When reading large indexes from disk, a portion of the time is
dominated in malloc() calls. This can be mitigated by allocating a
large block of memory and manage it ourselves via memory pools.
This change moves the cache entry allocation to be on top of memory
pools.
Design:
The index_state struct will gain a notion of an associated memory_pool
from which cache_entries will be allocated from. When reading in the
index from disk, we have information on the number of entries and
their size, which can guide us in deciding how large our initial
memory allocation should be. When an index is discarded, the
associated memory_pool will be discarded as well - so the lifetime of
a cache_entry is tied to the lifetime of the index_state that it was
allocated for.
In the case of a Split Index, the following rules are followed. 1st,
some terminology is defined:
Terminology:
- 'the_index': represents the logical view of the index
- 'split_index': represents the "base" cache entries. Read from the
split index file.
'the_index' can reference a single split_index, as well as
cache_entries from the split_index. `the_index` will be discarded
before the `split_index` is. This means that when we are allocating
cache_entries in the presence of a split index, we need to allocate
the entries from the `split_index`'s memory pool. This allows us to
follow the pattern that `the_index` can reference cache_entries from
the `split_index`, and that the cache_entries will not be freed while
they are still being referenced.
Managing transient cache_entry structs:
Cache entries are usually allocated for an index, but this is not always
the case. Cache entries are sometimes allocated because this is the
type that the existing checkout_entry function works with. Because of
this, the existing code needs to handle cache entries associated with an
index / memory pool, and those that only exist transiently. Several
strategies were contemplated around how to handle this:
Chosen approach:
An extra field was added to the cache_entry type to track whether the
cache_entry was allocated from a memory pool or not. This is currently
an int field, as there are no more available bits in the existing
ce_flags bit field. If / when more bits are needed, this new field can
be turned into a proper bit field.
Alternatives:
1) Do not include any information about how the cache_entry was
allocated. Calling code would be responsible for tracking whether the
cache_entry needed to be freed or not.
Pro: No extra memory overhead to track this state
Con: Extra complexity in callers to handle this correctly.
The extra complexity and burden to not regress this behavior in the
future was more than we wanted.
2) cache_entry would gain knowledge about which mem_pool allocated it
Pro: Could (potentially) do extra logic to know when a mem_pool no
longer had references to any cache_entry
Con: cache_entry would grow heavier by a pointer, instead of int
We didn't see a tangible benefit to this approach
3) Do not add any extra information to a cache_entry, but when freeing a
cache entry, check if the memory exists in a region managed by existing
mem_pools.
Pro: No extra memory overhead to track state
Con: Extra computation is performed when freeing cache entries
We decided tracking and iterating over known memory pool regions was
less desirable than adding an extra field to track this stae.
Signed-off-by: Jameson Miller <jamill@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add functions for:
- combining two memory pools
- determining if a memory address is within the range managed by a
memory pool
These functions will be used by future commits.
Signed-off-by: Jameson Miller <jamill@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add initialization and discard functions to mem_pool type. As the
memory allocated by mem_pool can now be freed, we also track the large
allocations.
If the there are existing mp_blocks in the mem_poo's linked list of
mp_blocksl, then the mp_block for a large allocation is inserted
behind the head block. This is because only the head mp_block is considered
when searching for availble space. This results in the following
desirable properties:
1) The mp_block allocated for the large request will not be included
not included in the search for available in future requests, the large
mp_block is sized for the specific request and does not contain any
spare space.
2) The head mp_block will not bumped from considation for future
memory requests just because a request for a large chunk of memory
came in.
These changes are in preparation for a future commit that will utilize
creating and discarding memory pool.
Signed-off-by: Jameson Miller <jamill@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Instead of searching all memory blocks for available space to fulfill
a memory request, only search the head block. If the head block does
not have space, assume that previous block would most likely not be
able to fulfill request either. This could potentially lead to more
memory fragmentation, but also avoids searching memory blocks that
probably will not be able to fulfill request.
This pattern will benefit consumers that are able to generate a good
estimate for how much memory will be needed, or if they are performing
fixed sized allocations, so that once a block is exhausted it will
never be able to fulfill a future request.
Signed-off-by: Jameson Miller <jamill@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
It has been observed that the time spent loading an index with a large
number of entries is partly dominated by malloc() calls. This change
is in preparation for using memory pools to reduce the number of
malloc() calls made to allocate cahce entries when loading an index.
Add an API to allocate and discard cache entries, abstracting the
details of managing the memory backing the cache entries. This commit
does actually change how memory is managed - this will be done in a
later commit in the series.
This change makes the distinction between cache entries that are
associated with an index and cache entries that are not associated with
an index. A main use of cache entries is with an index, and we can
optimize the memory management around this. We still have other cases
where a cache entry is not persisted with an index, and so we need to
handle the "transient" use case as well.
To keep the congnitive overhead of managing the cache entries, there
will only be a single discard function. This means there must be enough
information kept with the cache entry so that we know how to discard
them.
A summary of the main functions in the API is:
make_cache_entry: create cache entry for use in an index. Uses specified
parameters to populate cache_entry fields.
make_empty_cache_entry: Create an empty cache entry for use in an index.
Returns cache entry with empty fields.
make_transient_cache_entry: create cache entry that is not used in an
index. Uses specified parameters to populate
cache_entry fields.
make_empty_transient_cache_entry: create cache entry that is not used in
an index. Returns cache entry with
empty fields.
discard_cache_entry: A single function that knows how to discard a cache
entry regardless of how it was allocated.
Signed-off-by: Jameson Miller <jamill@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Refactor refresh_cache_entry() to work on a specific index, instead of
implicitly using the_index. This is in preparation for making the
make_cache_entry function apply to a specific index.
Signed-off-by: Jameson Miller <jamill@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When running t7400 in a shell you observe more output than expected:
...
ok 8 - setup - hide init subdirectory
ok 9 - setup - repository to add submodules to
ok 10 - submodule add
[master (root-commit) d79ce16] one
Author: A U Thor <author@example.com>
1 file changed, 1 insertion(+)
create mode 100644 one.t
ok 11 - redirected submodule add does not show progress
ok 12 - redirected submodule add --progress does show progress
ok 13 - submodule add to .gitignored path fails
...
Fix the output by encapsulating the setup code in test_expect_success
Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When testing a reworded root commit, ensure that the squash-onto commit
which is created and amended is still the root commit.
Suggested-by: Phillip Wood <phillip.wood@talktalk.net>
Helped-by: Johannes Schindelin <Johannes.Schindelin@gmx.de>
Signed-off-by: Todd Zullinger <tmz@pobox.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"make NO_ICONV=NoThanks" did not override NEEDS_LIBICONV
(i.e. linkage of -lintl, -liconv, etc. that are platform-specific
tweaks), which has been corrected.
* es/make-no-iconv:
Makefile: make NO_ICONV really mean "no iconv"
A regression to "rebase -i --root" introduced during this cycle has
been fixed.
* js/rebase-i-root-fix:
rebase --root: fix amending root commit messages
rebase --root: demonstrate a bug while amending root commit messages
The code to read compressed bitmap was not careful to avoid reading
past the end of the file, which has been corrected.
* jk/ewah-bounds-check:
ewah: adjust callers of ewah_read_mmap()
ewah_read_mmap: bounds-check mmap reads
Update the Korean translation and change the team leader to Gwan-gyeong
Mun.
Signed-off-by: Gwan-gyeong Mun <elongbug@gmail.com>
Signed-off-by: Changwoo Ryu <cwryu@debian.org>
Reviewed-by: Gwan-gyeong Mun <elongbug@gmail.com>
Newly added codepath in merge-recursive had potential buffer
overrun, which has been fixed.
* en/rename-directory-detection:
merge-recursive: use xstrdup() instead of fixed buffer
Make zlib inflate codepath more robust against versions of zlib
that clobber unused portion of outbuf.
* jl/zlib-restore-nul-termination:
packfile: correct zlib buffer handling
"git p4" updates.
* ld/git-p4-updates:
git-p4: auto-size the block
git-p4: narrow the scope of exceptions caught when parsing an int
git-p4: raise exceptions from p4CmdList based on error from p4 server
git-p4: better error reporting when p4 fails
git-p4: add option to disable syncing of p4/master with p4
git-p4: disable-rebase: allow setting this via configuration
git-p4: add options --commit and --disable-rebase
Paths can be longer than PATH_MAX. Avoid a buffer overrun in
check_dir_renamed() by using xstrdup() to make a private copy safely.
Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
It was not "newer versions of bash" but newer versions of
bash-completion that made commit 085e2ee0e6 (completion: load
completion file for external subcommand, 2018-04-29) both necessary
and possible.
Update the corresponding RelNotes entry accordingly.
Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Three tests in 't7406-submodule-update' contain broken &&-chains, but
since they are all in subshells, chain-lint couldn't notice them.
Signed-off-by: SZEDER Gábor <szeder.dev@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The code path that triggered that "BUG" really does not want to run
without an explicit commit message. In the case where we want to amend a
commit message, we have an *implicit* commit message, though: the one of
the commit to amend. Therefore, this code path should not even be
entered.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When splitting a repository, running `git rebase -i --root` to reword
the initial commit, Git dies with
BUG: sequencer.c:795: root commit without message.
Signed-off-by: Todd Zullinger <tmz@pobox.com>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The return value of ewah_read_mmap() is now an ssize_t,
since we could (in theory) process up to 32GB of data. This
would never happen in practice, but a corrupt or malicious
.bitmap or index file could convince us to do so.
Let's make sure that we don't stuff the value into an int,
which would cause us to incorrectly move our pointer
forward. We'd always move too little, since negative values
are used for reporting errors. So the worst case is just
that we end up reporting a corrupt file, not an
out-of-bounds read.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The on-disk ewah format tells us how big the ewah data is,
and we blindly read that much from the buffer without
considering whether the mmap'd data is long enough, which
can lead to out-of-bound reads.
Let's make sure we have data available before reading it,
both for the ewah header/footer as well as for the bit data
itself. In particular:
- keep our ptr/len pair in sync as we move through the
buffer, and check it before each read
- check the size for integer overflow (this should be
impossible on 64-bit, as the size is given as a 32-bit
count of 8-byte words, but is possible on a 32-bit
system)
- return the number of bytes read as an ssize_t instead of
an int, again to prevent integer overflow
- compute the return value using a pointer difference;
this should yield the same result as the existing code,
but makes it more obvious that we got our computations
right
The included test is far from comprehensive, as it just
picks a static point at which to truncate the generated
bitmap. But in practice this will hit in the middle of an
ewah and make sure we're at least exercising this code.
Reported-by: Luat Nguyen <root@l4w.io>
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Support for the --set-upstream option was removed in 52668846ea
(builtin/branch: stop supporting the "--set-upstream" option,
2017-08-17). The change did not completely remove the command
due to an issue noted in the commit's log message.
So, a test was added to ensure that a command which uses the
'--set-upstream' option fails instead of silently acting as an alias
for the '--set-upstream-to' option due to option parsing features.
To avoid confusion, clarify that the option is disabled intentionally
in the corresponding test description.
The test is expected to be around as long as we intentionally fail
on seeing the '--set-upstream' option which in turn we expect to
do for a period of time after which we can be sure that existing
users of '--set-upstream' are aware that the option is no
longer supported.
Signed-off-by: Kaartic Sivaraam <kaartic.sivaraam@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
