ; v46 = 0
    v65 = ushr v63, v57
    nop
    v66 = ushr v65, v46  ; v46 = 0
    v67 = ushr v66, v46  ; v46 = 0
    v68 = ushr v67, v46  ; v46 = 0
    v69 = fcopysign v39, v39  ; v39 = 0x1.d7d7d7d7d006fp984, v39 = 0x1.d7d7d7d7d006fp984
    return v58, v13, v5, v1, v1, v1, v46, v14, v69, v45, v42, v1, v68, v1, v46, v61  ; v46 = 0, v46 = 0

block1(v15: i128, v16: i128, v17: i128, v18: i128, v19: b1, v20: b1) cold:
    return v70, v71, v72, v20, v20, v20, v73, v18, v74, v18, v75, v20, v76, v20, v77, v78  ; v70 = 0, v71 = 0.0, v72 = 0, v73 = 0, v74 = 0.0, v75 = 0, v76 = 0, v77 = 0, v78 = 0

block2:
    v165 = bconst.b1 false
    v112 -> v165
    v164 -> v165
    v163 = iconst.i128 0
    v110 -> v163
    v162 -> v163
    v140 = iconst.i128 0
    v126 -> v140
    v78 -> v126
    v139 = iconst.i8 0
    v125 -> v139
    v77 -> v125
    v138 = iconst.i128 0
    v124 -> v138
    v75 -> v124
    v137 = f64const 0.0
    v111 -> v137
    v123 -> v137
    v74 -> v123
    v136 = iconst.i16 0
    v109 -> v136
    v122 -> v136
    v72 -> v122
    v135 = f32const 0.0
    v108 -> v135
    v121 -> v135
    v71 -> v121
    v134 = iconst.i64 0
    v107 -> v134
    v120 -> v134
    v70 -> v120
    v133 = iconst.i32 0
    v85 -> v133
    v132 = iconst.i32 0
    v84 -> v132
    v76 -> v84
    v131 = iconst.i8 0
    v83 -> v131
    v73 -> v83
    v130 = iconst.i8 0
    v82 -> v130
    v129 = bconst.b1 false
    v81 -> v129
    v128 = iconst.i128 0
    v80 -> v128
    v127 = iconst.i128 0
    v79 -> v127
    br_icmp eq v82, v83, block1(v79, v79, v79, v80, v81, v81)  ; v82 = 0, v83 = 0, v79 = 0, v79 = 0, v79 = 0, v80 = 0, v81 = false, v81 = false
    jump block5(v84, v84, v84, v84, v85, v81, v81, v83)  ; v84 = 0, v84 = 0, v84 = 0, v84 = 0, v85 = 0, v81 = false, v81 = false, v83 = 0

block3:
    v150 = iconst.i8 0
    v95 -> v150
    v149 = iconst.i32 0
    v94 -> v149
    v148 = f64const 0.0
    v93 -> v148
    v147 = iconst.i128 0
    v92 -> v147
    v146 = iconst.i8 0
    v91 -> v146
    v145 = bconst.b1 false
    v90 -> v145
    v144 = bconst.b1 false
    v89 -> v144
    v143 = iconst.i16 0
    v88 -> v143
    v142 = f32const 0.0
    v87 -> v142
    v141 = iconst.i64 0
    v86 -> v141
    return v86, v87, v88, v89, v90, v90, v91, v92, v93, v92, v92, v90, v94, v90, v95, v92  ; v86 = 0, v87 = 0.0, v88 = 0, v89 = false, v90 = false, v90 = false, v91 = 0, v92 = 0, v93 = 0.0, v92 = 0, v92 = 0, v90 = false, v94 = 0, v90 = false, v95 = 0, v92 = 0

block4(v21: b1):
    v161 = bconst.b1 false
    v106 -> v161
    v160 = iconst.i32 0
    v105 -> v160
    v159 = iconst.i128 0
    v104 -> v159
    v158 = f64const 0.0
    v103 -> v158
    v157 = iconst.i128 0
    v102 -> v157
    v156 = iconst.i8 0
    v101 -> v156
    v155 = bconst.b1 false
    v100 -> v155
    v154 = bconst.b1 false
    v99 -> v154
    v153 = iconst.i16 0
    v98 -> v153
    v152 = f32const 0.0
    v97 -> v152
    v151 = iconst.i64 0
    v96 -> v151
    return v96, v97, v98, v99, v100, v21, v101, v102, v103, v104, v104, v21, v105, v106, v101, v104  ; v96 = 0, v97 = 0.0, v98 = 0, v99 = false, v100 = false, v101 = 0, v102 = 0, v103 = 0.0, v104 = 0, v104 = 0, v105 = 0, v106 = false, v101 = 0, v104 = 0

block5(v22: i32, v23: i32, v24: i32, v25: i32, v26: i32, v27: b1, v28: b1, v29: i8):
    return v107, v108, v109, v28, v28, v28, v29, v110, v111, v110, v110, v112, v26, v28, v29, v110  ; v107 = 0, v108 = 0.0, v109 = 0, v110 = 0, v111 = 0.0, v110 = 0, v110 = 0, v112 = false, v110 = 0

block6(v30: i64, v31: i64, v32: i8, v33: i32, v34: i32, v35: i32, v36: b1, v37: b1, v38: i128) cold:
    v172 = iconst.i8 0
    v119 -> v172
    v171 = iconst.i32 0
    v118 -> v171
    v170 = f64const 0.0
    v117 -> v170
    v169 = iconst.i128 0
    v116 -> v169
    v168 = bconst.b1 false
    v115 -> v168
    v167 = iconst.i16 0
    v114 -> v167
    v166 = f32const 0.0
    v113 -> v166
    return v31, v113, v114, v115, v37, v37, v32, v116, v117, v116, v116, v37, v118, v37, v119, v116  ; v113 = 0.0, v114 = 0, v115 = false, v116 = 0, v117 = 0.0, v116 = 0, v116 = 0, v118 = 0, v119 = 0, v116 = 0
}

; Note: the results in the below test cases are simply a placeholder and probably will be wrong

; run: u0:1(86737344494455290561902794539232559746, false, false, 0, false, 0, 0, 0, 0, 0, 0, 0, 0, 0.0, 0) == [0, 0.0, 0, false, false, false, 0, 0, 0.0, 0, 0, false, 0, false, 0, 0]

</details

Wasmtime GitHub notifications bot (Aug 24 2022 at 12:01):

afonso360 commented on issue #4761:

<details>
<summary> Minimized this to: </summary>
```
;; Fuzzgen test case

test interpret
test run
set enable_llvm_abi_extensions
target aarch64
target s390x
target x86_64

function %a(i128, b1, b1, i128, b1, i16 uext, i8 sext, i64 sext, i32, i64 sext, i128, i128 sext, i64 uext, f32, i128 sext) -> i64 sext, f32, i16, b1, b1 sext, b1 sext, i8, i128 sext, f64 sext, i128 sext, i128, b1, i32, b1 uext, i8, i128 sext windows_fastcall {
ss0 = explicit_slot 59

block0(v0: i128, v1: b1, v2: b1, v3: i128, v4: b1, v5: i16, v6: i8, v7: i64, v8: i32, v9: i64, v10: i128, v11: i128, v12: i64, v13: f32, v14: i128):
v39 = f64const 0x1.d7d7d7d7d006fp984
v40 = iconst.i64 0xa901_0024_3b00_0000
v41 = iconst.i64 0x7d7d_7d7d_7d7d_7d00
v42 = iconcat v41, v40 ; v41 = 0x7d7d_7d7d_7d7d_7d00, v40 = 0xa901_0024_3b00_0000
v43 = iconst.i64 0xffff_ffff_ffff_0e7d
v44 = iconst.i64 2303
v45 = iconcat v44, v43 ; v44 = 2303, v43 = 0xffff_ffff_ffff_0e7d
v46 = iconst.i8 0
v47 = bconst.b1 true
v48 = iconst.i32 0xffff_ffff_c3c3_c3c3
v49 = bconst.b1 false
v50 = iconst.i8 -1
v51 = iconst.i64 0xc3c3_c341_3c3c
v52 = iconst.i128 0
v53 = iconst.i64 0
v54 = iconst.i32 0
v55 = iconst.i16 0
v56 = iconst.i8 0
stack_store v52, ss0 ; v52 = 0
stack_store v52, ss0+16 ; v52 = 0
stack_store v52, ss0+32 ; v52 = 0
stack_store v53, ss0+48 ; v53 = 0
stack_store v55, ss0+56 ; v55 = 0
stack_store v56, ss0+58 ; v56 = 0
v57 = iadd v6, v6
stack_store v5, ss0
v58 = uextend.i64 v46 ; v46 = 0
v59 = udiv v51, v51 ; v51 = 0xc3c3_c341_3c3c, v51 = 0xc3c3_c341_3c3c
v60 = udiv v59, v59
v61 = iadd v0, v0
v62 = isub v8, v8
v63 = ushr v62, v46 ; v46 = 0
v64 = ushr v62, v46 ; v46 = 0
v65 = ushr v63, v57
v66 = ushr v65, v46 ; v46 = 0
v67 = ushr v66, v46 ; v46 = 0
v68 = ushr v67, v46 ; v46 = 0
v69 = fcopysign v39, v39 ; v39 = 0x1.d7d7d7d7d006fp984, v39 = 0x1.d7d7d7d7d006fp984
return v58, v13, v5, v1, v1, v1, v46, v14, v69, v45, v42, v1, v68, v1, v46, v61 ; v46 = 0, v46 = 0
}

; run: %a(86737344494455290561902794539232559746, false, false, 0, false, 0, 0, 0, 0, 0, 0, 0, 0, 0.0, 0) == [0, 0.0, 0, false, false, false, 0, 0, 0x1.d7d7d7d7d006fp984, -1140506845845240447760129, -115637640465955680012984649720956879616, false, 0, false, 0, -166807677932027882339569018353303091964]

```
</details>

And this stops failing if we remove the stack_store.i16, so I think its related to that.

Wasmtime GitHub notifications bot (Aug 24 2022 at 12:01):

afonso360 edited a comment on issue #4761:

<details>
<summary> Minimized this to: </summary>

```
;; Fuzzgen test case

test interpret
test run
set enable_llvm_abi_extensions
target aarch64
target s390x
target x86_64

function %a(i128, b1, b1, i128, b1, i16 uext, i8 sext, i64 sext, i32, i64 sext, i128, i128 sext, i64 uext, f32, i128 sext) -> i64 sext, f32, i16, b1, b1 sext, b1 sext, i8, i128 sext, f64 sext, i128 sext, i128, b1, i32, b1 uext, i8, i128 sext windows_fastcall {
ss0 = explicit_slot 59

block0(v0: i128, v1: b1, v2: b1, v3: i128, v4: b1, v5: i16, v6: i8, v7: i64, v8: i32, v9: i64, v10: i128, v11: i128, v12: i64, v13: f32, v14: i128):
v39 = f64const 0x1.d7d7d7d7d006fp984
v40 = iconst.i64 0xa901_0024_3b00_0000
v41 = iconst.i64 0x7d7d_7d7d_7d7d_7d00
v42 = iconcat v41, v40 ; v41 = 0x7d7d_7d7d_7d7d_7d00, v40 = 0xa901_0024_3b00_0000
v43 = iconst.i64 0xffff_ffff_ffff_0e7d
v44 = iconst.i64 2303
v45 = iconcat v44, v43 ; v44 = 2303, v43 = 0xffff_ffff_ffff_0e7d
v46 = iconst.i8 0
v47 = bconst.b1 true
v48 = iconst.i32 0xffff_ffff_c3c3_c3c3
v49 = bconst.b1 false
v50 = iconst.i8 -1
v51 = iconst.i64 0xc3c3_c341_3c3c
v52 = iconst.i128 0
v53 = iconst.i64 0
v54 = iconst.i32 0
v55 = iconst.i16 0
v56 = iconst.i8 0
stack_store v52, ss0 ; v52 = 0
stack_store v52, ss0+16 ; v52 = 0
stack_store v52, ss0+32 ; v52 = 0
stack_store v53, ss0+48 ; v53 = 0
stack_store v55, ss0+56 ; v55 = 0
stack_store v56, ss0+58 ; v56 = 0
v57 = iadd v6, v6
stack_store v5, ss0
v58 = uextend.i64 v46 ; v46 = 0
v59 = udiv v51, v51 ; v51 = 0xc3c3_c341_3c3c, v51 = 0xc3c3_c341_3c3c
v60 = udiv v59, v59
v61 = iadd v0, v0
v62 = isub v8, v8
v63 = ushr v62, v46 ; v46 = 0
v64 = ushr v62, v46 ; v46 = 0
v65 = ushr v63, v57
v66 = ushr v65, v46 ; v46 = 0
v67 = ushr v66, v46 ; v46 = 0
v68 = ushr v67, v46 ; v46 = 0
v69 = fcopysign v39, v39 ; v39 = 0x1.d7d7d7d7d006fp984, v39 = 0x1.d7d7d7d7d006fp984
return v58, v13, v5, v1, v1, v1, v46, v14, v69, v45, v42, v1, v68, v1, v46, v61 ; v46 = 0, v46 = 0
}

; run: %a(86737344494455290561902794539232559746, false, false, 0, false, 0, 0, 0, 0, 0, 0, 0, 0, 0.0, 0) == [0, 0.0, 0, false, false, false, 0, 0, 0x1.d7d7d7d7d006fp984, -1140506845845240447760129, -115637640465955680012984649720956879616, false, 0, false, 0, -166807677932027882339569018353303091964]

```

</details>

And this stops failing if we remove the stack_store.i16, so I think its related to that.

Wasmtime GitHub notifications bot (Aug 24 2022 at 12:07):

afonso360 edited a comment on issue #4761:

<details>
<summary> Minimized this to: </summary>

```
;; Fuzzgen test case

test interpret
test run
set enable_llvm_abi_extensions
target aarch64
target s390x
target x86_64

function %a(i128, b1, b1, i128, b1, i16 uext, i8 sext, i64 sext, i32, i64 sext, i128, i128 sext, i64 uext, f32, i128 sext) -> i64 sext, f32, i16, b1, b1 sext, b1 sext, i8, i128 sext, f64 sext, i128 sext, i128, b1, i32, b1 uext, i8, i128 sext windows_fastcall {
ss0 = explicit_slot 59

block0(v0: i128, v1: b1, v2: b1, v3: i128, v4: b1, v5: i16, v6: i8, v7: i64, v8: i32, v9: i64, v10: i128, v11: i128, v12: i64, v13: f32, v14: i128):
v39 = f64const 0x1.d7d7d7d7d006fp984
v40 = iconst.i64 0xa901_0024_3b00_0000
v41 = iconst.i64 0x7d7d_7d7d_7d7d_7d00
v42 = iconcat v41, v40 ; v41 = 0x7d7d_7d7d_7d7d_7d00, v40 = 0xa901_0024_3b00_0000
v43 = iconst.i64 0xffff_ffff_ffff_0e7d
v44 = iconst.i64 2303
v45 = iconcat v44, v43 ; v44 = 2303, v43 = 0xffff_ffff_ffff_0e7d
v46 = iconst.i8 0
v47 = bconst.b1 true
v48 = iconst.i32 0xffff_ffff_c3c3_c3c3
v49 = bconst.b1 false
v50 = iconst.i8 -1
v51 = iconst.i64 0xc3c3_c341_3c3c
v52 = iconst.i128 0
v53 = iconst.i64 0
v54 = iconst.i32 0
v55 = iconst.i16 0
v56 = iconst.i8 0
stack_store v52, ss0 ; v52 = 0
stack_store v52, ss0+16 ; v52 = 0
stack_store v52, ss0+32 ; v52 = 0
stack_store v53, ss0+48 ; v53 = 0
stack_store v55, ss0+56 ; v55 = 0
stack_store v56, ss0+58 ; v56 = 0
v57 = iadd v6, v6
stack_store v5, ss0
v58 = uextend.i64 v46 ; v46 = 0
v59 = udiv v51, v51 ; v51 = 0xc3c3_c341_3c3c, v51 = 0xc3c3_c341_3c3c
v60 = udiv v59, v59
v61 = iadd v0, v0
v62 = isub v8, v8
v63 = ushr v62, v46 ; v46 = 0
v64 = ushr v62, v46 ; v46 = 0
v65 = ushr v63, v57
v66 = ushr v65, v46 ; v46 = 0
v67 = ushr v66, v46 ; v46 = 0
v68 = ushr v67, v46 ; v46 = 0
v69 = fcopysign v39, v39 ; v39 = 0x1.d7d7d7d7d006fp984, v39 = 0x1.d7d7d7d7d006fp984
return v58, v13, v5, v1, v1, v1, v46, v14, v69, v45, v42, v1, v68, v1, v46, v61 ; v46 = 0, v46 = 0
}

; run: %a(86737344494455290561902794539232559746, false, false, 0, false, 0, 0, 0, 0, 0, 0, 0, 0, 0.0, 0) == [0, 0.0, 0, false, false, false, 0, 0, 0x1.d7d7d7d7d006fp984, -1140506845845240447760129, -115637640465955680012984649720956879616, false, 0, false, 0, -166807677932027882339569018353303091964]

```

</details>

And this stops failing if we remove the stack_store v5, ss0, so I think its related to that.

Wasmtime GitHub notifications bot (Aug 26 2022 at 14:52):

afonso360 commented on issue #4761:

I've tried to minimize this further, and here's what I got

<details>
<summary>Minimized CLIF</summary>

test run
set enable_llvm_abi_extensions
target x86_64

function %a(i128, b1, b1, i128, b1, i16 uext, i8 sext, i64 sext, i32, i64 sext, i128, i128 sext, i64 uext, f32, i128 sext) -> i64 sext, f32, i16, b1, b1 sext, b1 sext, i8, i128 sext, f64 sext, i128 sext, i128, b1, i32, b1 uext, i8, i128 sext windows_fastcall {
  ss0 = explicit_slot 59

block0(v0: i128, v1: b1, v2: b1, v3: i128, v4: b1, v5: i16, v6: i8, v7: i64, v8: i32, v9: i64, v10: i128, v11: i128, v12: i64, v13: f32, v14: i128):
  v39 = f64const 0x1.d7d7d7d7d006fp984
  v43 = iconst.i64 0xffff_ffff_ffff_0e7d
  v44 = iconst.i64 2303
  v45 = iconcat v44, v43  ; v44 = 2303, v43 = 0xffff_ffff_ffff_0e7d
  v46 = iconst.i8 0
  v51 = iconst.i64 0xc3c3_c341_3c3c
  v52 = iconst.i128 0
  v53 = iconst.i64 0
  v55 = iconst.i16 0
  v56 = iconst.i8 0
  stack_store v52, ss0  ; v52 = 0
  stack_store v52, ss0+16  ; v52 = 0
  stack_store v52, ss0+32  ; v52 = 0
  stack_store v53, ss0+48  ; v53 = 0
  stack_store v55, ss0+56  ; v55 = 0
  stack_store v56, ss0+58  ; v56 = 0
  stack_store v5, ss0
  v58 = uextend.i64 v46  ; v46 = 0
  v59 = udiv v51, v51  ; v51 = 0xc3c3_c341_3c3c, v51 = 0xc3c3_c341_3c3c
  v61 = iadd v0, v0
  v63 = ushr v8, v46  ; v46 = 0
  v69 = fcopysign v39, v39  ; v39 = 0x1.d7d7d7d7d006fp984, v39 = 0x1.d7d7d7d7d006fp984
  return v58, v13, v5, v1, v1, v1, v46, v14, v69, v45, v0, v1, v63, v1, v46, v61  ; v46 = 0, v46 = 0
}
; run: %a(86737344494455290561902794539232559746, false, false, 0, false, 0, 0, 0, 0, 0, 0, 0, 0, 0.0, 0) == [0, 0.0, 0, false, false, false, 0, 0, 0x1.d7d7d7d7d006fp984, -1140506845845240447760129, -115637640465955680012984649720956879616, false, 0, false, 0, -166807677932027882339569018353303091964]

</details>

Removing any of those instructions causes this to stop crashing.

<details>
<summary>Disassembly</summary>

Disassembly of 712 bytes:
   0:   55                      push    rbp
   1:   48 89 e5                mov     rbp, rsp
   4:   48 81 ec f0 00 00 00    sub     rsp, 0xf0
   b:   48 89 9c 24 a0 00 00 00 mov     qword ptr [rsp + 0xa0], rbx
  13:   48 89 b4 24 a8 00 00 00 mov     qword ptr [rsp + 0xa8], rsi
  1b:   48 89 bc 24 b0 00 00 00 mov     qword ptr [rsp + 0xb0], rdi
  23:   4c 89 a4 24 b8 00 00 00 mov     qword ptr [rsp + 0xb8], r12
  2b:   4c 89 ac 24 c0 00 00 00 mov     qword ptr [rsp + 0xc0], r13
  33:   4c 89 b4 24 c8 00 00 00 mov     qword ptr [rsp + 0xc8], r14
  3b:   4c 89 bc 24 d0 00 00 00 mov     qword ptr [rsp + 0xd0], r15
  43:   f3 44 0f 7f b4 24 e0 00 00 00
                                movdqu  xmmword ptr [rsp + 0xe0], xmm14
  4d:   48 89 4c 24 40          mov     qword ptr [rsp + 0x40], rcx
  52:   48 89 54 24 48          mov     qword ptr [rsp + 0x48], rdx
  57:   4c 89 44 24 50          mov     qword ptr [rsp + 0x50], r8
  5c:   48 8b 45 30             mov     rax, qword ptr [rbp + 0x30]
  60:   48 8b 75 38             mov     rsi, qword ptr [rbp + 0x38]
  64:   4c 8b 4d 40             mov     r9, qword ptr [rbp + 0x40]
  68:   48 8b 55 48             mov     rdx, qword ptr [rbp + 0x48]
  6c:   4c 8b 55 50             mov     r10, qword ptr [rbp + 0x50]
  70:   48 8b 5d 58             mov     rbx, qword ptr [rbp + 0x58]
  74:   4c 8b 5d 60             mov     r11, qword ptr [rbp + 0x60]
  78:   4c 89 9c 24 88 00 00 00 mov     qword ptr [rsp + 0x88], r11
  80:   4c 8b 65 68             mov     r12, qword ptr [rbp + 0x68]
  84:   4c 8b 6d 70             mov     r13, qword ptr [rbp + 0x70]
  88:   4c 8b 75 78             mov     r14, qword ptr [rbp + 0x78]
  8c:   4c 8b bd 80 00 00 00    mov     r15, qword ptr [rbp + 0x80]
  93:   48 8b bd 88 00 00 00    mov     rdi, qword ptr [rbp + 0x88]
  9a:   4c 8b 9d 90 00 00 00    mov     r11, qword ptr [rbp + 0x90]
  a1:   f3 0f 10 85 98 00 00 00 movss   xmm0, dword ptr [rbp + 0x98]
  a9:   48 8b bd a0 00 00 00    mov     rdi, qword ptr [rbp + 0xa0]
  b0:   48 89 7c 24 70          mov     qword ptr [rsp + 0x70], rdi
  b5:   48 8b 8d a8 00 00 00    mov     rcx, qword ptr [rbp + 0xa8]
  bc:   48 89 4c 24 78          mov     qword ptr [rsp + 0x78], rcx
  c1:   4c 8b 9d b0 00 00 00    mov     r11, qword ptr [rbp + 0xb0]
  c8:   b8 ff 08 00 00          mov     eax, 0x8ff
  cd:   48 89 84 24 98 00 00 00 mov     qword ptr [rsp + 0x98], rax
  d5:   48 c7 c0 7d 0e ff ff    mov     rax, 0xffffffffffff0e7d
  dc:   49 89 c2                mov     r10, rax
  df:   48 8d 34 24             lea     rsi, [rsp]
  e3:   4d 31 c9                xor     r9, r9
  e6:   48 31 ff                xor     rdi, rdi
  e9:   4c 89 0e                mov     qword ptr [rsi], r9
  ec:   48 89 7e 08             mov     qword ptr [rsi + 8], rdi
  f0:   48 8d 74 24 10          lea     rsi, [rsp + 0x10]
  f5:   48 31 c0                xor     rax, rax
  f8:   48 31 ff                xor     rdi, rdi
  fb:   48 89 06                mov     qword ptr [rsi], rax
  fe:   48 89 7e 08             mov     qword ptr [rsi + 8], rdi
 102:   48 8d 44 24 20          lea     rax, [rsp + 0x20]
 107:   48 31 ff                xor     rdi, rdi
 10a:   4d 31 c0                xor     r8, r8
 10d:   48 89 38                mov     qword ptr [rax], rdi
 110:   4c 89 40 08             mov     qword ptr [rax + 8], r8
 114:   4c 8d 44 24 30          lea     r8, [rsp + 0x30]
 119:   4d 31 c9                xor     r9, r9
 11c:   4d 89 08                mov     qword ptr [r8], r9
 11f:   4c 8d 44 24 38          lea     r8, [rsp + 0x38]
 124:   45 31 c9                xor     r9d, r9d
 127:   66 45 89 08             mov     word ptr [r8], r9w
 12b:   4c 8d 4c 24 3a          lea     r9, [rsp + 0x3a]
 130:   31 f6                   xor     esi, esi
 132:   41 88 31                mov     byte ptr [r9], sil
 135:   48 8d 34 24             lea     rsi, [rsp]
 139:   66 89 16                mov     word ptr [rsi], dx
 13c:   48 89 54 24 68          mov     qword ptr [rsp + 0x68], rdx
 141:   48 0f b6 05 77 01 00 00 movzx   rax, byte ptr [rip + 0x177]
 149:   49 89 c0                mov     r8, rax
 14c:   48 b8 3c 3c 41 c3 c3 c3 00 00
                                movabs  rax, 0xc3c3c3413c3c
 156:   48 be 3c 3c 41 c3 c3 c3 00 00
                                movabs  rsi, 0xc3c3c3413c3c
 160:   ba 00 00 00 00          mov     edx, 0
 165:   48 f7 f6                div     rsi
 168:   48 8b 4c 24 40          mov     rcx, qword ptr [rsp + 0x40]
 16d:   49 89 cc                mov     r12, rcx
 170:   49 01 cc                add     r12, rcx
 173:   4d 89 e1                mov     r9, r12
 176:   48 8b 5c 24 48          mov     rbx, qword ptr [rsp + 0x48]
 17b:   48 89 da                mov     rdx, rbx
 17e:   48 11 da                adc     rdx, rbx
 181:   48 89 94 24 90 00 00 00 mov     qword ptr [rsp + 0x90], rdx
 189:   48 8b 8c 24 88 00 00 00 mov     rcx, qword ptr [rsp + 0x88]
 191:   c1 e9 00                shr     ecx, 0
 194:   48 be 00 00 00 00 00 00 00 80
                                movabs  rsi, 0x8000000000000000
 19e:   66 4c 0f 6e f6          movq    xmm14, rsi
 1a3:   66 41 0f 6f ce          movdqa  xmm1, xmm14
 1a8:   66 0f 55 0d 08 01 00 00 andnpd  xmm1, xmmword ptr [rip + 0x108]
 1b0:   66 44 0f 54 35 ff 00 00 00
                                andpd   xmm14, xmmword ptr [rip + 0xff]
 1b9:   66 41 0f 56 ce          orpd    xmm1, xmm14
 1be:   4c 89 c0                mov     rax, r8
 1c1:   48 8b 54 24 68          mov     rdx, qword ptr [rsp + 0x68]
 1c6:   48 8b 7c 24 50          mov     rdi, qword ptr [rsp + 0x50]
 1cb:   48 89 7c 24 58          mov     qword ptr [rsp + 0x58], rdi
 1d0:   48 89 7c 24 60          mov     qword ptr [rsp + 0x60], rdi
 1d5:   48 89 bc 24 80 00 00 00 mov     qword ptr [rsp + 0x80], rdi
 1dd:   45 31 ed                xor     r13d, r13d
 1e0:   4c 8b 7c 24 70          mov     r15, qword ptr [rsp + 0x70]
 1e5:   4c 8b 64 24 78          mov     r12, qword ptr [rsp + 0x78]
 1ea:   4d 89 d6                mov     r14, r10
 1ed:   48 8b 74 24 40          mov     rsi, qword ptr [rsp + 0x40]
 1f2:   45 31 c0                xor     r8d, r8d
 1f5:   4c 8b 54 24 58          mov     r10, qword ptr [rsp + 0x58]
 1fa:   45 88 13                mov     byte ptr [r11], r10b
 1fd:   4c 8b 54 24 60          mov     r10, qword ptr [rsp + 0x60]
 202:   45 88 53 08             mov     byte ptr [r11 + 8], r10b
 206:   4c 8b 94 24 80 00 00 00 mov     r10, qword ptr [rsp + 0x80]
 20e:   45 88 53 10             mov     byte ptr [r11 + 0x10], r10b
 212:   45 88 6b 18             mov     byte ptr [r11 + 0x18], r13b
 216:   4d 89 7b 20             mov     qword ptr [r11 + 0x20], r15
 21a:   4d 89 63 28             mov     qword ptr [r11 + 0x28], r12
 21e:   f2 41 0f 11 4b 30       movsd   qword ptr [r11 + 0x30], xmm1
 224:   4c 8b 94 24 98 00 00 00 mov     r10, qword ptr [rsp + 0x98]
 22c:   4d 89 53 38             mov     qword ptr [r11 + 0x38], r10
 230:   4d 89 73 40             mov     qword ptr [r11 + 0x40], r14
 234:   49 89 73 48             mov     qword ptr [r11 + 0x48], rsi
 238:   49 89 5b 50             mov     qword ptr [r11 + 0x50], rbx
 23c:   41 88 7b 58             mov     byte ptr [r11 + 0x58], dil
 240:   41 89 4b 60             mov     dword ptr [r11 + 0x60], ecx
 244:   48 8b 74 24 50          mov     rsi, qword ptr [rsp + 0x50]
 249:   41 88 73 68             mov     byte ptr [r11 + 0x68], sil
 24d:   45 88 43 70             mov     byte ptr [r11 + 0x70], r8b
 251:   4d 89 4b 78             mov     qword ptr [r11 + 0x78], r9
 255:   4c 8b 84 24 90 00 00 00 mov     r8, qword ptr [rsp +
[message truncated]

Wasmtime GitHub notifications bot (Aug 26 2022 at 15:07):

afonso360 deleted a comment on issue #4761:

I've tried to minimize this further, and here's what I got

<details>
<summary>Minimized CLIF</summary>

test run
set enable_llvm_abi_extensions
target x86_64

function %a(i128, b1, b1, i128, b1, i16 uext, i8 sext, i64 sext, i32, i64 sext, i128, i128 sext, i64 uext, f32, i128 sext) -> i64 sext, f32, i16, b1, b1 sext, b1 sext, i8, i128 sext, f64 sext, i128 sext, i128, b1, i32, b1 uext, i8, i128 sext windows_fastcall {
  ss0 = explicit_slot 59

block0(v0: i128, v1: b1, v2: b1, v3: i128, v4: b1, v5: i16, v6: i8, v7: i64, v8: i32, v9: i64, v10: i128, v11: i128, v12: i64, v13: f32, v14: i128):
  v39 = f64const 0x1.d7d7d7d7d006fp984
  v43 = iconst.i64 0xffff_ffff_ffff_0e7d
  v44 = iconst.i64 2303
  v45 = iconcat v44, v43  ; v44 = 2303, v43 = 0xffff_ffff_ffff_0e7d
  v46 = iconst.i8 0
  v51 = iconst.i64 0xc3c3_c341_3c3c
  v52 = iconst.i128 0
  v53 = iconst.i64 0
  v55 = iconst.i16 0
  v56 = iconst.i8 0
  stack_store v52, ss0  ; v52 = 0
  stack_store v52, ss0+16  ; v52 = 0
  stack_store v52, ss0+32  ; v52 = 0
  stack_store v53, ss0+48  ; v53 = 0
  stack_store v55, ss0+56  ; v55 = 0
  stack_store v56, ss0+58  ; v56 = 0
  stack_store v5, ss0
  v58 = uextend.i64 v46  ; v46 = 0
  v59 = udiv v51, v51  ; v51 = 0xc3c3_c341_3c3c, v51 = 0xc3c3_c341_3c3c
  v61 = iadd v0, v0
  v63 = ushr v8, v46  ; v46 = 0
  v69 = fcopysign v39, v39  ; v39 = 0x1.d7d7d7d7d006fp984, v39 = 0x1.d7d7d7d7d006fp984
  return v58, v13, v5, v1, v1, v1, v46, v14, v69, v45, v0, v1, v63, v1, v46, v61  ; v46 = 0, v46 = 0
}
; run: %a(86737344494455290561902794539232559746, false, false, 0, false, 0, 0, 0, 0, 0, 0, 0, 0, 0.0, 0) == [0, 0.0, 0, false, false, false, 0, 0, 0x1.d7d7d7d7d006fp984, -1140506845845240447760129, -115637640465955680012984649720956879616, false, 0, false, 0, -166807677932027882339569018353303091964]

</details>

Removing any of those instructions causes this to stop crashing.

<details>
<summary>Disassembly</summary>

Disassembly of 712 bytes:
   0:   55                      push    rbp
   1:   48 89 e5                mov     rbp, rsp
   4:   48 81 ec f0 00 00 00    sub     rsp, 0xf0
   b:   48 89 9c 24 a0 00 00 00 mov     qword ptr [rsp + 0xa0], rbx
  13:   48 89 b4 24 a8 00 00 00 mov     qword ptr [rsp + 0xa8], rsi
  1b:   48 89 bc 24 b0 00 00 00 mov     qword ptr [rsp + 0xb0], rdi
  23:   4c 89 a4 24 b8 00 00 00 mov     qword ptr [rsp + 0xb8], r12
  2b:   4c 89 ac 24 c0 00 00 00 mov     qword ptr [rsp + 0xc0], r13
  33:   4c 89 b4 24 c8 00 00 00 mov     qword ptr [rsp + 0xc8], r14
  3b:   4c 89 bc 24 d0 00 00 00 mov     qword ptr [rsp + 0xd0], r15
  43:   f3 44 0f 7f b4 24 e0 00 00 00
                                movdqu  xmmword ptr [rsp + 0xe0], xmm14
  4d:   48 89 4c 24 40          mov     qword ptr [rsp + 0x40], rcx
  52:   48 89 54 24 48          mov     qword ptr [rsp + 0x48], rdx
  57:   4c 89 44 24 50          mov     qword ptr [rsp + 0x50], r8
  5c:   48 8b 45 30             mov     rax, qword ptr [rbp + 0x30]
  60:   48 8b 75 38             mov     rsi, qword ptr [rbp + 0x38]
  64:   4c 8b 4d 40             mov     r9, qword ptr [rbp + 0x40]
  68:   48 8b 55 48             mov     rdx, qword ptr [rbp + 0x48]
  6c:   4c 8b 55 50             mov     r10, qword ptr [rbp + 0x50]
  70:   48 8b 5d 58             mov     rbx, qword ptr [rbp + 0x58]
  74:   4c 8b 5d 60             mov     r11, qword ptr [rbp + 0x60]
  78:   4c 89 9c 24 88 00 00 00 mov     qword ptr [rsp + 0x88], r11
  80:   4c 8b 65 68             mov     r12, qword ptr [rbp + 0x68]
  84:   4c 8b 6d 70             mov     r13, qword ptr [rbp + 0x70]
  88:   4c 8b 75 78             mov     r14, qword ptr [rbp + 0x78]
  8c:   4c 8b bd 80 00 00 00    mov     r15, qword ptr [rbp + 0x80]
  93:   48 8b bd 88 00 00 00    mov     rdi, qword ptr [rbp + 0x88]
  9a:   4c 8b 9d 90 00 00 00    mov     r11, qword ptr [rbp + 0x90]
  a1:   f3 0f 10 85 98 00 00 00 movss   xmm0, dword ptr [rbp + 0x98]
  a9:   48 8b bd a0 00 00 00    mov     rdi, qword ptr [rbp + 0xa0]
  b0:   48 89 7c 24 70          mov     qword ptr [rsp + 0x70], rdi
  b5:   48 8b 8d a8 00 00 00    mov     rcx, qword ptr [rbp + 0xa8]
  bc:   48 89 4c 24 78          mov     qword ptr [rsp + 0x78], rcx
  c1:   4c 8b 9d b0 00 00 00    mov     r11, qword ptr [rbp + 0xb0]
  c8:   b8 ff 08 00 00          mov     eax, 0x8ff
  cd:   48 89 84 24 98 00 00 00 mov     qword ptr [rsp + 0x98], rax
  d5:   48 c7 c0 7d 0e ff ff    mov     rax, 0xffffffffffff0e7d
  dc:   49 89 c2                mov     r10, rax
  df:   48 8d 34 24             lea     rsi, [rsp]
  e3:   4d 31 c9                xor     r9, r9
  e6:   48 31 ff                xor     rdi, rdi
  e9:   4c 89 0e                mov     qword ptr [rsi], r9
  ec:   48 89 7e 08             mov     qword ptr [rsi + 8], rdi
  f0:   48 8d 74 24 10          lea     rsi, [rsp + 0x10]
  f5:   48 31 c0                xor     rax, rax
  f8:   48 31 ff                xor     rdi, rdi
  fb:   48 89 06                mov     qword ptr [rsi], rax
  fe:   48 89 7e 08             mov     qword ptr [rsi + 8], rdi
 102:   48 8d 44 24 20          lea     rax, [rsp + 0x20]
 107:   48 31 ff                xor     rdi, rdi
 10a:   4d 31 c0                xor     r8, r8
 10d:   48 89 38                mov     qword ptr [rax], rdi
 110:   4c 89 40 08             mov     qword ptr [rax + 8], r8
 114:   4c 8d 44 24 30          lea     r8, [rsp + 0x30]
 119:   4d 31 c9                xor     r9, r9
 11c:   4d 89 08                mov     qword ptr [r8], r9
 11f:   4c 8d 44 24 38          lea     r8, [rsp + 0x38]
 124:   45 31 c9                xor     r9d, r9d
 127:   66 45 89 08             mov     word ptr [r8], r9w
 12b:   4c 8d 4c 24 3a          lea     r9, [rsp + 0x3a]
 130:   31 f6                   xor     esi, esi
 132:   41 88 31                mov     byte ptr [r9], sil
 135:   48 8d 34 24             lea     rsi, [rsp]
 139:   66 89 16                mov     word ptr [rsi], dx
 13c:   48 89 54 24 68          mov     qword ptr [rsp + 0x68], rdx
 141:   48 0f b6 05 77 01 00 00 movzx   rax, byte ptr [rip + 0x177]
 149:   49 89 c0                mov     r8, rax
 14c:   48 b8 3c 3c 41 c3 c3 c3 00 00
                                movabs  rax, 0xc3c3c3413c3c
 156:   48 be 3c 3c 41 c3 c3 c3 00 00
                                movabs  rsi, 0xc3c3c3413c3c
 160:   ba 00 00 00 00          mov     edx, 0
 165:   48 f7 f6                div     rsi
 168:   48 8b 4c 24 40          mov     rcx, qword ptr [rsp + 0x40]
 16d:   49 89 cc                mov     r12, rcx
 170:   49 01 cc                add     r12, rcx
 173:   4d 89 e1                mov     r9, r12
 176:   48 8b 5c 24 48          mov     rbx, qword ptr [rsp + 0x48]
 17b:   48 89 da                mov     rdx, rbx
 17e:   48 11 da                adc     rdx, rbx
 181:   48 89 94 24 90 00 00 00 mov     qword ptr [rsp + 0x90], rdx
 189:   48 8b 8c 24 88 00 00 00 mov     rcx, qword ptr [rsp + 0x88]
 191:   c1 e9 00                shr     ecx, 0
 194:   48 be 00 00 00 00 00 00 00 80
                                movabs  rsi, 0x8000000000000000
 19e:   66 4c 0f 6e f6          movq    xmm14, rsi
 1a3:   66 41 0f 6f ce          movdqa  xmm1, xmm14
 1a8:   66 0f 55 0d 08 01 00 00 andnpd  xmm1, xmmword ptr [rip + 0x108]
 1b0:   66 44 0f 54 35 ff 00 00 00
                                andpd   xmm14, xmmword ptr [rip + 0xff]
 1b9:   66 41 0f 56 ce          orpd    xmm1, xmm14
 1be:   4c 89 c0                mov     rax, r8
 1c1:   48 8b 54 24 68          mov     rdx, qword ptr [rsp + 0x68]
 1c6:   48 8b 7c 24 50          mov     rdi, qword ptr [rsp + 0x50]
 1cb:   48 89 7c 24 58          mov     qword ptr [rsp + 0x58], rdi
 1d0:   48 89 7c 24 60          mov     qword ptr [rsp + 0x60], rdi
 1d5:   48 89 bc 24 80 00 00 00 mov     qword ptr [rsp + 0x80], rdi
 1dd:   45 31 ed                xor     r13d, r13d
 1e0:   4c 8b 7c 24 70          mov     r15, qword ptr [rsp + 0x70]
 1e5:   4c 8b 64 24 78          mov     r12, qword ptr [rsp + 0x78]
 1ea:   4d 89 d6                mov     r14, r10
 1ed:   48 8b 74 24 40          mov     rsi, qword ptr [rsp + 0x40]
 1f2:   45 31 c0                xor     r8d, r8d
 1f5:   4c 8b 54 24 58          mov     r10, qword ptr [rsp + 0x58]
 1fa:   45 88 13                mov     byte ptr [r11], r10b
 1fd:   4c 8b 54 24 60          mov     r10, qword ptr [rsp + 0x60]
 202:   45 88 53 08             mov     byte ptr [r11 + 8], r10b
 206:   4c 8b 94 24 80 00 00 00 mov     r10, qword ptr [rsp + 0x80]
 20e:   45 88 53 10             mov     byte ptr [r11 + 0x10], r10b
 212:   45 88 6b 18             mov     byte ptr [r11 + 0x18], r13b
 216:   4d 89 7b 20             mov     qword ptr [r11 + 0x20], r15
 21a:   4d 89 63 28             mov     qword ptr [r11 + 0x28], r12
 21e:   f2 41 0f 11 4b 30       movsd   qword ptr [r11 + 0x30], xmm1
 224:   4c 8b 94 24 98 00 00 00 mov     r10, qword ptr [rsp + 0x98]
 22c:   4d 89 53 38             mov     qword ptr [r11 + 0x38], r10
 230:   4d 89 73 40             mov     qword ptr [r11 + 0x40], r14
 234:   49 89 73 48             mov     qword ptr [r11 + 0x48], rsi
 238:   49 89 5b 50             mov     qword ptr [r11 + 0x50], rbx
 23c:   41 88 7b 58             mov     byte ptr [r11 + 0x58], dil
 240:   41 89 4b 60             mov     dword ptr [r11 + 0x60], ecx
 244:   48 8b 74 24 50          mov     rsi, qword ptr [rsp + 0x50]
 249:   41 88 73 68             mov     byte ptr [r11 + 0x68], sil
 24d:   45 88 43 70             mov     byte ptr [r11 + 0x70], r8b
 251:   4d 89 4b 78             mov     qword ptr [r11 + 0x78], r9
 255:   4c 8b 84 24 90 00 00 00 mov     r8, qword pt
[message truncated]

Wasmtime GitHub notifications bot (Aug 26 2022 at 15:08):

bjorn3 commented on issue #4761:

The address is 8 aligned (does it need to be 16?).

Yes, vector instructions that do a load or store as part of their operand need it to be 16 byte aligned. Only the unaligned variants of vector loads and stores (like movupd) allow arbitrary alignment.

Wasmtime GitHub notifications bot (Aug 26 2022 at 15:10):

afonso360 commented on issue #4761:

I've tried to minimize this further, and here's what I got

<details>
<summary>Minimized CLIF</summary>

test run
set enable_llvm_abi_extensions
target x86_64

function %a(i128, b1, b1, i128, b1, i16 uext, i8 sext, i64 sext, i32, i64 sext, i128, i128 sext, i64 uext, f32, i128 sext) -> i64 sext, f32, i16, b1, b1 sext, b1 sext, i8, i128 sext, f64 sext, i128 sext, i128, b1, i32, b1 uext, i8, i128 sext windows_fastcall {
  ss0 = explicit_slot 59

block0(v0: i128, v1: b1, v2: b1, v3: i128, v4: b1, v5: i16, v6: i8, v7: i64, v8: i32, v9: i64, v10: i128, v11: i128, v12: i64, v13: f32, v14: i128):
  v39 = f64const 0x1.d7d7d7d7d006fp984
  v43 = iconst.i64 0xffff_ffff_ffff_0e7d
  v44 = iconst.i64 2303
  v45 = iconcat v44, v43  ; v44 = 2303, v43 = 0xffff_ffff_ffff_0e7d
  v46 = iconst.i8 0
  v51 = iconst.i64 0xc3c3_c341_3c3c
  v52 = iconst.i128 0
  v53 = iconst.i64 0
  v55 = iconst.i16 0
  v56 = iconst.i8 0
  stack_store v52, ss0  ; v52 = 0
  stack_store v52, ss0+16  ; v52 = 0
  stack_store v52, ss0+32  ; v52 = 0
  stack_store v53, ss0+48  ; v53 = 0
  stack_store v55, ss0+56  ; v55 = 0
  stack_store v56, ss0+58  ; v56 = 0
  stack_store v5, ss0
  v58 = uextend.i64 v46  ; v46 = 0
  v59 = udiv v51, v51  ; v51 = 0xc3c3_c341_3c3c, v51 = 0xc3c3_c341_3c3c
  v61 = iadd v0, v0
  v63 = ushr v8, v46  ; v46 = 0
  v69 = fcopysign v39, v39  ; v39 = 0x1.d7d7d7d7d006fp984, v39 = 0x1.d7d7d7d7d006fp984
  return v58, v13, v5, v1, v1, v1, v46, v14, v69, v45, v0, v1, v63, v1, v46, v61  ; v46 = 0, v46 = 0
}
; run: %a(86737344494455290561902794539232559746, false, false, 0, false, 0, 0, 0, 0, 0, 0, 0, 0, 0.0, 0) == [0, 0.0, 0, false, false, false, 0, 0, 0x1.d7d7d7d7d006fp984, -1140506845845240447760129, -115637640465955680012984649720956879616, false, 0, false, 0, -166807677932027882339569018353303091964]

</details>

Removing any of those instructions causes this to stop crashing.

<details>
<summary>Disassembly</summary>

Disassembly of 712 bytes:
   0:   55                      push    rbp
   1:   48 89 e5                mov     rbp, rsp
   4:   48 81 ec f0 00 00 00    sub     rsp, 0xf0
   b:   48 89 9c 24 a0 00 00 00 mov     qword ptr [rsp + 0xa0], rbx
  13:   48 89 b4 24 a8 00 00 00 mov     qword ptr [rsp + 0xa8], rsi
  1b:   48 89 bc 24 b0 00 00 00 mov     qword ptr [rsp + 0xb0], rdi
  23:   4c 89 a4 24 b8 00 00 00 mov     qword ptr [rsp + 0xb8], r12
  2b:   4c 89 ac 24 c0 00 00 00 mov     qword ptr [rsp + 0xc0], r13
  33:   4c 89 b4 24 c8 00 00 00 mov     qword ptr [rsp + 0xc8], r14
  3b:   4c 89 bc 24 d0 00 00 00 mov     qword ptr [rsp + 0xd0], r15
  43:   f3 44 0f 7f b4 24 e0 00 00 00
                                movdqu  xmmword ptr [rsp + 0xe0], xmm14
  4d:   48 89 4c 24 40          mov     qword ptr [rsp + 0x40], rcx
  52:   48 89 54 24 48          mov     qword ptr [rsp + 0x48], rdx
  57:   4c 89 44 24 50          mov     qword ptr [rsp + 0x50], r8
  5c:   48 8b 45 30             mov     rax, qword ptr [rbp + 0x30]
  60:   48 8b 75 38             mov     rsi, qword ptr [rbp + 0x38]
  64:   4c 8b 4d 40             mov     r9, qword ptr [rbp + 0x40]
  68:   48 8b 55 48             mov     rdx, qword ptr [rbp + 0x48]
  6c:   4c 8b 55 50             mov     r10, qword ptr [rbp + 0x50]
  70:   48 8b 5d 58             mov     rbx, qword ptr [rbp + 0x58]
  74:   4c 8b 5d 60             mov     r11, qword ptr [rbp + 0x60]
  78:   4c 89 9c 24 88 00 00 00 mov     qword ptr [rsp + 0x88], r11
  80:   4c 8b 65 68             mov     r12, qword ptr [rbp + 0x68]
  84:   4c 8b 6d 70             mov     r13, qword ptr [rbp + 0x70]
  88:   4c 8b 75 78             mov     r14, qword ptr [rbp + 0x78]
  8c:   4c 8b bd 80 00 00 00    mov     r15, qword ptr [rbp + 0x80]
  93:   48 8b bd 88 00 00 00    mov     rdi, qword ptr [rbp + 0x88]
  9a:   4c 8b 9d 90 00 00 00    mov     r11, qword ptr [rbp + 0x90]
  a1:   f3 0f 10 85 98 00 00 00 movss   xmm0, dword ptr [rbp + 0x98]
  a9:   48 8b bd a0 00 00 00    mov     rdi, qword ptr [rbp + 0xa0]
  b0:   48 89 7c 24 70          mov     qword ptr [rsp + 0x70], rdi
  b5:   48 8b 8d a8 00 00 00    mov     rcx, qword ptr [rbp + 0xa8]
  bc:   48 89 4c 24 78          mov     qword ptr [rsp + 0x78], rcx
  c1:   4c 8b 9d b0 00 00 00    mov     r11, qword ptr [rbp + 0xb0]
  c8:   b8 ff 08 00 00          mov     eax, 0x8ff
  cd:   48 89 84 24 98 00 00 00 mov     qword ptr [rsp + 0x98], rax
  d5:   48 c7 c0 7d 0e ff ff    mov     rax, 0xffffffffffff0e7d
  dc:   49 89 c2                mov     r10, rax
  df:   48 8d 34 24             lea     rsi, [rsp]
  e3:   4d 31 c9                xor     r9, r9
  e6:   48 31 ff                xor     rdi, rdi
  e9:   4c 89 0e                mov     qword ptr [rsi], r9
  ec:   48 89 7e 08             mov     qword ptr [rsi + 8], rdi
  f0:   48 8d 74 24 10          lea     rsi, [rsp + 0x10]
  f5:   48 31 c0                xor     rax, rax
  f8:   48 31 ff                xor     rdi, rdi
  fb:   48 89 06                mov     qword ptr [rsi], rax
  fe:   48 89 7e 08             mov     qword ptr [rsi + 8], rdi
 102:   48 8d 44 24 20          lea     rax, [rsp + 0x20]
 107:   48 31 ff                xor     rdi, rdi
 10a:   4d 31 c0                xor     r8, r8
 10d:   48 89 38                mov     qword ptr [rax], rdi
 110:   4c 89 40 08             mov     qword ptr [rax + 8], r8
 114:   4c 8d 44 24 30          lea     r8, [rsp + 0x30]
 119:   4d 31 c9                xor     r9, r9
 11c:   4d 89 08                mov     qword ptr [r8], r9
 11f:   4c 8d 44 24 38          lea     r8, [rsp + 0x38]
 124:   45 31 c9                xor     r9d, r9d
 127:   66 45 89 08             mov     word ptr [r8], r9w
 12b:   4c 8d 4c 24 3a          lea     r9, [rsp + 0x3a]
 130:   31 f6                   xor     esi, esi
 132:   41 88 31                mov     byte ptr [r9], sil
 135:   48 8d 34 24             lea     rsi, [rsp]
 139:   66 89 16                mov     word ptr [rsi], dx
 13c:   48 89 54 24 68          mov     qword ptr [rsp + 0x68], rdx
 141:   48 0f b6 05 77 01 00 00 movzx   rax, byte ptr [rip + 0x177]
 149:   49 89 c0                mov     r8, rax
 14c:   48 b8 3c 3c 41 c3 c3 c3 00 00
                                movabs  rax, 0xc3c3c3413c3c
 156:   48 be 3c 3c 41 c3 c3 c3 00 00
                                movabs  rsi, 0xc3c3c3413c3c
 160:   ba 00 00 00 00          mov     edx, 0
 165:   48 f7 f6                div     rsi
 168:   48 8b 4c 24 40          mov     rcx, qword ptr [rsp + 0x40]
 16d:   49 89 cc                mov     r12, rcx
 170:   49 01 cc                add     r12, rcx
 173:   4d 89 e1                mov     r9, r12
 176:   48 8b 5c 24 48          mov     rbx, qword ptr [rsp + 0x48]
 17b:   48 89 da                mov     rdx, rbx
 17e:   48 11 da                adc     rdx, rbx
 181:   48 89 94 24 90 00 00 00 mov     qword ptr [rsp + 0x90], rdx
 189:   48 8b 8c 24 88 00 00 00 mov     rcx, qword ptr [rsp + 0x88]
 191:   c1 e9 00                shr     ecx, 0
 194:   48 be 00 00 00 00 00 00 00 80
                                movabs  rsi, 0x8000000000000000
 19e:   66 4c 0f 6e f6          movq    xmm14, rsi
 1a3:   66 41 0f 6f ce          movdqa  xmm1, xmm14
 1a8:   66 0f 55 0d 08 01 00 00 andnpd  xmm1, xmmword ptr [rip + 0x108]
 1b0:   66 44 0f 54 35 ff 00 00 00
                                andpd   xmm14, xmmword ptr [rip + 0xff]
 1b9:   66 41 0f 56 ce          orpd    xmm1, xmm14
 1be:   4c 89 c0                mov     rax, r8
 1c1:   48 8b 54 24 68          mov     rdx, qword ptr [rsp + 0x68]
 1c6:   48 8b 7c 24 50          mov     rdi, qword ptr [rsp + 0x50]
 1cb:   48 89 7c 24 58          mov     qword ptr [rsp + 0x58], rdi
 1d0:   48 89 7c 24 60          mov     qword ptr [rsp + 0x60], rdi
 1d5:   48 89 bc 24 80 00 00 00 mov     qword ptr [rsp + 0x80], rdi
 1dd:   45 31 ed                xor     r13d, r13d
 1e0:   4c 8b 7c 24 70          mov     r15, qword ptr [rsp + 0x70]
 1e5:   4c 8b 64 24 78          mov     r12, qword ptr [rsp + 0x78]
 1ea:   4d 89 d6                mov     r14, r10
 1ed:   48 8b 74 24 40          mov     rsi, qword ptr [rsp + 0x40]
 1f2:   45 31 c0                xor     r8d, r8d
 1f5:   4c 8b 54 24 58          mov     r10, qword ptr [rsp + 0x58]
 1fa:   45 88 13                mov     byte ptr [r11], r10b
 1fd:   4c 8b 54 24 60          mov     r10, qword ptr [rsp + 0x60]
 202:   45 88 53 08             mov     byte ptr [r11 + 8], r10b
 206:   4c 8b 94 24 80 00 00 00 mov     r10, qword ptr [rsp + 0x80]
 20e:   45 88 53 10             mov     byte ptr [r11 + 0x10], r10b
 212:   45 88 6b 18             mov     byte ptr [r11 + 0x18], r13b
 216:   4d 89 7b 20             mov     qword ptr [r11 + 0x20], r15
 21a:   4d 89 63 28             mov     qword ptr [r11 + 0x28], r12
 21e:   f2 41 0f 11 4b 30       movsd   qword ptr [r11 + 0x30], xmm1
 224:   4c 8b 94 24 98 00 00 00 mov     r10, qword ptr [rsp + 0x98]
 22c:   4d 89 53 38             mov     qword ptr [r11 + 0x38], r10
 230:   4d 89 73 40             mov     qword ptr [r11 + 0x40], r14
 234:   49 89 73 48             mov     qword ptr [r11 + 0x48], rsi
 238:   49 89 5b 50             mov     qword ptr [r11 + 0x50], rbx
 23c:   41 88 7b 58             mov     byte ptr [r11 + 0x58], dil
 240:   41 89 4b 60             mov     dword ptr [r11 + 0x60], ecx
 244:   48 8b 74 24 50          mov     rsi, qword ptr [rsp + 0x50]
 249:   41 88 73 68             mov     byte ptr [r11 + 0x68], sil
 24d:   45 88 43 70             mov     byte ptr [r11 + 0x70], r8b
 251:   4d 89 4b 78             mov     qword ptr [r11 + 0x78], r9
 255:   4c 8b 84 24 90 00 00 00 mov     r8, qword ptr [rsp +
[message truncated]

Wasmtime GitHub notifications bot (Aug 26 2022 at 15:21):

afonso360 commented on issue #4761:

The address is 8 aligned (does it need to be 16?).

Yes, vector instructions that do a load or store as part of their operand need it to be 16 byte aligned. Only the unaligned variants of vector loads and stores (like movupd) allow arbitrary alignment.

Yep, that does seem to fix it

Wasmtime GitHub notifications bot (Aug 26 2022 at 15:34):

afonso360 edited a comment on issue #4761:

The address is 8 aligned (does it need to be 16?).

Yes, vector instructions that do a load or store as part of their operand need it to be 16 byte aligned. Only the unaligned variants of vector loads and stores (like movupd) allow arbitrary alignment.

Yep, that does seem to fix it, thanks!

Wasmtime GitHub notifications bot (Aug 26 2022 at 16:52):

jameysharp commented on issue #4761:

@cfallin @abrown @elliottt: Is there a more targeted way to fix this than reserving 16 bytes for every entry in the constant pool, or should we just merge #4789?

This is cool: cranelift-fuzzgen found a codegen bug that involves interaction between distant parts of the backend, and requires an unlucky stack layout. That's exactly the kind of thing I'd hope for from it. And good work on minimizing the failing input, @afonso360!

Wasmtime GitHub notifications bot (Aug 26 2022 at 17:56):

elliottt commented on issue #4761:

I've managed to minimize the test down a bit:

test run
set enable_llvm_abi_extensions
target x86_64

function %a() -> f64 {
  ss0 = explicit_slot 59

block0:
  v0 = f64const 0x1.d7d7d7d7d006fp984
  v1 = fcopysign v0, v0
  return v1
}

; run: %a() == 0x0.0

Wasmtime GitHub notifications bot (Aug 26 2022 at 17:57):

elliottt edited a comment on issue #4761:

I've managed to minimize the test down a bit, this will segfault for any non-zero explicit_slot values greater than 0, but fail the test assertion otherwise.
test run
set enable_llvm_abi_extensions
target x86_64

function %a() -> f64 {
  ss0 = explicit_slot 59

block0:
  v0 = f64const 0x1.d7d7d7d7d006fp984
  v1 = fcopysign v0, v0
  return v1
}

; run: %a() == 0x0.0

Wasmtime GitHub notifications bot (Aug 26 2022 at 18:38):

jameysharp commented on issue #4761:

After digging into this, we've determined that the problem here is not the constant alignment, exactly. The root cause is that we're trying to load values from memory which are not 16 bytes long, using instructions that load 16 bytes.

This specific bug was introduced this week in cee4b209f346ea279490268fe434dc52d0e0680c, while migrating fcopysign to ISLE.

The old lowering always did an 8-byte (or 4-byte, for F32) load to put the operands into general-purpose registers first. Then it moved from the GPRs to XMM registers, and never used the memory-operand variants of these SIMD instructions.

The natural translation of that pattern to ISLE has revealed a footgun around implicit conversions to xmm-or-mem operands. It's only safe to use the memory operand variant of these instructions for types which really are 16 bytes in size.

@elliottt is working on fixing that implicit conversion, which should resolve this issue and avoid all future problems of this particular type.

Wasmtime GitHub notifications bot (Aug 26 2022 at 20:04):

elliottt closed issue #4761:

https://oss-fuzz.com/testcase-detail/4896860279537664

ERROR: AddressSanitizer: SEGV on unknown address (pc 0x6250000014d5 bp 0x7fff30133db0 sp 0x7fff30133ad0 T0)
--
  | ==743==The signal is caused by a READ memory access.
  | ==743==Hint: this fault was caused by a dereference of a high value address (see register values below).  Disassemble the provided pc to learn which register was used.
  | SCARINESS: 20 (wild-addr-read)
  | #0 0x6250000014d5  (<unknown module>)
  | #1 0x62500000175d  (<unknown module>)
  | #0 0x5579a6d0c6c4 in cranelift_filetests::function_runner::CompiledFunction::call::ha6b9162cd2e21784 wasmtime/cranelift/filetests/src/function_runner.rs:183:9

input:

IFWoAAAAAE0AAABSSUl9LCB7eAx6ZcYNR9T0RvCdMjI0MjaWMgAAAAA7ACypAH19f319fX0AAAA7
JAABqQB9fX19fX19fX0IAAAAAAAAAQAAwyvDw8PDAADJ/zw8PEHDw8MAAGUBb/PGANpGAAAAAAAj
OwAsqW9vb29vb28AfX19fX19fQAAADskAAGpAH19fX19fX19fQ7/////////CAAAAAAAAAEAAMMr
w8PDwwAAyf88PDxBw8PDAABlAW/zxgDaRgAAAAAAAAAAAAAAAACTgoL/////////MAAAAAAAAAUA
AAAACIKCgoKCgkEAgoKCgoJNAAAAgoKCgoKCgoKCgoKCwoKCgoKCgoKSioIEAAcAgoKCgoKCgoKC
goKCgoKCGIKC//8wAAAAAAAABQAAAAAIgoKCgoKCQQCCgoKCgk0AAACCgoKCgoKCgoKCgoLCeoKC
eYKCgpKKggQAAACCgoKCgoKCgoKCgoKCgoIYgoKCgoKCgoKCgoKCFACCgjOCgoKCgoL/////goKC
goKCgoKCgoKCgoJBAUFB

cc @afonso360

Wasmtime GitHub notifications bot (Aug 26 2022 at 20:36):

elliottt edited a comment on issue #4761:

I've managed to minimize the test down a bit, this will segfault for any explicit_slot values greater than 0, but fail the test assertion otherwise.
test run
set enable_llvm_abi_extensions
target x86_64

function %a() -> f64 {
  ss0 = explicit_slot 59

block0:
  v0 = f64const 0x1.d7d7d7d7d006fp984
  v1 = fcopysign v0, v0
  return v1
}

; run: %a() == 0x0.0

Last updated: Nov 22 2024 at 17:03 UTC