2 commits

Author	SHA1	Message	Date
Ryan Prichard	129f7a1d8e	Neon-optimized version of the GNU symbol calculation ... On 64-bit walleye, improves the linker relocation benchmark from 71.9ms to 70.7ms (1.7% of the run-time). On a 32-bit device, it improves the linker relocation benchmark from 205.5ms to 201.2ms (2.1% of the run-time). $ adb shell taskset 10 /data/benchmarktest64/linker-benchmarks/linker-benchmarks --benchmark_repetitions=100 --benchmark_display_aggregates_only -------------------------------------------------------------------- Benchmark Time CPU Iterations -------------------------------------------------------------------- BM_gnu_hash_simple_mean 15232 ns 15212 ns 100 BM_gnu_hash_simple_median 15176 ns 15159 ns 100 BM_gnu_hash_simple_stddev 111 ns 110 ns 100 BM_gnu_hash_neon_mean 10265 ns 10252 ns 100 BM_gnu_hash_neon_median 10261 ns 10249 ns 100 BM_gnu_hash_neon_stddev 28.1 ns 23.9 ns 100 Bug: none Test: linker-unit-tests Test: linker-benchmarks Change-Id: I3983bca1dddc9241bb70290ad3651d895f046660	2020-01-13 13:29:25 -08:00
Ryan Prichard	339ecef22d	Optimize GNU hash linking for large inputs ... Symbol lookup is O(L) where L is the number of libraries to search (e.g. in the global and local lookup groups). Factor out the per-DSO work into soinfo_do_lookup_impl, and optimize for the situation where all the DSOs are using DT_GNU_HASH (rather than SysV hashes). To load a set of libraries, the loader first constructs an auxiliary list of libraries (SymbolLookupList, containing SymbolLookupLib objects). The SymbolLookupList is reused for each DSO in a load group. (-Bsymbolic is accommodated by modifying the SymbolLookupLib at the front of the list.) To search for a symbol, soinfo_do_lookup_impl has a small loop that first scans a vector of GNU bloom filters looking for a possible match. There was a slight improvement from templatizing soinfo_do_lookup_impl and skipping the does-this-DSO-lack-GNU-hash check. Rewrite the relocation processing loop to be faster. There are specialized functions that handle the expected relocation types in normal relocation sections and in PLT relocation sections. This CL can reduce the initial link time of large programs by around 40-50% (e.g. audioserver, cameraserver, etc). On the linker relocation benchmark (64-bit walleye), it reduces the time from 131.6ms to 71.9ms. Bug: http://b/143577578 (incidentally fixed by this CL) Test: bionic-unit-tests Change-Id: If40a42fb6ff566570f7280b71d58f7fa290b9343	2020-01-13 13:29:25 -08:00