When writing out the index, the ondisk struct was being allocated and freed within the loop of cache entries. A better way would be to use a ondisk struct on the stack and reuse it avoiding the alloc and free calls.
A test has been added to measure the performance of writing the index (p0007-write-cache). Running this test on smaller repos showed no degradation in performance. On larger repos there was ~3-7% improvement. 0007.2: write_locked_index 10 times (1000001 files) 5.98(0.00+0.04) 5.75(0.01+0.04) -3.8% 0007.2: write_locked_index 10 times (1000001 files) 6.20(0.00+0.06) 5.86(0.01+0.03) -5.5% 0007.2: write_locked_index 3 times (4394531 files) 10.29(0.04+0.03) 9.75(0.04+0.01) -5.2% 0007.2: write_locked_index 3 times (4394531 files) 10.52(0.00+0.04) 9.79(0.03+0.03) -6.9% Kevin Willford (3): perf: add test for writing the index read-cache: fix memory leak in do_write_index read-cache: avoid allocating every ondisk entry when writing Makefile | 1 + read-cache.c | 62 +++++++++++++++++++++++++-------------------- t/helper/test-write-cache.c | 23 +++++++++++++++++ t/perf/p0007-write-cache.sh | 29 +++++++++++++++++++++ 4 files changed, 87 insertions(+), 28 deletions(-) create mode 100644 t/helper/test-write-cache.c create mode 100755 t/perf/p0007-write-cache.sh -- 2.14.1.205.g2812f3410d
