When using git gui blame on some files in one of my repositories, and
noticing some rather depressing performance, I decided to look for some
low-hanging fruit to optimize.
I found that a lot of the time is spent in xdiff_outf continually
xreallocing and freeing memory for a string. I decided to replace that
with a strbuf. Unfortunately this now means that there are resources
that need to be freed after using xdiff_outf, so a new interface for
doing that needed to be created first.
The performance difference on my repository is not amazing, but quite
noticeable. Before:
:; time git blame -M -C -C -p --incremental server.c >/dev/null
101.52user 0.17system 1:41.73elapsed 99%CPU (0avgtext+0avgdata 0maxresident)k
0inputs+0outputs (0major+39561minor)pagefaults 0swaps
After:
:; time git blame -M -C -C -p --incremental server.c >/dev/null
80.38user 0.30system 1:20.81elapsed 99%CPU (0avgtext+0avgdata 0maxresident)k
0inputs+0outputs (0major+50979minor)pagefaults 0swaps
Obviously, it should improve other uses of xdiff_outf as well, though I
doubt many things hit it as hard as blame does.
[PATCH 1/2] Make xdiff_outf_{init,release} interface
[PATCH 2/2] Use strbuf for struct xdiff_emit_state's remainder
builtin-blame.c | 4 ++--
combine-diff.c | 4 ++--
diff.c | 20 ++++++++++----------
xdiff-interface.c | 44 ++++++++++++++++++++++----------------------
xdiff-interface.h | 6 ++++--
5 files changed, 40 insertions(+), 38 deletions(-)
-bcd
--
To unsubscribe from this list: send the line "unsubscribe git" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html| david | Re: Dual-Licensing Linux Kernel with GPL V2 and GPL V3 |
| Greg Kroah-Hartman | [PATCH 005/196] Chinese: add translation of SubmittingDrivers |
| Vladislav Bolkhovitin | Re: Integration of SCST in the mainstream Linux kernel |
| Jan Engelhardt | intel iommu (Re: -mm merge plans for 2.6.23) |
git: | |
| David Miller | [GIT]: Networking |
| David Miller | Re: [PATCH] pkt_sched: Destroy gen estimators under rtnl_lock(). |
| Gerrit Renker | [PATCH 27/37] dccp: Integration of dynamic feature activation - part 2 (server side) |
| Natalie Protasevich | [BUG] New Kernel Bugs |
