Commit 777b420347 (dir: synchronize treat_leading_path() and
read_directory_recursive(), 2019-12-19) tried to add two warning
comments in those functions, pointing at each other. But the one in
treat_leading_path() just points at itself.
Let's fix that. Since the comment also redirects the reader for more
details to "the commit that added this warning", and since we're now
modifying the warning (creating a new commit without those details),
let's mention the actual commit id.
Signed-off-by: Jeff King <peff@peff.net>
Reviewed-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Restructure the code slightly to avoid passing around a struct dirent
anywhere, which also enables us to avoid trying to manufacture one.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
I was going to title this "dir: more synchronizing of
treat_leading_path() and read_directory_recursive()", a nod to commit
777b420347 ("dir: synchronize treat_leading_path() and
read_directory_recursive()", 2019-12-19), but the title was too long.
Anyway, first the backstory...
fill_directory() has always had a slightly error-prone interface: it
returns a subset of paths which *might* match the specified pathspec; it
was intended to prune away some paths which didn't match the specified
pathspec and keep at least all the ones that did match it. Given this
interface, callers were responsible to post-process the results and
check whether each actually matched the pathspec.
builtin/clean.c did this. It would first prune out duplicates (e.g. if
"dir" was returned as well as all files under "dir/", then it would
simplify this to just "dir"), and after pruning duplicates it would
compare the remaining paths to the specified pathspec(s). This
post-processing itself could run into problems, though, as noted in
commit 404ebceda0 ("dir: also check directories for matching
pathspecs", 2019-09-17):
For the case of git-clean and a set of pathspecs of "dir/file" and
"more", this caused a problem because we'd end up with dir entries
for both of
"dir"
"dir/file"
Then correct_untracked_entries() would try to helpfully prune
duplicates for us by removing "dir/file" since it's under "dir",
leaving us with
"dir"
Since the original pathspec only had "dir/file", the only entry left
doesn't match and leaves nothing to be removed. (Note that if only
one pathspec was specified, e.g. only "dir/file", then the
common_prefix_len optimizations in fill_directory would cause us to
bypass this problem, making it appear in simple tests that we could
correctly remove manually specified pathspecs.)
That commit fixed the issue -- when multiple pathspecs were specified --
by making sure fill_directory() wouldn't return both "dir" and
"dir/file" outside the common_prefix_len optimization path. This is
where it starts to get fun.
In commit b9670c1f5e ("dir: fix checks on common prefix directory",
2019-12-19), we noticed that the common_prefix_len wasn't doing
appropriate checks and letting all kinds of stuff through, resulting in
recursing into .git/ directories and other craziness. So it started
locking down and doing checks on pathnames within that code path. That
continued with commit 777b420347 ("dir: synchronize
treat_leading_path() and read_directory_recursive()", 2019-12-19), which
noted the following:
Our optimization to avoid calling into read_directory_recursive()
when all pathspecs have a common leading directory mean that we need
to match the logic that read_directory_recursive() would use if we
had just called it from the root. Since it does more than call
treat_path() we need to copy that same logic.
...and then it more forcefully addressed the issue with this wonderfully
ironic statement:
Needing to duplicate logic like this means it is guaranteed someone
will eventually need to make further changes and forget to update
both locations. It is tempting to just nuke the leading_directory
special casing to avoid such bugs and simplify the code, but
unpack_trees' verify_clean_subdirectory() also calls
read_directory() and does so with a non-empty leading path, so I'm
hesitant to try to restructure further. Add obnoxious warnings to
treat_leading_path() and read_directory_recursive() to try to warn
people of such problems.
You would think that with such a strongly worded description, that its
author would have actually ensured that the logic in
treat_leading_path() and read_directory_recursive() did actually match
and that *everything* that was needed had at least been copied over at
the time that this paragraph was written. But you'd be wrong, I messed
it up by missing part of the logic.
Copy the missing bits to fix the new final test in t7300.
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Assorted fixes to the directory traversal API.
* en/fill-directory-fixes:
dir.c: use st_add3() for allocation size
dir: consolidate similar code in treat_directory()
dir: synchronize treat_leading_path() and read_directory_recursive()
dir: fix checks on common prefix directory
dir: break part of read_directory_recursive() out for reuse
dir: exit before wildcard fall-through if there is no wildcard
dir: remove stray quote character in comment
Revert "dir.c: make 'git-status --ignored' work within leading directories"
t3011: demonstrate directory traversal failures
Management of sparsely checked-out working tree has gained a
dedicated "sparse-checkout" command.
* ds/sparse-cone: (21 commits)
sparse-checkout: improve OS ls compatibility
sparse-checkout: respect core.ignoreCase in cone mode
sparse-checkout: check for dirty status
sparse-checkout: update working directory in-process for 'init'
sparse-checkout: cone mode should not interact with .gitignore
sparse-checkout: write using lockfile
sparse-checkout: use in-process update for disable subcommand
sparse-checkout: update working directory in-process
sparse-checkout: sanitize for nested folders
unpack-trees: add progress to clear_ce_flags()
unpack-trees: hash less in cone mode
sparse-checkout: init and set in cone mode
sparse-checkout: use hashmaps for cone patterns
sparse-checkout: add 'cone' mode
trace2: add region in clear_ce_flags
sparse-checkout: create 'disable' subcommand
sparse-checkout: add '--stdin' option to set subcommand
sparse-checkout: 'set' subcommand
clone: add --sparse mode
sparse-checkout: create 'init' subcommand
...
When preparing a manufactured dirent instance, we add a length of
path to the size of struct to decide how many bytes to allocate.
Make sure this addition does not wrap-around to cause us
underallocate.
Suggested-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Both the DIR_SKIP_NESTED_GIT and DIR_NO_GITLINKS cases were checking for
whether a path was actually a nonbare repository. That code could be
shared, with just the result of how to act differing between the two
cases.
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Our optimization to avoid calling into read_directory_recursive() when
all pathspecs have a common leading directory mean that we need to match
the logic that read_directory_recursive() would use if we had just
called it from the root. Since it does more than call treat_path() we
need to copy that same logic.
Alternatively, we could try to change treat_path to return path_recurse
for an untracked directory under the given special circumstances that
this logic checks for, but a simple switch results in many test failures
such as 'git clean -d' not wiping out untracked but empty directories.
To work around that, we'd need the caller of treat_path to check for
path_recurse and sometimes special case it into path_untracked. In
other words, we'd still have extra logic in both places.
Needing to duplicate logic like this means it is guaranteed someone will
eventually need to make further changes and forget to update both
locations. It is tempting to just nuke the leading_directory special
casing to avoid such bugs and simplify the code, but unpack_trees'
verify_clean_subdirectory() also calls read_directory() and does so with
a non-empty leading path, so I'm hesitant to try to restructure further.
Add obnoxious warnings to treat_leading_path() and
read_directory_recursive() to try to warn people of such problems.
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Many years ago, the directory traversing logic had an optimization that
would always recurse into any directory that was a common prefix of all
the pathspecs without walking the leading directories to get down to
the desired directory. Thus,
git ls-files -o .git/ # case A
would notice that .git/ was a common prefix of all pathspecs (since
it is the only pathspec listed), and then traverse into it and start
showing unknown files under that directory. Unfortunately, .git/ is not
a directory we should be traversing into, which made this optimization
problematic. This also affected cases like
git ls-files -o --exclude-standard t/ # case B
where t/ was in the .gitignore file and thus isn't interesting and
shouldn't be recursed into. It also affected cases like
git ls-files -o --directory untracked_dir/ # case C
where untracked_dir/ is indeed untracked and thus interesting, but the
--directory flag means we only want to show the directory itself, not
recurse into it and start listing untracked files below it.
The case B class of bugs were noted and fixed in commits 16e2cfa909
("read_directory(): further split treat_path()", 2010-01-08) and
48ffef966c ("ls-files: fix overeager pathspec optimization",
2010-01-08), with the idea being that we first wanted to check whether
the common prefix was interesting. The former patch noted that
treat_path() couldn't be used when checking the common prefix because
treat_path() requires a dir_entry() and we haven't read any directories
at the point we are checking the common prefix. So, that patch split
treat_one_path() out of treat_path(). The latter patch then created a
new treat_leading_path() which duplicated by hand the bits of
treat_path() that couldn't be broken out and then called
treat_one_path() for the remainder. There were three problems with this
approach:
* The duplicated logic in treat_leading_path() accidentally missed the
check for special paths (such as is_dot_or_dotdot and matching
".git"), causing case A types of bugs to continue to be an issue.
* The treat_leading_path() logic assumed we should traverse into
anything where path_treatment was not path_none, i.e. it perpetuated
class C types of bugs.
* It meant we had split logic that needed to kept in sync, running the
risk that people introduced new inconsistencies (such as in commit
be8a84c526, which we reverted earlier in this series, or in commit
df5bcdf83a which we'll fix in a subsequent commit)
Fix most these problems by making treat_leading_path() not only loop
over each leading path component, but calling treat_path() directly on
each. To do so, we have to create a synthetic dir_entry, but that only
takes a few lines. Then, pay attention to the path_treatment result we
get from treat_path() and don't treat path_excluded, path_untracked, and
path_recurse all the same as path_recurse.
This leaves one remaining problem, the new inconsistency from commit
df5bcdf83a. That will be addressed in a subsequent commit.
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* hw/doc-in-header:
trace2: move doc to trace2.h
submodule-config: move doc to submodule-config.h
tree-walk: move doc to tree-walk.h
trace: move doc to trace.h
run-command: move doc to run-command.h
parse-options: add link to doc file in parse-options.h
credential: move doc to credential.h
argv-array: move doc to argv-array.h
cache: move doc to cache.h
sigchain: move doc to sigchain.h
pathspec: move doc to pathspec.h
revision: move doc to revision.h
attr: move doc to attr.h
refs: move doc to refs.h
remote: move doc to remote.h and refspec.h
sha1-array: move doc to sha1-array.h
merge: move doc to ll-merge.h
graph: move doc to graph.h and graph.c
dir: move doc to dir.h
diff: move doc to diff.h and diffcore.h
When a user uses the sparse-checkout feature in cone mode, they
add patterns using "git sparse-checkout set <dir1> <dir2> ..."
or by using "--stdin" to provide the directories line-by-line over
stdin. This behaviour naturally looks a lot like the way a user
would type "git add <dir1> <dir2> ..."
If core.ignoreCase is enabled, then "git add" will match the input
using a case-insensitive match. Do the same for the sparse-checkout
feature.
Perform case-insensitive checks while updating the skip-worktree
bits during unpack_trees(). This is done by changing the hash
algorithm and hashmap comparison methods to optionally use case-
insensitive methods.
When this is enabled, there is a small performance cost in the
hashing algorithm. To tease out the worst possible case, the
following was run on a repo with a deep directory structure:
git ls-tree -d -r --name-only HEAD |
git sparse-checkout set --stdin
The 'set' command was timed with core.ignoreCase disabled or
enabled. For the repo with a deep history, the numbers were
core.ignoreCase=false: 62s
core.ignoreCase=true: 74s (+19.3%)
For reproducibility, the equivalent test on the Linux kernel
repository had these numbers:
core.ignoreCase=false: 3.1s
core.ignoreCase=true: 3.6s (+16%)
Now, this is not an entirely fair comparison, as most users
will define their sparse cone using more shallow directories,
and the performance improvement from eb42feca97 ("unpack-trees:
hash less in cone mode" 2019-11-21) can remove most of the
hash cost. For a more realistic test, drop the "-r" from the
ls-tree command to store only the first-level directories.
In that case, the Linux kernel repository takes 0.2-0.25s in
each case, and the deep repository takes one second, plus or
minus 0.05s, in each case.
Thus, we _can_ demonstrate a cost to this change, but it is
unlikely to matter to any reasonable sparse-checkout cone.
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Create an add_path_to_appropriate_result_list() function from the code
at the end of read_directory_recursive() so we can use it elsewhere.
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The DO_MATCH_LEADING_PATHSPEC had a fall-through case for if there was a
wildcard, noting that we don't yet have enough information to determine
if a further paths under the current directory might match due to the
presence of wildcards. But if we have no wildcards in our pathspec,
then we shouldn't get to that fall-through case.
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Commit be8a84c526 ("dir.c: make 'git-status --ignored' work within
leading directories", 2013-04-15) noted that
git status --ignored <SOMEPATH>
would not list ignored files and directories within <SOMEPATH> if
<SOMEPATH> was untracked, and modified the behavior to make it show
them. However, it did so via a hack that broke consistency; it would
show paths under <SOMEPATH> differently than a simple
git status --ignored | grep <SOMEPATH>
would show them. A correct fix is slightly more involved, and
complicated slightly by this hack, so we revert this commit (but keep
corrected versions of the testcases) and will later fix the original
bug with a subsequent patch.
Some history may be helpful:
A very, very similar case to the commit we are reverting was raised in
commit 48ffef966c ("ls-files: fix overeager pathspec optimization",
2010-01-08); but it actually went in somewhat the opposite direction. In
that commit, it mentioned how
git ls-files -o --exclude-standard t/
used to show untracked files under t/ even when t/ was ignored, and then
changed the behavior to stop showing untracked files under an ignored
directory. More importantly, this commit considered keeping this
behavior but noted that it would be inconsistent with the behavior when
multiple pathspecs were specified and thus rejected it.
The reason for this whole inconsistency when one pathspec is specified
versus zero or two is because common prefixes of pathspecs are sent
through a different set of checks (in treat_leading_path()) than normal
file/directory traversal (those go through read_directory_recursive()
and treat_path()). As such, for consistency, one needs to check that
both codepaths produce the same result.
Revert commit be8a84c526, except instead
of removing the testcase it added, modify it to check for correct and
consistent behavior. A subsequent patch in this series will fix the
testcase.
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The sparse-checkout feature in "cone mode" can use the fact that
the recursive patterns are "connected" to the root via parent
patterns to decide if a directory is entirely contained in the
sparse-checkout or entirely removed.
In these cases, we can skip hashing the paths within those
directories and simply set the skipworktree bit to the correct
value.
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
To make the cone pattern set easy to use, update the behavior of
'git sparse-checkout (init|set)'.
Add '--cone' flag to 'git sparse-checkout init' to set the config
option 'core.sparseCheckoutCone=true'.
When running 'git sparse-checkout set' in cone mode, a user only
needs to supply a list of recursive folder matches. Git will
automatically add the necessary parent matches for the leading
directories.
When testing 'git sparse-checkout set' in cone mode, check the
error stream to ensure we do not see any errors. Specifically,
we want to avoid the warning that the patterns do not match
the cone-mode patterns.
Helped-by: Eric Wong <e@80x24.org>
Helped-by: Johannes Schindelin <Johannes.Schindelin@gmx.de>
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The parent and recursive patterns allowed by the "cone mode"
option in sparse-checkout are restrictive enough that we
can avoid using the regex parsing. Everything is based on
prefix matches, so we can use hashsets to store the prefixes
from the sparse-checkout file. When checking a path, we can
strip path entries from the path and check the hashset for
an exact match.
As a test, I created a cone-mode sparse-checkout file for the
Linux repository that actually includes every file. This was
constructed by taking every folder in the Linux repo and creating
the pattern pairs here:
/$folder/
!/$folder/*/
This resulted in a sparse-checkout file sith 8,296 patterns.
Running 'git read-tree -mu HEAD' on this file had the following
performance:
core.sparseCheckout=false: 0.21 s (0.00 s)
core.sparseCheckout=true: 3.75 s (3.50 s)
core.sparseCheckoutCone=true: 0.23 s (0.01 s)
The times in parentheses above correspond to the time spent
in the first clear_ce_flags() call, according to the trace2
performance traces.
While this example is contrived, it demonstrates how these
patterns can slow the sparse-checkout feature.
Helped-by: Eric Wong <e@80x24.org>
Helped-by: Johannes Schindelin <Johannes.Schindelin@gmx.de>
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Move the documentation from Documentation/technical/api-directory-listing.txt
to dir.h as it's easier for the developers to find the usage information
beside the code instead of looking for it in another doc file.
Also documentation/technical/api-directory-listing.txt is removed because
the information it has is now redundant and it'll be hard to keep it up to
date and synchronized with the documentation in the header files.
Signed-off-by: Heba Waly <heba.waly@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git clean" fixes.
* en/clean-nested-with-ignored:
dir: special case check for the possibility that pathspec is NULL
clean: fix theoretical path corruption
clean: rewrap overly long line
clean: avoid removing untracked files in a nested git repository
clean: disambiguate the definition of -d
git-clean.txt: do not claim we will delete files with -n/--dry-run
dir: add commentary explaining match_pathspec_item's return value
dir: if our pathspec might match files under a dir, recurse into it
dir: make the DO_MATCH_SUBMODULE code reusable for a non-submodule case
dir: also check directories for matching pathspecs
dir: fix off-by-one error in match_pathspec_item
dir: fix typo in comment
t7300: add testcases showing failure to clean specified pathspecs
Commits 404ebceda0 ("dir: also check directories for matching
pathspecs", 2019-09-17) and 89a1f4aaf7 ("dir: if our pathspec might
match files under a dir, recurse into it", 2019-09-17) added calls to
match_pathspec() and do_match_pathspec() passing along their pathspec
parameter. Both match_pathspec() and do_match_pathspec() assume the
pathspec argument they are given is non-NULL. It turns out that
unpack-tree.c's verify_clean_subdirectory() calls read_directory() with
pathspec == NULL, and it is possible on case insensitive filesystems for
that NULL to make it to these new calls to match_pathspec() and
do_match_pathspec(). Add appropriate checks on the NULLness of pathspec
to avoid a segfault.
In case the negation throws anyone off (one of the calls was to
do_match_pathspec() while the other was to !match_pathspec(), yet no
negation of the NULLness of pathspec is used), there are two ways to
understand the differences:
* The code already handled the pathspec == NULL cases before this
series, and this series only tried to change behavior when there was
a pathspec, thus we only want to go into the if-block if pathspec is
non-NULL.
* One of the calls is for whether to recurse into a subdirectory, the
other is for after we've recursed into it for whether we want to
remove the subdirectory itself (i.e. the subdirectory didn't match
but something under it could have). That difference in situation
leads to the slight differences in logic used (well, that and the
slightly unusual fact that we don't want empty pathspecs to remove
untracked directories by default).
Denton found and analyzed one issue and provided the patch for the
match_pathspec() call, SZEDER figured out why the issue only reproduced
for some folks and not others and provided the testcase, and I looked
through the remainder of the series and noted the do_match_pathspec()
call that should have the same check.
Co-authored-by: Denton Liu <liu.denton@gmail.com>
Co-authored-by: SZEDER Gábor <szeder.dev@gmail.com>
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The internal code originally invented for ".gitignore" processing
got reshuffled and renamed to make it less tied to "excluding" and
stress more that it is about "matching", as it has been reused for
things like sparse checkout specification that want to check if a
path is "included".
* ds/include-exclude:
unpack-trees: rename 'is_excluded_from_list()'
treewide: rename 'exclude' methods to 'pattern'
treewide: rename 'EXCL_FLAG_' to 'PATTERN_FLAG_'
treewide: rename 'struct exclude_list' to 'struct pattern_list'
treewide: rename 'struct exclude' to 'struct path_pattern'
Users expect files in a nested git repository to be left alone unless
sufficiently forced (with two -f's). Unfortunately, in certain
circumstances, git would delete both tracked (and possibly dirty) files
and untracked files within a nested repository. To explain how this
happens, let's contrast a couple cases. First, take the following
example setup (which assumes we are already within a git repo):
git init nested
cd nested
>tracked
git add tracked
git commit -m init
>untracked
cd ..
In this setup, everything works as expected; running 'git clean -fd'
will result in fill_directory() returning the following paths:
nested/
nested/tracked
nested/untracked
and then correct_untracked_entries() would notice this can be compressed
to
nested/
and then since "nested/" is a directory, we would call
remove_dirs("nested/", ...), which would
check is_nonbare_repository_dir() and then decide to skip it.
However, if someone also creates an ignored file:
>nested/ignored
then running 'git clean -fd' would result in fill_directory() returning
the same paths:
nested/
nested/tracked
nested/untracked
but correct_untracked_entries() will notice that we had ignored entries
under nested/ and thus simplify this list to
nested/tracked
nested/untracked
Since these are not directories, we do not call remove_dirs() which was
the only place that had the is_nonbare_repository_dir() safety check --
resulting in us deleting both the untracked file and the tracked (and
possibly dirty) file.
One possible fix for this issue would be walking the parent directories
of each path and checking if they represent nonbare repositories, but
that would be wasteful. Even if we added caching of some sort, it's
still a waste because we should have been able to check that "nested/"
represented a nonbare repository before even descending into it in the
first place. Add a DIR_SKIP_NESTED_GIT flag to dir_struct.flags and use
it to prevent fill_directory() and friends from descending into nested
git repos.
With this change, we also modify two regression tests added in commit
91479b9c72 ("t7300: add tests to document behavior of clean and nested
git", 2015-06-15). That commit, nor its series, nor the six previous
iterations of that series on the mailing list discussed why those tests
coded the expectation they did. In fact, it appears their purpose was
simply to test _existing_ behavior to make sure that the performance
changes didn't change the behavior. However, these two tests directly
contradicted the manpage's claims that two -f's were required to delete
files/directories under a nested git repository. While one could argue
that the user gave an explicit path which matched files/directories that
were within a nested repository, there's a slippery slope that becomes
very difficult for users to understand once you go down that route (e.g.
what if they specified "git clean -f -d '*.c'"?) It would also be hard
to explain what the exact behavior was; avoid such problems by making it
really simple.
Also, clean up some grammar errors describing this functionality in the
git-clean manpage.
Finally, there are still a couple bugs with -ffd not cleaning out enough
(e.g. missing the nested .git) and with -ffdX possibly cleaning out the
wrong files (paying attention to outer .gitignore instead of inner).
This patch does not address these cases at all (and does not change the
behavior relative to those flags), it only fixes the handling when given
a single -f. See
https://public-inbox.org/git/20190905212043.GC32087@szeder.dev/ for more
discussion of the -ffd[X?] bugs.
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The way match_pathspec_item() handles names and pathspecs with trailing
slash characters, in conjunction with special options like
DO_MATCH_DIRECTORY and DO_MATCH_LEADING_PATHSPEC were non-obvious, and
broken until this patch series. Add a table in a comment explaining the
intent of how these work.
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
For git clean, if a directory is entirely untracked and the user did not
specify -d (corresponding to DIR_SHOW_IGNORED_TOO), then we usually do
not want to remove that directory and thus do not recurse into it.
However, if the user manually specified specific (or even globbed) paths
somewhere under that directory to remove, then we need to recurse into
the directory to make sure we remove the relevant paths under that
directory as the user requested.
Note that this does not mean that the recursed-into directory will be
added to dir->entries for later removal; as of a few commits earlier in
this series, there is another more strict match check that is run after
returning from a recursed-into directory before deciding to add it to the
list of entries. Therefore, this will only result in files underneath
the given directory which match one of the pathspecs being added to the
entries list.
Two notes of potential interest to future readers:
* If we wanted to only recurse into a directory when it is specifically
matched rather than matched-via-glob (e.g. '*.c'), then we could do
so via making the final non-zero return in match_pathspec_item be
MATCHED_RECURSIVELY instead of MATCHED_RECURSIVELY_LEADING_PATHSPEC.
(Note that the relative order of MATCHED_RECURSIVELY_LEADING_PATHSPEC
and MATCHED_RECURSIVELY are important for such a change.) I was
leaving open that possibility while writing an RFC asking for the
behavior we want, but even though we don't want it, that knowledge
might help you understand the code flow better.
* There is a growing amount of logic in read_directory_recursive() for
deciding whether to recurse into a subdirectory. However, there is a
comment immediately preceding this logic that says to recurse if
instructed by treat_path(). It may be better for the logic in
read_directory_recursive() to ultimately be moved to treat_path() (or
another function it calls, such as treat_directory()), but I have
left that for someone else to tackle in the future.
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The specific checks done in match_pathspec_item for the DO_MATCH_SUBMODULE
case are useful for other cases which have nothing to do with submodules.
Rename this constant; a subsequent commit will make use of this change.
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Even if a directory doesn't match a pathspec, it is possible, depending
on the precise pathspecs, that some file underneath it might. So we
special case and recurse into the directory for such situations. However,
we previously always added any untracked directory that we recursed into
to the list of untracked paths, regardless of whether the directory
itself matched the pathspec.
For the case of git-clean and a set of pathspecs of "dir/file" and "more",
this caused a problem because we'd end up with dir entries for both of
"dir"
"dir/file"
Then correct_untracked_entries() would try to helpfully prune duplicates
for us by removing "dir/file" since it's under "dir", leaving us with
"dir"
Since the original pathspec only had "dir/file", the only entry left
doesn't match and leaves nothing to be removed. (Note that if only one
pathspec was specified, e.g. only "dir/file", then the common_prefix_len
optimizations in fill_directory would cause us to bypass this problem,
making it appear in simple tests that we could correctly remove manually
specified pathspecs.)
Fix this by actually checking whether the directory we are about to add
to the list of dir entries actually matches the pathspec; only do this
matching check after we have already returned from recursing into the
directory.
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
For a pathspec like 'foo/bar' comparing against a path named "foo/",
namelen will be 4, and match[namelen] will be 'b'. The correct location
of the directory separator is namelen-1.
However, other callers of match_pathspec_item() such as builtin/grep.c's
submodule_path_match() will compare against a path named "foo" instead of
"foo/". It might be better to change all the callers to be consistent,
as discussed at
https://public-inbox.org/git/xmqq7e6cdnkr.fsf@gitster-ct.c.googlers.com/
and
https://public-inbox.org/git/CABPp-BERWUPCPq-9fVW1LNocqkrfsoF4BPj3gJd9+En43vEkTQ@mail.gmail.com/
but there are many cases to audit, so for now just make sure we handle
both cases with and without a trailing slash.
The reason the code worked despite this sometimes-off-by-one error was
that the subsequent code immediately checked whether the first matchlen
characters matched (which they do) and then bailed and return
MATCHED_RECURSIVELY anyway since wildmatch doesn't have the ability to
check if "name" can be matched as a directory (or prefix) against the
pathspec.
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The first consumer of pattern-matching filenames was the
.gitignore feature. In that context, storing a list of patterns
as a 'struct exclude_list' makes sense. However, the
sparse-checkout feature then adopted these structures and methods,
but with the opposite meaning: these patterns match the files
that should be included!
Now that this library is renamed to use 'struct pattern_list'
and 'struct pattern', we can now rename the method used by
the sparse-checkout feature to determine which paths should
appear in the working directory.
The method is_excluded_from_list() is only used by the
sparse-checkout logic in unpack-trees and list-objects-filter.
The confusing part is that it returned 1 for "excluded" (i.e.
it matches the list of exclusions) but that really manes that
the path matched the list of patterns for _inclusion_ in the
working directory.
Rename the method to be path_matches_pattern_list() and have
it return an explicit 'enum pattern_match_result'. Here, the
values MATCHED = 1, UNMATCHED = 0, and UNDECIDED = -1 agree
with the previous integer values. This shift allows future
consumers to better understand what the retur values mean,
and provides more type checking for handling those values.
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The first consumer of pattern-matching filenames was the
.gitignore feature. In that context, storing a list of patterns
as a 'struct exclude_list' makes sense. However, the
sparse-checkout feature then adopted these structures and methods,
but with the opposite meaning: these patterns match the files
that should be included!
It would be clearer to rename this entire library as a "pattern
matching" library, and the callers apply exclusion/inclusion
logic accordingly based on their needs.
This commit renames several methods defined in dir.h to make
more sense with the renamed 'struct exclude_list' to 'struct
pattern_list' and 'struct exclude' to 'struct path_pattern':
* last_exclude_matching() -> last_matching_pattern()
* parse_exclude() -> parse_path_pattern()
In addition, the word 'exclude' was replaced with 'pattern'
in the methods below:
* add_exclude_list()
* add_excludes_from_file_to_list()
* add_excludes_from_file()
* add_excludes_from_blob_to_list()
* add_exclude()
* clear_exclude_list()
A few methods with the word "exclude" remain. These will
be handled seperately. In particular, the method
"is_excluded()" is concretely about the .gitignore file
relative to a specific directory. This is the important
boundary between library and consumer: is_excluded() cares
about .gitignore, but is_excluded() calls
last_matching_pattern() to make that decision.
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The first consumer of pattern-matching filenames was the
.gitignore feature. In that context, storing a list of patterns
as a 'struct exclude_list' makes sense. However, the
sparse-checkout feature then adopted these structures and methods,
but with the opposite meaning: these patterns match the files
that should be included!
It would be clearer to rename this entire library as a "pattern
matching" library, and the callers apply exclusion/inclusion
logic accordingly based on their needs.
This commit replaces 'EXCL_FLAG_' to 'PATTERN_FLAG_' in the
names of the flags used on 'struct path_pattern'.
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The first consumer of pattern-matching filenames was the
.gitignore feature. In that context, storing a list of patterns
as a 'struct exclude_list' makes sense. However, the
sparse-checkout feature then adopted these structures and methods,
but with the opposite meaning: these patterns match the files
that should be included!
It would be clearer to rename this entire library as a "pattern
matching" library, and the callers apply exclusion/inclusion
logic accordingly based on their needs.
This commit renames 'struct exclude_list' to 'struct pattern_list'
and renames several variables called 'el' to 'pl'.
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The first consumer of pattern-matching filenames was the
.gitignore feature. In that context, storing a list of patterns
as a list of 'struct exclude' items makes sense. However, the
sparse-checkout feature then adopted these structures and methods,
but with the opposite meaning: these patterns match the files
that should be included!
It would be clearer to rename this entire library as a "pattern
matching" library, and the callers apply exclusion/inclusion
logic accordingly based on their needs.
This commit renames 'struct exclude' to 'struct path_pattern'
and renames several variable names to match. 'struct pattern'
was already taken by attr.c, and this more completely describes
that the patterns are specific to file paths.
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Calculating the sum of two array indexes to find the midpoint between
them can overflow, i.e. code like this is unsafe for big arrays:
mid = (first + last) >> 1;
Make sure the intermediate value stays within the boundaries instead,
like this:
mid = first + ((last - first) >> 1);
The loop condition of the binary search makes sure that 'last' is
always greater than 'first', so this is safe as long as 'first' is
not negative. And that can be verified easily using the pre-context
of each change, except for name-hash.c, so add an assertion to that
effect there.
The unsafe calculations were found with:
git grep '(.*+.*) *>> *1'
This is a continuation of 19716b21a4 (cleanup: fix possible overflow
errors in binary search, 2017-10-08).
Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Code clean-up.
* jk/untracked-cache-more-fixes:
untracked-cache: simplify parsing by dropping "len"
untracked-cache: simplify parsing by dropping "next"
untracked-cache: be defensive about missing NULs in index
Further code clean-up to allow the lowest level of name-to-object
mapping layer to work with a passed-in repository other than the
default one.
* nd/sha1-name-c-wo-the-repository: (34 commits)
sha1-name.c: remove the_repo from get_oid_mb()
sha1-name.c: remove the_repo from other get_oid_*
sha1-name.c: remove the_repo from maybe_die_on_misspelt_object_name
submodule-config.c: use repo_get_oid for reading .gitmodules
sha1-name.c: add repo_get_oid()
sha1-name.c: remove the_repo from get_oid_with_context_1()
sha1-name.c: remove the_repo from resolve_relative_path()
sha1-name.c: remove the_repo from diagnose_invalid_index_path()
sha1-name.c: remove the_repo from handle_one_ref()
sha1-name.c: remove the_repo from get_oid_1()
sha1-name.c: remove the_repo from get_oid_basic()
sha1-name.c: remove the_repo from get_describe_name()
sha1-name.c: remove the_repo from get_oid_oneline()
sha1-name.c: add repo_interpret_branch_name()
sha1-name.c: remove the_repo from interpret_branch_mark()
sha1-name.c: remove the_repo from interpret_nth_prior_checkout()
sha1-name.c: remove the_repo from get_short_oid()
sha1-name.c: add repo_for_each_abbrev()
sha1-name.c: store and use repo in struct disambiguate_state
sha1-name.c: add repo_find_unique_abbrev_r()
...
Running "git add" on a repository created inside the current
repository is an explicit indication that the user wants to add it
as a submodule, but when the HEAD of the inner repository is on an
unborn branch, it cannot be added as a submodule. Worse, the files
in its working tree can be added as if they are a part of the outer
repository, which is not what the user wants. These problems are
being addressed.
* km/empty-repo-is-still-a-repo:
add: error appropriately on repository with no commits
dir: do not traverse repositories with no commits
submodule: refuse to add repository with no commits
An underallocation in the code to read the untracked cache
extension has been corrected.
* js/untracked-cache-allocfix:
untracked cache: fix off-by-one
Conversion from unsigned char[20] to struct object_id continues.
* bc/hash-transition-16: (35 commits)
gitweb: make hash size independent
Git.pm: make hash size independent
read-cache: read data in a hash-independent way
dir: make untracked cache extension hash size independent
builtin/difftool: use parse_oid_hex
refspec: make hash size independent
archive: convert struct archiver_args to object_id
builtin/get-tar-commit-id: make hash size independent
get-tar-commit-id: parse comment record
hash: add a function to lookup hash algorithm by length
remote-curl: make hash size independent
http: replace sha1_to_hex
http: compute hash of downloaded objects using the_hash_algo
http: replace hard-coded constant with the_hash_algo
http-walker: replace sha1_to_hex
http-push: remove remaining uses of sha1_to_hex
http-backend: allow 64-character hex names
http-push: convert to use the_hash_algo
builtin/pull: make hash-size independent
builtin/am: make hash size independent
...
The code which parses untracked-cache extensions from disk keeps a "len"
variable, which is the size of the string we are parsing. But since we
now have an "end of string" variable, we can just use that to get the
length when we need it. This eliminates the need to keep "len" up to
date (and removes the possibility of any errors where "len" and "eos"
get out of sync).
As a bonus, it means we are not storing a string length in an "int",
which is a potential source of overflows (though in this case it seems
fairly unlikely for that to cause any memory problems).
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When we parse an on-disk untracked cache, we have two pointers, "data"
and "next". As we parse, we point "next" to the end of an element, and
then later update "data" to match.
But we actually don't need two pointers. Each parsing step can just
update "data" directly from other variables we hold (and we don't have
to worry about bailing in an intermediate state, since any parsing
failure causes us to immediately discard "data" and return).
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The on-disk format for the untracked-cache extension contains
NUL-terminated filenames. We parse these from the mmap'd file using
string functions like strlen(). This works fine in the normal case, but
if we see a malformed or corrupted index, we might read off the end of
our mmap.
Instead, let's use memchr() to find the trailing NUL within the bytes we
know are available, and return an error if it's missing.
Note that we can further simplify by folding another range check into
our conditional. After we find the end of the string, we set "next" to
the byte after the string and treat it as an error if there are no such
bytes left. That saves us from having to do a range check at the
beginning of each subsequent string (and works because there is always
data after each string). We can do both range checks together by
checking "!eos" (we didn't find a NUL) and "eos == end" (it was on the
last available byte, meaning there's nothing after). This replaces the
existing "next > end" checks.
Note also that the decode_varint() calls have a similar problem (we
don't even pass them "end"; they just keep parsing). These are probably
OK in practice since varints have a finite length (we stop parsing when
we'd overflow a uintmax_t), so the worst case is that we'd overflow into
reading the trailing bytes of the index.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In f9e6c64958 (untracked cache: load from UNTR index extension,
2015-03-08), code was added to read back the untracked cache from an
index extension.
Probably in the endeavor to avoid the `calloc()` implied by
`FLEX_ALLOC_STR()` (it is hard to know why exactly, the commit message
of that commit is a bit parsimonious with information), it calls
`malloc()` manually and then `memcpy()`s the bits and pieces into place.
It allocates the size of `struct untracked_cache_dir` plus the string
length of the untracked file name, then copies the information in two
steps: first the fixed-size metadata, then the name. And here lies the
rub: it includes the trailing NUL byte in the name.
If `FLEX_ARRAY` is defined as 0, this results in a buffer overrun.
To fix this, let's just add 1, for the trailing NUL byte. Technically,
this overallocates on platforms where `FLEX_ARRAY` is 1, but it should
not matter much in reality, as `malloc()` usually overallocates anyway,
unless the size to allocate aligns exactly with some internal chunk size
(see below for more on that).
The real strange thing is that neither valgrind nor DrMemory catches
this bug. In this developer's tests, a `memcpy()` (but not a
`memset()`!) could write up to 4 bytes after the allocated memory range
before valgrind would start reporting an issue.
However, when running Git built with nedmalloc as allocator, under rare
conditions (and inconsistently at that), this bug triggered an `abort()`
because nedmalloc rounds up the size to be `malloc()`ed to a multiple of
a certain chunk size, then adds a few bytes to be used for storing some
internal state. If there is no rounding up to do (because the size is
already a multiple of that chunk size), and if the buffer is overrun as
in the code patched in this commit, the internal state is corrupted.
The scenario that triggered this here bug fix entailed a git.git
checkout with an extra copy of the source code in an untracked
subdirectory, meaning that there was an untracked subdirectory called
"thunderbird-patch-inline" whose name's length is exactly 24 bytes,
which, added to the size of above-mentioned `struct untracked_cache_dir`
that weighs in with 104 bytes on a 64-bit system, amounts to 128,
aligning perfectly with nedmalloc's chunk size.
As there is no obvious way to trigger this bug reliably, on all
platforms supported by Git, and as the bug is obvious enough, this patch
comes without a regression test.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When treat_directory() encounters a directory that is not in the index
and DIR_NO_GITLINKS is unset, it calls resolve_gitlink_ref() to decide
if a directory looks like a repository, in which case the directory
won't be traversed. As a result, 'status -uall' and 'ls-files -o'
will show only the directory, even when there are untracked files
within the directory.
For the unusual case where a repository doesn't have a commit checked
out, resolve_gitlink_ref() returns -1 because HEAD cannot be resolved,
and the directory is treated as a normal directory (i.e. traversal
does not stop at the repository boundary). The status and ls-files
commands above list untracked files within the repository rather than
showing only the top-level directory. And if 'git add' is called on a
repository with no commit checked out, any untracked files under the
repository are added as blobs in the top-level project, a behavior
that is unlikely to be what the caller intended.
The above case is a corner case in an already unusual situation of the
working tree containing a repository that is not a tracked submodule,
but we might as well treat anything that looks like a repository
consistently. Loosen the "looks like a repository" criteria in
treat_directory() by replacing resolve_gitlink_ref() with
is_nonbare_repository_dir(), one of the checks that is performed
downstream when resolve_gitlink_ref() is called.
As the required update to t3700-add shows, calling 'git add' on a
repository with no commit checked out will now raise an error. While
this is the desired behavior, note that the output isn't yet
appropriate. The next commit will improve this output.
Signed-off-by: Kyle Meyer <kyle@kyleam.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Instead of using a struct with a flex array member to read and write the
untracked cache extension, use a shorter, fixed-length struct and add
the name and hash data explicitly.
Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This hasn't been used since 17ddc66e70 (convert report_path_error to
take struct pathspec, 2013-07-14), as the names in the struct will have
already been prefixed when they were parsed.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The assumption to work on the single "in-core index" instance has
been reduced from the library-ish part of the codebase.
* nd/the-index-final:
cache.h: flip NO_THE_INDEX_COMPATIBILITY_MACROS switch
read-cache.c: remove the_* from index_has_changes()
merge-recursive.c: remove implicit dependency on the_repository
merge-recursive.c: remove implicit dependency on the_index
sha1-name.c: remove implicit dependency on the_index
read-cache.c: replace update_index_if_able with repo_&
read-cache.c: kill read_index()
checkout: avoid the_index when possible
repository.c: replace hold_locked_index() with repo_hold_locked_index()
notes-utils.c: remove the_repository references
grep: use grep_opt->repo instead of explict repo argument
Jeff King
Elijah Newren
Junio C Hamano
Derrick Stolee
Heba Waly
René Scharfe
Nguyễn Thái Ngọc Duy
Johannes Schindelin
Kyle Meyer
brian m. carlson