Fix performance regression in v1.8.4.x and later.
* jk/mark-edges-uninteresting:
list-objects: only look at cmdline trees with edge_hint
t/perf: time rev-list with UNINTERESTING commits
When rev-list is given a command-line like:
git rev-list --objects $commit --not --all
the most accurate answer is the difference between the set
of objects reachable from $commit and the set reachable from
all of the existing refs. However, we have not historically
provided that answer, because it is very expensive to
calculate. We would have to open every tree of every commit
in the entire history.
Instead, we find the accurate set difference of the
reachable commits, and then mark the trees at the boundaries
as uninteresting. This misses objects which appear in the
trees of both the interesting commits and deep within the
uninteresting history.
Commit fbd4a70 (list-objects: mark more commits as edges in
mark_edges_uninteresting, 2013-08-16) noticed that we miss
those objects during pack-objects, and added code to examine
the trees of all of the "--not" refs given on the
command-line. Note that this is still not the complete set
difference, because we look only at the tips of the
command-line arguments, not all of their reachable commits.
But it increases the set of boundary objects we consider,
which is especially important for shallow fetches. So we
are trading extra CPU time for a larger set of boundary
objects, which can improve the resulting pack size for a
--thin pack.
This tradeoff probably makes sense in the context of
pack-objects, where we have set revs->edge_hint to have the
traversal feed us the set of boundary objects. For a
regular rev-list, though, it is probably not a good
tradeoff. It is true that it makes our list slightly closer
to a true set difference, but it is a rare case where this
is important. And because we do not have revs->edge_hint
set, we do nothing useful with the larger set of boundary
objects.
This patch therefore ties the extra tree examination to the
revs->edge_hint flag; it is the presence of that flag that
makes the tradeoff worthwhile.
Here is output from the p0001-rev-list showing the
improvement in performance:
Test HEAD^ HEAD
-----------------------------------------------------------------------------------------
0001.1: rev-list --all 0.69(0.65+0.02) 0.69(0.66+0.02) +0.0%
0001.2: rev-list --all --objects 3.22(3.19+0.03) 3.23(3.20+0.03) +0.3%
0001.4: rev-list $commit --not --all 0.04(0.04+0.00) 0.04(0.04+0.00) +0.0%
0001.5: rev-list --objects $commit --not --all 0.27(0.26+0.01) 0.04(0.04+0.00) -85.2%
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When there is no sufficient overlap between old and new history
during a fetch into a shallow repository, we unnecessarily sent
objects the sending side knows the receiving end has.
* nd/fetch-into-shallow:
Add testcase for needless objects during a shallow fetch
list-objects: mark more commits as edges in mark_edges_uninteresting
list-objects: reduce one argument in mark_edges_uninteresting
upload-pack: delegate rev walking in shallow fetch to pack-objects
shallow: add setup_temporary_shallow()
shallow: only add shallow graft points to new shallow file
move setup_alternate_shallow and write_shallow_commits to shallow.c
The purpose of edge commits is to let pack-objects know what objects
it can use as base, but does not need to include in the thin pack
because the other side is supposed to already have them. So far we
mark uninteresting parents of interesting commits as edges. But even
an unrelated uninteresting commit (that the other side has) may
become a good base for pack-objects and help produce more efficient
packs.
This is especially true for shallow clone, when the client issues a
fetch with a depth smaller or equal to the number of commits the
server is ahead of the client. For example, in this commit history
the client has up to "A" and the server has up to "B":
-------A---B
have--^ ^
/
want--+
If depth 1 is requested, the commit list to send to the client
includes only B. The way m_e_u is working, it checks if parent
commits of B are uninteresting, if so mark them as edges. Due to
shallow effect, commit B is grafted to have no parents and the
revision walker never sees A as the parent of B. In fact it marks no
edges at all in this simple case and sends everything B has to the
client even if it could have excluded what A and also the client
already have.
In a slightly different case where A is not a direct parent of B
(iow there are commits in between A and B), marking A as an edge can
still save some because B may still have stuff from the far ancestor
A.
There is another case from the earlier patch, when we deepen a ref
from C->E to A->E:
---A---B C---D---E
want--^ ^ ^
shallow-+ /
have-------+
In this case we need to send A and B to the client, and C (i.e. the
current shallow point that the client informs the server) is a very
good base because it's closet to A and B. Normal m_e_u won't recognize
C as an edge because it only looks back to parents (i.e. A<-B) not the
opposite way B->C even if C is already marked as uninteresting commit
by the previous patch.
This patch includes all uninteresting commits from command line as
edges and lets pack-objects decide what's best to do. The upside is we
have better chance of producing better packs in certain cases. The
downside is we may need to process some extra objects on the server
side.
For the shallow case on git.git, when the client is 5 commits behind
and does "fetch --depth=3", the result pack is 99.26 KiB instead of
4.92 MiB.
Reported-and-analyzed-by: Matthijs Kooijman <matthijs@stdin.nl>
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
mark_edges_uninteresting() is always called with this form
mark_edges_uninteresting(revs->commits, revs, ...);
Remove the first argument and let mark_edges_uninteresting figure that
out by itself. It helps answer the question "are this commit list and
revs related in any way?" when looking at mark_edges_uninteresting
implementation.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Many code paths will free a tree object's buffer and set it
to NULL after finishing with it in order to keep memory
usage down during a traversal. However, out of 8 sites that
do this, only one actually unsets the "parsed" flag back.
Those sites that don't are setting a trap for later users of
the tree object; even after calling parse_tree, the buffer
will remain NULL, causing potential segfaults.
It is not known whether this is triggerable in the current
code. Most commands do not do an in-memory traversal
followed by actually using the objects again. However, it
does not hurt to be safe for future callers.
In most cases, we can abstract this out to a
"free_tree_buffer" helper. However, there are two
exceptions:
1. The fsck code relies on the parsed flag to know that we
were able to parse the object at one point. We can
switch this to using a flag in the "flags" field.
2. The index-pack code sets the buffer to NULL but does
not free it (it is freed by a caller). We should still
unset the parsed flag here, but we cannot use our
helper, as we do not want to free the buffer.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
It is a basic code hygiene to avoid magic constants that are unnamed.
Besides, this helps extending the value later on for "interesting, but
cannot decide if the entry truely matches yet" (ie. prefix matches)
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The traverse_commit_list() API takes two callback functions, one to show
commit objects, and the other to show other kinds of objects. Even though
the former has a callback data parameter, so that the callback does not
have to rely on global state, the latter does not.
Give the show_objects() callback the same callback data parameter.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* nd/struct-pathspec:
pathspec: rename per-item field has_wildcard to use_wildcard
Improve tree_entry_interesting() handling code
Convert read_tree{,_recursive} to support struct pathspec
Reimplement read_tree_recursive() using tree_entry_interesting()
t_e_i() can return -1 or 2 to early shortcut a search. Current code
may use up to two variables to handle it. One for saving return value
from t_e_i temporarily, one for saving return code 2.
The second variable is not needed. If we make sure the first variable
does not change until the next t_e_i() call, then we can do something
like this:
int ret = 0;
while (...) {
if (ret != 2) {
ret = t_e_i();
if (ret < 0) /* no longer interesting */
break;
if (ret == 0) /* skip this round */
continue;
}
/* ret > 0, interesting */
}
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git rev-list --first-parent --boundary $commit^..$commit" segfaults on a
merge commit since 8d2dfc4 (process_{tree,blob}: show objects without
buffering, 2009-04-10), as it tried to dereference a commit that was
discarded as UNINTERESTING without being parsed (hence lacking "tree").
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When traversing commits, the selection of commits would heed the list of
pathspecs passed, but subsequent walking of the trees of those commits
would not. This resulted in 'rev-list --objects HEAD -- <paths>'
displaying objects at unwanted paths.
Have process_tree() call tree_entry_interesting() to determine which paths
are interesting and should be walked.
Naturally, this change can provide a large speedup when paths are specified
together with --objects, since many tree entries are now correctly ignored.
Interestingly, though, this change also gives me a small (~1%) but
repeatable speedup even when no paths are specified with --objects.
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* lt/pack-object-memuse:
show_object(): push path_name() call further down
process_{tree,blob}: show objects without buffering
Conflicts:
builtin-pack-objects.c
builtin-rev-list.c
list-objects.c
list-objects.h
upload-pack.c
Here's a less trivial thing, and slightly more dubious one.
I was looking at that "struct object_array objects", and wondering why we
do that. I have honestly totally forgotten. Why not just call the "show()"
function as we encounter the objects? Rather than add the objects to the
object_array, and then at the very end going through the array and doing a
'show' on all, just do things more incrementally.
Now, there are possible downsides to this:
- the "buffer using object_array" _can_ in theory result in at least
better I-cache usage (two tight loops rather than one more spread out
one). I don't think this is a real issue, but in theory..
- this _does_ change the order of the objects printed. Instead of doing a
"process_tree(revs, commit->tree, &objects, NULL, "");" in the loop
over the commits (which puts all the root trees _first_ in the object
list, this patch just adds them to the list of pending objects, and
then we'll traverse them in that order (and thus show each root tree
object together with the objects we discover under it)
I _think_ the new ordering actually makes more sense, but the object
ordering is actually a subtle thing when it comes to packing
efficiency, so any change in order is going to have implications for
packing. Good or bad, I dunno.
- There may be some reason why we did it that odd way with the object
array, that I have simply forgotten.
Anyway, now that we don't buffer up the objects before showing them
that may actually result in lower memory usage during that whole
traverse_commit_list() phase.
This is seriously not very deeply tested. It makes sense to me, it seems
to pass all the tests, it looks ok, but...
Does anybody remember why we did that "object_array" thing? It used to be
an "object_list" a long long time ago, but got changed into the array due
to better memory usage patterns (those linked lists of obejcts are
horrible from a memory allocation standpoint). But I wonder why we didn't
do this back then. Maybe there's a reason for it.
Or maybe there _used_ to be a reason, and no longer is.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In particular, pushing the "path_name()" call _into_ the show() function
would seem to allow
- more clarity into who "owns" the name (ie now when we free the name in
the show_object callback, it's because we generated it ourselves by
calling path_name())
- not calling path_name() at all, either because we don't care about the
name in the first place, or because we are actually happy walking the
linked list of "struct name_path *" and the last component.
Now, I didn't do that latter optimization, because it would require some
more coding, but especially looking at "builtin-pack-objects.c", we really
don't even want the whole pathname, we really would be better off with the
list of path components.
Why? We use that name for two things:
- add_preferred_base_object(), which actually _wants_ to traverse the
path, and now does it by looking for '/' characters!
- for 'name_hash()', which only cares about the last 16 characters of a
name, so again, generating the full name seems to be just unnecessary
work.
Anyway, so I didn't look any closer at those things, but it did convince
me that the "show_object()" calling convention was crazy, and we're
actually better off doing _less_ in list-objects.c, and giving people
access to the internal data structures so that they can decide whether
they want to generate a path-name or not.
This patch does that, and then for people who did use the name (even if
they might do something more clever in the future), it just does the
straightforward "name = path_name(path, component); .. free(name);" thing.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* cc/bisect-filter: (21 commits)
rev-list: add "int bisect_show_flags" in "struct rev_list_info"
rev-list: remove last static vars used in "show_commit"
list-objects: add "void *data" parameter to show functions
bisect--helper: string output variables together with "&&"
rev-list: pass "int flags" as last argument of "show_bisect_vars"
t6030: test bisecting with paths
bisect: use "bisect--helper" and remove "filter_skipped" function
bisect: implement "read_bisect_paths" to read paths in "$GIT_DIR/BISECT_NAMES"
bisect--helper: implement "git bisect--helper"
bisect: use the new generic "sha1_pos" function to lookup sha1
rev-list: call new "filter_skip" function
patch-ids: use the new generic "sha1_pos" function to lookup sha1
sha1-lookup: add new "sha1_pos" function to efficiently lookup sha1
rev-list: pass "revs" to "show_bisect_vars"
rev-list: make "show_bisect_vars" non static
rev-list: move code to show bisect vars into its own function
rev-list: move bisect related code into its own file
rev-list: make "bisect_list" variable local to "cmd_rev_list"
refs: add "for_each_ref_in" function to refactor "for_each_*_ref" functions
quote: add "sq_dequote_to_argv" to put unwrapped args in an argv array
...
The name of the processed object was duplicated for passing it to
add_object(), but that already calls path_name, which allocates a new
string anyway. So the memory allocated by the xstrdup calls just went
nowhere, leaking memory.
This reduces the RSS usage for a "rev-list --all --objects" by about 10% on
the gentoo repo (fully packed) as well as linux-2.6.git:
gentoo:
| old | new
----------------|-------------------------------
RSS | 1537284 | 1388408
VSZ | 1816852 | 1667952
time elapsed | 1:49.62 | 1:48.99
min. page faults| 417178 | 379919
linux-2.6.git:
| old | new
----------------|-------------------------------
RSS | 324452 | 292996
VSZ | 491792 | 460376
time elapsed | 0:14.53 | 0:14.28
min. page faults| 89360 | 81613
Signed-off-by: Björn Steinbrink <B.Steinbrink@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The goal of this patch is to get rid of the "static struct rev_info
revs" static variable in "builtin-rev-list.c".
To do that, we need to pass the revs to the "show_commit" function
in "builtin-rev-list.c" and this in turn means that the
"traverse_commit_list" function in "list-objects.c" must be passed
functions pointers to functions with 2 parameters instead of one.
So we have to change all the callers and all the functions passed
to "traverse_commit_list".
Anyway this makes the code more clean and more generic, so it
should be a good thing in the long run.
Signed-off-by: Christian Couder <chriscool@tuxfamily.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
As these functions are directly called with the result
from lookup_tree/blob, they must handle NULL.
Signed-off-by: Martin Koegler <mkoegler@auto.tuwien.ac.at>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
If we were listing objects too then the objects were buffered in an
array only reachable from a stack allocated structure. When this
function returns that array would be leaked as nobody would have
a reference to it anymore.
Historically this hasn't been a problem as the primary user of
traverse_commit_list() (the noble git-rev-list) would terminate
as soon as the function was finished, thus allowing the operating
system to cleanup memory. However we have been leaking this data
in git-pack-objects ever since that program learned how to run the
revision listing internally, rather than relying on reading object
names from git-rev-list.
To better facilitate reuse of traverse_commit_list during other
builtin tools (such as git-fetch) we shouldn't leak temporary memory
like this and instead we need to clean up properly after ourselves.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This allows us to pack superprojects and thus clone them (but not yet
check them out on the receiving side.. That's the next patch)
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
This removes slightly more lines than it adds, but the real reason for
doing this is that future optimizations will require more setup of the
tree descriptor, and so we want to do it in one place.
Also renamed the "desc.buf" field to "desc.buffer" just to trigger
compiler errors for old-style manual initializations, making sure I
didn't miss anything.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
This teaches the internal rev-list logic to understand options
that are needed for pack handling: --all, --unpacked, and --thin.
It also moves two functions from builtin-rev-list to list-objects
so that the two programs can share more code.
Signed-off-by: Junio C Hamano <junkio@cox.net>
Create a separate file, list-objects.c, and move object listing
routines from rev-list to it. The next round will use it in
pack-objects directly.
Signed-off-by: Junio C Hamano <junkio@cox.net>
Junio C Hamano
Jeff King
Nguyễn Thái Ngọc Duy
Elijah Newren
Linus Torvalds
Björn Steinbrink
Christian Couder
Martin Koegler
Shawn O. Pearce
Martin Waitz
Junio C Hamano