projects / haproxy-2.5.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 2235b26) | patch
OPTIM: http: improve parsing performance of long header lines
authorWilly Tarreau <w@1wt.eu>
Sat, 5 Nov 2016 16:35:40 +0000 (17:35 +0100)
committerWilly Tarreau <w@1wt.eu>
Sat, 5 Nov 2016 17:00:17 +0000 (18:00 +0100)
commit0431f9d476a1237f76e360998bc7605506651b84
tree9e4450a8c5935570268d4c4751e90ea357856fdetree | snapshot
parent2235b261b6163cf20325e519c13c3f93f16b4d3acommit | diff
OPTIM: http: improve parsing performance of long header lines

A performance test with 1kB cookies was capping at 194k req/s. After
implementing multi-byte skipping, the performance increased to 212k req/s,
or 9.2% faster. This patch implements this for architectures supporting
unaligned accesses (x86_64, x86, armv7a). Maybe other architectures can
benefit from this but they were not tested yet.
src/proto_http.c diff | blob | history
haproxy-2.5 public tree. Stable production code.