We've had this notion of a "object_list" for a long time, which eventually
grew a "name" member because some users (notably git-rev-list) wanted to
name each object as it is generated.
That object_list is great for some things, but it isn't all that wonderful
for others, and the "name" member is generally not used by everybody.
This patch splits the users of the object_list array up into two: the
traditional list users, who want the list-like format, and who don't
actually use or want the name. And another class of users that really used
the list as an extensible array, and generally wanted to name the objects.
The patch is fairly straightforward, but it's also biggish. Most of it
really just cleans things up: switching the revision parsing and listing
over to the array makes things like the builtin-diff usage much simpler
(we now see exactly how many members the array has, and we don't get the
objects reversed from the order they were on the command line).
One of the main reasons for doing this at all is that the malloc overhead
of the simple object list was actually pretty high, and the array is just
a lot denser. So this patch brings down memory usage by git-rev-list by
just under 3% (on top of all the other memory use optimizations) on the
mozilla archive.
It does add more lines than it removes, and more importantly, it adds a
whole new infrastructure for maintaining lists of objects, but on the
other hand, the new dynamic array code is pretty obvious. The change to
builtin-diff-tree.c shows a fairly good example of why an array interface
is sometimes more natural, and just much simpler for everybody.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
This shrinks "struct object" to the absolutely minimal size possible.
It now contains /only/ the object flags and the SHA1 hash name of the
object.
The "refs" field, which is really needed only for fsck, is maintained in
a separate hashed lookup-table, allowing all normal users to totally
ignore it.
This helps memory usage, although not as much as I hoped: it looks like
the allocation overhead of malloc (and the alignment constraints in
particular) means that while the structure size shrinks, the actual
allocation overhead mostly does not.
[ That said: memory usage is actually down, but not as much as it should
be: I suspect just one of the object types actually ended up shrinking
its effective allocation size.
To get to the next level, we probably need specialized allocators that
don't pad the allocation more than necessary. ]
The separation makes for some code cleanup, though, and makes the ref
tracking that fsck wants a clearly separate thing.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
This shrinks "struct object" by a small amount, by getting rid of the
"struct type *" pointer and replacing it with a 3-bit bitfield instead.
In addition, we merge the bitfields and the "flags" field, which
incidentally should also remove a useless 4-byte padding from the object
when in 64-bit mode.
Now, our "struct object" is still too damn large, but it's now less
obviously bloated, and of the remaining fields, only the "util" (which is
not used by most things) is clearly something that should be eventually
discarded.
This shrinks the "git-rev-list --all" memory use by about 2.5% on the
kernel archive (and, perhaps more importantly, on the larger mozilla
archive). That may not sound like much, but I suspect it's more on a
64-bit platform.
There are other remaining inefficiencies (the parent lists, for example,
probably have horrible malloc overhead), but this was pretty obvious.
Most of the patch is just changing the comparison of the "type" pointer
from one of the constant string pointers to the appropriate new TYPE_xxx
small integer constant.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
This is preparatory work for further cleanups, where we try to make
tree_entry look more like the more efficient tree-walk descriptor.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
This allows us to avoid allocating information for names etc, because
we can just use the information from the tree buffer directly.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
This replaces occurences of "blob", "commit", "tag", and "tree",
where they're really used as type specifiers, which we already
have defined global constants for.
Signed-off-by: Peter Eriksen <s022018@student.dtu.dk>
Signed-off-by: Junio C Hamano <junkio@cox.net>
The hashed object lookup had a subtle bug in re-hashing: it did
for (i = 0; i < count; i++)
if (objs[i]) {
.. rehash ..
where "count" was the old hash couny. Oon the face of it is obvious, since
it clearly re-hashes all the old objects.
However, it's wrong.
If the last old hash entry before re-hashing was in use (or became in use
by the re-hashing), then when re-hashing could have inserted an object
into the hash entries with idx >= count due to overflow. When we then
rehash the last old entry, that old entry might become empty, which means
that the overflow entries should be re-hashed again.
In other words, the loop has to be fixed to either traverse the whole
array, rather than just the old count.
(There's room for a slight optimization: instead of counting all the way
up, we can break when we see the first empty slot that is above the old
"count". At that point we know we don't have any collissions that we might
have to fix up any more. This patch only does the trivial fix)
[jc: with trivial fix on trivial fix]
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
Calling hashtable_index from find_object before objs is created
would result in division by zero failure. Avoid it.
Also the given object name may not be aligned suitably for
unsigned int; avoid dereferencing casted pointer.
Signed-off-by: Junio C Hamano <junkio@cox.net>
In a simple test, this brings down the CPU time from 47 sec to 22 sec.
Signed-off-by: Johannes Schindelin <Johannes.Schindelin@gmx.de>
Signed-off-by: Junio C Hamano <junkio@cox.net>
ISO C99 (and GCC 3.x or later) lets you write a flexible array
at the end of a structure, like this:
struct frotz {
int xyzzy;
char nitfol[]; /* more */
};
GCC 2.95 and 2.96 let you to do this with "char nitfol[0]";
unfortunately this is not allowed by ISO C90.
This declares such construct like this:
struct frotz {
int xyzzy;
char nitfol[FLEX_ARRAY]; /* more */
};
and git-compat-util.h defines FLEX_ARRAY to 0 for gcc 2.95 and
empty for others.
If you are using a C90 C compiler, you should be able
to override this with CFLAGS=-DFLEX_ARRAY=1 from the
command line of "make".
Signed-off-by: Junio C Hamano <junkio@cox.net>
Morten Welinder <mwelinder@gmail.com> writes:
> The code looks wrong. It assumes that pointers are no larger than ints.
> If pointers are larger than ints, the code does not necessarily compute
> a consistent ordering and qsort is allowed to do whatever it wants.
>
> Morten
>
> static int compare_object_pointers(const void *a, const void *b)
> {
> const struct object * const *pa = a;
> const struct object * const *pb = b;
> return *pa - *pb;
> }
Signed-off-by: Junio C Hamano <junkio@cox.net>
Store pointers to referenced objects in a variable sized array instead
of linked list. This cuts down memory usage of utilities which use
object references; e.g., git-fsck-objects --full on the git.git
repository consumes about 2 MB of memory tracked by Massif instead of
7 MB before the change. Object refs are still the biggest consumer of
memory (57%), but the malloc overhead for a single block instead of a
linked list is substantially smaller.
Signed-off-by: Sergey Vlasov <vsu@altlinux.ru>
Signed-off-by: Junio C Hamano <junkio@cox.net>
The object parsing code builds a generic "this object references that
object" because doing a full connectivity check for fsck requires it.
However, nothing else really needs it, and it's quite expensive for
git-rev-list that can have tons of objects in flight.
So, exactly like the commit buffer save thing, add a global flag to
disable it, and use it in git-rev-list.
Before:
$ /usr/bin/time git-rev-list --objects v2.6.12..HEAD | wc -l
12.28user 0.29system 0:12.57elapsed 99%CPU (0avgtext+0avgdata 0maxresident)k
0inputs+0outputs (0major+26718minor)pagefaults 0swaps
59124
After this change:
$ /usr/bin/time git-rev-list --objects v2.6.12..HEAD | wc -l
10.33user 0.18system 0:10.54elapsed 99%CPU (0avgtext+0avgdata 0maxresident)k
0inputs+0outputs (0major+18509minor)pagefaults 0swaps
59124
and note how the number of pages touched by git-rev-list for this
particular object list has shrunk from 26,718 (104 MB) to 18,509 (72 MB).
Calculating the total object difference between two git revisions is still
clearly the most expensive git operation (both in memory and CPU time),
but it's now less than 40% of what it used to be.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
Add function to look up an object which is entirely unknown, so that
it can be put in a list. Various other functions related to lists of
objects.
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
Packed delta files created by git-pack-objects seems to be the
way to go, and existing "delta" object handling code has exposed
the object representation details to too many places. Remove it
while we refactor code to come up with a proper interface in
sha1_file.c.
Signed-off-by: Junio C Hamano <junkio@cox.net>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Handle parsing a tag for a non-present object. This adds a function to lookup
an object with lookup_* for * in a string, so that it can get the right storage
based on the "type" line in the tag.
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Make 'sha1' parameters const where possible
Signed-off-by: Jason McMullan <jason.mcmullan@timesys.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Add <limits.h> to the include files handled by "cache.h", and remove
extraneous #include directives from various .c files. The rule is that
"cache.h" gets all the basic stuff, so that we'll have as few system
dependencies as possible.
This adds knowledge of delta objects to fsck-cache and various object
parsing code. A new switch to git-fsck-cache is provided to display the
maximum delta depth found in a repository.
Signed-off-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
It turns out that parse_object() is loading and decompressing given
object to free it just before calling the specific object parsing
function which does mmap and decompress the same object again. This
patch introduces the ability to parse specific objects directly from a
memory buffer.
Without this patch, running git-fsck-cache on the kernel repositorytake:
real 0m13.006s
user 0m11.421s
sys 0m1.218s
With this patch applied:
real 0m8.060s
user 0m7.071s
sys 0m0.710s
The performance increase is significant, and this is kind of a
prerequisite for sane delta object support with fsck.
Signed-off-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
This patch fixes memory leaks in parse_object() and related functions;
these leaks were very noticeable when running git-fsck-cache.
Signed-off-by: Sergey Vlasov <vsu@altlinux.ru>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
This adds a function that parses an object from the database when we have
to look up its actual type. It also checks the hash of the file, due to
its heritage as part of fsck-cache.
Signed-Off-By: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Introduce xmalloc and xrealloc to die gracefully with a descriptive
message when out of memory, rather than taking a SIGSEGV.
Signed-off-by: Christopher Li<chrislgit@chrisli.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
