https://gcc.gnu.org/bugzilla/show_bug.cgi?id=116743
--- Comment #18 from Eugene Rozenfeld <erozen at microsoft dot com> ---
Hi Rama.
I started looking at this. Below is what the info looks like for
ha_innobase::build_template in the gcov file you shared. If you look near the
end, you can see counts for calls to build_template_needs_field_in_icp but
nothing for the basic blocks that call build_template_needs_field. This is
probably why the compiler thinks that calls to build_template_needs_field are
unlikely and decides not to inline them. Is this gcov file generated by
profiling the same scenario that you then use to get your throughput numbers?
_ZN11ha_innobase14build_templateEb total:66636 head:0
0: 986
4: 103
5: 22
8: 68
13: 0
14: 1288
15: 1383
19: 0
27: 0
31: 162
46.2: 30
48.2: 33
48.4: 163
57.4: 916
59.4: 498
60: 0
61: 0
69: 202
71: 816
74: 1461
75: 1143
76: 1384
77: 67
78: 266
83: 76
83.1: 63
84: 27
87: 107
87.2: 35
90: 434
90.1: 0
91: 0
94: 184
112: 307
112.1: 0
113: 164
119: 18
120: 0
122: 39
123: 15
125: 1
126: 13
133: 9
138: 17
143: 122
169: 150
169.1: 0
170.1: 0
177: 0
181: 2
181.2: 77
185: 347
185.1: 0
186: 0
189: 167
192: 250
192.1: 0
194: 128
199: 12
200: 8
202: 33
203: 42
205: 2
206: 106
206.1: 0
207.1: 0
213: 10
216: 34
217: 12
222: 2
225: 82
227: 58
229: 60
229.2: 753
231: 13252
231.1: 0
233: 3813
240: 343
240.1: 0
240.3: 0
241.3: 0
244: 0
245: 0
246: 0
249: 0
254: 0
256: 574
259: 839
260: 760
262: 363
264: 0
270: 1287
276: 3074
277: 11192
279: 475
280: 0
283: 10535
284: 522
289: 191
289.1: 62
292: 43
292.3: 45
295.3: 71
298: 875
59611: 0
44: _ZN12dict_table_t11first_indexEv total:70
2: _ZNK12dict_table_t11first_indexEv total:70
2: 70
60: _ZN2ut14malloc_withkeyENS_16PSI_memory_key_tEm total:0
3:
_ZN2ut6detail6Alloc_INS0_9Alloc_pfsEE5allocILb0ES2_EENSt9enable_ifIXsrT0_21is_pfs_instrumented_vEPvE4typeEmj
total:0
1: _ZN2ut6detail9Alloc_pfs5allocILb0EEEPvmj total:0
2: 0
4: 0
10: 0
21: 0
8199: 0
3: _ZN2ut6detail8Alloc_fn5allocILb0EEEPvm total:0
4: _ZN2ut6detail8Alloc_fn6mallocEm total:0
0: _ZN2ut6detail6mallocEm total:0
0: 0
15: _ZN2ut6detail12PFS_metadata17pfs_owning_threadEPvP10PSI_thread
total:0
2: 0
16: _ZN2ut6detail12PFS_metadata11pfs_datalenEPvm total:0
3: 0
17: _ZN2ut6detail12PFS_metadata7pfs_keyEPvj total:0
2: 0
18: _ZN2ut6detail12PFS_metadata14pfs_metaoffsetEPvm total:0
3: 0
113:
_Z33build_template_needs_field_in_icpPK12dict_index_tPK14row_prebuilt_tbmb
total:25
11.4: 25
12: 0
194:
_Z33build_template_needs_field_in_icpPK12dict_index_tPK14row_prebuilt_tbmb
total:33
11.4: 33
12: 0
277: _ZNK12dict_index_t12is_clusteredEv total:3235
3: 3235
