bdiff: replace hash algorithm
This patch replaces lyhash with the hash algorithm used by diffutils.
The algorithm has its origins in Git commit
2e9d1410, which is all the
way back from 1992. The license header in the code at that revision
in GPL v2.
I have not performed an extensive analysis of the distribution
(and therefore buckets) of hash output. However, `hg perfbdiff`
gives some clear wins. I'd like to think that if it is good enough
for diffutils it is good enough for us?
From the mozilla-unified repository:
$ perfbdiff -m
3041e4d59df2
! wall 0.053271 comb 0.060000 user 0.060000 sys 0.000000 (best of 100)
! wall 0.035827 comb 0.040000 user 0.040000 sys 0.000000 (best of 100)
$ perfbdiff
0e9928989e9c --alldata --count 100
! wall 6.204277 comb 6.200000 user 6.200000 sys 0.000000 (best of 3)
! wall 4.309710 comb 4.300000 user 4.300000 sys 0.000000 (best of 3)
From the hg repo:
$ perfbdiff 35000 --alldata --count 1000
! wall 0.660358 comb 0.660000 user 0.660000 sys 0.000000 (best of 15)
! wall 0.534092 comb 0.530000 user 0.530000 sys 0.000000 (best of 19)
Looking at the generated assembly and statistical profiler output
from the kernel level, I believe there is room to make this function
even faster. Namely, we're still consuming data character by character
instead of at the word level. This translates to more loop iterations
and more instructions.
At this juncture though, the real performance killer is that we're
hashing every line. We should get a significant speedup if we change
the algorithm to find the longest prefix, longest suffix, treat those
as single "lines" and then only do the line splitting and hashing on
the parts that are different. That will require a lot of C code,
however. I'm optimistic this approach could result in a ~2x speedup.
profiling: make statprof the default profiler (BC)
The statprof sampling profiler runs with significantly less overhead.
Its data is therefore more useful. Furthermore, its default output
shows the hotpath by default, which I've found to be way more useful
than the default profiler's function time table.
There is one behavioral regression with this change worth noting:
the statprof profiler currently doesn't profile individual hgweb
requests like lsprof does. This is because the current implementation
of statprof only profiles the thread that started profiling.
The ability for lsprof to profile individual hgweb requests is
relatively new and likely not widely used. Furthermore, I have plans
to modify statprof to support profiling multiple threads. I expect
that change to go through several iterations. I'm submitting this
patch first so there is more time to test statprof. Perfect is the
enemy of good.
profiling: use vendored statprof and upstream enhancements (BC)
Now that the statprof module is vendored and suitable for use, we
switch our statprof profiler to use it. This required some minor
changes because of drift between the official statprof profiler
and the vendored copy.
We also incorporate Facebook's improvements from the "statprofext"
extension at
https://bitbucket.org/facebook/hg-experimental, notably support for
different display formats.
Because statprof output is different, this is marked as BC. Although
most users likely won't notice since most users don't profile.