On Thu, 27 Jun 2024 11:14:30 GMT, Shaojin Wen <d...@openjdk.org> wrote:
> Currently, the java.util.Formatter$Conversion::isValid method is implemented
> based on switch, which cannot be inlined because codeSize > 325. This problem
> can be avoided by implementing it with ImmutableBitSetPredicate.
>
> use `-XX:+UnlockDiagnosticVMOptions -XX:+PrintInlining` to see the master
> branch:
>
> @ 109 java.util.Formatter$Conversion::isValid (358 bytes) failed to
> inline: hot method too big
>
>
> current version
>
> @ 109 java.util.Formatter$Conversion::isValid (10 bytes) inline (hot)
> @ 4
> jdk.internal.util.ImmutableBitSetPredicate$SmallImmutableBitSetPredicate::test
> (50 bytes) inline (hot)
@cl4es This is a scenario optimized using ImmutableBitSetPredicate
Execute the following command, we can see that the
`java.util.Formatter$Conversion.isValid` method generates a tableswitch of size
85. This large tableswitch makes the code size 358 bytes. Can C2 be optimized
to use
`jdk.internal.util.ImmutableBitSetPredicate$SmallImmutableBitSetPredicate`?
javap -c java.util.Formatter.Conversion
* byte codes of `Conversion.isValid`
static boolean isValid(char);
Code:
0: iload_0
1: tableswitch { // 37 to 120
37: 352
38: 356
39: 356
40: 356
41: 356
42: 356
43: 356
44: 356
45: 356
46: 356
47: 356
48: 356
49: 356
50: 356
51: 356
52: 356
53: 356
54: 356
55: 356
56: 356
57: 356
58: 356
59: 356
60: 356
61: 356
62: 356
63: 356
64: 356
65: 352
66: 352
67: 352
68: 356
69: 352
70: 356
71: 352
72: 352
73: 356
74: 356
75: 356
76: 356
77: 356
78: 356
79: 356
80: 356
81: 356
82: 356
83: 352
84: 356
85: 356
86: 356
87: 356
88: 352
89: 356
90: 356
91: 356
92: 356
93: 356
94: 356
95: 356
96: 356
97: 352
98: 352
99: 352
100: 352
101: 352
102: 352
103: 352
104: 352
105: 356
106: 356
107: 356
108: 356
109: 356
110: 352
111: 352
112: 356
113: 356
114: 356
115: 352
116: 356
117: 356
118: 356
119: 356
120: 352
default: 356
}
352: iconst_1
353: goto 357
356: iconst_0
357: ireturn
-------------
PR Comment: https://git.openjdk.org/jdk/pull/19926#issuecomment-2194416329
PR Comment: https://git.openjdk.org/jdk/pull/19926#issuecomment-2194470276
