git-commit-vandalism/commit.c

1777 lines
44 KiB
C
Raw Normal View History

#include "cache.h"
#include "tag.h"
#include "commit.h"
#include "commit-graph.h"
#include "hex.h"
#include "repository.h"
#include "object-store.h"
#include "pkt-line.h"
#include "utf8.h"
#include "diff.h"
#include "revision.h"
#include "notes.h"
#include "alloc.h"
commit: teach --gpg-sign option This uses the gpg-interface.[ch] to allow signing the commit, i.e. $ git commit --gpg-sign -m foo You need a passphrase to unlock the secret key for user: "Junio C Hamano <gitster@pobox.com>" 4096-bit RSA key, ID 96AFE6CB, created 2011-10-03 (main key ID 713660A7) [master 8457d13] foo 1 files changed, 1 insertions(+), 0 deletions(-) The lines of GPG detached signature are placed in a new multi-line header field, instead of tucking the signature block at the end of the commit log message text (similar to how signed tag is done), for multiple reasons: - The signature won't clutter output from "git log" and friends if it is in the extra header. If we place it at the end of the log message, we would need to teach "git log" and friends to strip the signature block with an option. - Teaching new versions of "git log" and "gitk" to optionally verify and show signatures is cleaner if we structurally know where the signature block is (instead of scanning in the commit log message). - The signature needs to be stripped upon various commit rewriting operations, e.g. rebase, filter-branch, etc. They all already ignore unknown headers, but if we place signature in the log message, all of these tools (and third-party tools) also need to learn how a signature block would look like. - When we added the optional encoding header, all the tools (both in tree and third-party) that acts on the raw commit object should have been fixed to ignore headers they do not understand, so it is not like that new header would be more likely to break than extra text in the commit. A commit made with the above sample sequence would look like this: $ git cat-file commit HEAD tree 3cd71d90e3db4136e5260ab54599791c4f883b9d parent b87755351a47b09cb27d6913e6e0e17e6254a4d4 author Junio C Hamano <gitster@pobox.com> 1317862251 -0700 committer Junio C Hamano <gitster@pobox.com> 1317862251 -0700 gpgsig -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.10 (GNU/Linux) iQIcBAABAgAGBQJOjPtrAAoJELC16IaWr+bL4TMP/RSe2Y/jYnCkds9unO5JEnfG ... =dt98 -----END PGP SIGNATURE----- foo but "git log" (unless you ask for it with --pretty=raw) output is not cluttered with the signature information. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2011-10-06 02:23:20 +02:00
#include "gpg-interface.h"
#include "mergesort.h"
commit-slab: introduce a macro to define a slab for new type Introduce a header file to define a macro that can define the struct type, initializer, accessor and cleanup functions to manage a commit slab. Update the "indegree" topological sort facility using it. To associate 32 flag bits with each commit, you can write: define_commit_slab(flag32, uint32); to declare "struct flag32" type, define an instance of it with struct flag32 flags; and initialize it by calling init_flag32(&flags); After that, a call to flag32_at() function uint32 *fp = flag32_at(&flags, commit); will return a pointer pointing at a uint32 for that commit. Once you are done with these flags, clean them up with clear_flag32(&flags); Callers that cannot hard-code how wide the data to be associated with the commit be at compile time can use the "_with_stride" variant to initialize the slab. Suppose you want to give one bit per existing ref, and paint commits down to find which refs are descendants of each commit. Saying typedef uint32 bits320[5]; define_commit_slab(flagbits, bits320); at compile time will still limit your code with hard-coded limit, because you may find that you have more than 320 refs at runtime. The code can declare a commit slab "struct flagbits" like this instead: define_commit_slab(flagbits, unsigned char); struct flagbits flags; and initialize it by: nrefs = ... count number of refs ... init_flagbits_with_stride(&flags, (nrefs + 7) / 8); so that unsigned char *fp = flagbits_at(&flags, commit); will return a pointer pointing at an array of 40 "unsigned char"s associated with the commit, once you figure out nrefs is 320 at runtime. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-04-13 20:56:41 +02:00
#include "commit-slab.h"
#include "prio-queue.h"
#include "hash-lookup.h"
interpret-trailers: honor the cut line If a commit message is edited with the "verbose" option, the buffer will have a cut line and diff after the log message, like so: my subject # ------------------------ >8 ------------------------ # Do not touch the line above. # Everything below will be removed. diff --git a/foo.txt b/foo.txt index 5716ca5..7601807 100644 --- a/foo.txt +++ b/foo.txt @@ -1 +1 @@ -bar +baz "git interpret-trailers" is unaware of the cut line, and assumes the trailer block would be at the end of the whole thing. This can easily be seen with: $ GIT_EDITOR='git interpret-trailers --in-place --trailer Acked-by:me' \ git commit --amend -v Teach "git interpret-trailers" to notice the cut-line and ignore the remainder of the input when looking for a place to add new trailer block. This makes it consistent with how "git commit -v -s" inserts a new Signed-off-by: line. This can be done by the same logic as the existing helper function, wt_status_truncate_message_at_cut_line(), uses, but it wants the caller to pass a strbuf to it. Because the function ignore_non_trailer() used by the command takes a <pointer, length> pair, not a strbuf, steal the logic from wt_status_truncate_message_at_cut_line() to create a new wt_status_locate_end() helper function that takes <pointer, length> pair, and make ignore_non_trailer() call it to help "interpret-trailers". Since there is only one caller of wt_status_truncate_message_at_cut_line() in cmd_commit(), rewrite it to call wt_status_locate_end() helper instead and remove the old helper that no longer has any caller. Signed-off-by: Brian Malehorn <bmalehorn@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2017-05-16 08:06:49 +02:00
#include "wt-status.h"
#include "advice.h"
#include "refs.h"
#include "commit-reach.h"
#include "run-command.h"
#include "shallow.h"
#include "hook.h"
static struct commit_extra_header *read_commit_extra_header_lines(const char *buf, size_t len, const char **);
[PATCH] Avoid wasting memory in git-rev-list As pointed out on the list, git-rev-list can use a lot of memory. One low-hanging fruit is to free the commit buffer for commits that we parse. By default, parse_commit() will save away the buffer, since a lot of cases do want it, and re-reading it continually would be unnecessary. However, in many cases the buffer isn't actually necessary and saving it just wastes memory. We could just free the buffer ourselves, but especially in git-rev-list, we actually end up using the helper functions that automatically add parent commits to the commit lists, so we don't actually control the commit parsing directly. Instead, just make this behaviour of "parse_commit()" a global flag. Maybe this is a bit tasteless, but it's very simple, and it makes a noticable difference in memory usage. Before the change: [torvalds@g5 linux]$ /usr/bin/time git-rev-list v2.6.12..HEAD > /dev/null 0.26user 0.02system 0:00.28elapsed 99%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+3714minor)pagefaults 0swaps after the change: [torvalds@g5 linux]$ /usr/bin/time git-rev-list v2.6.12..HEAD > /dev/null 0.26user 0.00system 0:00.27elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+2433minor)pagefaults 0swaps note how the minor faults have decreased from 3714 pages to 2433 pages. That's all due to the fewer anonymous pages allocated to hold the comment buffers and their metadata. Signed-off-by: Linus Torvalds <torvalds@osdl.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
2005-09-15 23:43:17 +02:00
int save_commit_buffer = 1;
int no_graft_file_deprecated_advice;
[PATCH] Avoid wasting memory in git-rev-list As pointed out on the list, git-rev-list can use a lot of memory. One low-hanging fruit is to free the commit buffer for commits that we parse. By default, parse_commit() will save away the buffer, since a lot of cases do want it, and re-reading it continually would be unnecessary. However, in many cases the buffer isn't actually necessary and saving it just wastes memory. We could just free the buffer ourselves, but especially in git-rev-list, we actually end up using the helper functions that automatically add parent commits to the commit lists, so we don't actually control the commit parsing directly. Instead, just make this behaviour of "parse_commit()" a global flag. Maybe this is a bit tasteless, but it's very simple, and it makes a noticable difference in memory usage. Before the change: [torvalds@g5 linux]$ /usr/bin/time git-rev-list v2.6.12..HEAD > /dev/null 0.26user 0.02system 0:00.28elapsed 99%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+3714minor)pagefaults 0swaps after the change: [torvalds@g5 linux]$ /usr/bin/time git-rev-list v2.6.12..HEAD > /dev/null 0.26user 0.00system 0:00.27elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+2433minor)pagefaults 0swaps note how the minor faults have decreased from 3714 pages to 2433 pages. That's all due to the fewer anonymous pages allocated to hold the comment buffers and their metadata. Signed-off-by: Linus Torvalds <torvalds@osdl.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
2005-09-15 23:43:17 +02:00
const char *commit_type = "commit";
struct commit *lookup_commit_reference_gently(struct repository *r,
const struct object_id *oid, int quiet)
{
struct object *obj = deref_tag(r,
parse_object(r, oid),
NULL, 0);
if (!obj)
return NULL;
return object_as_type(obj, OBJ_COMMIT, quiet);
}
struct commit *lookup_commit_reference(struct repository *r, const struct object_id *oid)
{
return lookup_commit_reference_gently(r, oid, 0);
}
Convert lookup_commit* to struct object_id Convert lookup_commit, lookup_commit_or_die, lookup_commit_reference, and lookup_commit_reference_gently to take struct object_id arguments. Introduce a temporary in parse_object buffer in order to convert this function. This is required since in order to convert parse_object and parse_object_buffer, lookup_commit_reference_gently and lookup_commit_or_die would need to be converted. Not introducing a temporary would therefore require that lookup_commit_or_die take a struct object_id *, but lookup_commit would take unsigned char *, leaving a confusing and hard-to-use interface. parse_object_buffer will lose this temporary in a later patch. This commit was created with manual changes to commit.c, commit.h, and object.c, plus the following semantic patch: @@ expression E1, E2; @@ - lookup_commit_reference_gently(E1.hash, E2) + lookup_commit_reference_gently(&E1, E2) @@ expression E1, E2; @@ - lookup_commit_reference_gently(E1->hash, E2) + lookup_commit_reference_gently(E1, E2) @@ expression E1; @@ - lookup_commit_reference(E1.hash) + lookup_commit_reference(&E1) @@ expression E1; @@ - lookup_commit_reference(E1->hash) + lookup_commit_reference(E1) @@ expression E1; @@ - lookup_commit(E1.hash) + lookup_commit(&E1) @@ expression E1; @@ - lookup_commit(E1->hash) + lookup_commit(E1) @@ expression E1, E2; @@ - lookup_commit_or_die(E1.hash, E2) + lookup_commit_or_die(&E1, E2) @@ expression E1, E2; @@ - lookup_commit_or_die(E1->hash, E2) + lookup_commit_or_die(E1, E2) Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2017-05-07 00:10:10 +02:00
struct commit *lookup_commit_or_die(const struct object_id *oid, const char *ref_name)
{
struct commit *c = lookup_commit_reference(the_repository, oid);
if (!c)
die(_("could not parse %s"), ref_name);
if (!oideq(oid, &c->object.oid)) {
warning(_("%s %s is not a commit!"),
Convert lookup_commit* to struct object_id Convert lookup_commit, lookup_commit_or_die, lookup_commit_reference, and lookup_commit_reference_gently to take struct object_id arguments. Introduce a temporary in parse_object buffer in order to convert this function. This is required since in order to convert parse_object and parse_object_buffer, lookup_commit_reference_gently and lookup_commit_or_die would need to be converted. Not introducing a temporary would therefore require that lookup_commit_or_die take a struct object_id *, but lookup_commit would take unsigned char *, leaving a confusing and hard-to-use interface. parse_object_buffer will lose this temporary in a later patch. This commit was created with manual changes to commit.c, commit.h, and object.c, plus the following semantic patch: @@ expression E1, E2; @@ - lookup_commit_reference_gently(E1.hash, E2) + lookup_commit_reference_gently(&E1, E2) @@ expression E1, E2; @@ - lookup_commit_reference_gently(E1->hash, E2) + lookup_commit_reference_gently(E1, E2) @@ expression E1; @@ - lookup_commit_reference(E1.hash) + lookup_commit_reference(&E1) @@ expression E1; @@ - lookup_commit_reference(E1->hash) + lookup_commit_reference(E1) @@ expression E1; @@ - lookup_commit(E1.hash) + lookup_commit(&E1) @@ expression E1; @@ - lookup_commit(E1->hash) + lookup_commit(E1) @@ expression E1, E2; @@ - lookup_commit_or_die(E1.hash, E2) + lookup_commit_or_die(&E1, E2) @@ expression E1, E2; @@ - lookup_commit_or_die(E1->hash, E2) + lookup_commit_or_die(E1, E2) Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2017-05-07 00:10:10 +02:00
ref_name, oid_to_hex(oid));
}
return c;
}
struct commit *lookup_commit_object(struct repository *r,
const struct object_id *oid)
{
struct object *obj = parse_object(r, oid);
return obj ? object_as_type(obj, OBJ_COMMIT, 0) : NULL;
}
struct commit *lookup_commit(struct repository *r, const struct object_id *oid)
{
struct object *obj = lookup_object(r, oid);
if (!obj)
return create_object(r, oid, alloc_commit_node(r));
return object_as_type(obj, OBJ_COMMIT, 0);
}
struct commit *lookup_commit_reference_by_name(const char *name)
{
struct object_id oid;
struct commit *commit;
sha1_name: convert get_sha1* to get_oid* Now that all the callers of get_sha1 directly or indirectly use struct object_id, rename the functions starting with get_sha1 to start with get_oid. Convert the internals in sha1_name.c to use struct object_id as well, and eliminate explicit length checks where possible. Convert a use of 40 in get_oid_basic to GIT_SHA1_HEXSZ. Outside of sha1_name.c and cache.h, this transition was made with the following semantic patch: @@ expression E1, E2; @@ - get_sha1(E1, E2.hash) + get_oid(E1, &E2) @@ expression E1, E2; @@ - get_sha1(E1, E2->hash) + get_oid(E1, E2) @@ expression E1, E2; @@ - get_sha1_committish(E1, E2.hash) + get_oid_committish(E1, &E2) @@ expression E1, E2; @@ - get_sha1_committish(E1, E2->hash) + get_oid_committish(E1, E2) @@ expression E1, E2; @@ - get_sha1_treeish(E1, E2.hash) + get_oid_treeish(E1, &E2) @@ expression E1, E2; @@ - get_sha1_treeish(E1, E2->hash) + get_oid_treeish(E1, E2) @@ expression E1, E2; @@ - get_sha1_commit(E1, E2.hash) + get_oid_commit(E1, &E2) @@ expression E1, E2; @@ - get_sha1_commit(E1, E2->hash) + get_oid_commit(E1, E2) @@ expression E1, E2; @@ - get_sha1_tree(E1, E2.hash) + get_oid_tree(E1, &E2) @@ expression E1, E2; @@ - get_sha1_tree(E1, E2->hash) + get_oid_tree(E1, E2) @@ expression E1, E2; @@ - get_sha1_blob(E1, E2.hash) + get_oid_blob(E1, &E2) @@ expression E1, E2; @@ - get_sha1_blob(E1, E2->hash) + get_oid_blob(E1, E2) @@ expression E1, E2, E3, E4; @@ - get_sha1_with_context(E1, E2, E3.hash, E4) + get_oid_with_context(E1, E2, &E3, E4) @@ expression E1, E2, E3, E4; @@ - get_sha1_with_context(E1, E2, E3->hash, E4) + get_oid_with_context(E1, E2, E3, E4) Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2017-07-14 01:49:28 +02:00
if (get_oid_committish(name, &oid))
return NULL;
commit = lookup_commit_reference(the_repository, &oid);
if (parse_commit(commit))
return NULL;
return commit;
}
static timestamp_t parse_commit_date(const char *buf, const char *tail)
{
const char *dateptr;
if (buf + 6 >= tail)
return 0;
if (memcmp(buf, "author", 6))
return 0;
while (buf < tail && *buf++ != '\n')
/* nada */;
if (buf + 9 >= tail)
return 0;
if (memcmp(buf, "committer", 9))
return 0;
while (buf < tail && *buf++ != '>')
/* nada */;
if (buf >= tail)
return 0;
dateptr = buf;
while (buf < tail && *buf++ != '\n')
/* nada */;
if (buf >= tail)
return 0;
/* dateptr < buf && buf[-1] == '\n', so parsing will stop at buf-1 */
return parse_timestamp(dateptr, NULL, 10);
}
static const struct object_id *commit_graft_oid_access(size_t index, const void *table)
{
const struct commit_graft * const *commit_graft_table = table;
return &commit_graft_table[index]->oid;
}
int commit_graft_pos(struct repository *r, const struct object_id *oid)
{
return oid_pos(oid, r->parsed_objects->grafts,
r->parsed_objects->grafts_nr,
commit_graft_oid_access);
}
commit,shallow: unparse commits if grafts changed When a commit is parsed, it pretends to have a different (possibly empty) list of parents if there is graft information for that commit. But there is a bug that could occur when a commit is parsed, the graft information is updated (for example, when a shallow file is rewritten), and the same commit is subsequently used: the parents of the commit do not conform to the updated graft information, but the information at the time of parsing. This is usually not an issue, as a commit is usually introduced into the repository at the same time as its graft information. That means that when we try to parse that commit, we already have its graft information. But it is an issue when fetching a shallow point directly into a repository with submodules. The function assign_shallow_commits_to_refs() parses all sought objects (including the shallow point, which we are directly fetching). In update_shallow() in fetch-pack.c, assign_shallow_commits_to_refs() is called before commit_shallow_file(), which means that the shallow point would have been parsed before graft information is updated. Once a commit is parsed, it is no longer sensitive to any graft information updates. This parsed commit is subsequently used when we do a revision walk to search for submodules to fetch, meaning that the commit is considered to have parents even though it is a shallow point (and therefore should be treated as having no parents). Therefore, whenever graft information is updated, mark the commits that were previously grafts and the commits that are newly grafts as unparsed. Signed-off-by: Jonathan Tan <jonathantanmy@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-06-06 19:54:37 +02:00
static void unparse_commit(struct repository *r, const struct object_id *oid)
{
struct commit *c = lookup_commit(r, oid);
if (!c->object.parsed)
return;
free_commit_list(c->parents);
c->parents = NULL;
c->object.parsed = 0;
}
int register_commit_graft(struct repository *r, struct commit_graft *graft,
int ignore_dups)
{
int pos = commit_graft_pos(r, &graft->oid);
if (0 <= pos) {
if (ignore_dups)
free(graft);
else {
free(r->parsed_objects->grafts[pos]);
r->parsed_objects->grafts[pos] = graft;
}
return 1;
}
pos = -pos - 1;
ALLOC_GROW(r->parsed_objects->grafts,
r->parsed_objects->grafts_nr + 1,
r->parsed_objects->grafts_alloc);
r->parsed_objects->grafts_nr++;
if (pos < r->parsed_objects->grafts_nr)
memmove(r->parsed_objects->grafts + pos + 1,
r->parsed_objects->grafts + pos,
(r->parsed_objects->grafts_nr - pos - 1) *
sizeof(*r->parsed_objects->grafts));
r->parsed_objects->grafts[pos] = graft;
commit,shallow: unparse commits if grafts changed When a commit is parsed, it pretends to have a different (possibly empty) list of parents if there is graft information for that commit. But there is a bug that could occur when a commit is parsed, the graft information is updated (for example, when a shallow file is rewritten), and the same commit is subsequently used: the parents of the commit do not conform to the updated graft information, but the information at the time of parsing. This is usually not an issue, as a commit is usually introduced into the repository at the same time as its graft information. That means that when we try to parse that commit, we already have its graft information. But it is an issue when fetching a shallow point directly into a repository with submodules. The function assign_shallow_commits_to_refs() parses all sought objects (including the shallow point, which we are directly fetching). In update_shallow() in fetch-pack.c, assign_shallow_commits_to_refs() is called before commit_shallow_file(), which means that the shallow point would have been parsed before graft information is updated. Once a commit is parsed, it is no longer sensitive to any graft information updates. This parsed commit is subsequently used when we do a revision walk to search for submodules to fetch, meaning that the commit is considered to have parents even though it is a shallow point (and therefore should be treated as having no parents). Therefore, whenever graft information is updated, mark the commits that were previously grafts and the commits that are newly grafts as unparsed. Signed-off-by: Jonathan Tan <jonathantanmy@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-06-06 19:54:37 +02:00
unparse_commit(r, &graft->oid);
return 0;
}
struct commit_graft *read_graft_line(struct strbuf *line)
{
/* The format is just "Commit Parent1 Parent2 ...\n" */
int i, phase;
const char *tail = NULL;
struct commit_graft *graft = NULL;
struct object_id dummy_oid, *oid;
strbuf_rtrim(line);
if (!line->len || line->buf[0] == '#')
return NULL;
/*
* phase 0 verifies line, counts hashes in line and allocates graft
* phase 1 fills graft
*/
for (phase = 0; phase < 2; phase++) {
oid = graft ? &graft->oid : &dummy_oid;
if (parse_oid_hex(line->buf, oid, &tail))
goto bad_graft_data;
for (i = 0; *tail != '\0'; i++) {
oid = graft ? &graft->parent[i] : &dummy_oid;
if (!isspace(*tail++) || parse_oid_hex(tail, oid, &tail))
goto bad_graft_data;
}
if (!graft) {
graft = xmalloc(st_add(sizeof(*graft),
st_mult(sizeof(struct object_id), i)));
graft->nr_parent = i;
}
}
return graft;
bad_graft_data:
error("bad graft data: %s", line->buf);
assert(!graft);
return NULL;
}
static int read_graft_file(struct repository *r, const char *graft_file)
{
FILE *fp = fopen_or_warn(graft_file, "r");
struct strbuf buf = STRBUF_INIT;
if (!fp)
return -1;
if (!no_graft_file_deprecated_advice &&
advice_enabled(ADVICE_GRAFT_FILE_DEPRECATED))
advise(_("Support for <GIT_DIR>/info/grafts is deprecated\n"
"and will be removed in a future Git version.\n"
"\n"
"Please use \"git replace --convert-graft-file\"\n"
"to convert the grafts into replace refs.\n"
"\n"
"Turn this message off by running\n"
"\"git config advice.graftFileDeprecated false\""));
while (!strbuf_getwholeline(&buf, fp, '\n')) {
/* The format is just "Commit Parent1 Parent2 ...\n" */
struct commit_graft *graft = read_graft_line(&buf);
if (!graft)
continue;
if (register_commit_graft(r, graft, 1))
error("duplicate graft data: %s", buf.buf);
}
fclose(fp);
strbuf_release(&buf);
return 0;
}
void prepare_commit_graft(struct repository *r)
{
char *graft_file;
if (r->parsed_objects->commit_graft_prepared)
return;
prepare_commit_graft: treat non-repository as a noop The parse_commit_buffer() function consults lookup_commit_graft() to see if we need to rewrite parents. The latter will look at $GIT_DIR/info/grafts. If you're outside of a repository, then this will trigger a BUG() as of b1ef400eec (setup_git_env: avoid blind fall-back to ".git", 2016-10-20). It's probably uncommon to actually parse a commit outside of a repository, but you can see it in action with: cd /not/a/git/repo git index-pack --strict /some/file.pack This works fine without --strict, but the fsck checks will try to parse any commits, triggering the BUG(). We can fix that by teaching the graft code to behave as if there are no grafts when we aren't in a repository. Arguably index-pack (and fsck) are wrong to consider grafts at all. So another solution is to disable grafts entirely for those commands. But given that the graft feature is deprecated anyway, it's not worth even thinking through the ramifications that might have. There is one other corner case I considered here. What should: cd /not/a/git/repo export GIT_GRAFT_FILE=/file/with/grafts git index-pack --strict /some/file.pack do? We don't have a repository, but the user has pointed us directly at a graft file, which we could respect. I believe this case did work that way prior to b1ef400eec. However, fixing it now would be pretty invasive. Back then we would just call into setup_git_env() even without a repository. But these days it actually takes a git_dir argument. So there would be a fair bit of refactoring of the setup code involved. Given the obscurity of this case, plus the fact that grafts are deprecated and probably shouldn't work under index-pack anyway, it's not worth pursuing further. This patch at least un-breaks the common case where you're _not_ using grafts, but we BUG() anyway trying to even find that out. Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2018-06-01 00:42:53 +02:00
if (!startup_info->have_repository)
return;
graft_file = get_graft_file(r);
read_graft_file(r, graft_file);
/* make sure shallows are read */
is_repository_shallow(r);
r->parsed_objects->commit_graft_prepared = 1;
}
struct commit_graft *lookup_commit_graft(struct repository *r, const struct object_id *oid)
{
int pos;
prepare_commit_graft(r);
pos = commit_graft_pos(r, oid);
if (pos < 0)
return NULL;
return r->parsed_objects->grafts[pos];
}
int for_each_commit_graft(each_commit_graft_fn fn, void *cb_data)
{
int i, ret;
for (i = ret = 0; i < the_repository->parsed_objects->grafts_nr && !ret; i++)
ret = fn(the_repository->parsed_objects->grafts[i], cb_data);
return ret;
}
void reset_commit_grafts(struct repository *r)
{
int i;
commit,shallow: unparse commits if grafts changed When a commit is parsed, it pretends to have a different (possibly empty) list of parents if there is graft information for that commit. But there is a bug that could occur when a commit is parsed, the graft information is updated (for example, when a shallow file is rewritten), and the same commit is subsequently used: the parents of the commit do not conform to the updated graft information, but the information at the time of parsing. This is usually not an issue, as a commit is usually introduced into the repository at the same time as its graft information. That means that when we try to parse that commit, we already have its graft information. But it is an issue when fetching a shallow point directly into a repository with submodules. The function assign_shallow_commits_to_refs() parses all sought objects (including the shallow point, which we are directly fetching). In update_shallow() in fetch-pack.c, assign_shallow_commits_to_refs() is called before commit_shallow_file(), which means that the shallow point would have been parsed before graft information is updated. Once a commit is parsed, it is no longer sensitive to any graft information updates. This parsed commit is subsequently used when we do a revision walk to search for submodules to fetch, meaning that the commit is considered to have parents even though it is a shallow point (and therefore should be treated as having no parents). Therefore, whenever graft information is updated, mark the commits that were previously grafts and the commits that are newly grafts as unparsed. Signed-off-by: Jonathan Tan <jonathantanmy@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-06-06 19:54:37 +02:00
for (i = 0; i < r->parsed_objects->grafts_nr; i++) {
unparse_commit(r, &r->parsed_objects->grafts[i]->oid);
free(r->parsed_objects->grafts[i]);
commit,shallow: unparse commits if grafts changed When a commit is parsed, it pretends to have a different (possibly empty) list of parents if there is graft information for that commit. But there is a bug that could occur when a commit is parsed, the graft information is updated (for example, when a shallow file is rewritten), and the same commit is subsequently used: the parents of the commit do not conform to the updated graft information, but the information at the time of parsing. This is usually not an issue, as a commit is usually introduced into the repository at the same time as its graft information. That means that when we try to parse that commit, we already have its graft information. But it is an issue when fetching a shallow point directly into a repository with submodules. The function assign_shallow_commits_to_refs() parses all sought objects (including the shallow point, which we are directly fetching). In update_shallow() in fetch-pack.c, assign_shallow_commits_to_refs() is called before commit_shallow_file(), which means that the shallow point would have been parsed before graft information is updated. Once a commit is parsed, it is no longer sensitive to any graft information updates. This parsed commit is subsequently used when we do a revision walk to search for submodules to fetch, meaning that the commit is considered to have parents even though it is a shallow point (and therefore should be treated as having no parents). Therefore, whenever graft information is updated, mark the commits that were previously grafts and the commits that are newly grafts as unparsed. Signed-off-by: Jonathan Tan <jonathantanmy@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-06-06 19:54:37 +02:00
}
r->parsed_objects->grafts_nr = 0;
r->parsed_objects->commit_graft_prepared = 0;
}
struct commit_buffer {
void *buffer;
unsigned long size;
};
define_commit_slab(buffer_slab, struct commit_buffer);
struct buffer_slab *allocate_commit_buffer_slab(void)
{
struct buffer_slab *bs = xmalloc(sizeof(*bs));
init_buffer_slab(bs);
return bs;
}
void free_commit_buffer_slab(struct buffer_slab *bs)
{
clear_buffer_slab(bs);
free(bs);
}
void set_commit_buffer(struct repository *r, struct commit *commit, void *buffer, unsigned long size)
{
struct commit_buffer *v = buffer_slab_at(
r->parsed_objects->buffer_slab, commit);
v->buffer = buffer;
v->size = size;
}
const void *get_cached_commit_buffer(struct repository *r, const struct commit *commit, unsigned long *sizep)
{
struct commit_buffer *v = buffer_slab_peek(
r->parsed_objects->buffer_slab, commit);
if (!v) {
if (sizep)
*sizep = 0;
return NULL;
}
if (sizep)
*sizep = v->size;
return v->buffer;
}
const void *repo_get_commit_buffer(struct repository *r,
const struct commit *commit,
unsigned long *sizep)
{
const void *ret = get_cached_commit_buffer(r, commit, sizep);
if (!ret) {
enum object_type type;
unsigned long size;
ret = repo_read_object_file(r, &commit->object.oid, &type, &size);
if (!ret)
die("cannot read commit object %s",
oid_to_hex(&commit->object.oid));
if (type != OBJ_COMMIT)
die("expected commit for %s, got %s",
oid_to_hex(&commit->object.oid), type_name(type));
if (sizep)
*sizep = size;
}
return ret;
}
void repo_unuse_commit_buffer(struct repository *r,
const struct commit *commit,
const void *buffer)
{
struct commit_buffer *v = buffer_slab_peek(
r->parsed_objects->buffer_slab, commit);
if (!(v && v->buffer == buffer))
free((void *)buffer);
}
void free_commit_buffer(struct parsed_object_pool *pool, struct commit *commit)
{
struct commit_buffer *v = buffer_slab_peek(
pool->buffer_slab, commit);
if (v) {
FREE_AND_NULL(v->buffer);
v->size = 0;
}
}
static inline void set_commit_tree(struct commit *c, struct tree *t)
{
c->maybe_tree = t;
}
struct tree *repo_get_commit_tree(struct repository *r,
const struct commit *commit)
{
if (commit->maybe_tree || !commit->object.parsed)
return commit->maybe_tree;
if (commit_graph_position(commit) != COMMIT_NOT_FROM_GRAPH)
return get_commit_tree_in_graph(r, commit);
return NULL;
}
struct object_id *get_commit_tree_oid(const struct commit *commit)
{
struct tree *tree = get_commit_tree(commit);
return tree ? &tree->object.oid : NULL;
}
void release_commit_memory(struct parsed_object_pool *pool, struct commit *c)
{
set_commit_tree(c, NULL);
free_commit_buffer(pool, c);
c->index = 0;
free_commit_list(c->parents);
c->object.parsed = 0;
}
const void *detach_commit_buffer(struct commit *commit, unsigned long *sizep)
{
struct commit_buffer *v = buffer_slab_peek(
the_repository->parsed_objects->buffer_slab, commit);
void *ret;
if (!v) {
if (sizep)
*sizep = 0;
return NULL;
}
ret = v->buffer;
if (sizep)
*sizep = v->size;
v->buffer = NULL;
v->size = 0;
return ret;
}
int parse_commit_buffer(struct repository *r, struct commit *item, const void *buffer, unsigned long size, int check_graph)
{
const char *tail = buffer;
const char *bufptr = buffer;
struct object_id parent;
struct commit_list **pptr;
struct commit_graft *graft;
const int tree_entry_len = the_hash_algo->hexsz + 5;
const int parent_entry_len = the_hash_algo->hexsz + 7;
struct tree *tree;
if (item->object.parsed)
return 0;
/*
* Presumably this is leftover from an earlier failed parse;
* clear it out in preparation for us re-parsing (we'll hit the
* same error, but that's good, since it lets our caller know
* the result cannot be trusted.
*/
free_commit_list(item->parents);
item->parents = NULL;
commit, tag: don't set parsed bit for parse failures If we can't parse a commit, then parse_commit() will return an error code. But it _also_ sets the "parsed" flag, which tells us not to bother trying to re-parse the object. That means that subsequent parses have no idea that the information in the struct may be bogus. I.e., doing this: parse_commit(commit); ... if (parse_commit(commit) < 0) die("commit is broken"); will never trigger the die(). The second parse_commit() will see the "parsed" flag and quietly return success. There are two obvious ways to fix this: 1. Stop setting "parsed" until we've successfully parsed. 2. Keep a second "corrupt" flag to indicate that we saw an error (and when the parsed flag is set, return 0/-1 depending on the corrupt flag). This patch does option 1. The obvious downside versus option 2 is that we might continually re-parse a broken object. But in practice, corruption like this is rare, and we typically die() or return an error in the caller. So it's OK not to worry about optimizing for corruption. And it's much simpler: we don't need to use an extra bit in the object struct, and callers which check the "parsed" flag don't need to learn about the corrupt bit, too. There's no new test here, because this case is already covered in t5318. Note that we do need to update the expected message there, because we now detect the problem in the return from "parse_commit()", and not with a separate check for a NULL tree. In fact, we can now ditch that explicit tree check entirely, as we're covered robustly by this change (and the previous recent change to treat a NULL tree as a parse error). We'll also give tags the same treatment. I don't know offhand of any cases where the problem can be triggered (it implies somebody ignoring a parse error earlier in the process), but consistently returning an error should cause the least surprise. Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2019-10-25 23:20:20 +02:00
tail += size;
if (tail <= bufptr + tree_entry_len + 1 || memcmp(bufptr, "tree ", 5) ||
bufptr[tree_entry_len] != '\n')
return error("bogus commit object %s", oid_to_hex(&item->object.oid));
if (get_oid_hex(bufptr + 5, &parent) < 0)
return error("bad tree pointer in commit %s",
oid_to_hex(&item->object.oid));
tree = lookup_tree(r, &parent);
if (!tree)
return error("bad tree pointer %s in commit %s",
oid_to_hex(&parent),
oid_to_hex(&item->object.oid));
set_commit_tree(item, tree);
bufptr += tree_entry_len + 1; /* "tree " + "hex sha1" + "\n" */
pptr = &item->parents;
graft = lookup_commit_graft(r, &item->object.oid);
commit.c: don't persist substituted parents when unshallowing Since 37b9dcabfc (shallow.c: use '{commit,rollback}_shallow_file', 2020-04-22), Git knows how to reset stat-validity checks for the $GIT_DIR/shallow file, allowing it to change between a shallow and non-shallow state in the same process (e.g., in the case of 'git fetch --unshallow'). However, when $GIT_DIR/shallow changes, Git does not alter or remove any grafts (nor substituted parents) in memory. This comes up in a "git fetch --unshallow" with fetch.writeCommitGraph set to true. Ordinarily in a shallow repository (and before 37b9dcabfc, even in this case), commit_graph_compatible() would return false, indicating that the repository should not be used to write a commit-graphs (since commit-graph files cannot represent a shallow history). But since 37b9dcabfc, in an --unshallow operation that check succeeds. Thus even though the repository isn't shallow any longer (that is, we have all of the objects), the in-core representation of those objects still has munged parents at the shallow boundaries. When the commit-graph write proceeds, we use the incorrect parentage, producing wrong results. There are two ways for a user to work around this: either (1) set 'fetch.writeCommitGraph' to 'false', or (2) drop the commit-graph after unshallowing. One way to fix this would be to reset the parsed object pool entirely (flushing the cache and thus preventing subsequent reads from modifying their parents) after unshallowing. That would produce a problem when callers have a now-stale reference to the old pool, and so this patch implements a different approach. Instead, attach a new bit to the pool, 'substituted_parent', which indicates if the repository *ever* stored a commit which had its parents modified (i.e., the shallow boundary prior to unshallowing). This bit needs to be sticky because all reads subsequent to modifying a commit's parents are unreliable when unshallowing. Modify the check in 'commit_graph_compatible' to take this bit into account, and correctly avoid generating commit-graphs in this case, thus solving the bug. Helped-by: Derrick Stolee <dstolee@microsoft.com> Helped-by: Jonathan Nieder <jrnieder@gmail.com> Reported-by: Jay Conrod <jayconrod@google.com> Reviewed-by: Jonathan Nieder <jrnieder@gmail.com> Signed-off-by: Taylor Blau <me@ttaylorr.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2020-07-08 23:10:53 +02:00
if (graft)
r->parsed_objects->substituted_parent = 1;
while (bufptr + parent_entry_len < tail && !memcmp(bufptr, "parent ", 7)) {
struct commit *new_parent;
if (tail <= bufptr + parent_entry_len + 1 ||
get_oid_hex(bufptr + 7, &parent) ||
bufptr[parent_entry_len] != '\n')
return error("bad parents in commit %s", oid_to_hex(&item->object.oid));
bufptr += parent_entry_len + 1;
/*
* The clone is shallow if nr_parent < 0, and we must
* not traverse its real parents even when we unhide them.
*/
if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
continue;
new_parent = lookup_commit(r, &parent);
if (!new_parent)
return error("bad parent %s in commit %s",
oid_to_hex(&parent),
oid_to_hex(&item->object.oid));
pptr = &commit_list_insert(new_parent, pptr)->next;
}
if (graft) {
int i;
struct commit *new_parent;
for (i = 0; i < graft->nr_parent; i++) {
new_parent = lookup_commit(r,
&graft->parent[i]);
if (!new_parent)
return error("bad graft parent %s in commit %s",
oid_to_hex(&graft->parent[i]),
oid_to_hex(&item->object.oid));
pptr = &commit_list_insert(new_parent, pptr)->next;
}
}
item->date = parse_commit_date(bufptr, tail);
if (check_graph)
load_commit_graph_info(r, item);
commit, tag: don't set parsed bit for parse failures If we can't parse a commit, then parse_commit() will return an error code. But it _also_ sets the "parsed" flag, which tells us not to bother trying to re-parse the object. That means that subsequent parses have no idea that the information in the struct may be bogus. I.e., doing this: parse_commit(commit); ... if (parse_commit(commit) < 0) die("commit is broken"); will never trigger the die(). The second parse_commit() will see the "parsed" flag and quietly return success. There are two obvious ways to fix this: 1. Stop setting "parsed" until we've successfully parsed. 2. Keep a second "corrupt" flag to indicate that we saw an error (and when the parsed flag is set, return 0/-1 depending on the corrupt flag). This patch does option 1. The obvious downside versus option 2 is that we might continually re-parse a broken object. But in practice, corruption like this is rare, and we typically die() or return an error in the caller. So it's OK not to worry about optimizing for corruption. And it's much simpler: we don't need to use an extra bit in the object struct, and callers which check the "parsed" flag don't need to learn about the corrupt bit, too. There's no new test here, because this case is already covered in t5318. Note that we do need to update the expected message there, because we now detect the problem in the return from "parse_commit()", and not with a separate check for a NULL tree. In fact, we can now ditch that explicit tree check entirely, as we're covered robustly by this change (and the previous recent change to treat a NULL tree as a parse error). We'll also give tags the same treatment. I don't know offhand of any cases where the problem can be triggered (it implies somebody ignoring a parse error earlier in the process), but consistently returning an error should cause the least surprise. Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2019-10-25 23:20:20 +02:00
item->object.parsed = 1;
return 0;
}
int repo_parse_commit_internal(struct repository *r,
struct commit *item,
int quiet_on_missing,
int use_commit_graph)
{
enum object_type type;
void *buffer;
unsigned long size;
struct object_info oi = {
.typep = &type,
.sizep = &size,
.contentp = &buffer,
};
/*
* Git does not support partial clones that exclude commits, so set
* OBJECT_INFO_SKIP_FETCH_OBJECT to fail fast when an object is missing.
*/
int flags = OBJECT_INFO_LOOKUP_REPLACE | OBJECT_INFO_SKIP_FETCH_OBJECT |
OBJECT_INFO_DIE_IF_CORRUPT;
int ret;
if (!item)
return -1;
if (item->object.parsed)
return 0;
if (use_commit_graph && parse_commit_in_graph(r, item))
return 0;
if (oid_object_info_extended(r, &item->object.oid, &oi, flags) < 0)
return quiet_on_missing ? -1 :
error("Could not read %s",
oid_to_hex(&item->object.oid));
if (type != OBJ_COMMIT) {
free(buffer);
return error("Object %s not a commit",
oid_to_hex(&item->object.oid));
}
ret = parse_commit_buffer(r, item, buffer, size, 0);
[PATCH] Avoid wasting memory in git-rev-list As pointed out on the list, git-rev-list can use a lot of memory. One low-hanging fruit is to free the commit buffer for commits that we parse. By default, parse_commit() will save away the buffer, since a lot of cases do want it, and re-reading it continually would be unnecessary. However, in many cases the buffer isn't actually necessary and saving it just wastes memory. We could just free the buffer ourselves, but especially in git-rev-list, we actually end up using the helper functions that automatically add parent commits to the commit lists, so we don't actually control the commit parsing directly. Instead, just make this behaviour of "parse_commit()" a global flag. Maybe this is a bit tasteless, but it's very simple, and it makes a noticable difference in memory usage. Before the change: [torvalds@g5 linux]$ /usr/bin/time git-rev-list v2.6.12..HEAD > /dev/null 0.26user 0.02system 0:00.28elapsed 99%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+3714minor)pagefaults 0swaps after the change: [torvalds@g5 linux]$ /usr/bin/time git-rev-list v2.6.12..HEAD > /dev/null 0.26user 0.00system 0:00.27elapsed 100%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+2433minor)pagefaults 0swaps note how the minor faults have decreased from 3714 pages to 2433 pages. That's all due to the fewer anonymous pages allocated to hold the comment buffers and their metadata. Signed-off-by: Linus Torvalds <torvalds@osdl.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
2005-09-15 23:43:17 +02:00
if (save_commit_buffer && !ret) {
set_commit_buffer(r, item, buffer, size);
return 0;
}
free(buffer);
return ret;
}
int repo_parse_commit_gently(struct repository *r,
struct commit *item, int quiet_on_missing)
{
return repo_parse_commit_internal(r, item, quiet_on_missing, 1);
}
void parse_commit_or_die(struct commit *item)
{
if (parse_commit(item))
die("unable to parse commit %s",
item ? oid_to_hex(&item->object.oid) : "(null)");
}
int find_commit_subject(const char *commit_buffer, const char **subject)
{
const char *eol;
const char *p = commit_buffer;
while (*p && (*p != '\n' || p[1] != '\n'))
p++;
if (*p) {
p = skip_blank_lines(p + 2);
eol = strchrnul(p, '\n');
} else
eol = p;
*subject = p;
return eol - p;
}
size_t commit_subject_length(const char *body)
{
const char *p = body;
while (*p) {
const char *next = skip_blank_lines(p);
if (next != p)
break;
p = strchrnul(p, '\n');
if (*p)
p++;
}
return p - body;
}
struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
{
struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
new_list->item = item;
new_list->next = *list_p;
*list_p = new_list;
return new_list;
}
int commit_list_contains(struct commit *item, struct commit_list *list)
{
while (list) {
if (list->item == item)
return 1;
list = list->next;
}
return 0;
}
unsigned commit_list_count(const struct commit_list *l)
{
unsigned c = 0;
for (; l; l = l->next )
c++;
return c;
}
log: use true parents for diff even when rewriting When using pathspec filtering in combination with diff-based log output, parent simplification happens before the diff is computed. The diff is therefore against the *simplified* parents. This works okay, arguably by accident, in the normal case: simplification reduces to one parent as long as the commit is TREESAME to it. So the simplified parent of any given commit must have the same tree contents on the filtered paths as its true (unfiltered) parent. However, --full-diff breaks this guarantee, and indeed gives pretty spectacular results when comparing the output of git log --graph --stat ... git log --graph --full-diff --stat ... (--graph internally kicks in parent simplification, much like --parents). To fix it, store a copy of the parent list before simplification (in a slab) whenever --full-diff is in effect. Then use the stored parents instead of the simplified ones in the commit display code paths. The latter do not actually check for --full-diff to avoid duplicated code; they just grab the original parents if save_parents() has not been called for this revision walk. For ordinary commits it should be obvious that this is the right thing to do. Merge commits are a bit subtle. Observe that with default simplification, merge simplification is an all-or-nothing decision: either the merge is TREESAME to one parent and disappears, or it is different from all parents and the parent list remains intact. Redundant parents are not pruned, so the existing code also shows them as a merge. So if we do show a merge commit, the parent list just consists of the rewrite result on each parent. Running, e.g., --cc on this in --full-diff mode is not very useful: if any commits were skipped, some hunks will disagree with all sides of the merge (with one side, because commits were skipped; with the others, because they didn't have those changes in the first place). This triggers --cc showing these hunks spuriously. Therefore I believe that even for merge commits it is better to show the diffs wrt. the original parents. Reported-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de> Helped-by: Junio C Hamano <gitster@pobox.com> Helped-by: Ramsay Jones <ramsay@ramsay1.demon.co.uk> Signed-off-by: Thomas Rast <trast@inf.ethz.ch> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-07-31 22:13:20 +02:00
struct commit_list *copy_commit_list(struct commit_list *list)
{
struct commit_list *head = NULL;
struct commit_list **pp = &head;
while (list) {
pp = commit_list_append(list->item, pp);
log: use true parents for diff even when rewriting When using pathspec filtering in combination with diff-based log output, parent simplification happens before the diff is computed. The diff is therefore against the *simplified* parents. This works okay, arguably by accident, in the normal case: simplification reduces to one parent as long as the commit is TREESAME to it. So the simplified parent of any given commit must have the same tree contents on the filtered paths as its true (unfiltered) parent. However, --full-diff breaks this guarantee, and indeed gives pretty spectacular results when comparing the output of git log --graph --stat ... git log --graph --full-diff --stat ... (--graph internally kicks in parent simplification, much like --parents). To fix it, store a copy of the parent list before simplification (in a slab) whenever --full-diff is in effect. Then use the stored parents instead of the simplified ones in the commit display code paths. The latter do not actually check for --full-diff to avoid duplicated code; they just grab the original parents if save_parents() has not been called for this revision walk. For ordinary commits it should be obvious that this is the right thing to do. Merge commits are a bit subtle. Observe that with default simplification, merge simplification is an all-or-nothing decision: either the merge is TREESAME to one parent and disappears, or it is different from all parents and the parent list remains intact. Redundant parents are not pruned, so the existing code also shows them as a merge. So if we do show a merge commit, the parent list just consists of the rewrite result on each parent. Running, e.g., --cc on this in --full-diff mode is not very useful: if any commits were skipped, some hunks will disagree with all sides of the merge (with one side, because commits were skipped; with the others, because they didn't have those changes in the first place). This triggers --cc showing these hunks spuriously. Therefore I believe that even for merge commits it is better to show the diffs wrt. the original parents. Reported-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de> Helped-by: Junio C Hamano <gitster@pobox.com> Helped-by: Ramsay Jones <ramsay@ramsay1.demon.co.uk> Signed-off-by: Thomas Rast <trast@inf.ethz.ch> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-07-31 22:13:20 +02:00
list = list->next;
}
return head;
}
struct commit_list *reverse_commit_list(struct commit_list *list)
{
struct commit_list *next = NULL, *current, *backup;
for (current = list; current; current = backup) {
backup = current->next;
current->next = next;
next = current;
}
return next;
}
void free_commit_list(struct commit_list *list)
{
while (list)
pop_commit(&list);
}
struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
{
struct commit_list **pp = list;
struct commit_list *p;
while ((p = *pp) != NULL) {
if (p->item->date < item->date) {
break;
}
pp = &p->next;
}
return commit_list_insert(item, pp);
}
static int commit_list_compare_by_date(const struct commit_list *a,
const struct commit_list *b)
{
timestamp_t a_date = a->item->date;
timestamp_t b_date = b->item->date;
if (a_date < b_date)
return 1;
if (a_date > b_date)
return -1;
return 0;
}
DEFINE_LIST_SORT(static, commit_list_sort, struct commit_list, next);
void commit_list_sort_by_date(struct commit_list **list)
{
commit_list_sort(list, commit_list_compare_by_date);
}
struct commit *pop_most_recent_commit(struct commit_list **list,
unsigned int mark)
{
struct commit *ret = pop_commit(list);
struct commit_list *parents = ret->parents;
while (parents) {
struct commit *commit = parents->item;
if (!parse_commit(commit) && !(commit->object.flags & mark)) {
commit->object.flags |= mark;
commit_list_insert_by_date(commit, list);
}
parents = parents->next;
}
return ret;
}
static void clear_commit_marks_1(struct commit_list **plist,
struct commit *commit, unsigned int mark)
{
while (commit) {
struct commit_list *parents;
if (!(mark & commit->object.flags))
return;
commit->object.flags &= ~mark;
parents = commit->parents;
if (!parents)
return;
while ((parents = parents->next)) {
if (parents->item->object.flags & mark)
commit_list_insert(parents->item, plist);
}
commit = commit->parents->item;
}
}
void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
{
struct commit_list *list = NULL;
while (nr--) {
clear_commit_marks_1(&list, *commit, mark);
commit++;
}
while (list)
clear_commit_marks_1(&list, pop_commit(&list), mark);
}
void clear_commit_marks(struct commit *commit, unsigned int mark)
{
clear_commit_marks_many(1, &commit, mark);
}
[PATCH] Modify git-rev-list to linearise the commit history in merge order. This patch linearises the GIT commit history graph into merge order which is defined by invariants specified in Documentation/git-rev-list.txt. The linearisation produced by this patch is superior in an objective sense to that produced by the existing git-rev-list implementation in that the linearisation produced is guaranteed to have the minimum number of discontinuities, where a discontinuity is defined as an adjacent pair of commits in the output list which are not related in a direct child-parent relationship. With this patch a graph like this: a4 --- | \ \ | b4 | |/ | | a3 | | | | | a2 | | | | c3 | | | | | c2 | b3 | | | /| | b2 | | | c1 | | / | b1 a1 | | | a0 | | / root Sorts like this: = a4 | c3 | c2 | c1 ^ b4 | b3 | b2 | b1 ^ a3 | a2 | a1 | a0 = root Instead of this: = a4 | c3 ^ b4 | a3 ^ c2 ^ b3 ^ a2 ^ b2 ^ c1 ^ a1 ^ b1 ^ a0 = root A test script, t/t6000-rev-list.sh, includes a test which demonstrates that the linearisation produced by --merge-order has less discontinuities than the linearisation produced by git-rev-list without the --merge-order flag specified. To see this, do the following: cd t ./t6000-rev-list.sh cd trash cat actual-default-order cat actual-merge-order The existing behaviour of git-rev-list is preserved, by default. To obtain the modified behaviour, specify --merge-order or --merge-order --show-breaks on the command line. This version of the patch has been tested on the git repository and also on the linux-2.6 repository and has reasonable performance on both - ~50-100% slower than the original algorithm. This version of the patch has incorporated a functional equivalent of the Linus' output limiting algorithm into the merge-order algorithm itself. This operates per the notes associated with Linus' commit 337cb3fb8da45f10fe9a0c3cf571600f55ead2ce. This version has incorporated Linus' feedback regarding proposed changes to rev-list.c. (see: [PATCH] Factor out filtering in rev-list.c) This version has improved the way sort_first_epoch marks commits as uninteresting. For more details about this change, refer to Documentation/git-rev-list.txt and http://blackcubes.dyndns.org/epoch/. Signed-off-by: Jon Seymour <jon.seymour@gmail.com> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-06 17:39:40 +02:00
struct commit *pop_commit(struct commit_list **stack)
{
struct commit_list *top = *stack;
struct commit *item = top ? top->item : NULL;
if (top) {
*stack = top->next;
free(top);
}
return item;
}
commit-slab: introduce a macro to define a slab for new type Introduce a header file to define a macro that can define the struct type, initializer, accessor and cleanup functions to manage a commit slab. Update the "indegree" topological sort facility using it. To associate 32 flag bits with each commit, you can write: define_commit_slab(flag32, uint32); to declare "struct flag32" type, define an instance of it with struct flag32 flags; and initialize it by calling init_flag32(&flags); After that, a call to flag32_at() function uint32 *fp = flag32_at(&flags, commit); will return a pointer pointing at a uint32 for that commit. Once you are done with these flags, clean them up with clear_flag32(&flags); Callers that cannot hard-code how wide the data to be associated with the commit be at compile time can use the "_with_stride" variant to initialize the slab. Suppose you want to give one bit per existing ref, and paint commits down to find which refs are descendants of each commit. Saying typedef uint32 bits320[5]; define_commit_slab(flagbits, bits320); at compile time will still limit your code with hard-coded limit, because you may find that you have more than 320 refs at runtime. The code can declare a commit slab "struct flagbits" like this instead: define_commit_slab(flagbits, unsigned char); struct flagbits flags; and initialize it by: nrefs = ... count number of refs ... init_flagbits_with_stride(&flags, (nrefs + 7) / 8); so that unsigned char *fp = flagbits_at(&flags, commit); will return a pointer pointing at an array of 40 "unsigned char"s associated with the commit, once you figure out nrefs is 320 at runtime. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-04-13 20:56:41 +02:00
/*
* Topological sort support
*/
commit-slab: introduce a macro to define a slab for new type Introduce a header file to define a macro that can define the struct type, initializer, accessor and cleanup functions to manage a commit slab. Update the "indegree" topological sort facility using it. To associate 32 flag bits with each commit, you can write: define_commit_slab(flag32, uint32); to declare "struct flag32" type, define an instance of it with struct flag32 flags; and initialize it by calling init_flag32(&flags); After that, a call to flag32_at() function uint32 *fp = flag32_at(&flags, commit); will return a pointer pointing at a uint32 for that commit. Once you are done with these flags, clean them up with clear_flag32(&flags); Callers that cannot hard-code how wide the data to be associated with the commit be at compile time can use the "_with_stride" variant to initialize the slab. Suppose you want to give one bit per existing ref, and paint commits down to find which refs are descendants of each commit. Saying typedef uint32 bits320[5]; define_commit_slab(flagbits, bits320); at compile time will still limit your code with hard-coded limit, because you may find that you have more than 320 refs at runtime. The code can declare a commit slab "struct flagbits" like this instead: define_commit_slab(flagbits, unsigned char); struct flagbits flags; and initialize it by: nrefs = ... count number of refs ... init_flagbits_with_stride(&flags, (nrefs + 7) / 8); so that unsigned char *fp = flagbits_at(&flags, commit); will return a pointer pointing at an array of 40 "unsigned char"s associated with the commit, once you figure out nrefs is 320 at runtime. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-04-13 20:56:41 +02:00
/* count number of children that have not been emitted */
define_commit_slab(indegree_slab, int);
define_commit_slab(author_date_slab, timestamp_t);
void record_author_date(struct author_date_slab *author_date,
struct commit *commit)
{
const char *buffer = get_commit_buffer(commit, NULL);
struct ident_split ident;
const char *ident_line;
size_t ident_len;
char *date_end;
timestamp_t date;
ident_line = find_commit_header(buffer, "author", &ident_len);
if (!ident_line)
goto fail_exit; /* no author line */
if (split_ident_line(&ident, ident_line, ident_len) ||
!ident.date_begin || !ident.date_end)
goto fail_exit; /* malformed "author" line */
date = parse_timestamp(ident.date_begin, &date_end, 10);
if (date_end != ident.date_end)
goto fail_exit; /* malformed date */
*(author_date_slab_at(author_date, commit)) = date;
fail_exit:
unuse_commit_buffer(commit, buffer);
}
int compare_commits_by_author_date(const void *a_, const void *b_,
void *cb_data)
{
const struct commit *a = a_, *b = b_;
struct author_date_slab *author_date = cb_data;
timestamp_t a_date = *(author_date_slab_at(author_date, a));
timestamp_t b_date = *(author_date_slab_at(author_date, b));
/* newer commits with larger date first */
if (a_date < b_date)
return 1;
else if (a_date > b_date)
return -1;
return 0;
}
int compare_commits_by_gen_then_commit_date(const void *a_, const void *b_,
void *unused UNUSED)
{
const struct commit *a = a_, *b = b_;
const timestamp_t generation_a = commit_graph_generation(a),
generation_b = commit_graph_generation(b);
/* newer commits first */
if (generation_a < generation_b)
return 1;
else if (generation_a > generation_b)
return -1;
/* use date as a heuristic when generations are equal */
if (a->date < b->date)
return 1;
else if (a->date > b->date)
return -1;
return 0;
}
int compare_commits_by_commit_date(const void *a_, const void *b_,
void *unused UNUSED)
{
const struct commit *a = a_, *b = b_;
/* newer commits with larger date first */
if (a->date < b->date)
return 1;
else if (a->date > b->date)
return -1;
return 0;
}
/*
* Performs an in-place topological sort on the list supplied.
*/
void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
{
struct commit_list *next, *orig = *list;
struct commit_list **pptr;
commit-slab: introduce a macro to define a slab for new type Introduce a header file to define a macro that can define the struct type, initializer, accessor and cleanup functions to manage a commit slab. Update the "indegree" topological sort facility using it. To associate 32 flag bits with each commit, you can write: define_commit_slab(flag32, uint32); to declare "struct flag32" type, define an instance of it with struct flag32 flags; and initialize it by calling init_flag32(&flags); After that, a call to flag32_at() function uint32 *fp = flag32_at(&flags, commit); will return a pointer pointing at a uint32 for that commit. Once you are done with these flags, clean them up with clear_flag32(&flags); Callers that cannot hard-code how wide the data to be associated with the commit be at compile time can use the "_with_stride" variant to initialize the slab. Suppose you want to give one bit per existing ref, and paint commits down to find which refs are descendants of each commit. Saying typedef uint32 bits320[5]; define_commit_slab(flagbits, bits320); at compile time will still limit your code with hard-coded limit, because you may find that you have more than 320 refs at runtime. The code can declare a commit slab "struct flagbits" like this instead: define_commit_slab(flagbits, unsigned char); struct flagbits flags; and initialize it by: nrefs = ... count number of refs ... init_flagbits_with_stride(&flags, (nrefs + 7) / 8); so that unsigned char *fp = flagbits_at(&flags, commit); will return a pointer pointing at an array of 40 "unsigned char"s associated with the commit, once you figure out nrefs is 320 at runtime. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-04-13 20:56:41 +02:00
struct indegree_slab indegree;
struct prio_queue queue;
struct commit *commit;
struct author_date_slab author_date;
if (!orig)
return;
*list = NULL;
commit-slab: introduce a macro to define a slab for new type Introduce a header file to define a macro that can define the struct type, initializer, accessor and cleanup functions to manage a commit slab. Update the "indegree" topological sort facility using it. To associate 32 flag bits with each commit, you can write: define_commit_slab(flag32, uint32); to declare "struct flag32" type, define an instance of it with struct flag32 flags; and initialize it by calling init_flag32(&flags); After that, a call to flag32_at() function uint32 *fp = flag32_at(&flags, commit); will return a pointer pointing at a uint32 for that commit. Once you are done with these flags, clean them up with clear_flag32(&flags); Callers that cannot hard-code how wide the data to be associated with the commit be at compile time can use the "_with_stride" variant to initialize the slab. Suppose you want to give one bit per existing ref, and paint commits down to find which refs are descendants of each commit. Saying typedef uint32 bits320[5]; define_commit_slab(flagbits, bits320); at compile time will still limit your code with hard-coded limit, because you may find that you have more than 320 refs at runtime. The code can declare a commit slab "struct flagbits" like this instead: define_commit_slab(flagbits, unsigned char); struct flagbits flags; and initialize it by: nrefs = ... count number of refs ... init_flagbits_with_stride(&flags, (nrefs + 7) / 8); so that unsigned char *fp = flagbits_at(&flags, commit); will return a pointer pointing at an array of 40 "unsigned char"s associated with the commit, once you figure out nrefs is 320 at runtime. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-04-13 20:56:41 +02:00
init_indegree_slab(&indegree);
memset(&queue, '\0', sizeof(queue));
switch (sort_order) {
default: /* REV_SORT_IN_GRAPH_ORDER */
queue.compare = NULL;
break;
case REV_SORT_BY_COMMIT_DATE:
queue.compare = compare_commits_by_commit_date;
break;
case REV_SORT_BY_AUTHOR_DATE:
init_author_date_slab(&author_date);
queue.compare = compare_commits_by_author_date;
queue.cb_data = &author_date;
break;
}
/* Mark them and clear the indegree */
for (next = orig; next; next = next->next) {
struct commit *commit = next->item;
commit-slab: introduce a macro to define a slab for new type Introduce a header file to define a macro that can define the struct type, initializer, accessor and cleanup functions to manage a commit slab. Update the "indegree" topological sort facility using it. To associate 32 flag bits with each commit, you can write: define_commit_slab(flag32, uint32); to declare "struct flag32" type, define an instance of it with struct flag32 flags; and initialize it by calling init_flag32(&flags); After that, a call to flag32_at() function uint32 *fp = flag32_at(&flags, commit); will return a pointer pointing at a uint32 for that commit. Once you are done with these flags, clean them up with clear_flag32(&flags); Callers that cannot hard-code how wide the data to be associated with the commit be at compile time can use the "_with_stride" variant to initialize the slab. Suppose you want to give one bit per existing ref, and paint commits down to find which refs are descendants of each commit. Saying typedef uint32 bits320[5]; define_commit_slab(flagbits, bits320); at compile time will still limit your code with hard-coded limit, because you may find that you have more than 320 refs at runtime. The code can declare a commit slab "struct flagbits" like this instead: define_commit_slab(flagbits, unsigned char); struct flagbits flags; and initialize it by: nrefs = ... count number of refs ... init_flagbits_with_stride(&flags, (nrefs + 7) / 8); so that unsigned char *fp = flagbits_at(&flags, commit); will return a pointer pointing at an array of 40 "unsigned char"s associated with the commit, once you figure out nrefs is 320 at runtime. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-04-13 20:56:41 +02:00
*(indegree_slab_at(&indegree, commit)) = 1;
/* also record the author dates, if needed */
if (sort_order == REV_SORT_BY_AUTHOR_DATE)
record_author_date(&author_date, commit);
}
/* update the indegree */
for (next = orig; next; next = next->next) {
struct commit_list *parents = next->item->parents;
while (parents) {
struct commit *parent = parents->item;
commit-slab: introduce a macro to define a slab for new type Introduce a header file to define a macro that can define the struct type, initializer, accessor and cleanup functions to manage a commit slab. Update the "indegree" topological sort facility using it. To associate 32 flag bits with each commit, you can write: define_commit_slab(flag32, uint32); to declare "struct flag32" type, define an instance of it with struct flag32 flags; and initialize it by calling init_flag32(&flags); After that, a call to flag32_at() function uint32 *fp = flag32_at(&flags, commit); will return a pointer pointing at a uint32 for that commit. Once you are done with these flags, clean them up with clear_flag32(&flags); Callers that cannot hard-code how wide the data to be associated with the commit be at compile time can use the "_with_stride" variant to initialize the slab. Suppose you want to give one bit per existing ref, and paint commits down to find which refs are descendants of each commit. Saying typedef uint32 bits320[5]; define_commit_slab(flagbits, bits320); at compile time will still limit your code with hard-coded limit, because you may find that you have more than 320 refs at runtime. The code can declare a commit slab "struct flagbits" like this instead: define_commit_slab(flagbits, unsigned char); struct flagbits flags; and initialize it by: nrefs = ... count number of refs ... init_flagbits_with_stride(&flags, (nrefs + 7) / 8); so that unsigned char *fp = flagbits_at(&flags, commit); will return a pointer pointing at an array of 40 "unsigned char"s associated with the commit, once you figure out nrefs is 320 at runtime. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-04-13 20:56:41 +02:00
int *pi = indegree_slab_at(&indegree, parent);
if (*pi)
(*pi)++;
parents = parents->next;
}
}
/*
* find the tips
*
* tips are nodes not reachable from any other node in the list
*
* the tips serve as a starting set for the work queue.
*/
for (next = orig; next; next = next->next) {
struct commit *commit = next->item;
commit-slab: introduce a macro to define a slab for new type Introduce a header file to define a macro that can define the struct type, initializer, accessor and cleanup functions to manage a commit slab. Update the "indegree" topological sort facility using it. To associate 32 flag bits with each commit, you can write: define_commit_slab(flag32, uint32); to declare "struct flag32" type, define an instance of it with struct flag32 flags; and initialize it by calling init_flag32(&flags); After that, a call to flag32_at() function uint32 *fp = flag32_at(&flags, commit); will return a pointer pointing at a uint32 for that commit. Once you are done with these flags, clean them up with clear_flag32(&flags); Callers that cannot hard-code how wide the data to be associated with the commit be at compile time can use the "_with_stride" variant to initialize the slab. Suppose you want to give one bit per existing ref, and paint commits down to find which refs are descendants of each commit. Saying typedef uint32 bits320[5]; define_commit_slab(flagbits, bits320); at compile time will still limit your code with hard-coded limit, because you may find that you have more than 320 refs at runtime. The code can declare a commit slab "struct flagbits" like this instead: define_commit_slab(flagbits, unsigned char); struct flagbits flags; and initialize it by: nrefs = ... count number of refs ... init_flagbits_with_stride(&flags, (nrefs + 7) / 8); so that unsigned char *fp = flagbits_at(&flags, commit); will return a pointer pointing at an array of 40 "unsigned char"s associated with the commit, once you figure out nrefs is 320 at runtime. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-04-13 20:56:41 +02:00
if (*(indegree_slab_at(&indegree, commit)) == 1)
prio_queue_put(&queue, commit);
}
/*
* This is unfortunate; the initial tips need to be shown
* in the order given from the revision traversal machinery.
*/
if (sort_order == REV_SORT_IN_GRAPH_ORDER)
prio_queue_reverse(&queue);
/* We no longer need the commit list */
free_commit_list(orig);
pptr = list;
*list = NULL;
while ((commit = prio_queue_get(&queue)) != NULL) {
struct commit_list *parents;
for (parents = commit->parents; parents ; parents = parents->next) {
struct commit *parent = parents->item;
commit-slab: introduce a macro to define a slab for new type Introduce a header file to define a macro that can define the struct type, initializer, accessor and cleanup functions to manage a commit slab. Update the "indegree" topological sort facility using it. To associate 32 flag bits with each commit, you can write: define_commit_slab(flag32, uint32); to declare "struct flag32" type, define an instance of it with struct flag32 flags; and initialize it by calling init_flag32(&flags); After that, a call to flag32_at() function uint32 *fp = flag32_at(&flags, commit); will return a pointer pointing at a uint32 for that commit. Once you are done with these flags, clean them up with clear_flag32(&flags); Callers that cannot hard-code how wide the data to be associated with the commit be at compile time can use the "_with_stride" variant to initialize the slab. Suppose you want to give one bit per existing ref, and paint commits down to find which refs are descendants of each commit. Saying typedef uint32 bits320[5]; define_commit_slab(flagbits, bits320); at compile time will still limit your code with hard-coded limit, because you may find that you have more than 320 refs at runtime. The code can declare a commit slab "struct flagbits" like this instead: define_commit_slab(flagbits, unsigned char); struct flagbits flags; and initialize it by: nrefs = ... count number of refs ... init_flagbits_with_stride(&flags, (nrefs + 7) / 8); so that unsigned char *fp = flagbits_at(&flags, commit); will return a pointer pointing at an array of 40 "unsigned char"s associated with the commit, once you figure out nrefs is 320 at runtime. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-04-13 20:56:41 +02:00
int *pi = indegree_slab_at(&indegree, parent);
if (!*pi)
continue;
/*
* parents are only enqueued for emission
* when all their children have been emitted thereby
* guaranteeing topological order.
*/
if (--(*pi) == 1)
prio_queue_put(&queue, parent);
}
/*
* all children of commit have already been
* emitted. we can emit it now.
*/
commit-slab: introduce a macro to define a slab for new type Introduce a header file to define a macro that can define the struct type, initializer, accessor and cleanup functions to manage a commit slab. Update the "indegree" topological sort facility using it. To associate 32 flag bits with each commit, you can write: define_commit_slab(flag32, uint32); to declare "struct flag32" type, define an instance of it with struct flag32 flags; and initialize it by calling init_flag32(&flags); After that, a call to flag32_at() function uint32 *fp = flag32_at(&flags, commit); will return a pointer pointing at a uint32 for that commit. Once you are done with these flags, clean them up with clear_flag32(&flags); Callers that cannot hard-code how wide the data to be associated with the commit be at compile time can use the "_with_stride" variant to initialize the slab. Suppose you want to give one bit per existing ref, and paint commits down to find which refs are descendants of each commit. Saying typedef uint32 bits320[5]; define_commit_slab(flagbits, bits320); at compile time will still limit your code with hard-coded limit, because you may find that you have more than 320 refs at runtime. The code can declare a commit slab "struct flagbits" like this instead: define_commit_slab(flagbits, unsigned char); struct flagbits flags; and initialize it by: nrefs = ... count number of refs ... init_flagbits_with_stride(&flags, (nrefs + 7) / 8); so that unsigned char *fp = flagbits_at(&flags, commit); will return a pointer pointing at an array of 40 "unsigned char"s associated with the commit, once you figure out nrefs is 320 at runtime. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-04-13 20:56:41 +02:00
*(indegree_slab_at(&indegree, commit)) = 0;
pptr = &commit_list_insert(commit, pptr)->next;
}
commit-slab: introduce a macro to define a slab for new type Introduce a header file to define a macro that can define the struct type, initializer, accessor and cleanup functions to manage a commit slab. Update the "indegree" topological sort facility using it. To associate 32 flag bits with each commit, you can write: define_commit_slab(flag32, uint32); to declare "struct flag32" type, define an instance of it with struct flag32 flags; and initialize it by calling init_flag32(&flags); After that, a call to flag32_at() function uint32 *fp = flag32_at(&flags, commit); will return a pointer pointing at a uint32 for that commit. Once you are done with these flags, clean them up with clear_flag32(&flags); Callers that cannot hard-code how wide the data to be associated with the commit be at compile time can use the "_with_stride" variant to initialize the slab. Suppose you want to give one bit per existing ref, and paint commits down to find which refs are descendants of each commit. Saying typedef uint32 bits320[5]; define_commit_slab(flagbits, bits320); at compile time will still limit your code with hard-coded limit, because you may find that you have more than 320 refs at runtime. The code can declare a commit slab "struct flagbits" like this instead: define_commit_slab(flagbits, unsigned char); struct flagbits flags; and initialize it by: nrefs = ... count number of refs ... init_flagbits_with_stride(&flags, (nrefs + 7) / 8); so that unsigned char *fp = flagbits_at(&flags, commit); will return a pointer pointing at an array of 40 "unsigned char"s associated with the commit, once you figure out nrefs is 320 at runtime. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-04-13 20:56:41 +02:00
clear_indegree_slab(&indegree);
clear_prio_queue(&queue);
if (sort_order == REV_SORT_BY_AUTHOR_DATE)
clear_author_date_slab(&author_date);
}
struct rev_collect {
struct commit **commit;
int nr;
int alloc;
unsigned int initial : 1;
};
static void add_one_commit(struct object_id *oid, struct rev_collect *revs)
{
struct commit *commit;
if (is_null_oid(oid))
return;
commit = lookup_commit(the_repository, oid);
if (!commit ||
(commit->object.flags & TMP_MARK) ||
parse_commit(commit))
return;
ALLOC_GROW(revs->commit, revs->nr + 1, revs->alloc);
revs->commit[revs->nr++] = commit;
commit->object.flags |= TMP_MARK;
}
static int collect_one_reflog_ent(struct object_id *ooid, struct object_id *noid,
const char *ident UNUSED,
timestamp_t timestamp UNUSED, int tz UNUSED,
const char *message UNUSED, void *cbdata)
{
struct rev_collect *revs = cbdata;
if (revs->initial) {
revs->initial = 0;
add_one_commit(ooid, revs);
}
add_one_commit(noid, revs);
return 0;
}
struct commit *get_fork_point(const char *refname, struct commit *commit)
{
struct object_id oid;
struct rev_collect revs;
struct commit_list *bases;
int i;
struct commit *ret = NULL;
char *full_refname;
switch (dwim_ref(refname, strlen(refname), &oid, &full_refname, 0)) {
case 0:
die("No such ref: '%s'", refname);
case 1:
break; /* good */
default:
die("Ambiguous refname: '%s'", refname);
}
memset(&revs, 0, sizeof(revs));
revs.initial = 1;
for_each_reflog_ent(full_refname, collect_one_reflog_ent, &revs);
if (!revs.nr)
add_one_commit(&oid, &revs);
for (i = 0; i < revs.nr; i++)
revs.commit[i]->object.flags &= ~TMP_MARK;
bases = get_merge_bases_many(commit, revs.nr, revs.commit);
/*
* There should be one and only one merge base, when we found
* a common ancestor among reflog entries.
*/
if (!bases || bases->next)
goto cleanup_return;
/* And the found one must be one of the reflog entries */
for (i = 0; i < revs.nr; i++)
if (&bases->item->object == &revs.commit[i]->object)
break; /* found */
if (revs.nr <= i)
goto cleanup_return;
ret = bases->item;
cleanup_return:
free(revs.commit);
free_commit_list(bases);
free(full_refname);
return ret;
}
/*
* Indexed by hash algorithm identifier.
*/
static const char *gpg_sig_headers[] = {
NULL,
"gpgsig",
"gpgsig-sha256",
};
commit: teach --gpg-sign option This uses the gpg-interface.[ch] to allow signing the commit, i.e. $ git commit --gpg-sign -m foo You need a passphrase to unlock the secret key for user: "Junio C Hamano <gitster@pobox.com>" 4096-bit RSA key, ID 96AFE6CB, created 2011-10-03 (main key ID 713660A7) [master 8457d13] foo 1 files changed, 1 insertions(+), 0 deletions(-) The lines of GPG detached signature are placed in a new multi-line header field, instead of tucking the signature block at the end of the commit log message text (similar to how signed tag is done), for multiple reasons: - The signature won't clutter output from "git log" and friends if it is in the extra header. If we place it at the end of the log message, we would need to teach "git log" and friends to strip the signature block with an option. - Teaching new versions of "git log" and "gitk" to optionally verify and show signatures is cleaner if we structurally know where the signature block is (instead of scanning in the commit log message). - The signature needs to be stripped upon various commit rewriting operations, e.g. rebase, filter-branch, etc. They all already ignore unknown headers, but if we place signature in the log message, all of these tools (and third-party tools) also need to learn how a signature block would look like. - When we added the optional encoding header, all the tools (both in tree and third-party) that acts on the raw commit object should have been fixed to ignore headers they do not understand, so it is not like that new header would be more likely to break than extra text in the commit. A commit made with the above sample sequence would look like this: $ git cat-file commit HEAD tree 3cd71d90e3db4136e5260ab54599791c4f883b9d parent b87755351a47b09cb27d6913e6e0e17e6254a4d4 author Junio C Hamano <gitster@pobox.com> 1317862251 -0700 committer Junio C Hamano <gitster@pobox.com> 1317862251 -0700 gpgsig -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.10 (GNU/Linux) iQIcBAABAgAGBQJOjPtrAAoJELC16IaWr+bL4TMP/RSe2Y/jYnCkds9unO5JEnfG ... =dt98 -----END PGP SIGNATURE----- foo but "git log" (unless you ask for it with --pretty=raw) output is not cluttered with the signature information. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2011-10-06 02:23:20 +02:00
int sign_with_header(struct strbuf *buf, const char *keyid)
commit: teach --gpg-sign option This uses the gpg-interface.[ch] to allow signing the commit, i.e. $ git commit --gpg-sign -m foo You need a passphrase to unlock the secret key for user: "Junio C Hamano <gitster@pobox.com>" 4096-bit RSA key, ID 96AFE6CB, created 2011-10-03 (main key ID 713660A7) [master 8457d13] foo 1 files changed, 1 insertions(+), 0 deletions(-) The lines of GPG detached signature are placed in a new multi-line header field, instead of tucking the signature block at the end of the commit log message text (similar to how signed tag is done), for multiple reasons: - The signature won't clutter output from "git log" and friends if it is in the extra header. If we place it at the end of the log message, we would need to teach "git log" and friends to strip the signature block with an option. - Teaching new versions of "git log" and "gitk" to optionally verify and show signatures is cleaner if we structurally know where the signature block is (instead of scanning in the commit log message). - The signature needs to be stripped upon various commit rewriting operations, e.g. rebase, filter-branch, etc. They all already ignore unknown headers, but if we place signature in the log message, all of these tools (and third-party tools) also need to learn how a signature block would look like. - When we added the optional encoding header, all the tools (both in tree and third-party) that acts on the raw commit object should have been fixed to ignore headers they do not understand, so it is not like that new header would be more likely to break than extra text in the commit. A commit made with the above sample sequence would look like this: $ git cat-file commit HEAD tree 3cd71d90e3db4136e5260ab54599791c4f883b9d parent b87755351a47b09cb27d6913e6e0e17e6254a4d4 author Junio C Hamano <gitster@pobox.com> 1317862251 -0700 committer Junio C Hamano <gitster@pobox.com> 1317862251 -0700 gpgsig -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.10 (GNU/Linux) iQIcBAABAgAGBQJOjPtrAAoJELC16IaWr+bL4TMP/RSe2Y/jYnCkds9unO5JEnfG ... =dt98 -----END PGP SIGNATURE----- foo but "git log" (unless you ask for it with --pretty=raw) output is not cluttered with the signature information. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2011-10-06 02:23:20 +02:00
{
struct strbuf sig = STRBUF_INIT;
int inspos, copypos;
const char *eoh;
const char *gpg_sig_header = gpg_sig_headers[hash_algo_by_ptr(the_hash_algo)];
int gpg_sig_header_len = strlen(gpg_sig_header);
commit: teach --gpg-sign option This uses the gpg-interface.[ch] to allow signing the commit, i.e. $ git commit --gpg-sign -m foo You need a passphrase to unlock the secret key for user: "Junio C Hamano <gitster@pobox.com>" 4096-bit RSA key, ID 96AFE6CB, created 2011-10-03 (main key ID 713660A7) [master 8457d13] foo 1 files changed, 1 insertions(+), 0 deletions(-) The lines of GPG detached signature are placed in a new multi-line header field, instead of tucking the signature block at the end of the commit log message text (similar to how signed tag is done), for multiple reasons: - The signature won't clutter output from "git log" and friends if it is in the extra header. If we place it at the end of the log message, we would need to teach "git log" and friends to strip the signature block with an option. - Teaching new versions of "git log" and "gitk" to optionally verify and show signatures is cleaner if we structurally know where the signature block is (instead of scanning in the commit log message). - The signature needs to be stripped upon various commit rewriting operations, e.g. rebase, filter-branch, etc. They all already ignore unknown headers, but if we place signature in the log message, all of these tools (and third-party tools) also need to learn how a signature block would look like. - When we added the optional encoding header, all the tools (both in tree and third-party) that acts on the raw commit object should have been fixed to ignore headers they do not understand, so it is not like that new header would be more likely to break than extra text in the commit. A commit made with the above sample sequence would look like this: $ git cat-file commit HEAD tree 3cd71d90e3db4136e5260ab54599791c4f883b9d parent b87755351a47b09cb27d6913e6e0e17e6254a4d4 author Junio C Hamano <gitster@pobox.com> 1317862251 -0700 committer Junio C Hamano <gitster@pobox.com> 1317862251 -0700 gpgsig -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.10 (GNU/Linux) iQIcBAABAgAGBQJOjPtrAAoJELC16IaWr+bL4TMP/RSe2Y/jYnCkds9unO5JEnfG ... =dt98 -----END PGP SIGNATURE----- foo but "git log" (unless you ask for it with --pretty=raw) output is not cluttered with the signature information. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2011-10-06 02:23:20 +02:00
/* find the end of the header */
eoh = strstr(buf->buf, "\n\n");
if (!eoh)
inspos = buf->len;
else
inspos = eoh - buf->buf + 1;
commit: teach --gpg-sign option This uses the gpg-interface.[ch] to allow signing the commit, i.e. $ git commit --gpg-sign -m foo You need a passphrase to unlock the secret key for user: "Junio C Hamano <gitster@pobox.com>" 4096-bit RSA key, ID 96AFE6CB, created 2011-10-03 (main key ID 713660A7) [master 8457d13] foo 1 files changed, 1 insertions(+), 0 deletions(-) The lines of GPG detached signature are placed in a new multi-line header field, instead of tucking the signature block at the end of the commit log message text (similar to how signed tag is done), for multiple reasons: - The signature won't clutter output from "git log" and friends if it is in the extra header. If we place it at the end of the log message, we would need to teach "git log" and friends to strip the signature block with an option. - Teaching new versions of "git log" and "gitk" to optionally verify and show signatures is cleaner if we structurally know where the signature block is (instead of scanning in the commit log message). - The signature needs to be stripped upon various commit rewriting operations, e.g. rebase, filter-branch, etc. They all already ignore unknown headers, but if we place signature in the log message, all of these tools (and third-party tools) also need to learn how a signature block would look like. - When we added the optional encoding header, all the tools (both in tree and third-party) that acts on the raw commit object should have been fixed to ignore headers they do not understand, so it is not like that new header would be more likely to break than extra text in the commit. A commit made with the above sample sequence would look like this: $ git cat-file commit HEAD tree 3cd71d90e3db4136e5260ab54599791c4f883b9d parent b87755351a47b09cb27d6913e6e0e17e6254a4d4 author Junio C Hamano <gitster@pobox.com> 1317862251 -0700 committer Junio C Hamano <gitster@pobox.com> 1317862251 -0700 gpgsig -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.10 (GNU/Linux) iQIcBAABAgAGBQJOjPtrAAoJELC16IaWr+bL4TMP/RSe2Y/jYnCkds9unO5JEnfG ... =dt98 -----END PGP SIGNATURE----- foo but "git log" (unless you ask for it with --pretty=raw) output is not cluttered with the signature information. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2011-10-06 02:23:20 +02:00
if (!keyid || !*keyid)
keyid = get_signing_key();
if (sign_buffer(buf, &sig, keyid)) {
strbuf_release(&sig);
return -1;
}
for (copypos = 0; sig.buf[copypos]; ) {
const char *bol = sig.buf + copypos;
const char *eol = strchrnul(bol, '\n');
int len = (eol - bol) + !!*eol;
if (!copypos) {
strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
inspos += gpg_sig_header_len;
}
strbuf_insertstr(buf, inspos++, " ");
commit: teach --gpg-sign option This uses the gpg-interface.[ch] to allow signing the commit, i.e. $ git commit --gpg-sign -m foo You need a passphrase to unlock the secret key for user: "Junio C Hamano <gitster@pobox.com>" 4096-bit RSA key, ID 96AFE6CB, created 2011-10-03 (main key ID 713660A7) [master 8457d13] foo 1 files changed, 1 insertions(+), 0 deletions(-) The lines of GPG detached signature are placed in a new multi-line header field, instead of tucking the signature block at the end of the commit log message text (similar to how signed tag is done), for multiple reasons: - The signature won't clutter output from "git log" and friends if it is in the extra header. If we place it at the end of the log message, we would need to teach "git log" and friends to strip the signature block with an option. - Teaching new versions of "git log" and "gitk" to optionally verify and show signatures is cleaner if we structurally know where the signature block is (instead of scanning in the commit log message). - The signature needs to be stripped upon various commit rewriting operations, e.g. rebase, filter-branch, etc. They all already ignore unknown headers, but if we place signature in the log message, all of these tools (and third-party tools) also need to learn how a signature block would look like. - When we added the optional encoding header, all the tools (both in tree and third-party) that acts on the raw commit object should have been fixed to ignore headers they do not understand, so it is not like that new header would be more likely to break than extra text in the commit. A commit made with the above sample sequence would look like this: $ git cat-file commit HEAD tree 3cd71d90e3db4136e5260ab54599791c4f883b9d parent b87755351a47b09cb27d6913e6e0e17e6254a4d4 author Junio C Hamano <gitster@pobox.com> 1317862251 -0700 committer Junio C Hamano <gitster@pobox.com> 1317862251 -0700 gpgsig -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.10 (GNU/Linux) iQIcBAABAgAGBQJOjPtrAAoJELC16IaWr+bL4TMP/RSe2Y/jYnCkds9unO5JEnfG ... =dt98 -----END PGP SIGNATURE----- foo but "git log" (unless you ask for it with --pretty=raw) output is not cluttered with the signature information. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2011-10-06 02:23:20 +02:00
strbuf_insert(buf, inspos, bol, len);
inspos += len;
copypos += len;
}
strbuf_release(&sig);
return 0;
}
reuse cached commit buffer when parsing signatures When we call show_signature or show_mergetag, we read the commit object fresh via read_sha1_file and reparse its headers. However, in most cases we already have the object data available, attached to the "struct commit". This is partially laziness in dealing with the memory allocation issues, but partially defensive programming, in that we would always want to verify a clean version of the buffer (not one that might have been munged by other users of the commit). However, we do not currently ever munge the commit buffer, and not using the already-available buffer carries a fairly big performance penalty when we are looking at a large number of commits. Here are timings on linux.git: [baseline, no signatures] $ time git log >/dev/null real 0m4.902s user 0m4.784s sys 0m0.120s [before] $ time git log --show-signature >/dev/null real 0m14.735s user 0m9.964s sys 0m0.944s [after] $ time git log --show-signature >/dev/null real 0m9.981s user 0m5.260s sys 0m0.936s Note that our user CPU time drops almost in half, close to the non-signature case, but we do still spend more wall-clock and system time, presumably from dealing with gpg. An alternative to this is to note that most commits do not have signatures (less than 1% in this repo), yet we pay the re-parsing cost for every commit just to find out if it has a mergetag or signature. If we checked that when parsing the commit initially, we could avoid re-examining most commits later on. Even if we did pursue that direction, however, this would still speed up the cases where we _do_ have signatures. So it's probably worth doing either way. Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2014-06-13 08:32:11 +02:00
int parse_signed_commit(const struct commit *commit,
struct strbuf *payload, struct strbuf *signature,
const struct git_hash_algo *algop)
{
unsigned long size;
reuse cached commit buffer when parsing signatures When we call show_signature or show_mergetag, we read the commit object fresh via read_sha1_file and reparse its headers. However, in most cases we already have the object data available, attached to the "struct commit". This is partially laziness in dealing with the memory allocation issues, but partially defensive programming, in that we would always want to verify a clean version of the buffer (not one that might have been munged by other users of the commit). However, we do not currently ever munge the commit buffer, and not using the already-available buffer carries a fairly big performance penalty when we are looking at a large number of commits. Here are timings on linux.git: [baseline, no signatures] $ time git log >/dev/null real 0m4.902s user 0m4.784s sys 0m0.120s [before] $ time git log --show-signature >/dev/null real 0m14.735s user 0m9.964s sys 0m0.944s [after] $ time git log --show-signature >/dev/null real 0m9.981s user 0m5.260s sys 0m0.936s Note that our user CPU time drops almost in half, close to the non-signature case, but we do still spend more wall-clock and system time, presumably from dealing with gpg. An alternative to this is to note that most commits do not have signatures (less than 1% in this repo), yet we pay the re-parsing cost for every commit just to find out if it has a mergetag or signature. If we checked that when parsing the commit initially, we could avoid re-examining most commits later on. Even if we did pursue that direction, however, this would still speed up the cases where we _do_ have signatures. So it's probably worth doing either way. Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2014-06-13 08:32:11 +02:00
const char *buffer = get_commit_buffer(commit, &size);
int ret = parse_buffer_signed_by_header(buffer, size, payload, signature, algop);
unuse_commit_buffer(commit, buffer);
return ret;
}
int parse_buffer_signed_by_header(const char *buffer,
unsigned long size,
struct strbuf *payload,
struct strbuf *signature,
const struct git_hash_algo *algop)
{
int in_signature = 0, saw_signature = 0, other_signature = 0;
const char *line, *tail, *p;
const char *gpg_sig_header = gpg_sig_headers[hash_algo_by_ptr(algop)];
line = buffer;
tail = buffer + size;
while (line < tail) {
const char *sig = NULL;
reuse cached commit buffer when parsing signatures When we call show_signature or show_mergetag, we read the commit object fresh via read_sha1_file and reparse its headers. However, in most cases we already have the object data available, attached to the "struct commit". This is partially laziness in dealing with the memory allocation issues, but partially defensive programming, in that we would always want to verify a clean version of the buffer (not one that might have been munged by other users of the commit). However, we do not currently ever munge the commit buffer, and not using the already-available buffer carries a fairly big performance penalty when we are looking at a large number of commits. Here are timings on linux.git: [baseline, no signatures] $ time git log >/dev/null real 0m4.902s user 0m4.784s sys 0m0.120s [before] $ time git log --show-signature >/dev/null real 0m14.735s user 0m9.964s sys 0m0.944s [after] $ time git log --show-signature >/dev/null real 0m9.981s user 0m5.260s sys 0m0.936s Note that our user CPU time drops almost in half, close to the non-signature case, but we do still spend more wall-clock and system time, presumably from dealing with gpg. An alternative to this is to note that most commits do not have signatures (less than 1% in this repo), yet we pay the re-parsing cost for every commit just to find out if it has a mergetag or signature. If we checked that when parsing the commit initially, we could avoid re-examining most commits later on. Even if we did pursue that direction, however, this would still speed up the cases where we _do_ have signatures. So it's probably worth doing either way. Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2014-06-13 08:32:11 +02:00
const char *next = memchr(line, '\n', tail - line);
next = next ? next + 1 : tail;
if (in_signature && line[0] == ' ')
sig = line + 1;
else if (skip_prefix(line, gpg_sig_header, &p) &&
*p == ' ') {
sig = line + strlen(gpg_sig_header) + 1;
other_signature = 0;
}
else if (starts_with(line, "gpgsig"))
other_signature = 1;
else if (other_signature && line[0] != ' ')
other_signature = 0;
if (sig) {
strbuf_add(signature, sig, next - sig);
saw_signature = 1;
in_signature = 1;
} else {
if (*line == '\n')
/* dump the whole remainder of the buffer */
next = tail;
if (!other_signature)
strbuf_add(payload, line, next - line);
in_signature = 0;
}
line = next;
}
return saw_signature;
}
int remove_signature(struct strbuf *buf)
{
const char *line = buf->buf;
const char *tail = buf->buf + buf->len;
int in_signature = 0;
struct sigbuf {
const char *start;
const char *end;
} sigs[2], *sigp = &sigs[0];
int i;
const char *orig_buf = buf->buf;
memset(sigs, 0, sizeof(sigs));
while (line < tail) {
const char *next = memchr(line, '\n', tail - line);
next = next ? next + 1 : tail;
if (in_signature && line[0] == ' ')
sigp->end = next;
else if (starts_with(line, "gpgsig")) {
int i;
for (i = 1; i < GIT_HASH_NALGOS; i++) {
const char *p;
if (skip_prefix(line, gpg_sig_headers[i], &p) &&
*p == ' ') {
sigp->start = line;
sigp->end = next;
in_signature = 1;
}
}
} else {
if (*line == '\n')
/* dump the whole remainder of the buffer */
next = tail;
if (in_signature && sigp - sigs != ARRAY_SIZE(sigs))
sigp++;
in_signature = 0;
}
line = next;
}
for (i = ARRAY_SIZE(sigs) - 1; i >= 0; i--)
if (sigs[i].start)
strbuf_remove(buf, sigs[i].start - orig_buf, sigs[i].end - sigs[i].start);
return sigs[0].start != NULL;
}
static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
{
struct merge_remote_desc *desc;
struct commit_extra_header *mergetag;
char *buf;
unsigned long size;
enum object_type type;
struct strbuf payload = STRBUF_INIT;
struct strbuf signature = STRBUF_INIT;
desc = merge_remote_util(parent);
if (!desc || !desc->obj)
return;
buf = read_object_file(&desc->obj->oid, &type, &size);
if (!buf || type != OBJ_TAG)
goto free_return;
if (!parse_signature(buf, size, &payload, &signature))
goto free_return;
/*
* We could verify this signature and either omit the tag when
* it does not validate, but the integrator may not have the
* public key of the signer of the tag being merged, while a
* later auditor may have it while auditing, so let's not run
* verify-signed-buffer here for now...
*
* if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
* warn("warning: signed tag unverified.");
*/
CALLOC_ARRAY(mergetag, 1);
mergetag->key = xstrdup("mergetag");
mergetag->value = buf;
mergetag->len = size;
**tail = mergetag;
*tail = &mergetag->next;
strbuf_release(&payload);
strbuf_release(&signature);
return;
free_return:
free(buf);
}
int check_commit_signature(const struct commit *commit, struct signature_check *sigc)
{
struct strbuf payload = STRBUF_INIT;
struct strbuf signature = STRBUF_INIT;
int ret = 1;
sigc->result = 'N';
if (parse_signed_commit(commit, &payload, &signature, the_hash_algo) <= 0)
goto out;
ssh signing: make verify-commit consider key lifetime If valid-before/after dates are configured for this signatures key in the allowedSigners file then the verification should check if the key was valid at the time the commit was made. This allows for graceful key rollover and revoking keys without invalidating all previous commits. This feature needs openssh > 8.8. Older ssh-keygen versions will simply ignore this flag and use the current time. Strictly speaking this feature is available in 8.7, but since 8.7 has a bug that makes it unusable in another needed call we require 8.8. Timestamp information is present on most invocations of check_signature. However signer ident is not. We will need the signer email / name to be able to implement "Trust on first use" functionality later. Since the payload contains all necessary information we can parse it from there. The caller only needs to provide us some info about the payload by setting payload_type in the signature_check struct. - Add payload_type field & enum and payload_timestamp to struct signature_check - Populate the timestamp when not already set if we know about the payload type - Pass -Overify-time={payload_timestamp} in the users timezone to all ssh-keygen verification calls - Set the payload type when verifying commits - Add tests for expired, not yet valid and keys having a commit date outside of key validity as well as within Signed-off-by: Fabian Stelzer <fs@gigacodes.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-12-09 09:52:45 +01:00
sigc->payload_type = SIGNATURE_PAYLOAD_COMMIT;
sigc->payload = strbuf_detach(&payload, &sigc->payload_len);
ret = check_signature(sigc, signature.buf, signature.len);
out:
strbuf_release(&payload);
strbuf_release(&signature);
return ret;
}
gpg-interface: add minTrustLevel as a configuration option Previously, signature verification for merge and pull operations checked if the key had a trust-level of either TRUST_NEVER or TRUST_UNDEFINED in verify_merge_signature(). If that was the case, the process die()d. The other code paths that did signature verification relied entirely on the return code from check_commit_signature(). And signatures made with a good key, irregardless of its trust level, was considered valid by check_commit_signature(). This difference in behavior might induce users to erroneously assume that the trust level of a key in their keyring is always considered by Git, even for operations where it is not (e.g. during a verify-commit or verify-tag). The way it worked was by gpg-interface.c storing the result from the key/signature status *and* the lowest-two trust levels in the `result` member of the signature_check structure (the last of these status lines that were encountered got written to `result`). These are documented in GPG under the subsection `General status codes` and `Key related`, respectively [1]. The GPG documentation says the following on the TRUST_ status codes [1]: """ These are several similar status codes: - TRUST_UNDEFINED <error_token> - TRUST_NEVER <error_token> - TRUST_MARGINAL [0 [<validation_model>]] - TRUST_FULLY [0 [<validation_model>]] - TRUST_ULTIMATE [0 [<validation_model>]] For good signatures one of these status lines are emitted to indicate the validity of the key used to create the signature. The error token values are currently only emitted by gpgsm. """ My interpretation is that the trust level is conceptionally different from the validity of the key and/or signature. That seems to also have been the assumption of the old code in check_signature() where a result of 'G' (as in GOODSIG) and 'U' (as in TRUST_NEVER or TRUST_UNDEFINED) were both considered a success. The two cases where a result of 'U' had special meaning were in verify_merge_signature() (where this caused git to die()) and in format_commit_one() (where it affected the output of the %G? format specifier). I think it makes sense to refactor the processing of TRUST_ status lines such that users can configure a minimum trust level that is enforced globally, rather than have individual parts of git (e.g. merge) do it themselves (except for a grace period with backward compatibility). I also think it makes sense to not store the trust level in the same struct member as the key/signature status. While the presence of a TRUST_ status code does imply that the signature is good (see the first paragraph in the included snippet above), as far as I can tell, the order of the status lines from GPG isn't well-defined; thus it would seem plausible that the trust level could be overwritten with the key/signature status if they were stored in the same member of the signature_check structure. This patch introduces a new configuration option: gpg.minTrustLevel. It consolidates trust-level verification to gpg-interface.c and adds a new `trust_level` member to the signature_check structure. Backward-compatibility is maintained by introducing a special case in verify_merge_signature() such that if no user-configurable gpg.minTrustLevel is set, then the old behavior of rejecting TRUST_UNDEFINED and TRUST_NEVER is enforced. If, on the other hand, gpg.minTrustLevel is set, then that value overrides the old behavior. Similarly, the %G? format specifier will continue show 'U' for signatures made with a key that has a trust level of TRUST_UNDEFINED or TRUST_NEVER, even though the 'U' character no longer exist in the `result` member of the signature_check structure. A new format specifier, %GT, is also introduced for users that want to show all possible trust levels for a signature. Another approach would have been to simply drop the trust-level requirement in verify_merge_signature(). This would also have made the behavior consistent with other parts of git that perform signature verification. However, requiring a minimum trust level for signing keys does seem to have a real-world use-case. For example, the build system used by the Qubes OS project currently parses the raw output from verify-tag in order to assert a minimum trust level for keys used to sign git tags [2]. [1] https://git.gnupg.org/cgi-bin/gitweb.cgi?p=gnupg.git;a=blob;f=doc/doc/DETAILS;h=bd00006e933ac56719b1edd2478ecd79273eae72;hb=refs/heads/master [2] https://github.com/QubesOS/qubes-builder/blob/9674c1991deef45b1a1b1c71fddfab14ba50dccf/scripts/verify-git-tag#L43 Signed-off-by: Hans Jerry Illikainen <hji@dyntopia.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2019-12-27 14:55:57 +01:00
void verify_merge_signature(struct commit *commit, int verbosity,
int check_trust)
{
char hex[GIT_MAX_HEXSZ + 1];
struct signature_check signature_check;
gpg-interface: add minTrustLevel as a configuration option Previously, signature verification for merge and pull operations checked if the key had a trust-level of either TRUST_NEVER or TRUST_UNDEFINED in verify_merge_signature(). If that was the case, the process die()d. The other code paths that did signature verification relied entirely on the return code from check_commit_signature(). And signatures made with a good key, irregardless of its trust level, was considered valid by check_commit_signature(). This difference in behavior might induce users to erroneously assume that the trust level of a key in their keyring is always considered by Git, even for operations where it is not (e.g. during a verify-commit or verify-tag). The way it worked was by gpg-interface.c storing the result from the key/signature status *and* the lowest-two trust levels in the `result` member of the signature_check structure (the last of these status lines that were encountered got written to `result`). These are documented in GPG under the subsection `General status codes` and `Key related`, respectively [1]. The GPG documentation says the following on the TRUST_ status codes [1]: """ These are several similar status codes: - TRUST_UNDEFINED <error_token> - TRUST_NEVER <error_token> - TRUST_MARGINAL [0 [<validation_model>]] - TRUST_FULLY [0 [<validation_model>]] - TRUST_ULTIMATE [0 [<validation_model>]] For good signatures one of these status lines are emitted to indicate the validity of the key used to create the signature. The error token values are currently only emitted by gpgsm. """ My interpretation is that the trust level is conceptionally different from the validity of the key and/or signature. That seems to also have been the assumption of the old code in check_signature() where a result of 'G' (as in GOODSIG) and 'U' (as in TRUST_NEVER or TRUST_UNDEFINED) were both considered a success. The two cases where a result of 'U' had special meaning were in verify_merge_signature() (where this caused git to die()) and in format_commit_one() (where it affected the output of the %G? format specifier). I think it makes sense to refactor the processing of TRUST_ status lines such that users can configure a minimum trust level that is enforced globally, rather than have individual parts of git (e.g. merge) do it themselves (except for a grace period with backward compatibility). I also think it makes sense to not store the trust level in the same struct member as the key/signature status. While the presence of a TRUST_ status code does imply that the signature is good (see the first paragraph in the included snippet above), as far as I can tell, the order of the status lines from GPG isn't well-defined; thus it would seem plausible that the trust level could be overwritten with the key/signature status if they were stored in the same member of the signature_check structure. This patch introduces a new configuration option: gpg.minTrustLevel. It consolidates trust-level verification to gpg-interface.c and adds a new `trust_level` member to the signature_check structure. Backward-compatibility is maintained by introducing a special case in verify_merge_signature() such that if no user-configurable gpg.minTrustLevel is set, then the old behavior of rejecting TRUST_UNDEFINED and TRUST_NEVER is enforced. If, on the other hand, gpg.minTrustLevel is set, then that value overrides the old behavior. Similarly, the %G? format specifier will continue show 'U' for signatures made with a key that has a trust level of TRUST_UNDEFINED or TRUST_NEVER, even though the 'U' character no longer exist in the `result` member of the signature_check structure. A new format specifier, %GT, is also introduced for users that want to show all possible trust levels for a signature. Another approach would have been to simply drop the trust-level requirement in verify_merge_signature(). This would also have made the behavior consistent with other parts of git that perform signature verification. However, requiring a minimum trust level for signing keys does seem to have a real-world use-case. For example, the build system used by the Qubes OS project currently parses the raw output from verify-tag in order to assert a minimum trust level for keys used to sign git tags [2]. [1] https://git.gnupg.org/cgi-bin/gitweb.cgi?p=gnupg.git;a=blob;f=doc/doc/DETAILS;h=bd00006e933ac56719b1edd2478ecd79273eae72;hb=refs/heads/master [2] https://github.com/QubesOS/qubes-builder/blob/9674c1991deef45b1a1b1c71fddfab14ba50dccf/scripts/verify-git-tag#L43 Signed-off-by: Hans Jerry Illikainen <hji@dyntopia.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2019-12-27 14:55:57 +01:00
int ret;
memset(&signature_check, 0, sizeof(signature_check));
gpg-interface: add minTrustLevel as a configuration option Previously, signature verification for merge and pull operations checked if the key had a trust-level of either TRUST_NEVER or TRUST_UNDEFINED in verify_merge_signature(). If that was the case, the process die()d. The other code paths that did signature verification relied entirely on the return code from check_commit_signature(). And signatures made with a good key, irregardless of its trust level, was considered valid by check_commit_signature(). This difference in behavior might induce users to erroneously assume that the trust level of a key in their keyring is always considered by Git, even for operations where it is not (e.g. during a verify-commit or verify-tag). The way it worked was by gpg-interface.c storing the result from the key/signature status *and* the lowest-two trust levels in the `result` member of the signature_check structure (the last of these status lines that were encountered got written to `result`). These are documented in GPG under the subsection `General status codes` and `Key related`, respectively [1]. The GPG documentation says the following on the TRUST_ status codes [1]: """ These are several similar status codes: - TRUST_UNDEFINED <error_token> - TRUST_NEVER <error_token> - TRUST_MARGINAL [0 [<validation_model>]] - TRUST_FULLY [0 [<validation_model>]] - TRUST_ULTIMATE [0 [<validation_model>]] For good signatures one of these status lines are emitted to indicate the validity of the key used to create the signature. The error token values are currently only emitted by gpgsm. """ My interpretation is that the trust level is conceptionally different from the validity of the key and/or signature. That seems to also have been the assumption of the old code in check_signature() where a result of 'G' (as in GOODSIG) and 'U' (as in TRUST_NEVER or TRUST_UNDEFINED) were both considered a success. The two cases where a result of 'U' had special meaning were in verify_merge_signature() (where this caused git to die()) and in format_commit_one() (where it affected the output of the %G? format specifier). I think it makes sense to refactor the processing of TRUST_ status lines such that users can configure a minimum trust level that is enforced globally, rather than have individual parts of git (e.g. merge) do it themselves (except for a grace period with backward compatibility). I also think it makes sense to not store the trust level in the same struct member as the key/signature status. While the presence of a TRUST_ status code does imply that the signature is good (see the first paragraph in the included snippet above), as far as I can tell, the order of the status lines from GPG isn't well-defined; thus it would seem plausible that the trust level could be overwritten with the key/signature status if they were stored in the same member of the signature_check structure. This patch introduces a new configuration option: gpg.minTrustLevel. It consolidates trust-level verification to gpg-interface.c and adds a new `trust_level` member to the signature_check structure. Backward-compatibility is maintained by introducing a special case in verify_merge_signature() such that if no user-configurable gpg.minTrustLevel is set, then the old behavior of rejecting TRUST_UNDEFINED and TRUST_NEVER is enforced. If, on the other hand, gpg.minTrustLevel is set, then that value overrides the old behavior. Similarly, the %G? format specifier will continue show 'U' for signatures made with a key that has a trust level of TRUST_UNDEFINED or TRUST_NEVER, even though the 'U' character no longer exist in the `result` member of the signature_check structure. A new format specifier, %GT, is also introduced for users that want to show all possible trust levels for a signature. Another approach would have been to simply drop the trust-level requirement in verify_merge_signature(). This would also have made the behavior consistent with other parts of git that perform signature verification. However, requiring a minimum trust level for signing keys does seem to have a real-world use-case. For example, the build system used by the Qubes OS project currently parses the raw output from verify-tag in order to assert a minimum trust level for keys used to sign git tags [2]. [1] https://git.gnupg.org/cgi-bin/gitweb.cgi?p=gnupg.git;a=blob;f=doc/doc/DETAILS;h=bd00006e933ac56719b1edd2478ecd79273eae72;hb=refs/heads/master [2] https://github.com/QubesOS/qubes-builder/blob/9674c1991deef45b1a1b1c71fddfab14ba50dccf/scripts/verify-git-tag#L43 Signed-off-by: Hans Jerry Illikainen <hji@dyntopia.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2019-12-27 14:55:57 +01:00
ret = check_commit_signature(commit, &signature_check);
find_unique_abbrev_r(hex, &commit->object.oid, DEFAULT_ABBREV);
switch (signature_check.result) {
case 'G':
gpg-interface: add minTrustLevel as a configuration option Previously, signature verification for merge and pull operations checked if the key had a trust-level of either TRUST_NEVER or TRUST_UNDEFINED in verify_merge_signature(). If that was the case, the process die()d. The other code paths that did signature verification relied entirely on the return code from check_commit_signature(). And signatures made with a good key, irregardless of its trust level, was considered valid by check_commit_signature(). This difference in behavior might induce users to erroneously assume that the trust level of a key in their keyring is always considered by Git, even for operations where it is not (e.g. during a verify-commit or verify-tag). The way it worked was by gpg-interface.c storing the result from the key/signature status *and* the lowest-two trust levels in the `result` member of the signature_check structure (the last of these status lines that were encountered got written to `result`). These are documented in GPG under the subsection `General status codes` and `Key related`, respectively [1]. The GPG documentation says the following on the TRUST_ status codes [1]: """ These are several similar status codes: - TRUST_UNDEFINED <error_token> - TRUST_NEVER <error_token> - TRUST_MARGINAL [0 [<validation_model>]] - TRUST_FULLY [0 [<validation_model>]] - TRUST_ULTIMATE [0 [<validation_model>]] For good signatures one of these status lines are emitted to indicate the validity of the key used to create the signature. The error token values are currently only emitted by gpgsm. """ My interpretation is that the trust level is conceptionally different from the validity of the key and/or signature. That seems to also have been the assumption of the old code in check_signature() where a result of 'G' (as in GOODSIG) and 'U' (as in TRUST_NEVER or TRUST_UNDEFINED) were both considered a success. The two cases where a result of 'U' had special meaning were in verify_merge_signature() (where this caused git to die()) and in format_commit_one() (where it affected the output of the %G? format specifier). I think it makes sense to refactor the processing of TRUST_ status lines such that users can configure a minimum trust level that is enforced globally, rather than have individual parts of git (e.g. merge) do it themselves (except for a grace period with backward compatibility). I also think it makes sense to not store the trust level in the same struct member as the key/signature status. While the presence of a TRUST_ status code does imply that the signature is good (see the first paragraph in the included snippet above), as far as I can tell, the order of the status lines from GPG isn't well-defined; thus it would seem plausible that the trust level could be overwritten with the key/signature status if they were stored in the same member of the signature_check structure. This patch introduces a new configuration option: gpg.minTrustLevel. It consolidates trust-level verification to gpg-interface.c and adds a new `trust_level` member to the signature_check structure. Backward-compatibility is maintained by introducing a special case in verify_merge_signature() such that if no user-configurable gpg.minTrustLevel is set, then the old behavior of rejecting TRUST_UNDEFINED and TRUST_NEVER is enforced. If, on the other hand, gpg.minTrustLevel is set, then that value overrides the old behavior. Similarly, the %G? format specifier will continue show 'U' for signatures made with a key that has a trust level of TRUST_UNDEFINED or TRUST_NEVER, even though the 'U' character no longer exist in the `result` member of the signature_check structure. A new format specifier, %GT, is also introduced for users that want to show all possible trust levels for a signature. Another approach would have been to simply drop the trust-level requirement in verify_merge_signature(). This would also have made the behavior consistent with other parts of git that perform signature verification. However, requiring a minimum trust level for signing keys does seem to have a real-world use-case. For example, the build system used by the Qubes OS project currently parses the raw output from verify-tag in order to assert a minimum trust level for keys used to sign git tags [2]. [1] https://git.gnupg.org/cgi-bin/gitweb.cgi?p=gnupg.git;a=blob;f=doc/doc/DETAILS;h=bd00006e933ac56719b1edd2478ecd79273eae72;hb=refs/heads/master [2] https://github.com/QubesOS/qubes-builder/blob/9674c1991deef45b1a1b1c71fddfab14ba50dccf/scripts/verify-git-tag#L43 Signed-off-by: Hans Jerry Illikainen <hji@dyntopia.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2019-12-27 14:55:57 +01:00
if (ret || (check_trust && signature_check.trust_level < TRUST_MARGINAL))
die(_("Commit %s has an untrusted GPG signature, "
"allegedly by %s."), hex, signature_check.signer);
break;
case 'B':
die(_("Commit %s has a bad GPG signature "
"allegedly by %s."), hex, signature_check.signer);
default: /* 'N' */
die(_("Commit %s does not have a GPG signature."), hex);
}
if (verbosity >= 0 && signature_check.result == 'G')
printf(_("Commit %s has a good GPG signature by %s\n"),
hex, signature_check.signer);
signature_check_clear(&signature_check);
}
void append_merge_tag_headers(struct commit_list *parents,
struct commit_extra_header ***tail)
{
while (parents) {
struct commit *parent = parents->item;
handle_signed_tag(parent, tail);
parents = parents->next;
}
}
static void add_extra_header(struct strbuf *buffer,
struct commit_extra_header *extra)
{
strbuf_addstr(buffer, extra->key);
if (extra->len)
strbuf_add_lines(buffer, " ", extra->value, extra->len);
else
strbuf_addch(buffer, '\n');
}
struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
const char **exclude)
{
struct commit_extra_header *extra = NULL;
unsigned long size;
reuse cached commit buffer when parsing signatures When we call show_signature or show_mergetag, we read the commit object fresh via read_sha1_file and reparse its headers. However, in most cases we already have the object data available, attached to the "struct commit". This is partially laziness in dealing with the memory allocation issues, but partially defensive programming, in that we would always want to verify a clean version of the buffer (not one that might have been munged by other users of the commit). However, we do not currently ever munge the commit buffer, and not using the already-available buffer carries a fairly big performance penalty when we are looking at a large number of commits. Here are timings on linux.git: [baseline, no signatures] $ time git log >/dev/null real 0m4.902s user 0m4.784s sys 0m0.120s [before] $ time git log --show-signature >/dev/null real 0m14.735s user 0m9.964s sys 0m0.944s [after] $ time git log --show-signature >/dev/null real 0m9.981s user 0m5.260s sys 0m0.936s Note that our user CPU time drops almost in half, close to the non-signature case, but we do still spend more wall-clock and system time, presumably from dealing with gpg. An alternative to this is to note that most commits do not have signatures (less than 1% in this repo), yet we pay the re-parsing cost for every commit just to find out if it has a mergetag or signature. If we checked that when parsing the commit initially, we could avoid re-examining most commits later on. Even if we did pursue that direction, however, this would still speed up the cases where we _do_ have signatures. So it's probably worth doing either way. Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2014-06-13 08:32:11 +02:00
const char *buffer = get_commit_buffer(commit, &size);
extra = read_commit_extra_header_lines(buffer, size, exclude);
unuse_commit_buffer(commit, buffer);
return extra;
}
int for_each_mergetag(each_mergetag_fn fn, struct commit *commit, void *data)
{
struct commit_extra_header *extra, *to_free;
int res = 0;
to_free = read_commit_extra_headers(commit, NULL);
for (extra = to_free; !res && extra; extra = extra->next) {
if (strcmp(extra->key, "mergetag"))
continue; /* not a merge tag */
res = fn(commit, extra, data);
}
free_commit_extra_headers(to_free);
return res;
}
static inline int standard_header_field(const char *field, size_t len)
{
return ((len == 4 && !memcmp(field, "tree", 4)) ||
(len == 6 && !memcmp(field, "parent", 6)) ||
(len == 6 && !memcmp(field, "author", 6)) ||
(len == 9 && !memcmp(field, "committer", 9)) ||
(len == 8 && !memcmp(field, "encoding", 8)));
}
static int excluded_header_field(const char *field, size_t len, const char **exclude)
{
if (!exclude)
return 0;
while (*exclude) {
size_t xlen = strlen(*exclude);
if (len == xlen && !memcmp(field, *exclude, xlen))
return 1;
exclude++;
}
return 0;
}
static struct commit_extra_header *read_commit_extra_header_lines(
const char *buffer, size_t size,
const char **exclude)
{
struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
const char *line, *next, *eof, *eob;
struct strbuf buf = STRBUF_INIT;
for (line = buffer, eob = line + size;
line < eob && *line != '\n';
line = next) {
next = memchr(line, '\n', eob - line);
next = next ? next + 1 : eob;
if (*line == ' ') {
/* continuation */
if (it)
strbuf_add(&buf, line + 1, next - (line + 1));
continue;
}
if (it)
it->value = strbuf_detach(&buf, &it->len);
strbuf_reset(&buf);
it = NULL;
eof = memchr(line, ' ', next - line);
if (!eof)
eof = next;
else if (standard_header_field(line, eof - line) ||
excluded_header_field(line, eof - line, exclude))
continue;
CALLOC_ARRAY(it, 1);
it->key = xmemdupz(line, eof-line);
*tail = it;
tail = &it->next;
if (eof + 1 < next)
strbuf_add(&buf, eof + 1, next - (eof + 1));
}
if (it)
it->value = strbuf_detach(&buf, &it->len);
return extra;
}
void free_commit_extra_headers(struct commit_extra_header *extra)
{
while (extra) {
struct commit_extra_header *next = extra->next;
free(extra->key);
free(extra->value);
free(extra);
extra = next;
}
}
int commit_tree(const char *msg, size_t msg_len, const struct object_id *tree,
struct commit_list *parents, struct object_id *ret,
commit: teach --gpg-sign option This uses the gpg-interface.[ch] to allow signing the commit, i.e. $ git commit --gpg-sign -m foo You need a passphrase to unlock the secret key for user: "Junio C Hamano <gitster@pobox.com>" 4096-bit RSA key, ID 96AFE6CB, created 2011-10-03 (main key ID 713660A7) [master 8457d13] foo 1 files changed, 1 insertions(+), 0 deletions(-) The lines of GPG detached signature are placed in a new multi-line header field, instead of tucking the signature block at the end of the commit log message text (similar to how signed tag is done), for multiple reasons: - The signature won't clutter output from "git log" and friends if it is in the extra header. If we place it at the end of the log message, we would need to teach "git log" and friends to strip the signature block with an option. - Teaching new versions of "git log" and "gitk" to optionally verify and show signatures is cleaner if we structurally know where the signature block is (instead of scanning in the commit log message). - The signature needs to be stripped upon various commit rewriting operations, e.g. rebase, filter-branch, etc. They all already ignore unknown headers, but if we place signature in the log message, all of these tools (and third-party tools) also need to learn how a signature block would look like. - When we added the optional encoding header, all the tools (both in tree and third-party) that acts on the raw commit object should have been fixed to ignore headers they do not understand, so it is not like that new header would be more likely to break than extra text in the commit. A commit made with the above sample sequence would look like this: $ git cat-file commit HEAD tree 3cd71d90e3db4136e5260ab54599791c4f883b9d parent b87755351a47b09cb27d6913e6e0e17e6254a4d4 author Junio C Hamano <gitster@pobox.com> 1317862251 -0700 committer Junio C Hamano <gitster@pobox.com> 1317862251 -0700 gpgsig -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.10 (GNU/Linux) iQIcBAABAgAGBQJOjPtrAAoJELC16IaWr+bL4TMP/RSe2Y/jYnCkds9unO5JEnfG ... =dt98 -----END PGP SIGNATURE----- foo but "git log" (unless you ask for it with --pretty=raw) output is not cluttered with the signature information. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2011-10-06 02:23:20 +02:00
const char *author, const char *sign_commit)
{
struct commit_extra_header *extra = NULL, **tail = &extra;
int result;
append_merge_tag_headers(parents, &tail);
result = commit_tree_extended(msg, msg_len, tree, parents, ret, author,
NULL, sign_commit, extra);
free_commit_extra_headers(extra);
return result;
}
static int find_invalid_utf8(const char *buf, int len)
{
int offset = 0;
static const unsigned int max_codepoint[] = {
0x7f, 0x7ff, 0xffff, 0x10ffff
};
while (len) {
unsigned char c = *buf++;
int bytes, bad_offset;
unsigned int codepoint;
unsigned int min_val, max_val;
len--;
offset++;
/* Simple US-ASCII? No worries. */
if (c < 0x80)
continue;
bad_offset = offset-1;
/*
* Count how many more high bits set: that's how
* many more bytes this sequence should have.
*/
bytes = 0;
while (c & 0x40) {
c <<= 1;
bytes++;
}
/*
* Must be between 1 and 3 more bytes. Longer sequences result in
* codepoints beyond U+10FFFF, which are guaranteed never to exist.
*/
if (bytes < 1 || 3 < bytes)
return bad_offset;
/* Do we *have* that many bytes? */
if (len < bytes)
return bad_offset;
/*
* Place the encoded bits at the bottom of the value and compute the
* valid range.
*/
codepoint = (c & 0x7f) >> bytes;
min_val = max_codepoint[bytes-1] + 1;
max_val = max_codepoint[bytes];
offset += bytes;
len -= bytes;
/* And verify that they are good continuation bytes */
do {
codepoint <<= 6;
codepoint |= *buf & 0x3f;
if ((*buf++ & 0xc0) != 0x80)
return bad_offset;
} while (--bytes);
/* Reject codepoints that are out of range for the sequence length. */
if (codepoint < min_val || codepoint > max_val)
return bad_offset;
/* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
if ((codepoint & 0x1ff800) == 0xd800)
return bad_offset;
/* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
if ((codepoint & 0xfffe) == 0xfffe)
return bad_offset;
/* So are anything in the range U+FDD0..U+FDEF. */
if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
return bad_offset;
}
return -1;
}
/*
* This verifies that the buffer is in proper utf8 format.
*
* If it isn't, it assumes any non-utf8 characters are Latin1,
* and does the conversion.
*/
static int verify_utf8(struct strbuf *buf)
{
int ok = 1;
long pos = 0;
for (;;) {
int bad;
unsigned char c;
unsigned char replace[2];
bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
if (bad < 0)
return ok;
pos += bad;
ok = 0;
c = buf->buf[pos];
strbuf_remove(buf, pos, 1);
/* We know 'c' must be in the range 128-255 */
replace[0] = 0xc0 + (c >> 6);
replace[1] = 0x80 + (c & 0x3f);
strbuf_insert(buf, pos, replace, 2);
pos += 2;
}
}
static const char commit_utf8_warn[] =
N_("Warning: commit message did not conform to UTF-8.\n"
"You may want to amend it after fixing the message, or set the config\n"
"variable i18n.commitEncoding to the encoding your project uses.\n");
int commit_tree_extended(const char *msg, size_t msg_len,
const struct object_id *tree,
struct commit_list *parents, struct object_id *ret,
const char *author, const char *committer,
const char *sign_commit,
commit: teach --gpg-sign option This uses the gpg-interface.[ch] to allow signing the commit, i.e. $ git commit --gpg-sign -m foo You need a passphrase to unlock the secret key for user: "Junio C Hamano <gitster@pobox.com>" 4096-bit RSA key, ID 96AFE6CB, created 2011-10-03 (main key ID 713660A7) [master 8457d13] foo 1 files changed, 1 insertions(+), 0 deletions(-) The lines of GPG detached signature are placed in a new multi-line header field, instead of tucking the signature block at the end of the commit log message text (similar to how signed tag is done), for multiple reasons: - The signature won't clutter output from "git log" and friends if it is in the extra header. If we place it at the end of the log message, we would need to teach "git log" and friends to strip the signature block with an option. - Teaching new versions of "git log" and "gitk" to optionally verify and show signatures is cleaner if we structurally know where the signature block is (instead of scanning in the commit log message). - The signature needs to be stripped upon various commit rewriting operations, e.g. rebase, filter-branch, etc. They all already ignore unknown headers, but if we place signature in the log message, all of these tools (and third-party tools) also need to learn how a signature block would look like. - When we added the optional encoding header, all the tools (both in tree and third-party) that acts on the raw commit object should have been fixed to ignore headers they do not understand, so it is not like that new header would be more likely to break than extra text in the commit. A commit made with the above sample sequence would look like this: $ git cat-file commit HEAD tree 3cd71d90e3db4136e5260ab54599791c4f883b9d parent b87755351a47b09cb27d6913e6e0e17e6254a4d4 author Junio C Hamano <gitster@pobox.com> 1317862251 -0700 committer Junio C Hamano <gitster@pobox.com> 1317862251 -0700 gpgsig -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.10 (GNU/Linux) iQIcBAABAgAGBQJOjPtrAAoJELC16IaWr+bL4TMP/RSe2Y/jYnCkds9unO5JEnfG ... =dt98 -----END PGP SIGNATURE----- foo but "git log" (unless you ask for it with --pretty=raw) output is not cluttered with the signature information. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2011-10-06 02:23:20 +02:00
struct commit_extra_header *extra)
{
int result;
int encoding_is_utf8;
struct strbuf buffer;
assert_oid_type(tree, OBJ_TREE);
if (memchr(msg, '\0', msg_len))
return error("a NUL byte in commit log message not allowed.");
/* Not having i18n.commitencoding is the same as having utf-8 */
encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
strbuf_addf(&buffer, "tree %s\n", oid_to_hex(tree));
/*
* NOTE! This ordering means that the same exact tree merged with a
* different order of parents will be a _different_ changeset even
* if everything else stays the same.
*/
while (parents) {
struct commit *parent = pop_commit(&parents);
strbuf_addf(&buffer, "parent %s\n",
oid_to_hex(&parent->object.oid));
}
/* Person/date information */
if (!author)
author = git_author_info(IDENT_STRICT);
strbuf_addf(&buffer, "author %s\n", author);
if (!committer)
committer = git_committer_info(IDENT_STRICT);
strbuf_addf(&buffer, "committer %s\n", committer);
if (!encoding_is_utf8)
strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
while (extra) {
add_extra_header(&buffer, extra);
extra = extra->next;
}
strbuf_addch(&buffer, '\n');
/* And add the comment */
strbuf_add(&buffer, msg, msg_len);
/* And check the encoding */
if (encoding_is_utf8 && !verify_utf8(&buffer))
fprintf(stderr, _(commit_utf8_warn));
if (sign_commit && sign_with_header(&buffer, sign_commit)) {
result = -1;
goto out;
}
commit: teach --gpg-sign option This uses the gpg-interface.[ch] to allow signing the commit, i.e. $ git commit --gpg-sign -m foo You need a passphrase to unlock the secret key for user: "Junio C Hamano <gitster@pobox.com>" 4096-bit RSA key, ID 96AFE6CB, created 2011-10-03 (main key ID 713660A7) [master 8457d13] foo 1 files changed, 1 insertions(+), 0 deletions(-) The lines of GPG detached signature are placed in a new multi-line header field, instead of tucking the signature block at the end of the commit log message text (similar to how signed tag is done), for multiple reasons: - The signature won't clutter output from "git log" and friends if it is in the extra header. If we place it at the end of the log message, we would need to teach "git log" and friends to strip the signature block with an option. - Teaching new versions of "git log" and "gitk" to optionally verify and show signatures is cleaner if we structurally know where the signature block is (instead of scanning in the commit log message). - The signature needs to be stripped upon various commit rewriting operations, e.g. rebase, filter-branch, etc. They all already ignore unknown headers, but if we place signature in the log message, all of these tools (and third-party tools) also need to learn how a signature block would look like. - When we added the optional encoding header, all the tools (both in tree and third-party) that acts on the raw commit object should have been fixed to ignore headers they do not understand, so it is not like that new header would be more likely to break than extra text in the commit. A commit made with the above sample sequence would look like this: $ git cat-file commit HEAD tree 3cd71d90e3db4136e5260ab54599791c4f883b9d parent b87755351a47b09cb27d6913e6e0e17e6254a4d4 author Junio C Hamano <gitster@pobox.com> 1317862251 -0700 committer Junio C Hamano <gitster@pobox.com> 1317862251 -0700 gpgsig -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.10 (GNU/Linux) iQIcBAABAgAGBQJOjPtrAAoJELC16IaWr+bL4TMP/RSe2Y/jYnCkds9unO5JEnfG ... =dt98 -----END PGP SIGNATURE----- foo but "git log" (unless you ask for it with --pretty=raw) output is not cluttered with the signature information. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2011-10-06 02:23:20 +02:00
result = write_object_file(buffer.buf, buffer.len, OBJ_COMMIT, ret);
out:
strbuf_release(&buffer);
return result;
}
define_commit_slab(merge_desc_slab, struct merge_remote_desc *);
static struct merge_desc_slab merge_desc_slab = COMMIT_SLAB_INIT(1, merge_desc_slab);
struct merge_remote_desc *merge_remote_util(struct commit *commit)
{
return *merge_desc_slab_at(&merge_desc_slab, commit);
}
void set_merge_remote_desc(struct commit *commit,
const char *name, struct object *obj)
{
struct merge_remote_desc *desc;
FLEX_ALLOC_STR(desc, name, name);
desc->obj = obj;
*merge_desc_slab_at(&merge_desc_slab, commit) = desc;
}
struct commit *get_merge_parent(const char *name)
{
struct object *obj;
struct commit *commit;
struct object_id oid;
sha1_name: convert get_sha1* to get_oid* Now that all the callers of get_sha1 directly or indirectly use struct object_id, rename the functions starting with get_sha1 to start with get_oid. Convert the internals in sha1_name.c to use struct object_id as well, and eliminate explicit length checks where possible. Convert a use of 40 in get_oid_basic to GIT_SHA1_HEXSZ. Outside of sha1_name.c and cache.h, this transition was made with the following semantic patch: @@ expression E1, E2; @@ - get_sha1(E1, E2.hash) + get_oid(E1, &E2) @@ expression E1, E2; @@ - get_sha1(E1, E2->hash) + get_oid(E1, E2) @@ expression E1, E2; @@ - get_sha1_committish(E1, E2.hash) + get_oid_committish(E1, &E2) @@ expression E1, E2; @@ - get_sha1_committish(E1, E2->hash) + get_oid_committish(E1, E2) @@ expression E1, E2; @@ - get_sha1_treeish(E1, E2.hash) + get_oid_treeish(E1, &E2) @@ expression E1, E2; @@ - get_sha1_treeish(E1, E2->hash) + get_oid_treeish(E1, E2) @@ expression E1, E2; @@ - get_sha1_commit(E1, E2.hash) + get_oid_commit(E1, &E2) @@ expression E1, E2; @@ - get_sha1_commit(E1, E2->hash) + get_oid_commit(E1, E2) @@ expression E1, E2; @@ - get_sha1_tree(E1, E2.hash) + get_oid_tree(E1, &E2) @@ expression E1, E2; @@ - get_sha1_tree(E1, E2->hash) + get_oid_tree(E1, E2) @@ expression E1, E2; @@ - get_sha1_blob(E1, E2.hash) + get_oid_blob(E1, &E2) @@ expression E1, E2; @@ - get_sha1_blob(E1, E2->hash) + get_oid_blob(E1, E2) @@ expression E1, E2, E3, E4; @@ - get_sha1_with_context(E1, E2, E3.hash, E4) + get_oid_with_context(E1, E2, &E3, E4) @@ expression E1, E2, E3, E4; @@ - get_sha1_with_context(E1, E2, E3->hash, E4) + get_oid_with_context(E1, E2, E3, E4) Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2017-07-14 01:49:28 +02:00
if (get_oid(name, &oid))
return NULL;
obj = parse_object(the_repository, &oid);
commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
if (commit && !merge_remote_util(commit))
set_merge_remote_desc(commit, name, obj);
return commit;
}
/*
* Append a commit to the end of the commit_list.
*
* next starts by pointing to the variable that holds the head of an
* empty commit_list, and is updated to point to the "next" field of
* the last item on the list as new commits are appended.
*
* Usage example:
*
* struct commit_list *list;
* struct commit_list **next = &list;
*
* next = commit_list_append(c1, next);
* next = commit_list_append(c2, next);
* assert(commit_list_count(list) == 2);
* return list;
*/
struct commit_list **commit_list_append(struct commit *commit,
struct commit_list **next)
{
struct commit_list *new_commit = xmalloc(sizeof(struct commit_list));
new_commit->item = commit;
*next = new_commit;
new_commit->next = NULL;
return &new_commit->next;
}
const char *find_header_mem(const char *msg, size_t len,
const char *key, size_t *out_len)
{
int key_len = strlen(key);
const char *line = msg;
/*
* NEEDSWORK: It's possible for strchrnul() to scan beyond the range
* given by len. However, current callers are safe because they compute
* len by scanning a NUL-terminated block of memory starting at msg.
* Nonetheless, it would be better to ensure the function does not look
* at msg beyond the len provided by the caller.
*/
while (line && line < msg + len) {
const char *eol = strchrnul(line, '\n');
if (line == eol)
return NULL;
if (eol - line > key_len &&
!strncmp(line, key, key_len) &&
line[key_len] == ' ') {
*out_len = eol - line - key_len - 1;
return line + key_len + 1;
}
line = *eol ? eol + 1 : NULL;
}
return NULL;
}
const char *find_commit_header(const char *msg, const char *key, size_t *out_len)
{
return find_header_mem(msg, strlen(msg), key, out_len);
}
/*
* Inspect the given string and determine the true "end" of the log message, in
Documentation: stylistically normalize references to Signed-off-by: Ted reported an old typo in the git-commit.txt and merge-options.txt. Namely, the phrase "Signed-off-by line" was used without either a definite nor indefinite article. Upon examination, it seems that the documentation (including items in Documentation/, but also option help strings) have been quite inconsistent on usage when referring to `Signed-off-by`. First, very few places used a definite or indefinite article with the phrase "Signed-off-by line", but that was the initial typo that led to this investigation. So, normalize using either an indefinite or definite article consistently. The original phrasing, in Commit 3f971fc425b (Documentation updates, 2005-08-14), is "Add Signed-off-by line". Commit 6f855371a53 (Add --signoff, --check, and long option-names. 2005-12-09) switched to using "Add `Signed-off-by:` line", but didn't normalize the former commit to match. Later commits seem to have cut and pasted from one or the other, which is likely how the usage became so inconsistent. Junio stated on the git mailing list in <xmqqy2k1dfoh.fsf@gitster.c.googlers.com> a preference to leave off the colon. Thus, prefer `Signed-off-by` (with backticks) for the documentation files and Signed-off-by (without backticks) for option help strings. Additionally, Junio argued that "trailer" is now the standard term to refer to `Signed-off-by`, saying that "becomes plenty clear that we are not talking about any random line in the log message". As such, prefer "trailer" over "line" anywhere the former word fits. However, leave alone those few places in documentation that use Signed-off-by to refer to the process (rather than the specific trailer), or in places where mail headers are generally discussed in comparison with Signed-off-by. Reported-by: "Theodore Y. Ts'o" <tytso@mit.edu> Signed-off-by: Bradley M. Kuhn <bkuhn@sfconservancy.org> Acked-by: Taylor Blau <me@ttaylorr.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2020-10-20 03:03:55 +02:00
* order to find where to put a new Signed-off-by trailer. Ignored are
interpret-trailers: honor the cut line If a commit message is edited with the "verbose" option, the buffer will have a cut line and diff after the log message, like so: my subject # ------------------------ >8 ------------------------ # Do not touch the line above. # Everything below will be removed. diff --git a/foo.txt b/foo.txt index 5716ca5..7601807 100644 --- a/foo.txt +++ b/foo.txt @@ -1 +1 @@ -bar +baz "git interpret-trailers" is unaware of the cut line, and assumes the trailer block would be at the end of the whole thing. This can easily be seen with: $ GIT_EDITOR='git interpret-trailers --in-place --trailer Acked-by:me' \ git commit --amend -v Teach "git interpret-trailers" to notice the cut-line and ignore the remainder of the input when looking for a place to add new trailer block. This makes it consistent with how "git commit -v -s" inserts a new Signed-off-by: line. This can be done by the same logic as the existing helper function, wt_status_truncate_message_at_cut_line(), uses, but it wants the caller to pass a strbuf to it. Because the function ignore_non_trailer() used by the command takes a <pointer, length> pair, not a strbuf, steal the logic from wt_status_truncate_message_at_cut_line() to create a new wt_status_locate_end() helper function that takes <pointer, length> pair, and make ignore_non_trailer() call it to help "interpret-trailers". Since there is only one caller of wt_status_truncate_message_at_cut_line() in cmd_commit(), rewrite it to call wt_status_locate_end() helper instead and remove the old helper that no longer has any caller. Signed-off-by: Brian Malehorn <bmalehorn@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2017-05-16 08:06:49 +02:00
* trailing comment lines and blank lines. To support "git commit -s
* --amend" on an existing commit, we also ignore "Conflicts:". To
* support "git commit -v", we truncate at cut lines.
*
* Returns the number of bytes from the tail to ignore, to be fed as
* the second parameter to append_signoff().
*/
size_t ignore_non_trailer(const char *buf, size_t len)
{
size_t boc = 0;
size_t bol = 0;
int in_old_conflicts_block = 0;
interpret-trailers: honor the cut line If a commit message is edited with the "verbose" option, the buffer will have a cut line and diff after the log message, like so: my subject # ------------------------ >8 ------------------------ # Do not touch the line above. # Everything below will be removed. diff --git a/foo.txt b/foo.txt index 5716ca5..7601807 100644 --- a/foo.txt +++ b/foo.txt @@ -1 +1 @@ -bar +baz "git interpret-trailers" is unaware of the cut line, and assumes the trailer block would be at the end of the whole thing. This can easily be seen with: $ GIT_EDITOR='git interpret-trailers --in-place --trailer Acked-by:me' \ git commit --amend -v Teach "git interpret-trailers" to notice the cut-line and ignore the remainder of the input when looking for a place to add new trailer block. This makes it consistent with how "git commit -v -s" inserts a new Signed-off-by: line. This can be done by the same logic as the existing helper function, wt_status_truncate_message_at_cut_line(), uses, but it wants the caller to pass a strbuf to it. Because the function ignore_non_trailer() used by the command takes a <pointer, length> pair, not a strbuf, steal the logic from wt_status_truncate_message_at_cut_line() to create a new wt_status_locate_end() helper function that takes <pointer, length> pair, and make ignore_non_trailer() call it to help "interpret-trailers". Since there is only one caller of wt_status_truncate_message_at_cut_line() in cmd_commit(), rewrite it to call wt_status_locate_end() helper instead and remove the old helper that no longer has any caller. Signed-off-by: Brian Malehorn <bmalehorn@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2017-05-16 08:06:49 +02:00
size_t cutoff = wt_status_locate_end(buf, len);
interpret-trailers: honor the cut line If a commit message is edited with the "verbose" option, the buffer will have a cut line and diff after the log message, like so: my subject # ------------------------ >8 ------------------------ # Do not touch the line above. # Everything below will be removed. diff --git a/foo.txt b/foo.txt index 5716ca5..7601807 100644 --- a/foo.txt +++ b/foo.txt @@ -1 +1 @@ -bar +baz "git interpret-trailers" is unaware of the cut line, and assumes the trailer block would be at the end of the whole thing. This can easily be seen with: $ GIT_EDITOR='git interpret-trailers --in-place --trailer Acked-by:me' \ git commit --amend -v Teach "git interpret-trailers" to notice the cut-line and ignore the remainder of the input when looking for a place to add new trailer block. This makes it consistent with how "git commit -v -s" inserts a new Signed-off-by: line. This can be done by the same logic as the existing helper function, wt_status_truncate_message_at_cut_line(), uses, but it wants the caller to pass a strbuf to it. Because the function ignore_non_trailer() used by the command takes a <pointer, length> pair, not a strbuf, steal the logic from wt_status_truncate_message_at_cut_line() to create a new wt_status_locate_end() helper function that takes <pointer, length> pair, and make ignore_non_trailer() call it to help "interpret-trailers". Since there is only one caller of wt_status_truncate_message_at_cut_line() in cmd_commit(), rewrite it to call wt_status_locate_end() helper instead and remove the old helper that no longer has any caller. Signed-off-by: Brian Malehorn <bmalehorn@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2017-05-16 08:06:49 +02:00
while (bol < cutoff) {
const char *next_line = memchr(buf + bol, '\n', len - bol);
if (!next_line)
next_line = buf + len;
else
next_line++;
if (buf[bol] == comment_line_char || buf[bol] == '\n') {
/* is this the first of the run of comments? */
if (!boc)
boc = bol;
/* otherwise, it is just continuing */
} else if (starts_with(buf + bol, "Conflicts:\n")) {
in_old_conflicts_block = 1;
if (!boc)
boc = bol;
} else if (in_old_conflicts_block && buf[bol] == '\t') {
; /* a pathname in the conflicts block */
} else if (boc) {
/* the previous was not trailing comment */
boc = 0;
in_old_conflicts_block = 0;
}
bol = next_line - buf;
}
interpret-trailers: honor the cut line If a commit message is edited with the "verbose" option, the buffer will have a cut line and diff after the log message, like so: my subject # ------------------------ >8 ------------------------ # Do not touch the line above. # Everything below will be removed. diff --git a/foo.txt b/foo.txt index 5716ca5..7601807 100644 --- a/foo.txt +++ b/foo.txt @@ -1 +1 @@ -bar +baz "git interpret-trailers" is unaware of the cut line, and assumes the trailer block would be at the end of the whole thing. This can easily be seen with: $ GIT_EDITOR='git interpret-trailers --in-place --trailer Acked-by:me' \ git commit --amend -v Teach "git interpret-trailers" to notice the cut-line and ignore the remainder of the input when looking for a place to add new trailer block. This makes it consistent with how "git commit -v -s" inserts a new Signed-off-by: line. This can be done by the same logic as the existing helper function, wt_status_truncate_message_at_cut_line(), uses, but it wants the caller to pass a strbuf to it. Because the function ignore_non_trailer() used by the command takes a <pointer, length> pair, not a strbuf, steal the logic from wt_status_truncate_message_at_cut_line() to create a new wt_status_locate_end() helper function that takes <pointer, length> pair, and make ignore_non_trailer() call it to help "interpret-trailers". Since there is only one caller of wt_status_truncate_message_at_cut_line() in cmd_commit(), rewrite it to call wt_status_locate_end() helper instead and remove the old helper that no longer has any caller. Signed-off-by: Brian Malehorn <bmalehorn@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2017-05-16 08:06:49 +02:00
return boc ? len - boc : len - cutoff;
}
int run_commit_hook(int editor_is_used, const char *index_file,
hooks: fix an obscure TOCTOU "did we just run a hook?" race Fix a Time-of-check to time-of-use (TOCTOU) race in code added in 680ee550d72 (commit: skip discarding the index if there is no pre-commit hook, 2017-08-14). This obscure race condition can occur if we e.g. ran the "pre-commit" hook and it modified the index, but hook_exists() returns false later on (e.g., because the hook itself went away, the directory became unreadable, etc.). Then we won't call discard_cache() when we should have. The race condition itself probably doesn't matter, and users would have been unlikely to run into it in practice. This problem has been noted on-list when 680ee550d72 was discussed[1], but had not been fixed. This change is mainly intended to improve the readability of the code involved, and to make reasoning about it more straightforward. It wasn't as obvious what we were trying to do here, but by having an "invoked_hook" it's clearer that e.g. our discard_cache() is happening because of the earlier hook execution. Let's also change this for the push-to-checkout hook. Now instead of checking if the hook exists and either doing a push to checkout or a push to deploy we'll always attempt a push to checkout. If the hook doesn't exist we'll fall back on push to deploy. The same behavior as before, without the TOCTOU race. See 0855331941b (receive-pack: support push-to-checkout hook, 2014-12-01) for the introduction of the previous behavior. This leaves uses of hook_exists() in two places that matter. The "reference-transaction" check in refs.c, see 67541597670 (refs: implement reference transaction hook, 2020-06-19), and the "prepare-commit-msg" hook, see 66618a50f9c (sequencer: run 'prepare-commit-msg' hook, 2018-01-24). In both of those cases we're saving ourselves CPU time by not preparing data for the hook that we'll then do nothing with if we don't have the hook. So using this "invoked_hook" pattern doesn't make sense in those cases. The "reference-transaction" and "prepare-commit-msg" hook also aren't racy. In those cases we'll skip the hook runs if we race with a new hook being added, whereas in the TOCTOU races being fixed here we were incorrectly skipping the required post-hook logic. 1. https://lore.kernel.org/git/20170810191613.kpmhzg4seyxy3cpq@sigill.intra.peff.net/ Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-03-07 13:33:46 +01:00
int *invoked_hook, const char *name, ...)
{
struct run_hooks_opt opt = RUN_HOOKS_OPT_INIT;
va_list args;
const char *arg;
strvec_pushf(&opt.env, "GIT_INDEX_FILE=%s", index_file);
/*
* Let the hook know that no editor will be launched.
*/
if (!editor_is_used)
strvec_push(&opt.env, "GIT_EDITOR=:");
va_start(args, name);
while ((arg = va_arg(args, const char *)))
strvec_push(&opt.args, arg);
va_end(args);
hooks: fix "invoked hook" regression in a8cc5943338 Fix a regression in a8cc5943338 (hooks: fix an obscure TOCTOU "did we just run a hook?" race, 2022-03-07): The "invoked_hook" variable passed to run_commit_hook() wasn't passed forward to run_hooks_opt(), as push_to_checkout() in that commit correctly did. Whether we ran the code contingent on having run the hook or not was thus undefined, but in practice on most (all?) modern platforms we'd have run it (almost?) all the time, since stack variables will get initialized to some random value, which most of the time isn't "0". This bug was revealed by running e.g. "t5537-fetch-shallow.sh" with the --valgrind option. Unfortunately running the whole test suite with --valgrind is really slow, so we didn't have a CI job that spotted this. The --valgrind output was: ==31275== Conditional jump or move depends on uninitialised value(s) ==31275== at 0x43C63F: prepare_to_commit (commit.c:1058) ==31275== by 0x4396A5: cmd_commit (commit.c:1722) ==31275== by 0x407C8A: run_builtin (git.c:465) ==31275== by 0x406741: handle_builtin (git.c:718) ==31275== by 0x407665: run_argv (git.c:785) ==31275== by 0x406500: cmd_main (git.c:916) ==31275== by 0x510839: main (common-main.c:56) ==31275== Uninitialised value was created by a stack allocation ==31275== at 0x43B344: prepare_to_commit (commit.c:719) Reported-by: Jonathan Tan <jonathantanmy@google.com> Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-03-22 00:15:13 +01:00
opt.invoked_hook = invoked_hook;
return run_hooks_opt(name, &opt);
}