Teach FSCACHE to remember "not found" directories.
This is a performance optimization.
FSCACHE is a performance optimization available for Windows. It
intercepts Posix-style lstat() calls into an in-memory directory
using FindFirst/FindNext. It improves performance on Windows by
catching the first lstat() call in a directory, using FindFirst/
FindNext to read the list of files (and attribute data) for the
entire directory into the cache, and short-cut subsequent lstat()
calls in the same directory. This gives a major performance
boost on Windows.
However, it does not remember "not found" directories. When STATUS
runs and there are missing directories, the lstat() interception
fails to find the parent directory and simply return ENOENT for the
file -- it does not remember that the FindFirst on the directory
failed. Thus subsequent lstat() calls in the same directory, each
re-attempt the FindFirst. This completely defeats any performance
gains.
This can be seen by doing a sparse-checkout on a large repo and
then doing a read-tree to reset the skip-worktree bits and then
running status.
This change reduced status times for my very large repo by 60%.
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Windows paths are typically limited to MAX_PATH = 260 characters, even
though the underlying NTFS file system supports paths up to 32,767 chars.
This limitation is also evident in Windows Explorer, cmd.exe and many
other applications (including IDEs).
Particularly annoying is that most Windows APIs return bogus error codes
if a relative path only barely exceeds MAX_PATH in conjunction with the
current directory, e.g. ERROR_PATH_NOT_FOUND / ENOENT instead of the
infinitely more helpful ERROR_FILENAME_EXCED_RANGE / ENAMETOOLONG.
Many Windows wide char APIs support longer than MAX_PATH paths through the
file namespace prefix ('\\?\' or '\\?\UNC\') followed by an absolute path.
Notable exceptions include functions dealing with executables and the
current directory (CreateProcess, LoadLibrary, Get/SetCurrentDirectory) as
well as the entire shell API (ShellExecute, SHGetSpecialFolderPath...).
Introduce a handle_long_path function to check the length of a specified
path properly (and fail with ENAMETOOLONG), and to optionally expand long
paths using the '\\?\' file namespace prefix. Short paths will not be
modified, so we don't need to worry about device names (NUL, CON, AUX).
Contrary to MSDN docs, the GetFullPathNameW function doesn't seem to be
limited to MAX_PATH (at least not on Win7), so we can use it to do the
heavy lifting of the conversion (translate '/' to '\', eliminate '.' and
'..', and make an absolute path).
Add long path error checking to xutftowcs_path for APIs with hard MAX_PATH
limit.
Add a new MAX_LONG_PATH constant and xutftowcs_long_path function for APIs
that support long paths.
While improved error checking is always active, long paths support must be
explicitly enabled via 'core.longpaths' option. This is to prevent end
users to shoot themselves in the foot by checking out files that Windows
Explorer, cmd/bash or their favorite IDE cannot handle.
Test suite:
Test the case is when the full pathname length of a dir is close
to 260 (MAX_PATH).
Bug report and an original reproducer by Andrey Rogozhnikov:
https://github.com/msysgit/git/pull/122#issuecomment-43604199
Note that the test cannot rely on the presence of short names, as they
are not enabled by default except on the system drive.
[jes: adjusted test number to avoid conflicts, reinstated && chain,
adjusted test to work without short names]
Thanks-to: Martin W. Kirst <maki@bitkings.de>
Thanks-to: Doug Kelly <dougk.ff7@gmail.com>
Signed-off-by: Karsten Blees <blees@dcon.de>
Original-test-by: Andrey Rogozhnikov <rogozhnikov.andrey@gmail.com>
Signed-off-by: Stepan Kasal <kasal@ucw.cz>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Windows paths are typically limited to MAX_PATH = 260 characters, even
though the underlying NTFS file system supports paths up to 32,767 chars.
This limitation is also evident in Windows Explorer, cmd.exe and many
other applications (including IDEs).
Particularly annoying is that most Windows APIs return bogus error codes
if a relative path only barely exceeds MAX_PATH in conjunction with the
current directory, e.g. ERROR_PATH_NOT_FOUND / ENOENT instead of the
infinitely more helpful ERROR_FILENAME_EXCED_RANGE / ENAMETOOLONG.
Many Windows wide char APIs support longer than MAX_PATH paths through the
file namespace prefix ('\\?\' or '\\?\UNC\') followed by an absolute path.
Notable exceptions include functions dealing with executables and the
current directory (CreateProcess, LoadLibrary, Get/SetCurrentDirectory) as
well as the entire shell API (ShellExecute, SHGetSpecialFolderPath...).
Introduce a handle_long_path function to check the length of a specified
path properly (and fail with ENAMETOOLONG), and to optionally expand long
paths using the '\\?\' file namespace prefix. Short paths will not be
modified, so we don't need to worry about device names (NUL, CON, AUX).
Contrary to MSDN docs, the GetFullPathNameW function doesn't seem to be
limited to MAX_PATH (at least not on Win7), so we can use it to do the
heavy lifting of the conversion (translate '/' to '\', eliminate '.' and
'..', and make an absolute path).
Add long path error checking to xutftowcs_path for APIs with hard MAX_PATH
limit.
Add a new MAX_LONG_PATH constant and xutftowcs_long_path function for APIs
that support long paths.
While improved error checking is always active, long paths support must be
explicitly enabled via 'core.longpaths' option. This is to prevent end
users to shoot themselves in the foot by checking out files that Windows
Explorer, cmd/bash or their favorite IDE cannot handle.
Test suite:
Test the case is when the full pathname length of a dir is close
to 260 (MAX_PATH).
Bug report and an original reproducer by Andrey Rogozhnikov:
https://github.com/msysgit/git/pull/122#issuecomment-43604199
[jes: adjusted test number to avoid conflicts]
Thanks-to: Martin W. Kirst <maki@bitkings.de>
Thanks-to: Doug Kelly <dougk.ff7@gmail.com>
Signed-off-by: Karsten Blees <blees@dcon.de>
Original-test-by: Andrey Rogozhnikov <rogozhnikov.andrey@gmail.com>
Signed-off-by: Stepan Kasal <kasal@ucw.cz>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
[jes: adjusted test number to avoid conflicts, fixed non-portable use of
the 'export' statement, fixed broken && chain]
Signed-off-by: Doug Kelly <dougk.ff7@gmail.com>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
If multiple threads access a directory that is not yet in the cache, the
directory will be loaded by each thread. Only one of the results is added
to the cache, all others are leaked. This wastes performance and memory.
On cache miss, add a future object to the cache to indicate that the
directory is currently being loaded. Subsequent threads register themselves
with the future object and wait. When the first thread has loaded the
directory, it replaces the future object with the result and notifies
waiting threads.
Signed-off-by: Karsten Blees <blees@dcon.de>
Checking the work tree status is quite slow on Windows, due to slow lstat
emulation (git calls lstat once for each file in the index). Windows
operating system APIs seem to be much better at scanning the status
of entire directories than checking single files.
Add an lstat implementation that uses a cache for lstat data. Cache misses
read the entire parent directory and add it to the cache. Subsequent lstat
calls for the same directory are served directly from the cache.
Also implement opendir / readdir / closedir so that they create and use
directory listings in the cache.
The cache doesn't track file system changes and doesn't plug into any
modifying file APIs, so it has to be explicitly enabled for git functions
that don't modify the working copy.
Note: in an earlier version of this patch, the cache was always active and
tracked file system changes via ReadDirectoryChangesW. However, this was
much more complex and had negative impact on the performance of modifying
git commands such as 'git checkout'.
Signed-off-by: Karsten Blees <blees@dcon.de>
Add a macro to mark code sections that only read from the file system,
along with a config option and documentation.
This facilitates implementation of relatively simple file system level
caches without the need to synchronize with the file system.
Enable read-only sections for 'git status' and preload_index.
Signed-off-by: Karsten Blees <blees@dcon.de>
Emulating the POSIX lstat API on Windows via GetFileAttributes[Ex] is quite
slow. Windows operating system APIs seem to be much better at scanning the
status of entire directories than checking single files. A caching
implementation may improve performance by bulk-reading entire directories
or reusing data obtained via opendir / readdir.
Make the lstat implementation pluggable so that it can be switched at
runtime, e.g. based on a config option.
Signed-off-by: Karsten Blees <blees@dcon.de>
Emulating the POSIX dirent API on Windows via FindFirstFile/FindNextFile is
pretty staightforward, however, most of the information provided in the
WIN32_FIND_DATA structure is thrown away in the process. A more
sophisticated implementation may cache this data, e.g. for later reuse in
calls to lstat.
Make the dirent implementation pluggable so that it can be switched at
runtime, e.g. based on a config option.
Define a base DIR structure with pointers to readdir/closedir that match
the opendir implementation (i.e. similar to vtable pointers in OOP).
Define readdir/closedir so that they call the function pointers in the DIR
structure. This allows to choose the opendir implementation on a
call-by-call basis.
Move the fixed sized dirent.d_name buffer to the dirent-specific DIR
structure, as d_name may be implementation specific (e.g. a caching
implementation may just set d_name to point into the cache instead of
copying the entire file name string).
Signed-off-by: Karsten Blees <blees@dcon.de>
Git for Windows ships with its own Perl interpreter, and insists on
using it, so it will most likely wreak havoc if PERL5LIB is set before
launching Git.
Let's just unset that environment variables when spawning processes.
To make this feature extensible (and overrideable), there is a new
config setting `core.unsetenvvars` that allows specifying a
comma-separated list of names to unset before spawning processes.
Reported by Gabriel Fuhrmann.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
In the Git for Windows project, we have ample precendent for config
settings that apply to Windows, and to Windows only.
Let's formalize this concept by introducing a platform_core_config()
function that can be #define'd in a platform-specific manner.
This will allow us to contain platform-specific code better, as the
corresponding variables no longer need to be exported so that they can
be defined in environment.c and be set in config.c
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
This is the convention elsewhere (and prepares for the case where we may
need to pass callback data).
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
The test linter code has learned that the end of here-doc mark
"EOF" can be quoted in a double-quote pair, not just in a
single-quote pair.
* es/chain-lint-more:
chainlint: match "quoted" here-doc tags
Portability fix.
* ab/portable-more:
tests: fix non-portable iconv invocation
tests: fix non-portable "${var:-"str"}" construct
tests: fix and add lint for non-portable grep --file
tests: fix version-specific portability issue in Perl JSON
tests: use shorter labels in chainlint.sed for AIX sed
tests: fix comment syntax in chainlint.sed for AIX sed
tests: fix and add lint for non-portable seq
tests: fix and add lint for non-portable head -c N
"git merge-base" in 2.19-rc1 has performance regression when the
(experimental) commit-graph feature is in use, which has been
mitigated.
* ds/commit-graph-lockfile-fix:
commit: don't use generation numbers if not needed
Recent addition of "directory rename" heuristics to the
merge-recursive backend makes the command susceptible to false
positives and false negatives. In the context of "git am -3",
which does not know about surrounding unmodified paths and thus
cannot inform the merge machinery about the full trees involved,
this risk is particularly severe. As such, the heuristic is
disabled for "git am -3" to keep the machinery "more stupid but
predictable".
* en/directory-renames-nothanks:
am: avoid directory rename detection when calling recursive merge machinery
merge-recursive: add ability to turn off directory rename detection
t3401: add another directory rename testcase for rebase and am
Recent "git rebase -i" update started to write bogusly formatted
author-script, with a matching broken reading code. These are
fixed.
* pw/rebase-i-author-script-fix:
sequencer: fix quoting in write_author_script
sequencer: handle errors from read_author_ident()
This trick was performed by rebasing the builtin-stash-rebase branch
thicket via `git rebase -kir v2.19.0-rc1`, replacing all branches that
made it into `pu` by their current versions (and also the builtin-stash
by the newest iteration as of ungps/git), and then calling these
commands:
# save current tip
tip=$(git rev-parse HEAD)
# revert previous merge
git reset --hard git-for-windows/master^0
git revert -n -m 1 HEAD
git commit --squash HEAD -s -m "Let's drop this"
# now perform the 3-way merge with v2.19.0-rc1 as base
git merge-recursive v2.19.0-rc1 -- HEAD $tip
git merge --ff-only \
$(git commit-tree -p HEAD -p $tip -m "Merge" \
$(git write-tree))
git commit -c HEAD^^ --amend -s
The merge-recursive dance is necessary because of the merging-rebases:
the fake merges with which these start are mistaken by `git merge` to
mean that the branches were already merged, when the fake merges undid
the corresponding changes.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Let's drop this branch thicket during the next merging rebase, in
preparation for merging a newer branch thicket with builtin-stash-v8 and
builtin-rebase-i-v7.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
This branch adds back the scripted versions, then adds the option to
use the builtin versions of `stash` and `rebase` by setting
`stash.useBuiltin=true` and `rebase.useBuiltin=true`, respectively,
(the latter already worked for the top-level `git rebase` command and
the `--am` backend, and now it also works for the interactive backend).
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
A couple of fixes that should be squashed during the next merging
rebase of Git for Windows.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
The upcoming Git for Windows v2.19.0 wants to ship with the builtin
versions of stash, rebase and rebase -i. The reason: these are just *so
much faster*: t3400 and t3404 run about 60-70 percent faster, and t3903
even more than 80% faster.
However, these are still all pretty fresh, still being reviewed and
iterated on the Git mailing list.
So let's try to give users a way to test these (or to boldly use them
for their mission-critical tasks, as this here developer plans on
doing), but stay with the safe option by default: use the scripted
versions (which might be slow, but they are well tested).
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
We recently converted the `git stash` command from Unix shell scripts
to builtins.
Just like we have `rebase.useBuiltin` to fall back to the scripted
rebase, to give end users a way out when they discover a bug in the
builtin command, this commit adds support for `stash.useBuiltin`.
This is necessary because Git for Windows wants to ship the builtin
stash earlier than core Git: Git for Windows v2.19.0 will come with
the option of a drastically faster (if a lot less battle-tested)
`git stash`.
As the file name `git-stash` is already in use, let's rename the
scripted backend to `git-legacy-stash`.
To make the test suite pass with `stash.useBuiltin=false`, this commit
also backports rudimentary support for `-q` (but only *just* enough
to appease the test suite), and adds a super-ugly hack to force exit
code 129 for `git stash -h`.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
This simply copies the version as of v2.19.0-rc0 verbatim. As of now,
it is not hooked up.
The next commit will change the builtin `stash` to hand off to the
scripted `git stash` when `stash.useBuiltin=false`.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
We recently converted both the `git rebase` and the `git rebase -i`
command from Unix shell scripts to builtins.
The former has a safety valve allowing to fall back to the scripted
`rebase`, just in case that there is a bug in the builtin `rebase`:
setting the config variable `rebase.useBuiltin` to `false` will
fall back to using the scripted version.
The latter did not have such a safety hatch.
Let's reinstate the scripted interactive rebase backend so that `rebase.useBuiltin=false` will not use the builtin interactive rebase,
just in case that an end user runs into a bug with the builtin version
and needs to get out of the fix really quickly.
This is necessary because Git for Windows wants to ship the builtin
rebase/interactive rebase earlier than core Git: Git for Windows
v2.19.0 will come with the option of a drastically faster (if a lot
less battle-tested) `git rebase`/`git rebase -i`.
As the file name `git-rebase--interactive` is already in use, let's
rename the scripted backend to `git-legacy-rebase--interactive`.
A couple of additional touch-ups are needed (such as teaching the
builtin `rebase--interactive`, which assumed the role of the
`rebase--helper`, to perform the two tricks to skip the unnecessary
picks and to generate a new todo list) to make things work again.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
This patch teaches the builtin rebase to avoid the scripted --am backend
and call `git format-patch` and `git am` directly.
Meaning: apart from the --merge and the --preserve-merges backends, `git
rebase` is now implemented in pure C, with no need to ask the Unix shell
interpreter for help.
This brings us really close to a fully builtin `git rebase`: the
--preserve-merges mode is about to be deprecated (as soon as the
--rebase-merges mode has proven stable and robust enough), and there are
plans to scrap the `git-rebase--merge` backend in favor of teaching the
interactive rebase enough tricks to run the --merge mode, too.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
This *would* be a fixup commit, except that we want to avoid rewriting
commits that we merged from upstream's `pu` branch. Instead, we want to
send a new iteration, and then re-merge the new iteration once it made
it into the `pu` branch.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
This simply copies the version as of v2.19.0-rc0 verbatim. As of now,
it is not hooked up (because it needs a couple more changes to work);
The next commit will use the scripted interactive rebase backend from
`git rebase` again when `rebase.useBuiltin=false`.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
This final patch flips the switch and makes the builtin rebase the
default. The old, Unix shell scripted version can still be called via
git -c rebase.useBuiltin=false rebase [...]
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
This branch first merges the builtin interactive rebase, and then
teaches the builtin rebase to hand off interactive rebases to the
builtin backend correctly.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
This fifth batch of builtin rebase patches concludes the conversion: the
builtin rebase is now feature-complete.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
This wave of built rebase patches implements the remaining rebase
options in the builtin rebase.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
This set of patches implements the actions (such as --continue, --skip,
etc) in the builtin rebase.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
This is the first batch of the patches that turn `git rebase` into
a builtin.
This not only helps performance on Windows, but *especially* makes
things more robust, as no MSYS2 Bash will be required to run this
command any longer.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
This merges the builtin stash.
Upstream Git did not integrate it into any stable integration branch
yet, but the performance improvements are substantial enough,
especially on Windows, that we really, really, really want to have it
early.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
While the scripted `git rebase` still has to rely on the
`git-rebase--am.sh` script to implement the glue between the `rebase`
and the `am` commands, we can go a more direct route in the builtin
rebase and avoid using a shell script altogether.
This reduces the chances of Git for Windows running into trouble due to
problems with the POSIX emulation layer (known as "MSYS2 runtime",
itself a derivative of the Cygwin runtime): when no shell script is
called, the POSIX emulation layer is avoided altogether.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
OLD_ICONV has long been needed by FreeBSD so config.mak.uname defines
it unconditionally. However, recent versions do not need it, and its
presence results in compilation warnings. Resolve this issue by defining
OLD_ICONV only for older FreeBSD versions.
Specifically, revision r281550[1], which is part of FreeBSD 11, removed
the need for OLD_ICONV, and r282275[2] back-ported that change to 10.2.
Versions prior to 10.2 do need it.
[1] b0813ee288
[2] b709ec868a
[es: commit message; tweak version check to distinguish 10.x versions]
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Eric Sunshine <sunshine@sunshineco.com>
Reviewed-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This commits introduces a optimization by avoiding calling the
same functions again. For example, `git stash push -u`
would call at some points the following functions:
* `check_changes()` (inside `do_push_stash()`)
* `do_create_stash()`, which calls: `check_changes()` and
`get_untracked_files()`
Note that `check_changes()` also calls `get_untracked_files()`.
So, `check_changes()` is called 2 times and `get_untracked_files()`
3 times.
`get_untracked_files()` has now only two parameters and it will
fill a global strbuf called `untracked_files`.
The old function `check_changes()` now consists of two functions:
`get_untracked_files()` and `check_changes_tracked_files()`.
These are the call chains for `push` and `create`:
* `push_stash()` -> `do_push_stash()` -> `do_create_stash()`
* `create_stash()` -> `do_create_stash()`
To prevent calling the same functions over and over again,
`check_changes()` inside `do_create_stash()` is now placed
in the caller functions (`create_stash()` and `do_push_stash()`).
This way `check_changes()` and `get_untracked files()` are called
only one time.
https://public-inbox.org/git/20180818223329.GJ11326@hank.intra.tgummerer.com/
Signed-off-by: Paul-Sebastian Ungureanu <ungureanupaulsebastian@gmail.com>
