|
270fd807
|
2019-07-18T13:44:10
|
|
azure: compile one Windows platform with the WinHTTP SHA1 backend
We currently have no job that compiles libgit2 with the WinHTTP backend
for SHA1. Due to this, a compile error has been introduced and not
noticed for several months. Change the x86 MSVC job to use the HTTPS
backend for SHA1. The x86 job was chosen with no particular reason.
|
|
|
7574564e
|
2019-07-18T13:40:34
|
|
sha1: win32: fix compilation due to unknown type
In commit bbf034ab9 (hash: move `git_hash_prov` into Win32 backend,
2019-02-22), the `git_hash_prov`'s structure name has been removed in
favour of its typedef'ed name. But as we have no CI that compiles with
the WinHTTPS hashing backend right now, it wasn't noticed that the
implementation that uses this struct wasn't changed correctly.
Fix the struct type to make it compile again.
|
|
|
b7c247b3
|
2019-07-18T13:37:02
|
|
cmake: include SHA1 headers into our source files
When selecting the SHA1 backend, we only include the respective C
implementation of the selected backend. But since commit bd48bf3fb
(hash: introduce source files to break include circles, 2019-06-14), we
have introduced separate headers and compilation units for all hashes.
So by not including the headers, we may not honor them to compute
whether a file needs to be recompiled and they also will not be
displayed in IDEs.
Add the header files to fix this problem.
|
|
|
368b9795
|
2019-07-18T11:27:21
|
|
Merge pull request #5168 from tiennou/clar/fix-data-suite-count
clar: correctly account for "data" suites when counting
|
|
51124a5b
|
2019-07-17T17:33:34
|
|
Merge pull request #5170 from bk2204/packbuilder-efficient-realloc
Allocate memory more efficiently when packing objects
|
|
c4df926b
|
2019-07-16T21:54:10
|
|
pack-objects: allocate memory more efficiently
The packbuilder code allocates memory in chunks. When it needs to
allocate, it tries to add 1024 to the number of objects and multiply by
3/2. However, it actually multiplies by 1 instead, since it performs an
integral division in the expression "3 / 2" and only then multiplies by
the increased number of objects.
The current behavior causes the code to waste massive amounts of time
copying memory when it reallocates, causing inserting all non-blob
objects in the Linux repository into a new pack to take some
indeterminate time greater than 5 minutes instead of 52 seconds.
Correct this error by first dividing by two, and only then multiplying
by 3. We still check for overflow for the multiplication, which is the
only part that can overflow. This appears to be the only place in the
code base which has this problem.
|
|
4cd8dfaa
|
2019-07-16T20:20:55
|
|
clar: correctly account for "data" suites when counting
Failing to do that makes clar miss the last of the suites, as all
duplicated "data" would have not been accounted for.
|
|
|
f92d495d
|
2019-07-12T10:48:14
|
|
Merge pull request #5131 from pks-t/pks/fileops-mkdir-in-root
fileops: fix creation of directory in filesystem root
|
|
|
dacac9e1
|
2019-07-12T08:30:07
|
|
Merge pull request #5160 from pks-t/pks/win32-fuzzers
win32: fix fuzzers and have CI build them
|
|
|
5ae22a63
|
2019-06-21T08:13:31
|
|
fileops: fix creation of directory in filesystem root
In commit 45f24e787 (git_repository_init: stop traversing at
windows root, 2019-04-12), we have fixed `git_futils_mkdir` to
correctly handle the case where we create a directory in
Windows-style filesystem roots like "C:\repo".
The problem here is an off-by-one: previously, to that commit,
we've been checking wether the parent directory's length is equal
to the root directory's length incremented by one. When we call
the function with "/example", then the parent directory's length
("/") is 1, but the root directory offset is 0 as the path is
directly rooted without a drive prefix. This resulted in `1 == 0 +
1`, which was true. With the change, we've stopped incrementing
the root directory length, and thus now compare `1 <= 0`, which
is false.
The previous way of doing it was kind of finicky any non-obvious,
which is also why the error was introduced. So instead of just
re-adding the increment, let's explicitly add a condition that
aborts finding the parent if the current parent path is "/".
Making this change causes Azure Pipelines to fail the testcase
repo::init::nonexistent_paths on Unix-based systems. This is because we
have just fixed creating directories in the filesystem root, which
previously didn't work. As Docker-based tests are running as root user,
we are thus able to create the non-existing path and will now succeed to
create the repository that was expected to actually fail.
Let's split this up into three different tests:
- A test to verify that we do not create repos in a non-existing parent
directoy if the flag `GIT_REPOSITORY_INIT_MKPATH` is not set.
- A test to verify that we fail if the root directory does not exist. As
there is a common root directory on Unix-based systems that always
exist, we can only test for this on Windows-based systems.
- A test to verify that we fail if trying to create a repository in an
unwriteable parent directory. We can only test this if not running
tests as root user, as CAP_DAC_OVERRIDE will cause us to ignore
permissions when creating files.
|
|
|
a6ad9e8a
|
2019-07-11T14:03:21
|
|
Merge pull request #5134 from pks-t/pks/config-parser-separation
Config parser separation
|
|
|
dbeadf8a
|
2019-07-11T10:56:05
|
|
config_parse: provide parser init and dispose functions
Right now, all configuration file backends are expected to
directly mess with the configuration parser's internals in order
to set it up. Let's avoid doing that by implementing both a
`git_config_parser_init` and `git_config_parser_dispose` function
to clearly define the interface between configuration backends
and the parser.
Ideally, we would make the `git_config_parser` structure
definition private to its implementation. But as that would
require an additional memory allocation that was not required
before we just live with it being visible to others.
|
|
|
32157526
|
2019-07-11T11:10:02
|
|
config_file: refactor error handling in `config_write`
Error handling in `config_write` is rather convoluted and does
not match our current code style. Refactor it to make it easier
to understand.
|
|
|
820fa1a3
|
2019-07-11T11:04:33
|
|
config_file: internalize `git_config_file` struct
With the previous commits, we have finally separated the config
parsing logic from the specific configuration file backend. Due
to that, we can now move the `git_config_file` structure into the
config file backend's implementation so that no other code may
accidentally start using it again. Furthermore, we rename the
structure to `diskfile` to make it obvious that it is internal,
only, and to unify it with naming scheme of the other diskfile
structures.
|
|
|
6e6da75f
|
2019-07-11T11:00:05
|
|
config_parse: remove use of `git_config_file`
The config parser code needs to keep track of the current parsed
file's name so that we are able to provide proper error messages
to the user. Right now, we do that by storing a `git_config_file`
in the parser structure, but as that is a specific backend and
the parser aims to be generic, it is a layering violation.
Switch over to use a simple string to fix that.
|
|
|
54d350e0
|
2019-06-21T12:53:43
|
|
config_file: embed file in diskfile parse data
The config file code needs to keep track of the actual
`git_config_file` structure, as it not only contains the path
of the current configuration file, but it also keeps tracks of
all includes of that file. Right now, we keep track of that
structure via the `git_config_parser`, but as that's supposed to
be a backend generic implementation of configuration parsing it's
a layering violation to have it in there.
Switch over the config file backend to use its own config file
structure that's embedded in the backend parse data. This allows
us to switch over the generic config parser to avoid using the
`git_config_file` structure.
|
|
|
76749dfb
|
2019-06-21T12:33:31
|
|
config_parse: rename `data` parameter to `payload` for clarity
By convention, parameters that get passed to callbacks are
usually named `payload` in our codebase. Rename the `data`
parameters in the configuration parser callbacks to `payload` to
avoid confusion.
|
|
|
ba9725a2
|
2019-07-11T10:48:49
|
|
Merge pull request #5132 from pks-t/pks/config-stat-cache
config_file: implement stat cache to avoid repeated rehashing
|
|
|
2ba7020f
|
2019-06-27T09:23:59
|
|
config_file: avoid re-reading files on write
When we rewrite the configuration file due to any of its values
being modified, we call `config_refresh` to update the in-memory
representation of our config file backend. This is needlessly
wasteful though, as `config_refresh` will always open the on-disk
representation to reads the file contents while we already know
the complete file contents at this point in time as we have just
written it to disk.
Implement a new function `config_refresh_from_buffer` that will
refresh the backend's config entries from a buffer instead of
from the config file itself. Note that this will thus _not_
update the backend's timestamp, which will cause us to re-read
the buffer when performing a read operation on it. But this is
still an improvement as we now lazily re-read the contents, and
most importantly we will avoid constantly re-reading the contents
if we perform multiple write operations.
The following strace demonstrates this if we're re-writing a key
multiple times. It uses our config example with `config_set`
changed to update the file 10 times with different keys:
$ strace lg2 config x.x z |& grep '^open.*config'
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 3
open("/home/pks/.config/git/config", O_RDONLY|O_CLOEXEC) = 3
open("/tmp/repo/.git/config.lock", O_WRONLY|O_CREAT|O_EXCL|O_CLOEXEC, 0666) = 3
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 4
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 3
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 3
open("/tmp/repo/.git/config.lock", O_WRONLY|O_CREAT|O_EXCL|O_CLOEXEC, 0666) = 3
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 4
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 3
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 3
open("/tmp/repo/.git/config.lock", O_WRONLY|O_CREAT|O_EXCL|O_CLOEXEC, 0666) = 3
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 4
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 3
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 3
open("/tmp/repo/.git/config.lock", O_WRONLY|O_CREAT|O_EXCL|O_CLOEXEC, 0666) = 3
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 4
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 3
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 3
open("/tmp/repo/.git/config.lock", O_WRONLY|O_CREAT|O_EXCL|O_CLOEXEC, 0666) = 3
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 4
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 3
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 3
open("/tmp/repo/.git/config.lock", O_WRONLY|O_CREAT|O_EXCL|O_CLOEXEC, 0666) = 3
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 4
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 3
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 3
open("/tmp/repo/.git/config.lock", O_WRONLY|O_CREAT|O_EXCL|O_CLOEXEC, 0666) = 3
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 4
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 3
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 3
open("/tmp/repo/.git/config.lock", O_WRONLY|O_CREAT|O_EXCL|O_CLOEXEC, 0666) = 3
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 4
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 3
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 3
open("/tmp/repo/.git/config.lock", O_WRONLY|O_CREAT|O_EXCL|O_CLOEXEC, 0666) = 3
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 4
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 3
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 3
open("/tmp/repo/.git/config.lock", O_WRONLY|O_CREAT|O_EXCL|O_CLOEXEC, 0666) = 3
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 4
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 3
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 3
And now with the optimization of `config_refresh_from_buffer`:
$ strace lg2 config x.x z |& grep '^open.*config'
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 3
open("/home/pks/.config/git/config", O_RDONLY|O_CLOEXEC) = 3
open("/tmp/repo/.git/config.lock", O_WRONLY|O_CREAT|O_EXCL|O_CLOEXEC, 0666) = 3
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 4
open("/tmp/repo/.git/config.lock", O_WRONLY|O_CREAT|O_EXCL|O_CLOEXEC, 0666) = 3
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 4
open("/tmp/repo/.git/config.lock", O_WRONLY|O_CREAT|O_EXCL|O_CLOEXEC, 0666) = 3
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 4
open("/tmp/repo/.git/config.lock", O_WRONLY|O_CREAT|O_EXCL|O_CLOEXEC, 0666) = 3
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 4
open("/tmp/repo/.git/config.lock", O_WRONLY|O_CREAT|O_EXCL|O_CLOEXEC, 0666) = 3
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 4
open("/tmp/repo/.git/config.lock", O_WRONLY|O_CREAT|O_EXCL|O_CLOEXEC, 0666) = 3
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 4
open("/tmp/repo/.git/config.lock", O_WRONLY|O_CREAT|O_EXCL|O_CLOEXEC, 0666) = 3
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 4
open("/tmp/repo/.git/config.lock", O_WRONLY|O_CREAT|O_EXCL|O_CLOEXEC, 0666) = 3
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 4
open("/tmp/repo/.git/config.lock", O_WRONLY|O_CREAT|O_EXCL|O_CLOEXEC, 0666) = 3
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 4
open("/tmp/repo/.git/config.lock", O_WRONLY|O_CREAT|O_EXCL|O_CLOEXEC, 0666) = 3
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 4
As can be seen, this is quite a lot of `open` calls less.
|
|
|
a0dc3027
|
2019-06-27T08:54:51
|
|
config_file: split out function that sets config entries
Updating a config file backend's config entries is a bit more
involved, as it requires clearing of the old config entries as
well as handling locking correctly. As we will need this
functionality in a future patch to refresh config entries from a
buffer, let's extract this into its own function
`config_set_entries`.
|
|
|
985f5cdf
|
2019-06-27T08:41:16
|
|
config_file: split out function that reads entries from a buffer
The `config_read` function currently performs both reading the
on-disk config file as well as parsing the retrieved buffer
contents. To optimize how we refresh our config entries from an
in-memory buffer, we need to be able to directly parse buffers,
though, without involving any on-disk files at all.
Extract a new function `config_read_buffer` that sets up the
parsing logic and then parses config entries from a buffer, only.
Have `config_read` use it to avoid duplicated logic.
|
|
|
3e1c137a
|
2019-06-27T08:24:21
|
|
config_file: move refresh into `write` function
We are quite lazy in how we refresh our config file backend when
updating any of its keys: instead of just updating our in-memory
representation of the keys, we just discard the old set of keys
and then re-read the config file contents from disk. This refresh
currently happens separately at every callsite of `config_write`,
but it is clear that we _always_ want to refresh if we have
written the config file to disk. If we didn't, then we'd run
around with an outdated config file backend that does not
represent what we have on disk.
By moving the refresh into `config_write`, we are also able to
optimize the case where the config file is currently locked.
Before, we would've tried to re-read the file even if we have
only updated its cached contents without touching the on-disk
file. Thus we'd have unnecessarily stat'd the file, even though
we know that it shouldn't have been modified in the meantime due
to its lock.
|
|
|
d7f58eab
|
2019-06-21T11:55:21
|
|
config_file: implement stat cache to avoid repeated rehashing
To decide whether a config file has changed, we always hash its
complete contents. This is unnecessarily expensive, as
well-behaved filesystems will always update stat information for
files which have changed. So before computing the hash, we should
first check whether the stat info has actually changed for either
the configuration file or any of its includes. This avoids having
to re-read the configuration file and its includes every time
when we check whether it's been modified.
Tracing the for-each-ref example previous to this commit, one can
see that we repeatedly re-open both the repo configuration as
well as the global configuration:
$ strace lg2 for-each-ref |& grep config
access("/home/pks/.gitconfig", F_OK) = -1 ENOENT (No such file or directory)
access("/home/pks/.config/git/config", F_OK) = 0
access("/etc/gitconfig", F_OK) = -1 ENOENT (No such file or directory)
stat("/tmp/repo/.git/config", {st_mode=S_IFREG|0644, st_size=92, ...}) = 0
access("/tmp/repo/.git/config", F_OK) = 0
stat("/tmp/repo/.git/config", {st_mode=S_IFREG|0644, st_size=92, ...}) = 0
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 3
stat("/home/pks/.gitconfig", 0x7ffd15c05290) = -1 ENOENT (No such file or directory)
access("/home/pks/.gitconfig", F_OK) = -1 ENOENT (No such file or directory)
stat("/home/pks/.config/git/config", {st_mode=S_IFREG|0644, st_size=1154, ...}) = 0
access("/home/pks/.config/git/config", F_OK) = 0
stat("/home/pks/.config/git/config", {st_mode=S_IFREG|0644, st_size=1154, ...}) = 0
open("/home/pks/.config/git/config", O_RDONLY|O_CLOEXEC) = 3
stat("/tmp/repo/.git/config", {st_mode=S_IFREG|0644, st_size=92, ...}) = 0
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 3
stat("/home/pks/.gitconfig", 0x7ffd15c051f0) = -1 ENOENT (No such file or directory)
stat("/home/pks/.config/git/config", {st_mode=S_IFREG|0644, st_size=1154, ...}) = 0
open("/home/pks/.config/git/config", O_RDONLY|O_CLOEXEC) = 3
stat("/tmp/repo/.git/config", {st_mode=S_IFREG|0644, st_size=92, ...}) = 0
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 3
stat("/home/pks/.gitconfig", 0x7ffd15c05090) = -1 ENOENT (No such file or directory)
stat("/home/pks/.config/git/config", {st_mode=S_IFREG|0644, st_size=1154, ...}) = 0
open("/home/pks/.config/git/config", O_RDONLY|O_CLOEXEC) = 3
stat("/tmp/repo/.git/config", {st_mode=S_IFREG|0644, st_size=92, ...}) = 0
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 3
stat("/home/pks/.gitconfig", 0x7ffd15c05090) = -1 ENOENT (No such file or directory)
stat("/home/pks/.config/git/config", {st_mode=S_IFREG|0644, st_size=1154, ...}) = 0
open("/home/pks/.config/git/config", O_RDONLY|O_CLOEXEC) = 3
stat("/tmp/repo/.git/config", {st_mode=S_IFREG|0644, st_size=92, ...}) = 0
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 3
stat("/home/pks/.gitconfig", 0x7ffd15c05090) = -1 ENOENT (No such file or directory)
stat("/home/pks/.config/git/config", {st_mode=S_IFREG|0644, st_size=1154, ...}) = 0
open("/home/pks/.config/git/config", O_RDONLY|O_CLOEXEC) = 3
With the change, we only do stats for those files and open them a
single time, only:
$ strace lg2 for-each-ref |& grep config
access("/home/pks/.gitconfig", F_OK) = -1 ENOENT (No such file or directory)
access("/home/pks/.config/git/config", F_OK) = 0
access("/etc/gitconfig", F_OK) = -1 ENOENT (No such file or directory)
stat("/tmp/repo/.git/config", {st_mode=S_IFREG|0644, st_size=92, ...}) = 0
access("/tmp/repo/.git/config", F_OK) = 0
stat("/tmp/repo/.git/config", {st_mode=S_IFREG|0644, st_size=92, ...}) = 0
stat("/tmp/repo/.git/config", {st_mode=S_IFREG|0644, st_size=92, ...}) = 0
open("/tmp/repo/.git/config", O_RDONLY|O_CLOEXEC) = 3
stat("/home/pks/.gitconfig", 0x7ffe70540d20) = -1 ENOENT (No such file or directory)
access("/home/pks/.gitconfig", F_OK) = -1 ENOENT (No such file or directory)
stat("/home/pks/.config/git/config", {st_mode=S_IFREG|0644, st_size=1154, ...}) = 0
access("/home/pks/.config/git/config", F_OK) = 0
stat("/home/pks/.config/git/config", {st_mode=S_IFREG|0644, st_size=1154, ...}) = 0
stat("/home/pks/.config/git/config", {st_mode=S_IFREG|0644, st_size=1154, ...}) = 0
open("/home/pks/.config/git/config", O_RDONLY|O_CLOEXEC) = 3
stat("/tmp/repo/.git/config", {st_mode=S_IFREG|0644, st_size=92, ...}) = 0
stat("/home/pks/.gitconfig", 0x7ffe70540ca0) = -1 ENOENT (No such file or directory)
stat("/home/pks/.gitconfig", 0x7ffe70540c80) = -1 ENOENT (No such file or directory)
stat("/home/pks/.config/git/config", {st_mode=S_IFREG|0644, st_size=1154, ...}) = 0
stat("/tmp/repo/.git/config", {st_mode=S_IFREG|0644, st_size=92, ...}) = 0
stat("/home/pks/.gitconfig", 0x7ffe70540b40) = -1 ENOENT (No such file or directory)
stat("/home/pks/.gitconfig", 0x7ffe70540b20) = -1 ENOENT (No such file or directory)
stat("/home/pks/.config/git/config", {st_mode=S_IFREG|0644, st_size=1154, ...}) = 0
stat("/tmp/repo/.git/config", {st_mode=S_IFREG|0644, st_size=92, ...}) = 0
stat("/home/pks/.gitconfig", 0x7ffe70540b40) = -1 ENOENT (No such file or directory)
stat("/home/pks/.gitconfig", 0x7ffe70540b20) = -1 ENOENT (No such file or directory)
stat("/home/pks/.config/git/config", {st_mode=S_IFREG|0644, st_size=1154, ...}) = 0
stat("/tmp/repo/.git/config", {st_mode=S_IFREG|0644, st_size=92, ...}) = 0
stat("/home/pks/.gitconfig", 0x7ffe70540b40) = -1 ENOENT (No such file or directory)
stat("/home/pks/.gitconfig", 0x7ffe70540b20) = -1 ENOENT (No such file or directory)
stat("/home/pks/.config/git/config", {st_mode=S_IFREG|0644, st_size=1154, ...}) = 0
The following benchmark has been performed with and without the
stat cache in a best-of-ten run:
```
int lg2_repro(git_repository *repo, int argc, char **argv)
{
git_config *cfg;
int32_t dummy;
int i;
UNUSED(argc);
UNUSED(argv);
check_lg2(git_repository_config(&cfg, repo),
"Could not obtain config", NULL);
for (i = 1; i < 100000; ++i)
git_config_get_int32(&dummy, cfg, "foo.bar");
git_config_free(cfg);
return 0;
}
```
Without stat cache:
$ time lg2 repro
real 0m1.528s
user 0m0.568s
sys 0m0.944s
With stat cache:
$ time lg2 repro
real 0m0.526s
user 0m0.268s
sys 0m0.258s
This benchmark shows a nearly three-fold performance improvement.
This change requires that we check our configuration stress tests
as we're now in fact becoming more racy. If somebody is writing a
configuration file at nearly the same time (there is a window of
100ns on Windows-based systems), then it might be that we realize
that this file has actually changed and thus may not re-read it.
This will only happen if either an external process is rewriting
the configuration file or if the same process has multiple
`git_config` structures pointing to the same time, where one of
both is being used to write and the other one is used to read
values.
|
|
|
d0868646
|
2019-06-21T11:43:09
|
|
config: use `git_config_file` in favor of `struct config_file`
|
|
|
8ee3d39a
|
2019-06-27T09:18:19
|
|
examples: implement config example
Implement a new example that resembles git-config(1). Right now,
this example can both read and set configuration keys, only.
|
|
|
df54c7fb
|
2019-06-27T07:34:43
|
|
cmake: report whether we are using sub-second stat information
Depending on the platform and on build options, we may or may not
build libgit2 with support for nanoseconds when using `stat`
calls. It's currently unclear though whether sub-second stat
information is used at all.
Add feature info for this to tell at configure time whether it's
being used or not.
|
|
|
eb27fb9b
|
2019-07-05T11:59:17
|
|
ci: build fuzzers on Powershell based build jobs
In order to guarantee that our fuzzers build just fine on the Windows
platform, let's enable building fuzzers on all Powershell-based builds.
|
|
|
3c966fb4
|
2019-06-28T10:53:03
|
|
fuzzers: clean up header includes
There's multiple headers included in our fuzzers that aren't required at
all. Furthermore, some of them are not available on Win32, causing
builds to fail. Remove them to fix this.
|
|
|
9d43d45b
|
2019-06-28T12:10:51
|
|
fuzzers: use `git_buf_printf` instead of `snprintf`
The `snprintf` function does not exist on Win32, it only has
`_snprintf_s` available. Let's just avoid any cross-platform hassle and
use our own `git_buf` functionality instead.
|
|
|
a6b2fffd
|
2019-06-28T11:04:21
|
|
fuzzers: use POSIX emulation layer to unlink files
Use `p_unlink` instead of `unlink` to remove the generated packfiles in
our packfile fuzzer. Like this, we do not have to worry about using
proper includes that are known on all platforms, especially Win32.
|
|
|
69055813
|
2019-06-28T10:50:01
|
|
fuzzers: make printf formatters cross-platform compatible
The `printf` formatters in our standalone fuzzing driver are currently
using the "%m" specifier, which is a GNU extension that prints the error
message for the error code in `errno`. As we're using libgit2 functions
in both cases anyway, let's just use `git_error_last` instead to make
this valid on all platforms.
|
|
|
48d56328
|
2019-06-28T10:47:37
|
|
fuzzers: implement `mkdtemp` alternative for Win32
The `mkdtemp` function is not available on Windows, so our download_refs
fuzzer will fail to compile on Windows. Provide an alternative
implementation to fix it.
|
|
|
398412cc
|
2019-07-05T11:56:16
|
|
Merge pull request #5143 from libgit2/ethomson/warnings
ci: build with ENABLE_WERROR on Windows
|
|
|
a3afda9f
|
2019-06-28T11:50:32
|
|
tests: trace: fix parameter type of aux callback
The function `git_win32__stack__set_aux_cb` expects the second parameter
to be a function callback of type `git_win32__stack__aux_cb_lookup`,
which expects a `size_t` parameter. In our test suite
trace::windows::stacktrace, we declare the callback with `unsigned int`
as parameter, though, causing a compiler warning.
Correct the parameter type to silence the warning.
|
|
|
2f14c4fc
|
2019-06-28T14:39:20
|
|
w32_stack: convert buffer length param to `size_t`
In both `git_win32__stack_format` and `git_win32__stack`, we handle
buffer lengths via an integer variable. As we only ever pass buffer
sizes to it, this should be a `size_t` though to avoid loss of
precision. As we also use it to compare with other `size_t` variables,
this also silences signed/unsigned comparison warnings.
|
|
|
77d7e5eb
|
2019-06-27T15:29:36
|
|
clar: use `size_t` to keep track of current line number
We use the `__LINE__` macro in several places throughout clar to allow
easier traceability when e.g. a test fails. While `__LINE__` is of type
`size_t`, the clar functions all accept an integer and thus may loose
precision. While unlikely that any file in our codebase will exceed a
linecount of `INT_MAX`, let's convert it anyway to silence any compiler
warnings.
|
|
|
2dea4736
|
2019-06-27T15:27:29
|
|
examples: avoid warning when iterating over index entries
When iterating over index entries, we store the indices in an unsigned
int. As the index entrycount is a `size_t` though, this may be a loss of
precision which a compiler might rightfully complain about.
Use `size_t` instead to fix any warnings.
|
|
|
abf24a30
|
2019-06-27T15:25:17
|
|
examples: avoid conversion warnings when calculating progress
When computing the progress, we perform some arithmetics that are
implicitly converting from `size_t` to `int`. In one case we're
calclulating a percentage, so we know that it should always be in the
range of [0,100] and thus we're fine. In the other case we convert from
bytes to kilobytes -- this should be stored in a `size_t` to avoid loss
of precision, even though it probably won't matter due to limited
download rates.
|
|
|
e7bb1fe8
|
2019-06-27T15:14:08
|
|
examples: avoid passing signed integer to `memchr`
The memchr(3P) function expects a `size_t` as its last parameter, but we
do pass it an object size, which is of signed type `git_off_t`. As we
can be sure that the result will be non-negative, let's just cast the
parameter to a `size_t`.
|
|
|
976eed80
|
2019-06-27T15:12:11
|
|
examples: cast away constness for reallocating head arrays
When reallocating commit arrays in `opts_add_commit` and
`opts_add_refish`, respectively, we simply pass the const pointer to
`xrealloc`. As `xrealloc` expects a non-const pointer, though, this will
generate a warning with some compilers.
Cast away the constness to silence compilers.
|
|
|
5c87b5a8
|
2019-07-04T12:19:07
|
|
Merge pull request #5152 from csware/attr-system-attr-file
Fix Regression: attr: Correctly load system attr file (on Windows)
|
|
|
c87abeca
|
2019-07-04T11:45:02
|
|
tests: attr: add tests for system-level attributes
There were no tests that verified that system-level gitattributes files
get handled correctly. In fact, we have recently introduced a regression
that caused us to abort if there was a system-level gitattributes file
available.
Add two tests that verify that we're able to handle system-level
gitattributes files. The test attr::repo::sysdir_with_session would've
failed without the fix to the described regression.
|
|
|
1bbec26d
|
2019-07-04T11:41:21
|
|
attr_file: completely initialize attribute sessions
The function `git_attr_session__init` is currently only initializing
setting up the attribute's session key by incrementing the repo-global
key by one. Most notably, all other members of the `git_attr_session`
struct are not getting initialized at all. So if one is to allocate a
session on the stack and then calls `git_attr_session__init`, the
session will still not be fully initialized. We have fared just fine
with that until now as all users of the function have allocated the
session structure as part of bigger structs with `calloc`, and thus its
contents have been zero-initialized implicitly already.
Fix this by explicitly zeroing out the session to enable allocation of
sessions on the stack.
|
|
18a6d9f3
|
2019-06-29T16:19:08
|
|
attr: Don't fail in attr_setup if there exists a system attributes file
Regression introduced in commit 5452e49fce21f726bec19519da7f012e3f19e736 on PR #4967.
Signed-off-by: Sven Strickroth <email@cs-ware.de>
|
|
|
c4c1500a
|
2019-06-27T14:18:19
|
|
Merge pull request #5145 from pks-t/pks/hash-algo-uninit-return
hash: fix missing error return on production builds
|
|
|
7fd3f32b
|
2019-06-27T13:54:55
|
|
hash: fix missing error return on production builds
When no hash algorithm has been initialized in a given hash context,
then we will simply `assert` and not return a value at all. This works
just fine in debug builds, but on non-debug builds the assert will be
converted to a no-op and thus we do not have a proper return value.
Fix this by returning an error code in addition to the asserts.
|
|
|
73427b94
|
2019-06-27T13:23:17
|
|
Merge pull request #5142 from scottfurry/StaticChkFixExamples
Resolve static check warnings in example code
|
|
2ba7dbbe
|
2019-06-24T14:55:15
|
|
Resolve static check warnings in example code
Using cppcheck on libgit2 sources indicated two warnings in
example code.
merge.c was reported as having a memory leak. Fix applied
was to `free()` memory pointed to by `parents`.
init.c was reported as having a null pointer dereference
on variable arg. Function 'usage' was being called with
a null variable. Changed supplied parameter to empty string.
|
|
|
e9102def
|
2019-06-27T11:38:04
|
|
Merge pull request #4438 from pks-t/pks/hash-algorithm
Multiple hash algorithms
|
|
|
b6625a3b
|
2019-06-27T10:12:16
|
|
Merge pull request #5128 from tiennou/fix/docs
More documentation
|
|
|
3d22394a
|
2019-06-27T10:11:23
|
|
Merge pull request #4967 from tiennou/fix/4671
Incomplete commondir support
|
|
|
33448b45
|
2019-06-19T19:46:12
|
|
docs: More of it
|
|
|
501c51b2
|
2019-06-26T14:49:50
|
|
repo: commondir resolution can sometimes fallback to the repodir
For example, https://git-scm.com/docs/gitrepository-layout says:
info
Additional information about the repository is recorded in this directory.
This directory is ignored if $GIT_COMMON_DIR is set
and "$GIT_COMMON_DIR/info" will be used instead.
So when looking for `info/attributes`, we need to check the commondir first,
or fallback to "our" `info/attributes`.
|
|
|
9f723c97
|
2019-06-26T14:49:37
|
|
docs: fixups
|
|
|
b883d370
|
2019-06-26T14:49:30
|
|
ignore: fix a missing commondir causing failures
As with the preceding commit, the ignore code tries to load code from
info/exclude, and we fail to ignore a non-existent file here.
|
|
|
82c7a9bc
|
2019-06-26T14:49:24
|
|
attr: fix attribute lookup if repo has no common directory
If creating a repository without a common directory (e.g. by
using `git_repository_new`), then `git_repository_item_path` will
return `GIT_ENOTFOUND` for every file that's usually located in
this directory. While we do not care for this case when looking
up the "info/attributes" file, we fail to properly ignore these
errors when setting up or collecting attributes files. Thus, the
gitattributes lookup is broken and will only ever return
`GIT_ENOTFOUND`.
Fix this issue by properly ignoring `GIT_ENOTFOUND` returned by
`git_repository_item_path`.
|
|
|
5452e49f
|
2019-06-26T14:49:17
|
|
attr: refactor setup to match current coding style
The code in the `attr_setup` function is not really matching our
current coding style. Besides alignment issues, it's also hard to
see what functions calls depend on one another because they're
split up over multiple conditional statements.
Fix these issues by grouping together dependent function calls
and adjusting the alignment.
|
|
|
b6b2d9d7
|
2019-06-25T15:05:23
|
|
examples: ssize_t is signed, not unsigned
|
|
|
cd67a903
|
2019-06-25T14:55:51
|
|
examples: cast away const-ness
|
|
|
1118dd9a
|
2019-06-25T14:50:12
|
|
examples: don't lose `const`
|
|
|
ede458b4
|
2019-06-25T14:48:10
|
|
example: use `git_off_t` for the object size
|
|
|
f48cf5b3
|
2019-06-25T14:46:31
|
|
w32_stack: treat a len as an size_t
|
|
|
6b224054
|
2019-06-25T14:41:36
|
|
ci: build with ENABLE_WERROR on Windows
Build with -Werror's equivalent (/WX) on MSVC
|
|
|
c5448501
|
2019-06-25T12:55:10
|
|
Merge pull request #5078 from libgit2/ethomson/warnings
Remove warnings
|
|
|
a064920e
|
2019-06-24T23:27:47
|
|
Merge pull request #5140 from libgit2/ethomson/flaky_ci
Re-run flaky tests
|
|
|
c7b4ce55
|
2019-06-23T20:22:29
|
|
ci: add flaky test re-execution on Windows
Our online tests are occasionally flaky since they hit real network
endpoints. Re-run them up to 5 times if they fail, to allow us to
avoid having to fail the whole build.
|
|
|
6d8a34ad
|
2019-06-23T19:57:05
|
|
ci: add flaky test re-execution on Unix
Our online tests are occasionally flaky since they hit real network
endpoints. Re-run them up to 5 times if they fail, to allow us to
avoid having to fail the whole build.
|
|
|
b7187ed7
|
2019-02-22T14:38:31
|
|
hash: add ability to distinguish algorithms
Create an enum that allows us to distinguish between different
hashing algorithms. This enum is embedded into each
`git_hash_ctx` and will instruct the code to which hashing
function the particular request shall be dispatched.
As we do not yet have multiple hashing algorithms, we simply
initialize the hash algorithm to always be SHA1. At a later
point, we will have to extend the `git_hash_init_ctx` function to
get as parameter which algorithm shall be used.
|
|
|
8832172e
|
2019-02-22T14:32:40
|
|
hash: move SHA1 implementations to its own hashing context
Create a separate `git_hash_sha1_ctx` structure that is specific
to the SHA1 implementation and move all SHA1 functions over to
use that one instead of the generic `git_hash_ctx`. The
`git_hash_ctx` for now simply has a union containing this single
SHA1 implementation, only, without any mechanism to distinguish
between different algortihms.
|
|
|
d46d3b53
|
2019-04-05T10:59:46
|
|
hash: split into generic and SHA1-specific interface
As a preparatory step to allow multiple hashing APIs to exist at
the same time, split the hashing functions into one layer for generic
hashing and one layer for SHA1-specific hashing. Right now, this is
simply an additional indirection layer that doesn't yet serve any
purpose. In the future, the generic API will be extended to allow for
choosing which hash to use, though, by simply passing an enum to the
hash context initialization function. This is necessary as a first step
to be ready for Git's move to SHA256.
|
|
|
fda20622
|
2019-06-14T14:22:19
|
|
hash: move SHA1 implementations into 'sha1/' folder
As we will include additional hash algorithms in the future due
to upstream git discussing a move away from SHA1, we should
accomodate for that and prepare for the move. As a first step,
move all SHA1 implementations into a common subdirectory.
Also, create a SHA1-specific header file that lives inside the
hash folder. This header will contain the SHA1-specific header
includes, function declarations and the SHA1 context structure.
|
|
|
e61b92e0
|
2019-06-24T15:22:39
|
|
clang: disable documentation-deprecated-sync
Add the `-Wno-documentation-deprecated-sync` switch when compiling with
clang, since our documentation adds `deprecated` markers, but we do not
add the deprecation attribute in the code itself. (ie, the code is out
of sync with the docs).
In fact, we do not _want_ to mark these items as deprecated in the code,
at least not yet, as we are not quite ready to bother our end-users with
this since they're not going away.
|
|
|
9f3441cc
|
2019-06-15T22:37:11
|
|
zlib: remove unused functions
|
|
|
c512d58f
|
2019-06-15T22:26:23
|
|
win32: cast WinAPI to void * before casting
GetProcAddress is prototyped to return a `FARPROC`, which is meant to be
a generic function pointer. It's literally `int (FAR WINAPI * FARPROC)()`
which gcc complains if you attempt to cast to a `void (*)(GIT_SRWLOCK *)`.
Cast to a `void *` before casting to avoid warnings about the arguments.
|
|
|
38e6b287
|
2019-06-15T22:25:10
|
|
mingw: use `struct stat`
|
|
|
759ec7d4
|
2019-06-15T22:01:00
|
|
win32: cast GetProcAddress to void * before casting
GetProcAddress is prototyped to return a `FARPROC`, which is meant to be
a generic function pointer. It's literally `int (FAR WINAPI * FARPROC)()`
which gcc complains if you attempt to cast to a `void (*)(GIT_SRWLOCK *)`.
Cast to a `void *` before casting to avoid warnings about the arguments.
|
|
|
3cd123e9
|
2019-06-15T21:56:53
|
|
win32: define DWORD_MAX if it's not defined
MinGW does not define DWORD_MAX. Specify it when it's not defined.
|
|
|
d93b0aa0
|
2019-06-15T21:47:40
|
|
win32: decorate unused parameters
|
|
|
54a60ced
|
2019-06-15T21:45:26
|
|
mingw: disable format specification warnings
MinGW uses gcc, which expects POSIX formatting for printf, but uses the
Windows C library, which uses its own format specifiers. Therefore, it
gets confused about format specifiers. Disable warnings for format
specifiers.
|
|
|
e2aba8ba
|
2019-06-15T20:45:22
|
|
wildmatch: explicitly cast to int
|
|
|
3dd1942b
|
2019-06-15T20:43:13
|
|
win32: don't re-define RtlCaptureStackBackTrace
RtlCaptureStackBackTrace is well-defined in Windows, no need to redefine
it.
|
|
|
cc9e47c9
|
2019-06-15T18:51:40
|
|
win32: support upgrading warnings to errors (/WX)
For MSVC, support warnings as errors by providing the /WX compiler
flags. (/WX is the moral equivalent of -Werror.)
Disable warnings as errors ass part of xdiff, since it contains
warnings. But as a component of git itself, we want to avoid skew and
keep our implementation as similar as possible to theirs. We'll work
with upstream to fix these issues, but in the meantime, simply let those
continue to warn.
|
|
|
f6530438
|
2019-05-25T16:44:59
|
|
win32: stop inlining file_attribute_to_stat
Move `git_win32__file_attribute_to_stat` to a regular function instead
of an inlined function. This helps avoid header ordering issues and
declarations.
|
|
|
bd48bf3f
|
2019-06-14T14:21:32
|
|
hash: introduce source files to break include circles
The hash source files have circular include dependencies right
now, which shows by our broken generic hash implementation. The
"hash.h" header declares two functions and the `git_hash_ctx`
typedef before actually including the hash backend header and can
only declare the remaining hash functions after the include due
to possibly static function declarations inside of the
implementation includes.
Let's break this cycle and help maintainability by creating a
real implementation file for each of the hash implementations.
Instead of relying on the exact include order, we now especially
avoid the use of `GIT_INLINE` for function declarations.
|
|
|
bbf034ab
|
2019-02-22T13:43:16
|
|
hash: move `git_hash_prov` into Win32 backend
The structure `git_hash_prov` is only ever used by the Win32 SHA1
backend. As such, it doesn't make much sense to expose it via the
generic "hash.h" header, as it is an implementation detail of the Win32
backend only. Move the typedef of `git_hash_prov` into
"hash/sha1/win32.h" to fix this.
|
|
|
b11eb08f
|
2019-05-21T14:39:55
|
|
config parse: safely cast to int
|
|
|
6b349ecc
|
2019-05-21T14:36:57
|
|
odb loose: only read at most INT_MAX
|
|
|
8c925ef8
|
2019-05-21T14:30:28
|
|
smart protocol: validate progress message length
Ensure that the server has not sent us overly-large sideband messages
(ensure that they are no more than `INT_MAX` bytes), then cast to `int`.
|
|
|
7afe788c
|
2019-05-21T14:27:46
|
|
smart transport: use size_t for sizes
|
|
|
db7f1d9b
|
2019-05-21T14:21:58
|
|
local transport: cast message size to int explicitly
Our progress information messages are short (and bounded by their format
string), cast the length to int for callers.
|
|
|
8048ba70
|
2019-05-21T14:18:40
|
|
winhttp: safely cast length to DWORD
|
|
|
db6b8f7d
|
2019-05-21T14:15:58
|
|
strtol: cast error message length to int
|
|
|
d103f008
|
2019-05-21T13:44:47
|
|
pool: use `size_t` for sizes
|
|
|
c4a64b1b
|
2019-05-21T13:27:39
|
|
tree-cache: safely cast to uint32_t
|
|
|
2375be48
|
2019-05-21T12:57:28
|
|
tree: return `size_t` for treebuilder entrycount
We keep the treebuilder entrycount as a `size_t` - return that instead
of downcasting to an `unsigned int`. Callers who were storing this
value in an `unsigned int` will continue to downcast themselves, so
there should be no behavior change for callers.
|
|
|
ad6f2153
|
2019-05-21T12:50:46
|
|
utf8: use size_t for length of buffer
The `git__utf8_charlen` now takes `size_t` as the buffer length, since
it contains the full length of the buffer at the current position. It
now returns `-1` in all cases where utf8 codepoints are invalid, since
callers only care about a valid length of a sequence of codepoints, or
if the current position is not valid utf8.
|
|
|
5d5b76df
|
2019-05-21T12:35:19
|
|
worktree: use size_t for sizes
|
|
|
f7597410
|
2019-05-21T10:57:30
|
|
netops: safely cast to int
Only read at most INT_MAX from the underlying stream, so that we can
accurately return the number of bytes read.
Since callers are not guaranteed to get as many bytes as requested (due
to availability of input), this is safe and callers should call in a
loop until EOF.
|
|
|
cfd44d6a
|
2019-05-20T07:57:46
|
|
trailer: use size_t for sizes
|
|
|
fc3a94ba
|
2019-05-20T07:13:42
|
|
repository: use size_t for length
|