git/builtin/merge.c

1792 строки
49 KiB
C
Исходник Обычный вид История

/*
* Builtin "git merge"
*
* Copyright (c) 2008 Miklos Vajna <vmiklos@frugalware.org>
*
* Based on git-merge.sh by Junio C Hamano.
*/
#define USE_THE_INDEX_COMPATIBILITY_MACROS
#include "cache.h"
#include "config.h"
#include "parse-options.h"
#include "builtin.h"
#include "lockfile.h"
#include "run-command.h"
#include "hook.h"
#include "diff.h"
#include "diff-merges.h"
#include "refs.h"
#include "refspec.h"
#include "commit.h"
#include "diffcore.h"
#include "revision.h"
#include "unpack-trees.h"
#include "cache-tree.h"
#include "dir.h"
#include "utf8.h"
#include "log-tree.h"
#include "color.h"
#include "rerere.h"
#include "help.h"
#include "merge-recursive.h"
#include "merge-ort-wrappers.h"
#include "resolve-undo.h"
#include "remote.h"
#include "fmt-merge-msg.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 04:23:20 +04:00
#include "gpg-interface.h"
#include "sequencer.h"
#include "string-list.h"
#include "packfile.h"
#include "tag.h"
#include "alias.h"
#include "branch.h"
#include "commit-reach.h"
#include "wt-status.h"
#include "commit-graph.h"
#define DEFAULT_TWOHEAD (1<<0)
#define DEFAULT_OCTOPUS (1<<1)
#define NO_FAST_FORWARD (1<<2)
#define NO_TRIVIAL (1<<3)
struct strategy {
const char *name;
unsigned attr;
};
static const char * const builtin_merge_usage[] = {
N_("git merge [<options>] [<commit>...]"),
"git merge --abort",
"git merge --continue",
NULL
};
static int show_diffstat = 1, shortlog_len = -1, squash;
static int option_commit = -1;
static int option_edit = -1;
static int allow_trivial = 1, have_message, verify_signatures;
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 16:55:57 +03:00
static int check_trust_level = 1;
static int overwrite_ignore = 1;
static struct strbuf merge_msg = STRBUF_INIT;
static struct strategy **use_strategies;
static size_t use_strategies_nr, use_strategies_alloc;
static const char **xopts;
static size_t xopts_nr, xopts_alloc;
static const char *branch;
static char *branch_mergeoptions;
static int verbosity;
static int allow_rerere_auto;
static int abort_current_merge;
static int quit_current_merge;
static int continue_current_merge;
merge: refuse to create too cool a merge by default While it makes sense to allow merging unrelated histories of two projects that started independently into one, in the way "gitk" was merged to "git" itself aka "the coolest merge ever", such a merge is still an unusual event. Worse, if somebody creates an independent history by starting from a tarball of an established project and sends a pull request to the original project, "git merge" however happily creates such a merge without any sign of something unusual is happening. Teach "git merge" to refuse to create such a merge by default, unless the user passes a new "--allow-unrelated-histories" option to tell it that the user is aware that two unrelated projects are merged. Because such a "two project merge" is a rare event, a configuration option to always allow such a merge is not added. We could add the same option to "git pull" and have it passed through to underlying "git merge". I do not have a fundamental opposition against such a feature, but this commit does not do so and instead leaves it as low-hanging fruit for others, because such a "two project merge" would be done after fetching the other project into some location in the working tree of an existing project and making sure how well they fit together, it is sufficient to allow a local merge without such an option pass-through from "git pull" to "git merge". Many tests that are updated by this patch does the pass-through manually by turning: git pull something into its equivalent: git fetch something && git merge --allow-unrelated-histories FETCH_HEAD If somebody is inclined to add such an option, updated tests in this change need to be adjusted back to: git pull --allow-unrelated-histories something Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-03-18 23:21:09 +03:00
static int allow_unrelated_histories;
static int show_progress = -1;
static int default_to_upstream = 1;
static int signoff;
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 04:23:20 +04:00
static const char *sign_commit;
static int autostash;
static int no_verify;
merge: allow to pretend a merge is made into a different branch When a series of patches for a topic-B depends on having topic-A, the workflow to prepare the topic-B branch would look like this: $ git checkout -b topic-B main $ git merge --no-ff --no-edit topic-A $ git am <mbox-for-topic-B When topic-A gets updated, recreating the first merge and rebasing the rest of the topic-B, all on detached HEAD, is a useful technique. After updating topic-A with its new round of patches: $ git checkout topic-B $ prev=$(git rev-parse 'HEAD^{/^Merge branch .topic-A. into}') $ git checkout --detach $prev^1 $ git merge --no-ff --no-edit topic-A $ git rebase --onto HEAD $prev @{-1}^0 $ git checkout -B @{-1} This will (0) check out the current topic-B. (1) find the previous merge of topic-A into topic-B. (2) detach the HEAD to the parent of the previous merge. (3) merge the updated topic-A to it. (4) reapply the patches to rebuild the rest of topic-B. (5) update topic-B with the result. without contaminating the reflog of topic-B too much. topic-B@{1} is the "logically previous" state before topic-A got updated, for example. At (4), comparison (e.g. range-diff) between HEAD and @{-1} is a meaningful way to sanity check the result, and the same can be done at (5) by comparing topic-B and topic-B@{1}. But there is one glitch. The merge into the detached HEAD done in the step (3) above gives us "Merge branch 'topic-A' into HEAD", and does not say "into topic-B". Teach the "--into-name=<branch>" option to "git merge" and its underlying "git fmt-merge-message", to pretend as if we were merging into <branch>, no matter what branch we are actually merging into, when they prepare the merge message. The pretend name honors the usual "into <target>" suppression mechanism, which can be seen in the tests added here. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-12-21 01:53:43 +03:00
static char *into_name;
static struct strategy all_strategy[] = {
Change default merge backend from recursive to ort There are a few reasons to switch the default: * Correctness * Extensibility * Performance I'll provide some summaries about each. === Correctness === The original impetus for a new merge backend was to fix issues that were difficult to fix within recursive's design. The success with this goal is perhaps most easily demonstrated by running the following: $ git grep -2 KNOWN_FAILURE t/ | grep -A 4 GIT_TEST_MERGE_ALGORITHM $ git grep test_expect_merge_algorithm.failure.success t/ $ git grep test_expect_merge_algorithm.success.failure t/ In order, these greps show: * Seven sets of submodule tests (10 total tests) that fail with recursive but succeed with ort * 22 other tests that fail with recursive, but succeed with ort * 0 tests that pass with recursive, but fail with ort === Extensibility === Being able to perform merges without touching the working tree or index makes it possible to create new features that were difficult with the old backend: * Merging, cherry-picking, rebasing, reverting in bare repositories... or just on branches that aren't checked out. * `git diff AUTO_MERGE` -- ability to see what changes the user has made to resolve conflicts so far (see commit 5291828df8 ("merge-ort: write $GIT_DIR/AUTO_MERGE whenever we hit a conflict", 2021-03-20) * A --remerge-diff option for log/show, used to show diffs for merges that display the difference between what an automatic merge would have created and what was recorded in the merge. (This option will often result in an empty diff because many merges are clean, but for the non-clean ones it will show how conflicts were fixed including the removal of conflict markers, and also show additional changes made outside of conflict regions to e.g. fix semantic conflicts.) * A --remerge-diff-only option for log/show, similar to --remerge-diff but also showing how cherry-picks or reverts differed from what an automatic cherry-pick or revert would provide. The last three have been implemented already (though only one has been submitted upstream so far; the others were waiting for performance work to complete), and I still plan to implement the first one. === Performance === I'll quote from the summary of my final optimization for merge-ort (while fixing the testcase name from 'no-renames' to 'few-renames'): Timings Infinite merge- merge- Parallelism recursive recursive of rename merge-ort v2.30.0 current detection current ---------- --------- ----------- --------- few-renames: 18.912 s 18.030 s 11.699 s 198.3 ms mega-renames: 5964.031 s 361.281 s 203.886 s 661.8 ms just-one-mega: 149.583 s 11.009 s 7.553 s 264.6 ms Speedup factors Infinite merge- merge- Parallelism recursive recursive of rename v2.30.0 current detection merge-ort ---------- --------- ----------- --------- few-renames: 1 1.05 1.6 95 mega-renames: 1 16.5 29 9012 just-one-mega: 1 13.6 20 565 And, for partial clone users: Factor reduction in number of objects needed Infinite merge- merge- Parallelism recursive recursive of rename v2.30.0 current detection merge-ort ---------- --------- ----------- --------- mega-renames: 1 1 1 181.3 Signed-off-by: Elijah Newren <newren@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-08-04 08:38:01 +03:00
{ "recursive", NO_TRIVIAL },
{ "octopus", DEFAULT_OCTOPUS },
Change default merge backend from recursive to ort There are a few reasons to switch the default: * Correctness * Extensibility * Performance I'll provide some summaries about each. === Correctness === The original impetus for a new merge backend was to fix issues that were difficult to fix within recursive's design. The success with this goal is perhaps most easily demonstrated by running the following: $ git grep -2 KNOWN_FAILURE t/ | grep -A 4 GIT_TEST_MERGE_ALGORITHM $ git grep test_expect_merge_algorithm.failure.success t/ $ git grep test_expect_merge_algorithm.success.failure t/ In order, these greps show: * Seven sets of submodule tests (10 total tests) that fail with recursive but succeed with ort * 22 other tests that fail with recursive, but succeed with ort * 0 tests that pass with recursive, but fail with ort === Extensibility === Being able to perform merges without touching the working tree or index makes it possible to create new features that were difficult with the old backend: * Merging, cherry-picking, rebasing, reverting in bare repositories... or just on branches that aren't checked out. * `git diff AUTO_MERGE` -- ability to see what changes the user has made to resolve conflicts so far (see commit 5291828df8 ("merge-ort: write $GIT_DIR/AUTO_MERGE whenever we hit a conflict", 2021-03-20) * A --remerge-diff option for log/show, used to show diffs for merges that display the difference between what an automatic merge would have created and what was recorded in the merge. (This option will often result in an empty diff because many merges are clean, but for the non-clean ones it will show how conflicts were fixed including the removal of conflict markers, and also show additional changes made outside of conflict regions to e.g. fix semantic conflicts.) * A --remerge-diff-only option for log/show, similar to --remerge-diff but also showing how cherry-picks or reverts differed from what an automatic cherry-pick or revert would provide. The last three have been implemented already (though only one has been submitted upstream so far; the others were waiting for performance work to complete), and I still plan to implement the first one. === Performance === I'll quote from the summary of my final optimization for merge-ort (while fixing the testcase name from 'no-renames' to 'few-renames'): Timings Infinite merge- merge- Parallelism recursive recursive of rename merge-ort v2.30.0 current detection current ---------- --------- ----------- --------- few-renames: 18.912 s 18.030 s 11.699 s 198.3 ms mega-renames: 5964.031 s 361.281 s 203.886 s 661.8 ms just-one-mega: 149.583 s 11.009 s 7.553 s 264.6 ms Speedup factors Infinite merge- merge- Parallelism recursive recursive of rename v2.30.0 current detection merge-ort ---------- --------- ----------- --------- few-renames: 1 1.05 1.6 95 mega-renames: 1 16.5 29 9012 just-one-mega: 1 13.6 20 565 And, for partial clone users: Factor reduction in number of objects needed Infinite merge- merge- Parallelism recursive recursive of rename v2.30.0 current detection merge-ort ---------- --------- ----------- --------- mega-renames: 1 1 1 181.3 Signed-off-by: Elijah Newren <newren@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-08-04 08:38:01 +03:00
{ "ort", DEFAULT_TWOHEAD | NO_TRIVIAL },
{ "resolve", 0 },
{ "ours", NO_FAST_FORWARD | NO_TRIVIAL },
{ "subtree", NO_FAST_FORWARD | NO_TRIVIAL },
};
static const char *pull_twohead, *pull_octopus;
enum ff_type {
FF_NO,
FF_ALLOW,
FF_ONLY
};
static enum ff_type fast_forward = FF_ALLOW;
static const char *cleanup_arg;
static enum commit_msg_cleanup_mode cleanup_mode;
static int option_parse_message(const struct option *opt,
const char *arg, int unset)
{
struct strbuf *buf = opt->value;
if (unset)
strbuf_setlen(buf, 0);
else if (arg) {
strbuf_addf(buf, "%s%s", buf->len ? "\n\n" : "", arg);
have_message = 1;
} else
return error(_("switch `m' requires a value"));
return 0;
}
static enum parse_opt_result option_read_message(struct parse_opt_ctx_t *ctx,
const struct option *opt,
const char *arg_not_used,
int unset)
{
struct strbuf *buf = opt->value;
const char *arg;
BUG_ON_OPT_ARG(arg_not_used);
if (unset)
BUG("-F cannot be negated");
if (ctx->opt) {
arg = ctx->opt;
ctx->opt = NULL;
} else if (ctx->argc > 1) {
ctx->argc--;
arg = *++ctx->argv;
} else
return error(_("option `%s' requires a value"), opt->long_name);
if (buf->len)
strbuf_addch(buf, '\n');
if (ctx->prefix && !is_absolute_path(arg))
arg = prefix_filename(ctx->prefix, arg);
if (strbuf_read_file(buf, arg, 0) < 0)
return error(_("could not read file '%s'"), arg);
have_message = 1;
return 0;
}
static struct strategy *get_strategy(const char *name)
{
int i;
struct strategy *ret;
static struct cmdnames main_cmds, other_cmds;
static int loaded;
char *default_strategy = getenv("GIT_TEST_MERGE_ALGORITHM");
if (!name)
return NULL;
if (default_strategy &&
!strcmp(default_strategy, "ort") &&
!strcmp(name, "recursive")) {
name = "ort";
}
for (i = 0; i < ARRAY_SIZE(all_strategy); i++)
if (!strcmp(name, all_strategy[i].name))
return &all_strategy[i];
if (!loaded) {
struct cmdnames not_strategies;
loaded = 1;
memset(&not_strategies, 0, sizeof(struct cmdnames));
load_command_list("git-merge-", &main_cmds, &other_cmds);
for (i = 0; i < main_cmds.cnt; i++) {
int j, found = 0;
struct cmdname *ent = main_cmds.names[i];
for (j = 0; j < ARRAY_SIZE(all_strategy); j++)
if (!strncmp(ent->name, all_strategy[j].name, ent->len)
&& !all_strategy[j].name[ent->len])
found = 1;
if (!found)
add_cmdname(&not_strategies, ent->name, ent->len);
}
exclude_cmds(&main_cmds, &not_strategies);
}
if (!is_in_cmdlist(&main_cmds, name) && !is_in_cmdlist(&other_cmds, name)) {
fprintf(stderr, _("Could not find merge strategy '%s'.\n"), name);
fprintf(stderr, _("Available strategies are:"));
for (i = 0; i < main_cmds.cnt; i++)
fprintf(stderr, " %s", main_cmds.names[i]->name);
fprintf(stderr, ".\n");
if (other_cmds.cnt) {
fprintf(stderr, _("Available custom strategies are:"));
for (i = 0; i < other_cmds.cnt; i++)
fprintf(stderr, " %s", other_cmds.names[i]->name);
fprintf(stderr, ".\n");
}
exit(1);
}
CALLOC_ARRAY(ret, 1);
ret->name = xstrdup(name);
ret->attr = NO_TRIVIAL;
return ret;
}
static void append_strategy(struct strategy *s)
{
ALLOC_GROW(use_strategies, use_strategies_nr + 1, use_strategies_alloc);
use_strategies[use_strategies_nr++] = s;
}
static int option_parse_strategy(const struct option *opt,
const char *name, int unset)
{
if (unset)
return 0;
append_strategy(get_strategy(name));
return 0;
}
static int option_parse_x(const struct option *opt,
const char *arg, int unset)
{
if (unset)
return 0;
ALLOC_GROW(xopts, xopts_nr + 1, xopts_alloc);
xopts[xopts_nr++] = xstrdup(arg);
return 0;
}
static int option_parse_n(const struct option *opt,
const char *arg, int unset)
{
assert NOARG/NONEG behavior of parse-options callbacks When we define a parse-options callback, the flags we put in the option struct must match what the callback expects. For example, a callback which does not handle the "unset" parameter should only be used with PARSE_OPT_NONEG. But since the callback and the option struct are not defined next to each other, it's easy to get this wrong (as earlier patches in this series show). Fortunately, the compiler can help us here: compiling with -Wunused-parameters can show us which callbacks ignore their "unset" parameters (and likewise, ones that ignore "arg" expect to be triggered with PARSE_OPT_NOARG). But after we've inspected a callback and determined that all of its callers use the right flags, what do we do next? We'd like to silence the compiler warning, but do so in a way that will catch any wrong calls in the future. We can do that by actually checking those variables and asserting that they match our expectations. Because this is such a common pattern, we'll introduce some helper macros. The resulting messages aren't as descriptive as we could make them, but the file/line information from BUG() is enough to identify the problem (and anyway, the point is that these should never be seen). Each of the annotated callbacks in this patch triggers -Wunused-parameters, and was manually inspected to make sure all callers use the correct options (so none of these BUGs should be triggerable). Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2018-11-05 09:45:42 +03:00
BUG_ON_OPT_ARG(arg);
show_diffstat = unset;
return 0;
}
static struct option builtin_merge_options[] = {
OPT_CALLBACK_F('n', NULL, NULL, NULL,
N_("do not show a diffstat at the end of the merge"),
PARSE_OPT_NOARG, option_parse_n),
OPT_BOOL(0, "stat", &show_diffstat,
N_("show a diffstat at the end of the merge")),
OPT_BOOL(0, "summary", &show_diffstat, N_("(synonym to --stat)")),
{ OPTION_INTEGER, 0, "log", &shortlog_len, N_("n"),
N_("add (at most <n>) entries from shortlog to merge commit message"),
PARSE_OPT_OPTARG, NULL, DEFAULT_MERGE_LOG_LEN },
OPT_BOOL(0, "squash", &squash,
N_("create a single commit instead of doing a merge")),
OPT_BOOL(0, "commit", &option_commit,
N_("perform a commit if the merge succeeds (default)")),
merge: use editor by default in interactive sessions Traditionally, a cleanly resolved merge was committed by "git merge" using the auto-generated merge commit log message without invoking the editor. After 5 years of use in the field, it turns out that people perform too many unjustified merges of the upstream history into their topic branches. These merges are not just useless, but they are often not explained well, and making the end result unreadable when it gets time for merging their history back to their upstream. Earlier we added the "--edit" option to the command, so that people can edit the log message to explain and justify their merge commits. Let's take it one step further and spawn the editor by default when we are in an interactive session (i.e. the standard input and the standard output are pointing at the same tty device). There may be existing scripts that leave the standard input and the standard output of the "git merge" connected to whatever environment the scripts were started, and such invocation might trigger the above "interactive session" heuristics. GIT_MERGE_AUTOEDIT environment variable can be set to "no" at the beginning of such scripts to use the historical behaviour while the script runs. Note that this backward compatibility is meant only for scripts, and we deliberately do *not* support "merge.edit = yes/no/auto" configuration option to allow people to keep the historical behaviour. Suggested-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2012-01-11 10:44:45 +04:00
OPT_BOOL('e', "edit", &option_edit,
N_("edit message before committing")),
OPT_CLEANUP(&cleanup_arg),
OPT_SET_INT(0, "ff", &fast_forward, N_("allow fast-forward (default)"), FF_ALLOW),
OPT_SET_INT_F(0, "ff-only", &fast_forward,
N_("abort if fast-forward is not possible"),
FF_ONLY, PARSE_OPT_NONEG),
OPT_RERERE_AUTOUPDATE(&allow_rerere_auto),
OPT_BOOL(0, "verify-signatures", &verify_signatures,
N_("verify that the named commit has a valid GPG signature")),
OPT_CALLBACK('s', "strategy", &use_strategies, N_("strategy"),
N_("merge strategy to use"), option_parse_strategy),
OPT_CALLBACK('X', "strategy-option", &xopts, N_("option=value"),
N_("option for selected merge strategy"), option_parse_x),
OPT_CALLBACK('m', "message", &merge_msg, N_("message"),
N_("merge commit message (for a non-fast-forward merge)"),
option_parse_message),
{ OPTION_LOWLEVEL_CALLBACK, 'F', "file", &merge_msg, N_("path"),
N_("read message from file"), PARSE_OPT_NONEG,
NULL, 0, option_read_message },
merge: allow to pretend a merge is made into a different branch When a series of patches for a topic-B depends on having topic-A, the workflow to prepare the topic-B branch would look like this: $ git checkout -b topic-B main $ git merge --no-ff --no-edit topic-A $ git am <mbox-for-topic-B When topic-A gets updated, recreating the first merge and rebasing the rest of the topic-B, all on detached HEAD, is a useful technique. After updating topic-A with its new round of patches: $ git checkout topic-B $ prev=$(git rev-parse 'HEAD^{/^Merge branch .topic-A. into}') $ git checkout --detach $prev^1 $ git merge --no-ff --no-edit topic-A $ git rebase --onto HEAD $prev @{-1}^0 $ git checkout -B @{-1} This will (0) check out the current topic-B. (1) find the previous merge of topic-A into topic-B. (2) detach the HEAD to the parent of the previous merge. (3) merge the updated topic-A to it. (4) reapply the patches to rebuild the rest of topic-B. (5) update topic-B with the result. without contaminating the reflog of topic-B too much. topic-B@{1} is the "logically previous" state before topic-A got updated, for example. At (4), comparison (e.g. range-diff) between HEAD and @{-1} is a meaningful way to sanity check the result, and the same can be done at (5) by comparing topic-B and topic-B@{1}. But there is one glitch. The merge into the detached HEAD done in the step (3) above gives us "Merge branch 'topic-A' into HEAD", and does not say "into topic-B". Teach the "--into-name=<branch>" option to "git merge" and its underlying "git fmt-merge-message", to pretend as if we were merging into <branch>, no matter what branch we are actually merging into, when they prepare the merge message. The pretend name honors the usual "into <target>" suppression mechanism, which can be seen in the tests added here. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-12-21 01:53:43 +03:00
OPT_STRING(0, "into-name", &into_name, N_("name"),
N_("use <name> instead of the real target")),
OPT__VERBOSITY(&verbosity),
OPT_BOOL(0, "abort", &abort_current_merge,
N_("abort the current in-progress merge")),
OPT_BOOL(0, "quit", &quit_current_merge,
N_("--abort but leave index and working tree alone")),
OPT_BOOL(0, "continue", &continue_current_merge,
N_("continue the current in-progress merge")),
merge: refuse to create too cool a merge by default While it makes sense to allow merging unrelated histories of two projects that started independently into one, in the way "gitk" was merged to "git" itself aka "the coolest merge ever", such a merge is still an unusual event. Worse, if somebody creates an independent history by starting from a tarball of an established project and sends a pull request to the original project, "git merge" however happily creates such a merge without any sign of something unusual is happening. Teach "git merge" to refuse to create such a merge by default, unless the user passes a new "--allow-unrelated-histories" option to tell it that the user is aware that two unrelated projects are merged. Because such a "two project merge" is a rare event, a configuration option to always allow such a merge is not added. We could add the same option to "git pull" and have it passed through to underlying "git merge". I do not have a fundamental opposition against such a feature, but this commit does not do so and instead leaves it as low-hanging fruit for others, because such a "two project merge" would be done after fetching the other project into some location in the working tree of an existing project and making sure how well they fit together, it is sufficient to allow a local merge without such an option pass-through from "git pull" to "git merge". Many tests that are updated by this patch does the pass-through manually by turning: git pull something into its equivalent: git fetch something && git merge --allow-unrelated-histories FETCH_HEAD If somebody is inclined to add such an option, updated tests in this change need to be adjusted back to: git pull --allow-unrelated-histories something Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-03-18 23:21:09 +03:00
OPT_BOOL(0, "allow-unrelated-histories", &allow_unrelated_histories,
N_("allow merging unrelated histories")),
OPT_SET_INT(0, "progress", &show_progress, N_("force progress reporting"), 1),
{ OPTION_STRING, 'S', "gpg-sign", &sign_commit, N_("key-id"),
N_("GPG sign commit"), PARSE_OPT_OPTARG, NULL, (intptr_t) "" },
OPT_AUTOSTASH(&autostash),
OPT_BOOL(0, "overwrite-ignore", &overwrite_ignore, N_("update ignored files (default)")),
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 04:03:55 +03:00
OPT_BOOL(0, "signoff", &signoff, N_("add a Signed-off-by trailer")),
OPT_BOOL(0, "no-verify", &no_verify, N_("bypass pre-merge-commit and commit-msg hooks")),
OPT_END()
};
static int save_state(struct object_id *stash)
{
int len;
struct child_process cp = CHILD_PROCESS_INIT;
struct strbuf buffer = STRBUF_INIT;
struct lock_file lock_file = LOCK_INIT;
int fd;
int rc = -1;
fd = repo_hold_locked_index(the_repository, &lock_file, 0);
refresh_cache(REFRESH_QUIET);
if (0 <= fd)
repo_update_index_if_able(the_repository, &lock_file);
rollback_lock_file(&lock_file);
strvec_pushl(&cp.args, "stash", "create", NULL);
cp.out = -1;
cp.git_cmd = 1;
if (start_command(&cp))
die(_("could not run stash."));
len = strbuf_read(&buffer, cp.out, 1024);
close(cp.out);
if (finish_command(&cp) || len < 0)
die(_("stash failed"));
else if (!len) /* no changes */
goto out;
strbuf_setlen(&buffer, buffer.len-1);
if (get_oid(buffer.buf, stash))
die(_("not a valid object: %s"), buffer.buf);
rc = 0;
out:
strbuf_release(&buffer);
return rc;
}
static void read_empty(const struct object_id *oid, int verbose)
{
int i = 0;
const char *args[7];
args[i++] = "read-tree";
if (verbose)
args[i++] = "-v";
args[i++] = "-m";
args[i++] = "-u";
args[i++] = empty_tree_oid_hex();
args[i++] = oid_to_hex(oid);
args[i] = NULL;
if (run_command_v_opt(args, RUN_GIT_CMD))
die(_("read-tree failed"));
}
static void reset_hard(const struct object_id *oid, int verbose)
{
int i = 0;
const char *args[6];
args[i++] = "read-tree";
if (verbose)
args[i++] = "-v";
args[i++] = "--reset";
args[i++] = "-u";
args[i++] = oid_to_hex(oid);
args[i] = NULL;
if (run_command_v_opt(args, RUN_GIT_CMD))
die(_("read-tree failed"));
}
static void restore_state(const struct object_id *head,
const struct object_id *stash)
{
merge: make restore_state() restore staged state too There are multiple issues at play here: 1) If `git merge` is invoked with staged changes, it should abort without doing any merging, and the user's working tree and index should be the same as before merge was invoked. 2) Merge strategies are responsible for enforcing the index == HEAD requirement. (See 9822175d2b ("Ensure index matches head before invoking merge machinery, round N", 2019-08-17) for some history around this.) 3) Merge strategies can bail saying they are not an appropriate handler for the merge in question (possibly allowing other strategies to be used instead). 4) Merge strategies can make changes to the index and working tree, and have no expectation to clean up after themselves, *even* if they bail out and say they are not an appropriate handler for the merge in question. (The `octopus` merge strategy does this, for example.) 5) Because of (3) and (4), builtin/merge.c stashes state before trying merge strategies and restores it afterward. Unfortunately, if users had staged changes before calling `git merge`, builtin/merge.c could do the following: * stash the changes, in order to clean up after the strategies * try all the merge strategies in turn, each of which report they cannot function due to the index not matching HEAD * restore the changes via "git stash apply" But that last step would have the net effect of unstaging the user's changes. Fix this by adding the "--index" option to "git stash apply". While at it, also squelch the stash apply output; we already report "Rewinding the tree to pristine..." and don't need a detailed `git status` report afterwards. Also while at it, switch to using strvec so folks don't have to count the arguments to ensure we avoided an off-by-one error, and so it's easier to add additional arguments to the command. Signed-off-by: Elijah Newren <newren@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-07-23 04:53:16 +03:00
struct strvec args = STRVEC_INIT;
if (is_null_oid(stash))
return;
reset_hard(head, 1);
merge: make restore_state() restore staged state too There are multiple issues at play here: 1) If `git merge` is invoked with staged changes, it should abort without doing any merging, and the user's working tree and index should be the same as before merge was invoked. 2) Merge strategies are responsible for enforcing the index == HEAD requirement. (See 9822175d2b ("Ensure index matches head before invoking merge machinery, round N", 2019-08-17) for some history around this.) 3) Merge strategies can bail saying they are not an appropriate handler for the merge in question (possibly allowing other strategies to be used instead). 4) Merge strategies can make changes to the index and working tree, and have no expectation to clean up after themselves, *even* if they bail out and say they are not an appropriate handler for the merge in question. (The `octopus` merge strategy does this, for example.) 5) Because of (3) and (4), builtin/merge.c stashes state before trying merge strategies and restores it afterward. Unfortunately, if users had staged changes before calling `git merge`, builtin/merge.c could do the following: * stash the changes, in order to clean up after the strategies * try all the merge strategies in turn, each of which report they cannot function due to the index not matching HEAD * restore the changes via "git stash apply" But that last step would have the net effect of unstaging the user's changes. Fix this by adding the "--index" option to "git stash apply". While at it, also squelch the stash apply output; we already report "Rewinding the tree to pristine..." and don't need a detailed `git status` report afterwards. Also while at it, switch to using strvec so folks don't have to count the arguments to ensure we avoided an off-by-one error, and so it's easier to add additional arguments to the command. Signed-off-by: Elijah Newren <newren@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-07-23 04:53:16 +03:00
strvec_pushl(&args, "stash", "apply", "--index", "--quiet", NULL);
strvec_push(&args, oid_to_hex(stash));
/*
* It is OK to ignore error here, for example when there was
* nothing to restore.
*/
merge: make restore_state() restore staged state too There are multiple issues at play here: 1) If `git merge` is invoked with staged changes, it should abort without doing any merging, and the user's working tree and index should be the same as before merge was invoked. 2) Merge strategies are responsible for enforcing the index == HEAD requirement. (See 9822175d2b ("Ensure index matches head before invoking merge machinery, round N", 2019-08-17) for some history around this.) 3) Merge strategies can bail saying they are not an appropriate handler for the merge in question (possibly allowing other strategies to be used instead). 4) Merge strategies can make changes to the index and working tree, and have no expectation to clean up after themselves, *even* if they bail out and say they are not an appropriate handler for the merge in question. (The `octopus` merge strategy does this, for example.) 5) Because of (3) and (4), builtin/merge.c stashes state before trying merge strategies and restores it afterward. Unfortunately, if users had staged changes before calling `git merge`, builtin/merge.c could do the following: * stash the changes, in order to clean up after the strategies * try all the merge strategies in turn, each of which report they cannot function due to the index not matching HEAD * restore the changes via "git stash apply" But that last step would have the net effect of unstaging the user's changes. Fix this by adding the "--index" option to "git stash apply". While at it, also squelch the stash apply output; we already report "Rewinding the tree to pristine..." and don't need a detailed `git status` report afterwards. Also while at it, switch to using strvec so folks don't have to count the arguments to ensure we avoided an off-by-one error, and so it's easier to add additional arguments to the command. Signed-off-by: Elijah Newren <newren@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-07-23 04:53:16 +03:00
run_command_v_opt(args.v, RUN_GIT_CMD);
strvec_clear(&args);
refresh_cache(REFRESH_QUIET);
}
/* This is called when no merge was necessary. */
static void finish_up_to_date(void)
{
if (verbosity >= 0) {
if (squash)
puts(_("Already up to date. (nothing to squash)"));
else
puts(_("Already up to date."));
}
remove_merge_branch_state(the_repository);
}
static void squash_message(struct commit *commit, struct commit_list *remoteheads)
{
struct rev_info rev;
struct strbuf out = STRBUF_INIT;
struct commit_list *j;
struct pretty_print_context ctx = {0};
printf(_("Squash commit -- not updating HEAD\n"));
repo_init_revisions(the_repository, &rev, NULL);
diff_merges_suppress(&rev);
rev.commit_format = CMIT_FMT_MEDIUM;
commit->object.flags |= UNINTERESTING;
add_pending_object(&rev, &commit->object, NULL);
for (j = remoteheads; j; j = j->next)
add_pending_object(&rev, &j->item->object, NULL);
setup_revisions(0, NULL, &rev, NULL);
if (prepare_revision_walk(&rev))
die(_("revision walk setup failed"));
ctx.abbrev = rev.abbrev;
ctx.date_mode = rev.date_mode;
ctx.fmt = rev.commit_format;
strbuf_addstr(&out, "Squashed commit of the following:\n");
while ((commit = get_revision(&rev)) != NULL) {
strbuf_addch(&out, '\n');
strbuf_addf(&out, "commit %s\n",
oid_to_hex(&commit->object.oid));
pretty_print_commit(&ctx, commit, &out);
}
write_file_buf(git_path_squash_msg(the_repository), out.buf, out.len);
strbuf_release(&out);
release_revisions(&rev);
}
static void finish(struct commit *head_commit,
struct commit_list *remoteheads,
const struct object_id *new_head, const char *msg)
{
struct strbuf reflog_message = STRBUF_INIT;
const struct object_id *head = &head_commit->object.oid;
if (!msg)
strbuf_addstr(&reflog_message, getenv("GIT_REFLOG_ACTION"));
else {
if (verbosity >= 0)
printf("%s\n", msg);
strbuf_addf(&reflog_message, "%s: %s",
getenv("GIT_REFLOG_ACTION"), msg);
}
if (squash) {
squash_message(head_commit, remoteheads);
} else {
if (verbosity >= 0 && !merge_msg.len)
printf(_("No merge message -- not updating HEAD\n"));
else {
update_ref(reflog_message.buf, "HEAD", new_head, head,
0, UPDATE_REFS_DIE_ON_ERR);
/*
* We ignore errors in 'gc --auto', since the
* user should see them.
*/
run_auto_maintenance(verbosity < 0);
}
}
if (new_head && show_diffstat) {
struct diff_options opts;
repo_diff_setup(the_repository, &opts);
opts.stat_width = -1; /* use full terminal width */
opts.stat_graph_width = -1; /* respect statGraphWidth config */
opts.output_format |=
DIFF_FORMAT_SUMMARY | DIFF_FORMAT_DIFFSTAT;
opts.detect_rename = DIFF_DETECT_RENAME;
diff_setup_done(&opts);
diff_tree_oid(head, new_head, "", &opts);
diffcore_std(&opts);
diff_flush(&opts);
}
/* Run a post-merge hook */
run_hooks_l("post-merge", squash ? "1" : "0", NULL);
apply_autostash(git_path_merge_autostash(the_repository));
strbuf_release(&reflog_message);
}
/* Get the name for the merge commit's message. */
static void merge_name(const char *remote, struct strbuf *msg)
{
struct commit *remote_head;
struct object_id branch_head;
struct strbuf bname = STRBUF_INIT;
struct merge_remote_desc *desc;
const char *ptr;
char *found_ref = NULL;
int len, early;
interpret_branch_name: allow callers to restrict expansions The interpret_branch_name() function converts names like @{-1} and @{upstream} into branch names. The expanded ref names are not fully qualified, and may be outside of the refs/heads/ namespace (e.g., "@" expands to "HEAD", and "@{upstream}" is likely to be in "refs/remotes/"). This is OK for callers like dwim_ref() which are primarily interested in resolving the resulting name, no matter where it is. But callers like "git branch" treat the result as a branch name in refs/heads/. When we expand to a ref outside that namespace, the results are very confusing (e.g., "git branch @" tries to create refs/heads/HEAD, which is nonsense). Callers can't know from the returned string how the expansion happened (e.g., did the user really ask for a branch named "HEAD", or did we do a bogus expansion?). One fix would be to return some out-parameters describing the types of expansion that occurred. This has the benefit that the caller can generate precise error messages ("I understood @{upstream} to mean origin/master, but that is a remote tracking branch, so you cannot create it as a local name"). However, out-parameters make the function interface somewhat cumbersome. Instead, let's do the opposite: let the caller tell us which elements to expand. That's easier to pass in, and none of the callers give more precise error messages than "@{upstream} isn't a valid branch name" anyway (which should be sufficient). The strbuf_branchname() function needs a similar parameter, as most of the callers access interpret_branch_name() through it. We can break the callers down into two groups: 1. Callers that are happy with any kind of ref in the result. We pass "0" here, so they continue to work without restrictions. This includes merge_name(), the reflog handling in add_pending_object_with_path(), and substitute_branch_name(). This last is what powers dwim_ref(). 2. Callers that have funny corner cases (mostly in git-branch and git-checkout). These need to make use of the new parameter, but I've left them as "0" in this patch, and will address them individually in follow-on patches. Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2017-03-02 11:23:01 +03:00
strbuf_branchname(&bname, remote, 0);
remote = bname.buf;
oidclr(&branch_head);
remote_head = get_merge_parent(remote);
if (!remote_head)
die(_("'%s' does not point to a commit"), remote);
if (dwim_ref(remote, strlen(remote), &branch_head, &found_ref, 0) > 0) {
if (starts_with(found_ref, "refs/heads/")) {
strbuf_addf(msg, "%s\t\tbranch '%s' of .\n",
oid_to_hex(&branch_head), remote);
goto cleanup;
}
if (starts_with(found_ref, "refs/tags/")) {
strbuf_addf(msg, "%s\t\ttag '%s' of .\n",
oid_to_hex(&branch_head), remote);
goto cleanup;
}
if (starts_with(found_ref, "refs/remotes/")) {
strbuf_addf(msg, "%s\t\tremote-tracking branch '%s' of .\n",
oid_to_hex(&branch_head), remote);
goto cleanup;
}
}
/* See if remote matches <name>^^^.. or <name>~<number> */
for (len = 0, ptr = remote + strlen(remote);
remote < ptr && ptr[-1] == '^';
ptr--)
len++;
if (len)
early = 1;
else {
early = 0;
ptr = strrchr(remote, '~');
if (ptr) {
int seen_nonzero = 0;
len++; /* count ~ */
while (*++ptr && isdigit(*ptr)) {
seen_nonzero |= (*ptr != '0');
len++;
}
if (*ptr)
len = 0; /* not ...~<number> */
else if (seen_nonzero)
early = 1;
else if (len == 1)
early = 1; /* "name~" is "name~1"! */
}
}
if (len) {
struct strbuf truname = STRBUF_INIT;
strbuf_addf(&truname, "refs/heads/%s", remote);
strbuf_setlen(&truname, truname.len - len);
if (ref_exists(truname.buf)) {
strbuf_addf(msg,
"%s\t\tbranch '%s'%s of .\n",
oid_to_hex(&remote_head->object.oid),
truname.buf + 11,
(early ? " (early part)" : ""));
strbuf_release(&truname);
goto cleanup;
}
strbuf_release(&truname);
}
desc = merge_remote_util(remote_head);
if (desc && desc->obj && desc->obj->type == OBJ_TAG) {
strbuf_addf(msg, "%s\t\t%s '%s'\n",
oid_to_hex(&desc->obj->oid),
type_name(desc->obj->type),
remote);
goto cleanup;
}
strbuf_addf(msg, "%s\t\tcommit '%s'\n",
oid_to_hex(&remote_head->object.oid), remote);
cleanup:
free(found_ref);
strbuf_release(&bname);
}
static void parse_branch_merge_options(char *bmo)
{
const char **argv;
int argc;
if (!bmo)
return;
argc = split_cmdline(bmo, &argv);
if (argc < 0)
die(_("Bad branch.%s.mergeoptions string: %s"), branch,
_(split_cmdline_strerror(argc)));
REALLOC_ARRAY(argv, argc + 2);
MOVE_ARRAY(argv + 1, argv, argc + 1);
argc++;
argv[0] = "branch.*.mergeoptions";
parse_options(argc, argv, NULL, builtin_merge_options,
builtin_merge_usage, 0);
free(argv);
}
static int git_merge_config(const char *k, const char *v, void *cb)
{
int status;
const char *str;
if (branch &&
skip_prefix(k, "branch.", &str) &&
skip_prefix(str, branch, &str) &&
!strcmp(str, ".mergeoptions")) {
free(branch_mergeoptions);
branch_mergeoptions = xstrdup(v);
return 0;
}
if (!strcmp(k, "merge.diffstat") || !strcmp(k, "merge.stat"))
show_diffstat = git_config_bool(k, v);
else if (!strcmp(k, "merge.verifysignatures"))
verify_signatures = git_config_bool(k, v);
else if (!strcmp(k, "pull.twohead"))
return git_config_string(&pull_twohead, k, v);
else if (!strcmp(k, "pull.octopus"))
return git_config_string(&pull_octopus, k, v);
else if (!strcmp(k, "commit.cleanup"))
return git_config_string(&cleanup_arg, k, v);
else if (!strcmp(k, "merge.ff")) {
int boolval = git_parse_maybe_bool(v);
if (0 <= boolval) {
fast_forward = boolval ? FF_ALLOW : FF_NO;
} else if (v && !strcmp(v, "only")) {
fast_forward = FF_ONLY;
} /* do not barf on values from future versions of git */
return 0;
} else if (!strcmp(k, "merge.defaulttoupstream")) {
default_to_upstream = git_config_bool(k, v);
return 0;
} else if (!strcmp(k, "commit.gpgsign")) {
sign_commit = git_config_bool(k, v) ? "" : NULL;
return 0;
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 16:55:57 +03:00
} else if (!strcmp(k, "gpg.mintrustlevel")) {
check_trust_level = 0;
} else if (!strcmp(k, "merge.autostash")) {
autostash = git_config_bool(k, v);
return 0;
}
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 04:23:20 +04:00
status = fmt_merge_msg_config(k, v, cb);
if (status)
return status;
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 04:23:20 +04:00
status = git_gpg_config(k, v, NULL);
if (status)
return status;
return git_diff_ui_config(k, v, cb);
}
static int read_tree_trivial(struct object_id *common, struct object_id *head,
struct object_id *one)
{
int i, nr_trees = 0;
struct tree *trees[MAX_UNPACK_TREES];
struct tree_desc t[MAX_UNPACK_TREES];
struct unpack_trees_options opts;
memset(&opts, 0, sizeof(opts));
opts.head_idx = 2;
opts.src_index = &the_index;
opts.dst_index = &the_index;
opts.update = 1;
opts.verbose_update = 1;
opts.trivial_merges_only = 1;
opts.merge = 1;
opts.preserve_ignored = 0; /* FIXME: !overwrite_ignore */
trees[nr_trees] = parse_tree_indirect(common);
if (!trees[nr_trees++])
return -1;
trees[nr_trees] = parse_tree_indirect(head);
if (!trees[nr_trees++])
return -1;
trees[nr_trees] = parse_tree_indirect(one);
if (!trees[nr_trees++])
return -1;
opts.fn = threeway_merge;
cache_tree_free(&active_cache_tree);
for (i = 0; i < nr_trees; i++) {
parse_tree(trees[i]);
init_tree_desc(t+i, trees[i]->buffer, trees[i]->size);
}
if (unpack_trees(nr_trees, t, &opts))
return -1;
return 0;
}
static void write_tree_trivial(struct object_id *oid)
{
if (write_cache_as_tree(oid, 0, NULL))
die(_("git write-tree failed to write a tree"));
}
static int try_merge_strategy(const char *strategy, struct commit_list *common,
struct commit_list *remoteheads,
struct commit *head)
{
const char *head_arg = "HEAD";
if (refresh_and_write_cache(REFRESH_QUIET, SKIP_IF_UNCHANGED, 0) < 0)
return error(_("Unable to write index."));
if (!strcmp(strategy, "recursive") || !strcmp(strategy, "subtree") ||
!strcmp(strategy, "ort")) {
struct lock_file lock = LOCK_INIT;
int clean, x;
struct commit *result;
struct commit_list *reversed = NULL;
struct merge_options o;
struct commit_list *j;
if (remoteheads->next) {
error(_("Not handling anything other than two heads merge."));
return 2;
}
init_merge_options(&o, the_repository);
if (!strcmp(strategy, "subtree"))
o.subtree_shift = "";
o.show_rename_progress =
show_progress == -1 ? isatty(2) : show_progress;
for (x = 0; x < xopts_nr; x++)
if (parse_merge_opt(&o, xopts[x]))
die(_("unknown strategy option: -X%s"), xopts[x]);
o.branch1 = head_arg;
o.branch2 = merge_remote_util(remoteheads->item)->name;
for (j = common; j; j = j->next)
commit_list_insert(j->item, &reversed);
hold_locked_index(): align error handling with hold_lockfile_for_update() Callers of the hold_locked_index() function pass 0 when they want to prepare to write a new version of the index file without wishing to die or emit an error message when the request fails (e.g. somebody else already held the lock), and pass 1 when they want the call to die upon failure. This option is called LOCK_DIE_ON_ERROR by the underlying lockfile API, and the hold_locked_index() function translates the paramter to LOCK_DIE_ON_ERROR when calling the hold_lock_file_for_update(). Replace these hardcoded '1' with LOCK_DIE_ON_ERROR and stop translating. Callers other than the ones that are replaced with this change pass '0' to the function; no behaviour change is intended with this patch. Signed-off-by: Junio C Hamano <gitster@pobox.com> --- Among the callers of hold_locked_index() that passes 0: - diff.c::refresh_index_quietly() at the end of "git diff" is an opportunistic update; it leaks the lockfile structure but it is just before the program exits and nobody should care. - builtin/describe.c::cmd_describe(), builtin/commit.c::cmd_status(), sequencer.c::read_and_refresh_cache() are all opportunistic updates and they are OK. - builtin/update-index.c::cmd_update_index() takes a lock upfront but we may end up not needing to update the index (i.e. the entries may be fully up-to-date), in which case we do not need to issue an error upon failure to acquire the lock. We do diagnose and die if we indeed need to update, so it is OK. - wt-status.c::require_clean_work_tree() IS BUGGY. It asks silence, does not check the returned value. Compare with callsites like cmd_describe() and cmd_status() to notice that it is wrong to call update_index_if_able() unconditionally.
2016-12-07 21:33:54 +03:00
hold_locked_index(&lock, LOCK_DIE_ON_ERROR);
if (!strcmp(strategy, "ort"))
clean = merge_ort_recursive(&o, head, remoteheads->item,
reversed, &result);
else
clean = merge_recursive(&o, head, remoteheads->item,
reversed, &result);
if (clean < 0) {
rollback_lock_file(&lock);
return 2;
}
if (write_locked_index(&the_index, &lock,
COMMIT_LOCK | SKIP_IF_UNCHANGED))
die(_("unable to write %s"), get_index_file());
return clean ? 0 : 1;
} else {
return try_merge_command(the_repository,
strategy, xopts_nr, xopts,
common, head_arg, remoteheads);
}
}
static void count_diff_files(struct diff_queue_struct *q,
struct diff_options *opt, void *data)
{
int *count = data;
(*count) += q->nr;
}
static int count_unmerged_entries(void)
{
int i, ret = 0;
for (i = 0; i < active_nr; i++)
if (ce_stage(active_cache[i]))
ret++;
return ret;
}
static void add_strategies(const char *string, unsigned attr)
{
int i;
if (string) {
struct string_list list = STRING_LIST_INIT_DUP;
struct string_list_item *item;
string_list_split(&list, string, ' ', -1);
for_each_string_list_item(item, &list)
append_strategy(get_strategy(item->string));
string_list_clear(&list, 0);
return;
}
for (i = 0; i < ARRAY_SIZE(all_strategy); i++)
if (all_strategy[i].attr & attr)
append_strategy(&all_strategy[i]);
}
static void read_merge_msg(struct strbuf *msg)
{
const char *filename = git_path_merge_msg(the_repository);
strbuf_reset(msg);
if (strbuf_read_file(msg, filename, 0) < 0)
die_errno(_("Could not read from '%s'"), filename);
}
static void write_merge_state(struct commit_list *);
static void abort_commit(struct commit_list *remoteheads, const char *err_msg)
{
if (err_msg)
error("%s", err_msg);
fprintf(stderr,
_("Not committing merge; use 'git commit' to complete the merge.\n"));
write_merge_state(remoteheads);
exit(1);
}
static const char merge_editor_comment[] =
N_("Please enter a commit message to explain why this merge is necessary,\n"
"especially if it merges an updated upstream into a topic branch.\n"
"\n");
static const char scissors_editor_comment[] =
N_("An empty message aborts the commit.\n");
static const char no_scissors_editor_comment[] =
N_("Lines starting with '%c' will be ignored, and an empty message aborts\n"
"the commit.\n");
static void write_merge_heads(struct commit_list *);
static void prepare_to_commit(struct commit_list *remoteheads)
{
struct strbuf msg = STRBUF_INIT;
const char *index_file = get_index_file();
if (!no_verify) {
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 15:33:46 +03:00
int invoked_hook;
if (run_commit_hook(0 < option_edit, index_file, &invoked_hook,
"pre-merge-commit", NULL))
abort_commit(remoteheads, NULL);
/*
* Re-read the index as pre-merge-commit hook could have updated it,
* and write it out as a tree. We must do this before we invoke
* the editor and after we invoke run_status above.
*/
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 15:33:46 +03:00
if (invoked_hook)
discard_cache();
}
read_cache_from(index_file);
strbuf_addbuf(&msg, &merge_msg);
if (squash)
BUG("the control must not reach here under --squash");
if (0 < option_edit) {
strbuf_addch(&msg, '\n');
if (cleanup_mode == COMMIT_MSG_CLEANUP_SCISSORS) {
wt_status_append_cut_line(&msg);
strbuf_commented_addf(&msg, "\n");
}
strbuf_commented_addf(&msg, _(merge_editor_comment));
if (cleanup_mode == COMMIT_MSG_CLEANUP_SCISSORS)
strbuf_commented_addf(&msg, _(scissors_editor_comment));
else
strbuf_commented_addf(&msg,
_(no_scissors_editor_comment), comment_line_char);
}
if (signoff)
append_signoff(&msg, ignore_non_trailer(msg.buf, msg.len), 0);
write_merge_heads(remoteheads);
write_file_buf(git_path_merge_msg(the_repository), msg.buf, msg.len);
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 15:33:46 +03:00
if (run_commit_hook(0 < option_edit, get_index_file(), NULL,
"prepare-commit-msg",
git_path_merge_msg(the_repository), "merge", NULL))
abort_commit(remoteheads, NULL);
merge: use editor by default in interactive sessions Traditionally, a cleanly resolved merge was committed by "git merge" using the auto-generated merge commit log message without invoking the editor. After 5 years of use in the field, it turns out that people perform too many unjustified merges of the upstream history into their topic branches. These merges are not just useless, but they are often not explained well, and making the end result unreadable when it gets time for merging their history back to their upstream. Earlier we added the "--edit" option to the command, so that people can edit the log message to explain and justify their merge commits. Let's take it one step further and spawn the editor by default when we are in an interactive session (i.e. the standard input and the standard output are pointing at the same tty device). There may be existing scripts that leave the standard input and the standard output of the "git merge" connected to whatever environment the scripts were started, and such invocation might trigger the above "interactive session" heuristics. GIT_MERGE_AUTOEDIT environment variable can be set to "no" at the beginning of such scripts to use the historical behaviour while the script runs. Note that this backward compatibility is meant only for scripts, and we deliberately do *not* support "merge.edit = yes/no/auto" configuration option to allow people to keep the historical behaviour. Suggested-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2012-01-11 10:44:45 +04:00
if (0 < option_edit) {
if (launch_editor(git_path_merge_msg(the_repository), NULL, NULL))
abort_commit(remoteheads, NULL);
}
builtin/merge: honor commit-msg hook for merges Similar to 65969d43d1 (merge: honor prepare-commit-msg hook, 2011-02-14) merge should also honor the commit-msg hook: When a merge is stopped due to conflicts or --no-commit, the subsequent commit calls the commit-msg hook. However, it is not called after a clean merge. Fix this inconsistency by invoking the hook after clean merges as well. This change is motivated by Gerrit's commit-msg hook to install a ChangeId trailer into the commit message. Without such a ChangeId, Gerrit refuses to accept any commit by default, such that the inconsistency of (not) running the commit-msg hook between commit and merge leads to confusion and might block people from getting their work done. As the githooks man page is very vocal about the possibility of skipping the commit-msg hook via the --no-verify option, implement the option in merge, too. 'git merge --continue' is currently implemented as calling cmd_commit with no further arguments. This works for most other merge related options, such as demonstrated via the --allow-unrelated-histories flag in the test. The --no-verify option however is not remembered across invocations of git-merge. Originally the author assumed an alternative in which the 'git merge --continue' command accepts the --no-verify flag, but that opens up the discussion which flags are allows to the continued merge command and which must be given in the first invocation. Signed-off-by: Stefan Beller <sbeller@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2017-09-08 01:04:29 +03:00
if (!no_verify && run_commit_hook(0 < option_edit, get_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 15:33:46 +03:00
NULL, "commit-msg",
git_path_merge_msg(the_repository), NULL))
builtin/merge: honor commit-msg hook for merges Similar to 65969d43d1 (merge: honor prepare-commit-msg hook, 2011-02-14) merge should also honor the commit-msg hook: When a merge is stopped due to conflicts or --no-commit, the subsequent commit calls the commit-msg hook. However, it is not called after a clean merge. Fix this inconsistency by invoking the hook after clean merges as well. This change is motivated by Gerrit's commit-msg hook to install a ChangeId trailer into the commit message. Without such a ChangeId, Gerrit refuses to accept any commit by default, such that the inconsistency of (not) running the commit-msg hook between commit and merge leads to confusion and might block people from getting their work done. As the githooks man page is very vocal about the possibility of skipping the commit-msg hook via the --no-verify option, implement the option in merge, too. 'git merge --continue' is currently implemented as calling cmd_commit with no further arguments. This works for most other merge related options, such as demonstrated via the --allow-unrelated-histories flag in the test. The --no-verify option however is not remembered across invocations of git-merge. Originally the author assumed an alternative in which the 'git merge --continue' command accepts the --no-verify flag, but that opens up the discussion which flags are allows to the continued merge command and which must be given in the first invocation. Signed-off-by: Stefan Beller <sbeller@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2017-09-08 01:04:29 +03:00
abort_commit(remoteheads, NULL);
read_merge_msg(&msg);
cleanup_message(&msg, cleanup_mode, 0);
if (!msg.len)
abort_commit(remoteheads, _("Empty commit message."));
strbuf_release(&merge_msg);
strbuf_addbuf(&merge_msg, &msg);
strbuf_release(&msg);
}
static int merge_trivial(struct commit *head, struct commit_list *remoteheads)
{
struct object_id result_tree, result_commit;
struct commit_list *parents, **pptr = &parents;
if (refresh_and_write_cache(REFRESH_QUIET, SKIP_IF_UNCHANGED, 0) < 0)
return error(_("Unable to write index."));
write_tree_trivial(&result_tree);
printf(_("Wonderful.\n"));
pptr = commit_list_append(head, pptr);
pptr = commit_list_append(remoteheads->item, pptr);
prepare_to_commit(remoteheads);
if (commit_tree(merge_msg.buf, merge_msg.len, &result_tree, parents,
&result_commit, NULL, sign_commit))
die(_("failed to write commit object"));
finish(head, remoteheads, &result_commit, "In-index merge");
remove_merge_branch_state(the_repository);
return 0;
}
static int finish_automerge(struct commit *head,
int head_subsumed,
struct commit_list *common,
struct commit_list *remoteheads,
struct object_id *result_tree,
const char *wt_strategy)
{
struct commit_list *parents = NULL;
struct strbuf buf = STRBUF_INIT;
struct object_id result_commit;
write_tree_trivial(result_tree);
free_commit_list(common);
parents = remoteheads;
if (!head_subsumed || fast_forward == FF_NO)
commit_list_insert(head, &parents);
prepare_to_commit(remoteheads);
if (commit_tree(merge_msg.buf, merge_msg.len, result_tree, parents,
&result_commit, NULL, sign_commit))
die(_("failed to write commit object"));
strbuf_addf(&buf, "Merge made by the '%s' strategy.", wt_strategy);
finish(head, remoteheads, &result_commit, buf.buf);
strbuf_release(&buf);
remove_merge_branch_state(the_repository);
return 0;
}
static int suggest_conflicts(void)
{
const char *filename;
FILE *fp;
struct strbuf msgbuf = STRBUF_INIT;
filename = git_path_merge_msg(the_repository);
fp = xfopen(filename, "a");
/*
* We can't use cleanup_mode because if we're not using the editor,
* get_cleanup_mode will return COMMIT_MSG_CLEANUP_SPACE instead, even
* though the message is meant to be processed later by git-commit.
* Thus, we will get the cleanup mode which is returned when we _are_
* using an editor.
*/
append_conflicts_hint(&the_index, &msgbuf,
get_cleanup_mode(cleanup_arg, 1));
fputs(msgbuf.buf, fp);
strbuf_release(&msgbuf);
fclose(fp);
repo_rerere(the_repository, allow_rerere_auto);
printf(_("Automatic merge failed; "
"fix conflicts and then commit the result.\n"));
return 1;
}
static int evaluate_result(void)
{
int cnt = 0;
struct rev_info rev;
/* Check how many files differ. */
repo_init_revisions(the_repository, &rev, "");
setup_revisions(0, NULL, &rev, NULL);
rev.diffopt.output_format |=
DIFF_FORMAT_CALLBACK;
rev.diffopt.format_callback = count_diff_files;
rev.diffopt.format_callback_data = &cnt;
run_diff_files(&rev, 0);
/*
* Check how many unmerged entries are
* there.
*/
cnt += count_unmerged_entries();
release_revisions(&rev);
return cnt;
}
/*
* Pretend as if the user told us to merge with the remote-tracking
* branch we have for the upstream of the current branch
*/
static int setup_with_upstream(const char ***argv)
{
struct branch *branch = branch_get(NULL);
int i;
const char **args;
if (!branch)
die(_("No current branch."));
remote.c: drop "remote" pointer from "struct branch" When we create each branch struct, we fill in the "remote_name" field from the config, and then fill in the actual "remote" field (with a "struct remote") based on that name. However, it turns out that nobody really cares about the latter field. The only two sites that access it at all are: 1. git-merge, which uses it to notice when the branch does not have a remote defined. But we can easily replace this with looking at remote_name instead. 2. remote.c itself, when setting up the @{upstream} merge config. But we don't need to save the "remote" in the "struct branch" for that; we can just look it up for the duration of the operation. So there is no need to have both fields; they are redundant with each other (the struct remote contains the name, or you can look up the struct from the name). It would be nice to simplify this, especially as we are going to add matching pushremote config in a future patch (and it would be nice to keep them consistent). So which one do we keep and which one do we get rid of? If we had a lot of callers accessing the struct, it would be more efficient to keep it (since you have to do a lookup to go from the name to the struct, but not vice versa). But we don't have a lot of callers; we have exactly one, so efficiency doesn't matter. We can decide this based on simplicity and readability. And the meaning of the struct value is somewhat unclear. Is it always the remote matching remote_name? If remote_name is NULL (i.e., no per-branch config), does the struct fall back to the "origin" remote, or is it also NULL? These questions will get even more tricky with pushremotes, whose fallback behavior is more complicated. So let's just store the name, which pretty clearly represents the branch.*.remote config. Any lookup or fallback behavior can then be implemented in helper functions. Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2015-05-21 07:45:13 +03:00
if (!branch->remote_name)
die(_("No remote for the current branch."));
if (!branch->merge_nr)
die(_("No default upstream defined for the current branch."));
args = xcalloc(st_add(branch->merge_nr, 1), sizeof(char *));
for (i = 0; i < branch->merge_nr; i++) {
if (!branch->merge[i]->dst)
die(_("No remote-tracking branch for %s from %s"),
branch->merge[i]->src, branch->remote_name);
args[i] = branch->merge[i]->dst;
}
args[i] = NULL;
*argv = args;
return i;
}
static void write_merge_heads(struct commit_list *remoteheads)
{
struct commit_list *j;
struct strbuf buf = STRBUF_INIT;
for (j = remoteheads; j; j = j->next) {
struct object_id *oid;
struct commit *c = j->item;
struct merge_remote_desc *desc;
desc = merge_remote_util(c);
if (desc && desc->obj) {
oid = &desc->obj->oid;
} else {
oid = &c->object.oid;
}
strbuf_addf(&buf, "%s\n", oid_to_hex(oid));
}
write_file_buf(git_path_merge_head(the_repository), buf.buf, buf.len);
strbuf_reset(&buf);
if (fast_forward == FF_NO)
strbuf_addstr(&buf, "no-ff");
write_file_buf(git_path_merge_mode(the_repository), buf.buf, buf.len);
strbuf_release(&buf);
}
static void write_merge_state(struct commit_list *remoteheads)
{
write_merge_heads(remoteheads);
strbuf_addch(&merge_msg, '\n');
write_file_buf(git_path_merge_msg(the_repository), merge_msg.buf,
merge_msg.len);
}
merge: use editor by default in interactive sessions Traditionally, a cleanly resolved merge was committed by "git merge" using the auto-generated merge commit log message without invoking the editor. After 5 years of use in the field, it turns out that people perform too many unjustified merges of the upstream history into their topic branches. These merges are not just useless, but they are often not explained well, and making the end result unreadable when it gets time for merging their history back to their upstream. Earlier we added the "--edit" option to the command, so that people can edit the log message to explain and justify their merge commits. Let's take it one step further and spawn the editor by default when we are in an interactive session (i.e. the standard input and the standard output are pointing at the same tty device). There may be existing scripts that leave the standard input and the standard output of the "git merge" connected to whatever environment the scripts were started, and such invocation might trigger the above "interactive session" heuristics. GIT_MERGE_AUTOEDIT environment variable can be set to "no" at the beginning of such scripts to use the historical behaviour while the script runs. Note that this backward compatibility is meant only for scripts, and we deliberately do *not* support "merge.edit = yes/no/auto" configuration option to allow people to keep the historical behaviour. Suggested-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2012-01-11 10:44:45 +04:00
static int default_edit_option(void)
{
static const char name[] = "GIT_MERGE_AUTOEDIT";
const char *e = getenv(name);
struct stat st_stdin, st_stdout;
if (have_message)
/* an explicit -m msg without --[no-]edit */
return 0;
if (e) {
int v = git_parse_maybe_bool(e);
merge: use editor by default in interactive sessions Traditionally, a cleanly resolved merge was committed by "git merge" using the auto-generated merge commit log message without invoking the editor. After 5 years of use in the field, it turns out that people perform too many unjustified merges of the upstream history into their topic branches. These merges are not just useless, but they are often not explained well, and making the end result unreadable when it gets time for merging their history back to their upstream. Earlier we added the "--edit" option to the command, so that people can edit the log message to explain and justify their merge commits. Let's take it one step further and spawn the editor by default when we are in an interactive session (i.e. the standard input and the standard output are pointing at the same tty device). There may be existing scripts that leave the standard input and the standard output of the "git merge" connected to whatever environment the scripts were started, and such invocation might trigger the above "interactive session" heuristics. GIT_MERGE_AUTOEDIT environment variable can be set to "no" at the beginning of such scripts to use the historical behaviour while the script runs. Note that this backward compatibility is meant only for scripts, and we deliberately do *not* support "merge.edit = yes/no/auto" configuration option to allow people to keep the historical behaviour. Suggested-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2012-01-11 10:44:45 +04:00
if (v < 0)
die(_("Bad value '%s' in environment '%s'"), e, name);
merge: use editor by default in interactive sessions Traditionally, a cleanly resolved merge was committed by "git merge" using the auto-generated merge commit log message without invoking the editor. After 5 years of use in the field, it turns out that people perform too many unjustified merges of the upstream history into their topic branches. These merges are not just useless, but they are often not explained well, and making the end result unreadable when it gets time for merging their history back to their upstream. Earlier we added the "--edit" option to the command, so that people can edit the log message to explain and justify their merge commits. Let's take it one step further and spawn the editor by default when we are in an interactive session (i.e. the standard input and the standard output are pointing at the same tty device). There may be existing scripts that leave the standard input and the standard output of the "git merge" connected to whatever environment the scripts were started, and such invocation might trigger the above "interactive session" heuristics. GIT_MERGE_AUTOEDIT environment variable can be set to "no" at the beginning of such scripts to use the historical behaviour while the script runs. Note that this backward compatibility is meant only for scripts, and we deliberately do *not* support "merge.edit = yes/no/auto" configuration option to allow people to keep the historical behaviour. Suggested-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2012-01-11 10:44:45 +04:00
return v;
}
/* Use editor if stdin and stdout are the same and is a tty */
return (!fstat(0, &st_stdin) &&
!fstat(1, &st_stdout) &&
isatty(0) && isatty(1) &&
merge: use editor by default in interactive sessions Traditionally, a cleanly resolved merge was committed by "git merge" using the auto-generated merge commit log message without invoking the editor. After 5 years of use in the field, it turns out that people perform too many unjustified merges of the upstream history into their topic branches. These merges are not just useless, but they are often not explained well, and making the end result unreadable when it gets time for merging their history back to their upstream. Earlier we added the "--edit" option to the command, so that people can edit the log message to explain and justify their merge commits. Let's take it one step further and spawn the editor by default when we are in an interactive session (i.e. the standard input and the standard output are pointing at the same tty device). There may be existing scripts that leave the standard input and the standard output of the "git merge" connected to whatever environment the scripts were started, and such invocation might trigger the above "interactive session" heuristics. GIT_MERGE_AUTOEDIT environment variable can be set to "no" at the beginning of such scripts to use the historical behaviour while the script runs. Note that this backward compatibility is meant only for scripts, and we deliberately do *not* support "merge.edit = yes/no/auto" configuration option to allow people to keep the historical behaviour. Suggested-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2012-01-11 10:44:45 +04:00
st_stdin.st_dev == st_stdout.st_dev &&
st_stdin.st_ino == st_stdout.st_ino &&
st_stdin.st_mode == st_stdout.st_mode);
}
static struct commit_list *reduce_parents(struct commit *head_commit,
int *head_subsumed,
struct commit_list *remoteheads)
{
struct commit_list *parents, **remotes;
/*
* Is the current HEAD reachable from another commit being
* merged? If so we do not want to record it as a parent of
* the resulting merge, unless --no-ff is given. We will flip
* this variable to 0 when we find HEAD among the independent
* tips being merged.
*/
*head_subsumed = 1;
/* Find what parents to record by checking independent ones. */
parents = reduce_heads(remoteheads);
free_commit_list(remoteheads);
remoteheads = NULL;
remotes = &remoteheads;
while (parents) {
struct commit *commit = pop_commit(&parents);
if (commit == head_commit)
*head_subsumed = 0;
else
remotes = &commit_list_insert(commit, remotes)->next;
}
return remoteheads;
}
merge: use editor by default in interactive sessions Traditionally, a cleanly resolved merge was committed by "git merge" using the auto-generated merge commit log message without invoking the editor. After 5 years of use in the field, it turns out that people perform too many unjustified merges of the upstream history into their topic branches. These merges are not just useless, but they are often not explained well, and making the end result unreadable when it gets time for merging their history back to their upstream. Earlier we added the "--edit" option to the command, so that people can edit the log message to explain and justify their merge commits. Let's take it one step further and spawn the editor by default when we are in an interactive session (i.e. the standard input and the standard output are pointing at the same tty device). There may be existing scripts that leave the standard input and the standard output of the "git merge" connected to whatever environment the scripts were started, and such invocation might trigger the above "interactive session" heuristics. GIT_MERGE_AUTOEDIT environment variable can be set to "no" at the beginning of such scripts to use the historical behaviour while the script runs. Note that this backward compatibility is meant only for scripts, and we deliberately do *not* support "merge.edit = yes/no/auto" configuration option to allow people to keep the historical behaviour. Suggested-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2012-01-11 10:44:45 +04:00
static void prepare_merge_message(struct strbuf *merge_names, struct strbuf *merge_msg)
{
struct fmt_merge_msg_opts opts;
memset(&opts, 0, sizeof(opts));
opts.add_title = !have_message;
opts.shortlog_len = shortlog_len;
opts.credit_people = (0 < option_edit);
merge: allow to pretend a merge is made into a different branch When a series of patches for a topic-B depends on having topic-A, the workflow to prepare the topic-B branch would look like this: $ git checkout -b topic-B main $ git merge --no-ff --no-edit topic-A $ git am <mbox-for-topic-B When topic-A gets updated, recreating the first merge and rebasing the rest of the topic-B, all on detached HEAD, is a useful technique. After updating topic-A with its new round of patches: $ git checkout topic-B $ prev=$(git rev-parse 'HEAD^{/^Merge branch .topic-A. into}') $ git checkout --detach $prev^1 $ git merge --no-ff --no-edit topic-A $ git rebase --onto HEAD $prev @{-1}^0 $ git checkout -B @{-1} This will (0) check out the current topic-B. (1) find the previous merge of topic-A into topic-B. (2) detach the HEAD to the parent of the previous merge. (3) merge the updated topic-A to it. (4) reapply the patches to rebuild the rest of topic-B. (5) update topic-B with the result. without contaminating the reflog of topic-B too much. topic-B@{1} is the "logically previous" state before topic-A got updated, for example. At (4), comparison (e.g. range-diff) between HEAD and @{-1} is a meaningful way to sanity check the result, and the same can be done at (5) by comparing topic-B and topic-B@{1}. But there is one glitch. The merge into the detached HEAD done in the step (3) above gives us "Merge branch 'topic-A' into HEAD", and does not say "into topic-B". Teach the "--into-name=<branch>" option to "git merge" and its underlying "git fmt-merge-message", to pretend as if we were merging into <branch>, no matter what branch we are actually merging into, when they prepare the merge message. The pretend name honors the usual "into <target>" suppression mechanism, which can be seen in the tests added here. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-12-21 01:53:43 +03:00
opts.into_name = into_name;
fmt_merge_msg(merge_names, merge_msg, &opts);
if (merge_msg->len)
strbuf_setlen(merge_msg, merge_msg->len - 1);
}
merge: handle FETCH_HEAD internally The collect_parents() function now is responsible for 1. parsing the commits given on the command line into a list of commits to be merged; 2. filtering these parents into independent ones; and 3. optionally calling fmt_merge_msg() via prepare_merge_message() to prepare an auto-generated merge log message, using fake contents that FETCH_HEAD would have had if these commits were fetched from the current repository with "git pull . $args..." Make "git merge FETCH_HEAD" to be the same as the traditional git merge "$(git fmt-merge-msg <.git/FETCH_HEAD)" $commits invocation of the command in "git pull", where $commits are the ones that appear in FETCH_HEAD that are not marked as not-for-merge, by making it do a bit more, specifically: - noticing "FETCH_HEAD" is the only "commit" on the command line and picking the commits that are not marked as not-for-merge as the list of commits to be merged (substitute for step #1 above); - letting the resulting list fed to step #2 above; - doing the step #3 above, using the contents of the FETCH_HEAD instead of fake contents crafted from the list of commits parsed in the step #1 above. Note that this changes the semantics. "git merge FETCH_HEAD" has always behaved as if the first commit in the FETCH_HEAD file were directly specified on the command line, creating a two-way merge whose auto-generated merge log said "merge commit xyz". With this change, if the previous fetch was to grab multiple branches (e.g. "git fetch $there topic-a topic-b"), the new world order is to create an octopus, behaving as if "git pull $there topic-a topic-b" were run. This is a deliberate change to make that happen, and can be seen in the changes to t3033 tests. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2015-04-26 04:47:21 +03:00
static void handle_fetch_head(struct commit_list **remotes, struct strbuf *merge_names)
{
const char *filename;
int fd, pos, npos;
struct strbuf fetch_head_file = STRBUF_INIT;
const unsigned hexsz = the_hash_algo->hexsz;
merge: handle FETCH_HEAD internally The collect_parents() function now is responsible for 1. parsing the commits given on the command line into a list of commits to be merged; 2. filtering these parents into independent ones; and 3. optionally calling fmt_merge_msg() via prepare_merge_message() to prepare an auto-generated merge log message, using fake contents that FETCH_HEAD would have had if these commits were fetched from the current repository with "git pull . $args..." Make "git merge FETCH_HEAD" to be the same as the traditional git merge "$(git fmt-merge-msg <.git/FETCH_HEAD)" $commits invocation of the command in "git pull", where $commits are the ones that appear in FETCH_HEAD that are not marked as not-for-merge, by making it do a bit more, specifically: - noticing "FETCH_HEAD" is the only "commit" on the command line and picking the commits that are not marked as not-for-merge as the list of commits to be merged (substitute for step #1 above); - letting the resulting list fed to step #2 above; - doing the step #3 above, using the contents of the FETCH_HEAD instead of fake contents crafted from the list of commits parsed in the step #1 above. Note that this changes the semantics. "git merge FETCH_HEAD" has always behaved as if the first commit in the FETCH_HEAD file were directly specified on the command line, creating a two-way merge whose auto-generated merge log said "merge commit xyz". With this change, if the previous fetch was to grab multiple branches (e.g. "git fetch $there topic-a topic-b"), the new world order is to create an octopus, behaving as if "git pull $there topic-a topic-b" were run. This is a deliberate change to make that happen, and can be seen in the changes to t3033 tests. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2015-04-26 04:47:21 +03:00
if (!merge_names)
merge_names = &fetch_head_file;
filename = git_path_fetch_head(the_repository);
fd = xopen(filename, O_RDONLY);
merge: handle FETCH_HEAD internally The collect_parents() function now is responsible for 1. parsing the commits given on the command line into a list of commits to be merged; 2. filtering these parents into independent ones; and 3. optionally calling fmt_merge_msg() via prepare_merge_message() to prepare an auto-generated merge log message, using fake contents that FETCH_HEAD would have had if these commits were fetched from the current repository with "git pull . $args..." Make "git merge FETCH_HEAD" to be the same as the traditional git merge "$(git fmt-merge-msg <.git/FETCH_HEAD)" $commits invocation of the command in "git pull", where $commits are the ones that appear in FETCH_HEAD that are not marked as not-for-merge, by making it do a bit more, specifically: - noticing "FETCH_HEAD" is the only "commit" on the command line and picking the commits that are not marked as not-for-merge as the list of commits to be merged (substitute for step #1 above); - letting the resulting list fed to step #2 above; - doing the step #3 above, using the contents of the FETCH_HEAD instead of fake contents crafted from the list of commits parsed in the step #1 above. Note that this changes the semantics. "git merge FETCH_HEAD" has always behaved as if the first commit in the FETCH_HEAD file were directly specified on the command line, creating a two-way merge whose auto-generated merge log said "merge commit xyz". With this change, if the previous fetch was to grab multiple branches (e.g. "git fetch $there topic-a topic-b"), the new world order is to create an octopus, behaving as if "git pull $there topic-a topic-b" were run. This is a deliberate change to make that happen, and can be seen in the changes to t3033 tests. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2015-04-26 04:47:21 +03:00
if (strbuf_read(merge_names, fd, 0) < 0)
die_errno(_("could not read '%s'"), filename);
if (close(fd) < 0)
die_errno(_("could not close '%s'"), filename);
for (pos = 0; pos < merge_names->len; pos = npos) {
struct object_id oid;
merge: handle FETCH_HEAD internally The collect_parents() function now is responsible for 1. parsing the commits given on the command line into a list of commits to be merged; 2. filtering these parents into independent ones; and 3. optionally calling fmt_merge_msg() via prepare_merge_message() to prepare an auto-generated merge log message, using fake contents that FETCH_HEAD would have had if these commits were fetched from the current repository with "git pull . $args..." Make "git merge FETCH_HEAD" to be the same as the traditional git merge "$(git fmt-merge-msg <.git/FETCH_HEAD)" $commits invocation of the command in "git pull", where $commits are the ones that appear in FETCH_HEAD that are not marked as not-for-merge, by making it do a bit more, specifically: - noticing "FETCH_HEAD" is the only "commit" on the command line and picking the commits that are not marked as not-for-merge as the list of commits to be merged (substitute for step #1 above); - letting the resulting list fed to step #2 above; - doing the step #3 above, using the contents of the FETCH_HEAD instead of fake contents crafted from the list of commits parsed in the step #1 above. Note that this changes the semantics. "git merge FETCH_HEAD" has always behaved as if the first commit in the FETCH_HEAD file were directly specified on the command line, creating a two-way merge whose auto-generated merge log said "merge commit xyz". With this change, if the previous fetch was to grab multiple branches (e.g. "git fetch $there topic-a topic-b"), the new world order is to create an octopus, behaving as if "git pull $there topic-a topic-b" were run. This is a deliberate change to make that happen, and can be seen in the changes to t3033 tests. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2015-04-26 04:47:21 +03:00
char *ptr;
struct commit *commit;
ptr = strchr(merge_names->buf + pos, '\n');
if (ptr)
npos = ptr - merge_names->buf + 1;
else
npos = merge_names->len;
if (npos - pos < hexsz + 2 ||
get_oid_hex(merge_names->buf + pos, &oid))
merge: handle FETCH_HEAD internally The collect_parents() function now is responsible for 1. parsing the commits given on the command line into a list of commits to be merged; 2. filtering these parents into independent ones; and 3. optionally calling fmt_merge_msg() via prepare_merge_message() to prepare an auto-generated merge log message, using fake contents that FETCH_HEAD would have had if these commits were fetched from the current repository with "git pull . $args..." Make "git merge FETCH_HEAD" to be the same as the traditional git merge "$(git fmt-merge-msg <.git/FETCH_HEAD)" $commits invocation of the command in "git pull", where $commits are the ones that appear in FETCH_HEAD that are not marked as not-for-merge, by making it do a bit more, specifically: - noticing "FETCH_HEAD" is the only "commit" on the command line and picking the commits that are not marked as not-for-merge as the list of commits to be merged (substitute for step #1 above); - letting the resulting list fed to step #2 above; - doing the step #3 above, using the contents of the FETCH_HEAD instead of fake contents crafted from the list of commits parsed in the step #1 above. Note that this changes the semantics. "git merge FETCH_HEAD" has always behaved as if the first commit in the FETCH_HEAD file were directly specified on the command line, creating a two-way merge whose auto-generated merge log said "merge commit xyz". With this change, if the previous fetch was to grab multiple branches (e.g. "git fetch $there topic-a topic-b"), the new world order is to create an octopus, behaving as if "git pull $there topic-a topic-b" were run. This is a deliberate change to make that happen, and can be seen in the changes to t3033 tests. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2015-04-26 04:47:21 +03:00
commit = NULL; /* bad */
else if (memcmp(merge_names->buf + pos + hexsz, "\t\t", 2))
merge: handle FETCH_HEAD internally The collect_parents() function now is responsible for 1. parsing the commits given on the command line into a list of commits to be merged; 2. filtering these parents into independent ones; and 3. optionally calling fmt_merge_msg() via prepare_merge_message() to prepare an auto-generated merge log message, using fake contents that FETCH_HEAD would have had if these commits were fetched from the current repository with "git pull . $args..." Make "git merge FETCH_HEAD" to be the same as the traditional git merge "$(git fmt-merge-msg <.git/FETCH_HEAD)" $commits invocation of the command in "git pull", where $commits are the ones that appear in FETCH_HEAD that are not marked as not-for-merge, by making it do a bit more, specifically: - noticing "FETCH_HEAD" is the only "commit" on the command line and picking the commits that are not marked as not-for-merge as the list of commits to be merged (substitute for step #1 above); - letting the resulting list fed to step #2 above; - doing the step #3 above, using the contents of the FETCH_HEAD instead of fake contents crafted from the list of commits parsed in the step #1 above. Note that this changes the semantics. "git merge FETCH_HEAD" has always behaved as if the first commit in the FETCH_HEAD file were directly specified on the command line, creating a two-way merge whose auto-generated merge log said "merge commit xyz". With this change, if the previous fetch was to grab multiple branches (e.g. "git fetch $there topic-a topic-b"), the new world order is to create an octopus, behaving as if "git pull $there topic-a topic-b" were run. This is a deliberate change to make that happen, and can be seen in the changes to t3033 tests. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2015-04-26 04:47:21 +03:00
continue; /* not-for-merge */
else {
char saved = merge_names->buf[pos + hexsz];
merge_names->buf[pos + hexsz] = '\0';
merge: handle FETCH_HEAD internally The collect_parents() function now is responsible for 1. parsing the commits given on the command line into a list of commits to be merged; 2. filtering these parents into independent ones; and 3. optionally calling fmt_merge_msg() via prepare_merge_message() to prepare an auto-generated merge log message, using fake contents that FETCH_HEAD would have had if these commits were fetched from the current repository with "git pull . $args..." Make "git merge FETCH_HEAD" to be the same as the traditional git merge "$(git fmt-merge-msg <.git/FETCH_HEAD)" $commits invocation of the command in "git pull", where $commits are the ones that appear in FETCH_HEAD that are not marked as not-for-merge, by making it do a bit more, specifically: - noticing "FETCH_HEAD" is the only "commit" on the command line and picking the commits that are not marked as not-for-merge as the list of commits to be merged (substitute for step #1 above); - letting the resulting list fed to step #2 above; - doing the step #3 above, using the contents of the FETCH_HEAD instead of fake contents crafted from the list of commits parsed in the step #1 above. Note that this changes the semantics. "git merge FETCH_HEAD" has always behaved as if the first commit in the FETCH_HEAD file were directly specified on the command line, creating a two-way merge whose auto-generated merge log said "merge commit xyz". With this change, if the previous fetch was to grab multiple branches (e.g. "git fetch $there topic-a topic-b"), the new world order is to create an octopus, behaving as if "git pull $there topic-a topic-b" were run. This is a deliberate change to make that happen, and can be seen in the changes to t3033 tests. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2015-04-26 04:47:21 +03:00
commit = get_merge_parent(merge_names->buf + pos);
merge_names->buf[pos + hexsz] = saved;
merge: handle FETCH_HEAD internally The collect_parents() function now is responsible for 1. parsing the commits given on the command line into a list of commits to be merged; 2. filtering these parents into independent ones; and 3. optionally calling fmt_merge_msg() via prepare_merge_message() to prepare an auto-generated merge log message, using fake contents that FETCH_HEAD would have had if these commits were fetched from the current repository with "git pull . $args..." Make "git merge FETCH_HEAD" to be the same as the traditional git merge "$(git fmt-merge-msg <.git/FETCH_HEAD)" $commits invocation of the command in "git pull", where $commits are the ones that appear in FETCH_HEAD that are not marked as not-for-merge, by making it do a bit more, specifically: - noticing "FETCH_HEAD" is the only "commit" on the command line and picking the commits that are not marked as not-for-merge as the list of commits to be merged (substitute for step #1 above); - letting the resulting list fed to step #2 above; - doing the step #3 above, using the contents of the FETCH_HEAD instead of fake contents crafted from the list of commits parsed in the step #1 above. Note that this changes the semantics. "git merge FETCH_HEAD" has always behaved as if the first commit in the FETCH_HEAD file were directly specified on the command line, creating a two-way merge whose auto-generated merge log said "merge commit xyz". With this change, if the previous fetch was to grab multiple branches (e.g. "git fetch $there topic-a topic-b"), the new world order is to create an octopus, behaving as if "git pull $there topic-a topic-b" were run. This is a deliberate change to make that happen, and can be seen in the changes to t3033 tests. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2015-04-26 04:47:21 +03:00
}
if (!commit) {
if (ptr)
*ptr = '\0';
die(_("not something we can merge in %s: %s"),
merge: handle FETCH_HEAD internally The collect_parents() function now is responsible for 1. parsing the commits given on the command line into a list of commits to be merged; 2. filtering these parents into independent ones; and 3. optionally calling fmt_merge_msg() via prepare_merge_message() to prepare an auto-generated merge log message, using fake contents that FETCH_HEAD would have had if these commits were fetched from the current repository with "git pull . $args..." Make "git merge FETCH_HEAD" to be the same as the traditional git merge "$(git fmt-merge-msg <.git/FETCH_HEAD)" $commits invocation of the command in "git pull", where $commits are the ones that appear in FETCH_HEAD that are not marked as not-for-merge, by making it do a bit more, specifically: - noticing "FETCH_HEAD" is the only "commit" on the command line and picking the commits that are not marked as not-for-merge as the list of commits to be merged (substitute for step #1 above); - letting the resulting list fed to step #2 above; - doing the step #3 above, using the contents of the FETCH_HEAD instead of fake contents crafted from the list of commits parsed in the step #1 above. Note that this changes the semantics. "git merge FETCH_HEAD" has always behaved as if the first commit in the FETCH_HEAD file were directly specified on the command line, creating a two-way merge whose auto-generated merge log said "merge commit xyz". With this change, if the previous fetch was to grab multiple branches (e.g. "git fetch $there topic-a topic-b"), the new world order is to create an octopus, behaving as if "git pull $there topic-a topic-b" were run. This is a deliberate change to make that happen, and can be seen in the changes to t3033 tests. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2015-04-26 04:47:21 +03:00
filename, merge_names->buf + pos);
}
remotes = &commit_list_insert(commit, remotes)->next;
}
if (merge_names == &fetch_head_file)
strbuf_release(&fetch_head_file);
}
static struct commit_list *collect_parents(struct commit *head_commit,
int *head_subsumed,
int argc, const char **argv,
struct strbuf *merge_msg)
{
int i;
struct commit_list *remoteheads = NULL;
struct commit_list **remotes = &remoteheads;
struct strbuf merge_names = STRBUF_INIT, *autogen = NULL;
if (merge_msg && (!have_message || shortlog_len))
autogen = &merge_names;
if (head_commit)
remotes = &commit_list_insert(head_commit, remotes)->next;
merge: handle FETCH_HEAD internally The collect_parents() function now is responsible for 1. parsing the commits given on the command line into a list of commits to be merged; 2. filtering these parents into independent ones; and 3. optionally calling fmt_merge_msg() via prepare_merge_message() to prepare an auto-generated merge log message, using fake contents that FETCH_HEAD would have had if these commits were fetched from the current repository with "git pull . $args..." Make "git merge FETCH_HEAD" to be the same as the traditional git merge "$(git fmt-merge-msg <.git/FETCH_HEAD)" $commits invocation of the command in "git pull", where $commits are the ones that appear in FETCH_HEAD that are not marked as not-for-merge, by making it do a bit more, specifically: - noticing "FETCH_HEAD" is the only "commit" on the command line and picking the commits that are not marked as not-for-merge as the list of commits to be merged (substitute for step #1 above); - letting the resulting list fed to step #2 above; - doing the step #3 above, using the contents of the FETCH_HEAD instead of fake contents crafted from the list of commits parsed in the step #1 above. Note that this changes the semantics. "git merge FETCH_HEAD" has always behaved as if the first commit in the FETCH_HEAD file were directly specified on the command line, creating a two-way merge whose auto-generated merge log said "merge commit xyz". With this change, if the previous fetch was to grab multiple branches (e.g. "git fetch $there topic-a topic-b"), the new world order is to create an octopus, behaving as if "git pull $there topic-a topic-b" were run. This is a deliberate change to make that happen, and can be seen in the changes to t3033 tests. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2015-04-26 04:47:21 +03:00
if (argc == 1 && !strcmp(argv[0], "FETCH_HEAD")) {
handle_fetch_head(remotes, autogen);
remoteheads = reduce_parents(head_commit, head_subsumed, remoteheads);
} else {
for (i = 0; i < argc; i++) {
struct commit *commit = get_merge_parent(argv[i]);
if (!commit)
help_unknown_ref(argv[i], "merge",
_("not something we can merge"));
merge: handle FETCH_HEAD internally The collect_parents() function now is responsible for 1. parsing the commits given on the command line into a list of commits to be merged; 2. filtering these parents into independent ones; and 3. optionally calling fmt_merge_msg() via prepare_merge_message() to prepare an auto-generated merge log message, using fake contents that FETCH_HEAD would have had if these commits were fetched from the current repository with "git pull . $args..." Make "git merge FETCH_HEAD" to be the same as the traditional git merge "$(git fmt-merge-msg <.git/FETCH_HEAD)" $commits invocation of the command in "git pull", where $commits are the ones that appear in FETCH_HEAD that are not marked as not-for-merge, by making it do a bit more, specifically: - noticing "FETCH_HEAD" is the only "commit" on the command line and picking the commits that are not marked as not-for-merge as the list of commits to be merged (substitute for step #1 above); - letting the resulting list fed to step #2 above; - doing the step #3 above, using the contents of the FETCH_HEAD instead of fake contents crafted from the list of commits parsed in the step #1 above. Note that this changes the semantics. "git merge FETCH_HEAD" has always behaved as if the first commit in the FETCH_HEAD file were directly specified on the command line, creating a two-way merge whose auto-generated merge log said "merge commit xyz". With this change, if the previous fetch was to grab multiple branches (e.g. "git fetch $there topic-a topic-b"), the new world order is to create an octopus, behaving as if "git pull $there topic-a topic-b" were run. This is a deliberate change to make that happen, and can be seen in the changes to t3033 tests. Signed-off-by: Junio C Hamano <gitster@pobox.com>
2015-04-26 04:47:21 +03:00
remotes = &commit_list_insert(commit, remotes)->next;
}
remoteheads = reduce_parents(head_commit, head_subsumed, remoteheads);
if (autogen) {
struct commit_list *p;
for (p = remoteheads; p; p = p->next)
merge_name(merge_remote_util(p->item)->name, autogen);
}
}
if (autogen) {
prepare_merge_message(autogen, merge_msg);
strbuf_release(autogen);
}
return remoteheads;
}
static int merging_a_throwaway_tag(struct commit *commit)
{
char *tag_ref;
struct object_id oid;
int is_throwaway_tag = 0;
/* Are we merging a tag? */
if (!merge_remote_util(commit) ||
!merge_remote_util(commit)->obj ||
merge_remote_util(commit)->obj->type != OBJ_TAG)
return is_throwaway_tag;
/*
* Now we know we are merging a tag object. Are we downstream
* and following the tags from upstream? If so, we must have
* the tag object pointed at by "refs/tags/$T" where $T is the
* tagname recorded in the tag object. We want to allow such
* a "just to catch up" merge to fast-forward.
*
* Otherwise, we are playing an integrator's role, making a
* merge with a throw-away tag from a contributor with
* something like "git pull $contributor $signed_tag".
* We want to forbid such a merge from fast-forwarding
* by default; otherwise we would not keep the signature
* anywhere.
*/
tag_ref = xstrfmt("refs/tags/%s",
((struct tag *)merge_remote_util(commit)->obj)->tag);
if (!read_ref(tag_ref, &oid) &&
oideq(&oid, &merge_remote_util(commit)->obj->oid))
is_throwaway_tag = 0;
else
is_throwaway_tag = 1;
free(tag_ref);
return is_throwaway_tag;
}
int cmd_merge(int argc, const char **argv, const char *prefix)
{
struct object_id result_tree, stash, head_oid;
struct commit *head_commit;
struct strbuf buf = STRBUF_INIT;
int i, ret = 0, head_subsumed;
int best_cnt = -1, merge_was_ok = 0, automerge_was_ok = 0;
struct commit_list *common = NULL;
const char *best_strategy = NULL, *wt_strategy = NULL;
merge: fix memory leaks in cmd_merge() There were two commit_lists created in cmd_merge() that were only conditionally free()'d. Add a quick conditional call to free_commit_list() for each of them at the end of the function. Testing this commit against t6404 under valgrind shows that this patch fixes the following two leaks: 16 bytes in 1 blocks are definitely lost in loss record 16 of 126 at 0x484086F: malloc (vg_replace_malloc.c:380) by 0x69FFEB: do_xmalloc (wrapper.c:41) by 0x6A0073: xmalloc (wrapper.c:62) by 0x52A72D: commit_list_insert (commit.c:556) by 0x47FC93: reduce_parents (merge.c:1114) by 0x4801EE: collect_parents (merge.c:1214) by 0x480B56: cmd_merge (merge.c:1465) by 0x40686E: run_builtin (git.c:464) by 0x406C51: handle_builtin (git.c:716) by 0x406E96: run_argv (git.c:783) by 0x40730A: cmd_main (git.c:914) by 0x4E7DFA: main (common-main.c:56) 8 (16 direct, 32 indirect) bytes in 1 blocks are definitely lost in \ loss record 61 of 126 at 0x484086F: malloc (vg_replace_malloc.c:380) by 0x69FFEB: do_xmalloc (wrapper.c:41) by 0x6A0073: xmalloc (wrapper.c:62) by 0x52A72D: commit_list_insert (commit.c:556) by 0x52A8F2: commit_list_insert_by_date (commit.c:620) by 0x5270AC: get_merge_bases_many_0 (commit-reach.c:413) by 0x52716C: repo_get_merge_bases (commit-reach.c:438) by 0x480E5A: cmd_merge (merge.c:1520) by 0x40686E: run_builtin (git.c:464) by 0x406C51: handle_builtin (git.c:716) by 0x406E96: run_argv (git.c:783) by 0x40730A: cmd_main (git.c:914) There are still 3 leaks in chdir_notify_register() after this, but chdir_notify_register() has been brought up on the list before and folks were not a fan of fixing those, so I'm not touching them. Signed-off-by: Elijah Newren <newren@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-01-20 10:47:15 +03:00
struct commit_list *remoteheads = NULL, *p;
void *branch_to_free;
int orig_argc = argc;
if (argc == 2 && !strcmp(argv[1], "-h"))
usage_with_options(builtin_merge_usage, builtin_merge_options);
prepare_repo_settings(the_repository);
the_repository->settings.command_requires_full_index = 0;
/*
* Check if we are _not_ on a detached HEAD, i.e. if there is a
* current branch.
*/
branch = branch_to_free = resolve_refdup("HEAD", 0, &head_oid, NULL);
if (branch)
skip_prefix(branch, "refs/heads/", &branch);
if (!pull_twohead) {
char *default_strategy = getenv("GIT_TEST_MERGE_ALGORITHM");
if (default_strategy && !strcmp(default_strategy, "ort"))
pull_twohead = "ort";
}
init_diff_ui_defaults();
git_config(git_merge_config, NULL);
if (!branch || is_null_oid(&head_oid))
head_commit = NULL;
else
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 01:10:10 +03:00
head_commit = lookup_commit_or_die(&head_oid, "HEAD");
if (branch_mergeoptions)
parse_branch_merge_options(branch_mergeoptions);
argc = parse_options(argc, argv, prefix, builtin_merge_options,
builtin_merge_usage, 0);
if (shortlog_len < 0)
shortlog_len = (merge_log_config > 0) ? merge_log_config : 0;
if (verbosity < 0 && show_progress == -1)
show_progress = 0;
if (abort_current_merge) {
int nargc = 2;
const char *nargv[] = {"reset", "--merge", NULL};
struct strbuf stash_oid = STRBUF_INIT;
if (orig_argc != 2)
usage_msg_opt(_("--abort expects no arguments"),
builtin_merge_usage, builtin_merge_options);
if (!file_exists(git_path_merge_head(the_repository)))
die(_("There is no merge to abort (MERGE_HEAD missing)."));
if (read_oneliner(&stash_oid, git_path_merge_autostash(the_repository),
READ_ONELINER_SKIP_IF_EMPTY))
unlink(git_path_merge_autostash(the_repository));
/* Invoke 'git reset --merge' */
ret = cmd_reset(nargc, nargv, prefix);
if (stash_oid.len)
apply_autostash_oid(stash_oid.buf);
strbuf_release(&stash_oid);
goto done;
}
if (quit_current_merge) {
if (orig_argc != 2)
usage_msg_opt(_("--quit expects no arguments"),
builtin_merge_usage,
builtin_merge_options);
remove_merge_branch_state(the_repository);
goto done;
}
if (continue_current_merge) {
int nargc = 1;
const char *nargv[] = {"commit", NULL};
if (orig_argc != 2)
usage_msg_opt(_("--continue expects no arguments"),
builtin_merge_usage, builtin_merge_options);
if (!file_exists(git_path_merge_head(the_repository)))
die(_("There is no merge in progress (MERGE_HEAD missing)."));
/* Invoke 'git commit' */
ret = cmd_commit(nargc, nargv, prefix);
goto done;
}
if (read_cache_unmerged())
die_resolve_conflict("merge");
if (file_exists(git_path_merge_head(the_repository))) {
/*
* There is no unmerged entry, don't advise 'git
* add/rm <file>', just 'git commit'.
*/
if (advice_enabled(ADVICE_RESOLVE_CONFLICT))
die(_("You have not concluded your merge (MERGE_HEAD exists).\n"
"Please, commit your changes before you merge."));
else
die(_("You have not concluded your merge (MERGE_HEAD exists)."));
}
if (ref_exists("CHERRY_PICK_HEAD")) {
if (advice_enabled(ADVICE_RESOLVE_CONFLICT))
die(_("You have not concluded your cherry-pick (CHERRY_PICK_HEAD exists).\n"
"Please, commit your changes before you merge."));
else
die(_("You have not concluded your cherry-pick (CHERRY_PICK_HEAD exists)."));
}
resolve_undo_clear();
if (option_edit < 0)
option_edit = default_edit_option();
cleanup_mode = get_cleanup_mode(cleanup_arg, 0 < option_edit);
if (verbosity < 0)
show_diffstat = 0;
if (squash) {
if (fast_forward == FF_NO)
die(_("options '%s' and '%s' cannot be used together"), "--squash", "--no-ff.");
if (option_commit > 0)
die(_("options '%s' and '%s' cannot be used together"), "--squash", "--commit.");
/*
* squash can now silently disable option_commit - this is not
* a problem as it is only overriding the default, not a user
* supplied option.
*/
option_commit = 0;
}
if (option_commit < 0)
option_commit = 1;
if (!argc) {
if (default_to_upstream)
argc = setup_with_upstream(&argv);
else
die(_("No commit specified and merge.defaultToUpstream not set."));
} else if (argc == 1 && !strcmp(argv[0], "-")) {
argv[0] = "@{-1}";
}
if (!argc)
usage_with_options(builtin_merge_usage,
builtin_merge_options);
if (!head_commit) {
/*
* If the merged head is a valid one there is no reason
* to forbid "git merge" into a branch yet to be born.
* We do the same for "git pull".
*/
struct object_id *remote_head_oid;
if (squash)
die(_("Squash commit into empty head not supported yet"));
if (fast_forward == FF_NO)
die(_("Non-fast-forward commit does not make sense into "
"an empty head"));
remoteheads = collect_parents(head_commit, &head_subsumed,
argc, argv, NULL);
if (!remoteheads)
die(_("%s - not something we can merge"), argv[0]);
if (remoteheads->next)
die(_("Can merge only exactly one commit into empty head"));
if (verify_signatures)
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 16:55:57 +03:00
verify_merge_signature(remoteheads->item, verbosity,
check_trust_level);
remote_head_oid = &remoteheads->item->object.oid;
read_empty(remote_head_oid, 0);
update_ref("initial pull", "HEAD", remote_head_oid, NULL, 0,
UPDATE_REFS_DIE_ON_ERR);
goto done;
}
/*
* All the rest are the commits being merged; prepare
* the standard merge summary message to be appended
* to the given message.
*/
remoteheads = collect_parents(head_commit, &head_subsumed,
argc, argv, &merge_msg);
if (!head_commit || !argc)
usage_with_options(builtin_merge_usage,
builtin_merge_options);
if (verify_signatures) {
for (p = remoteheads; p; p = p->next) {
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 16:55:57 +03:00
verify_merge_signature(p->item, verbosity,
check_trust_level);
}
}
strbuf_addstr(&buf, "merge");
for (p = remoteheads; p; p = p->next)
strbuf_addf(&buf, " %s", merge_remote_util(p->item)->name);
setenv("GIT_REFLOG_ACTION", buf.buf, 0);
strbuf_reset(&buf);
for (p = remoteheads; p; p = p->next) {
struct commit *commit = p->item;
strbuf_addf(&buf, "GITHEAD_%s",
oid_to_hex(&commit->object.oid));
setenv(buf.buf, merge_remote_util(commit)->name, 1);
strbuf_reset(&buf);
if (fast_forward != FF_ONLY && merging_a_throwaway_tag(commit))
fast_forward = FF_NO;
}
Change default merge backend from recursive to ort There are a few reasons to switch the default: * Correctness * Extensibility * Performance I'll provide some summaries about each. === Correctness === The original impetus for a new merge backend was to fix issues that were difficult to fix within recursive's design. The success with this goal is perhaps most easily demonstrated by running the following: $ git grep -2 KNOWN_FAILURE t/ | grep -A 4 GIT_TEST_MERGE_ALGORITHM $ git grep test_expect_merge_algorithm.failure.success t/ $ git grep test_expect_merge_algorithm.success.failure t/ In order, these greps show: * Seven sets of submodule tests (10 total tests) that fail with recursive but succeed with ort * 22 other tests that fail with recursive, but succeed with ort * 0 tests that pass with recursive, but fail with ort === Extensibility === Being able to perform merges without touching the working tree or index makes it possible to create new features that were difficult with the old backend: * Merging, cherry-picking, rebasing, reverting in bare repositories... or just on branches that aren't checked out. * `git diff AUTO_MERGE` -- ability to see what changes the user has made to resolve conflicts so far (see commit 5291828df8 ("merge-ort: write $GIT_DIR/AUTO_MERGE whenever we hit a conflict", 2021-03-20) * A --remerge-diff option for log/show, used to show diffs for merges that display the difference between what an automatic merge would have created and what was recorded in the merge. (This option will often result in an empty diff because many merges are clean, but for the non-clean ones it will show how conflicts were fixed including the removal of conflict markers, and also show additional changes made outside of conflict regions to e.g. fix semantic conflicts.) * A --remerge-diff-only option for log/show, similar to --remerge-diff but also showing how cherry-picks or reverts differed from what an automatic cherry-pick or revert would provide. The last three have been implemented already (though only one has been submitted upstream so far; the others were waiting for performance work to complete), and I still plan to implement the first one. === Performance === I'll quote from the summary of my final optimization for merge-ort (while fixing the testcase name from 'no-renames' to 'few-renames'): Timings Infinite merge- merge- Parallelism recursive recursive of rename merge-ort v2.30.0 current detection current ---------- --------- ----------- --------- few-renames: 18.912 s 18.030 s 11.699 s 198.3 ms mega-renames: 5964.031 s 361.281 s 203.886 s 661.8 ms just-one-mega: 149.583 s 11.009 s 7.553 s 264.6 ms Speedup factors Infinite merge- merge- Parallelism recursive recursive of rename v2.30.0 current detection merge-ort ---------- --------- ----------- --------- few-renames: 1 1.05 1.6 95 mega-renames: 1 16.5 29 9012 just-one-mega: 1 13.6 20 565 And, for partial clone users: Factor reduction in number of objects needed Infinite merge- merge- Parallelism recursive recursive of rename v2.30.0 current detection merge-ort ---------- --------- ----------- --------- mega-renames: 1 1 1 181.3 Signed-off-by: Elijah Newren <newren@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-08-04 08:38:01 +03:00
if (!use_strategies && !pull_twohead &&
remoteheads && !remoteheads->next) {
char *default_strategy = getenv("GIT_TEST_MERGE_ALGORITHM");
if (default_strategy)
append_strategy(get_strategy(default_strategy));
}
if (!use_strategies) {
if (!remoteheads)
; /* already up-to-date */
else if (!remoteheads->next)
add_strategies(pull_twohead, DEFAULT_TWOHEAD);
else
add_strategies(pull_octopus, DEFAULT_OCTOPUS);
}
for (i = 0; i < use_strategies_nr; i++) {
if (use_strategies[i]->attr & NO_FAST_FORWARD)
fast_forward = FF_NO;
if (use_strategies[i]->attr & NO_TRIVIAL)
allow_trivial = 0;
}
if (!remoteheads)
; /* already up-to-date */
else if (!remoteheads->next)
common = get_merge_bases(head_commit, remoteheads->item);
else {
struct commit_list *list = remoteheads;
commit_list_insert(head_commit, &list);
common = get_octopus_merge_bases(list);
free(list);
}
update_ref("updating ORIG_HEAD", "ORIG_HEAD",
&head_commit->object.oid, NULL, 0, UPDATE_REFS_DIE_ON_ERR);
merge: refuse to create too cool a merge by default While it makes sense to allow merging unrelated histories of two projects that started independently into one, in the way "gitk" was merged to "git" itself aka "the coolest merge ever", such a merge is still an unusual event. Worse, if somebody creates an independent history by starting from a tarball of an established project and sends a pull request to the original project, "git merge" however happily creates such a merge without any sign of something unusual is happening. Teach "git merge" to refuse to create such a merge by default, unless the user passes a new "--allow-unrelated-histories" option to tell it that the user is aware that two unrelated projects are merged. Because such a "two project merge" is a rare event, a configuration option to always allow such a merge is not added. We could add the same option to "git pull" and have it passed through to underlying "git merge". I do not have a fundamental opposition against such a feature, but this commit does not do so and instead leaves it as low-hanging fruit for others, because such a "two project merge" would be done after fetching the other project into some location in the working tree of an existing project and making sure how well they fit together, it is sufficient to allow a local merge without such an option pass-through from "git pull" to "git merge". Many tests that are updated by this patch does the pass-through manually by turning: git pull something into its equivalent: git fetch something && git merge --allow-unrelated-histories FETCH_HEAD If somebody is inclined to add such an option, updated tests in this change need to be adjusted back to: git pull --allow-unrelated-histories something Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-03-18 23:21:09 +03:00
if (remoteheads && !common) {
/* No common ancestors found. */
if (!allow_unrelated_histories)
die(_("refusing to merge unrelated histories"));
/* otherwise, we need a real merge. */
} else if (!remoteheads ||
(!remoteheads->next && !common->next &&
common->item == remoteheads->item)) {
/*
* If head can reach all the merge then we are up to date.
* but first the most common case of merging one remote.
*/
finish_up_to_date();
goto done;
} else if (fast_forward != FF_NO && !remoteheads->next &&
!common->next &&
oideq(&common->item->object.oid, &head_commit->object.oid)) {
/* Again the most common case of merging one remote. */
struct strbuf msg = STRBUF_INIT;
struct commit *commit;
if (verbosity >= 0) {
printf(_("Updating %s..%s\n"),
find_unique_abbrev(&head_commit->object.oid,
DEFAULT_ABBREV),
find_unique_abbrev(&remoteheads->item->object.oid,
DEFAULT_ABBREV));
}
strbuf_addstr(&msg, "Fast-forward");
if (have_message)
strbuf_addstr(&msg,
" (no commit created; -m option ignored)");
commit = remoteheads->item;
if (!commit) {
ret = 1;
goto done;
}
if (autostash)
create_autostash(the_repository,
git_path_merge_autostash(the_repository));
if (checkout_fast_forward(the_repository,
&head_commit->object.oid,
&commit->object.oid,
overwrite_ignore)) {
apply_autostash(git_path_merge_autostash(the_repository));
ret = 1;
goto done;
}
finish(head_commit, remoteheads, &commit->object.oid, msg.buf);
remove_merge_branch_state(the_repository);
strbuf_release(&msg);
goto done;
} else if (!remoteheads->next && common->next)
;
/*
* We are not doing octopus and not fast-forward. Need
* a real merge.
*/
else if (!remoteheads->next && !common->next && option_commit) {
/*
* We are not doing octopus, not fast-forward, and have
* only one common.
*/
refresh_cache(REFRESH_QUIET);
if (allow_trivial && fast_forward != FF_ONLY) {
/*
* Must first ensure that index matches HEAD before
* attempting a trivial merge.
*/
struct tree *head_tree = get_commit_tree(head_commit);
struct strbuf sb = STRBUF_INIT;
if (repo_index_has_changes(the_repository, head_tree,
&sb)) {
error(_("Your local changes to the following files would be overwritten by merge:\n %s"),
sb.buf);
strbuf_release(&sb);
return 2;
}
/* See if it is really trivial. */
git_committer_info(IDENT_STRICT);
printf(_("Trying really trivial in-index merge...\n"));
if (!read_tree_trivial(&common->item->object.oid,
&head_commit->object.oid,
&remoteheads->item->object.oid)) {
ret = merge_trivial(head_commit, remoteheads);
goto done;
}
printf(_("Nope.\n"));
}
} else {
/*
* An octopus. If we can reach all the remote we are up
* to date.
*/
int up_to_date = 1;
struct commit_list *j;
for (j = remoteheads; j; j = j->next) {
struct commit_list *common_one;
/*
* Here we *have* to calculate the individual
* merge_bases again, otherwise "git merge HEAD^
* HEAD^^" would be missed.
*/
common_one = get_merge_bases(head_commit, j->item);
if (!oideq(&common_one->item->object.oid, &j->item->object.oid)) {
up_to_date = 0;
break;
}
}
if (up_to_date) {
finish_up_to_date();
goto done;
}
}
if (fast_forward == FF_ONLY)
die_ff_impossible();
if (autostash)
create_autostash(the_repository,
git_path_merge_autostash(the_repository));
/* We are going to make a new commit. */
git_committer_info(IDENT_STRICT);
/*
* At this point, we need a real merge. No matter what strategy
* we use, it would operate on the index, possibly affecting the
* working tree, and when resolved cleanly, have the desired
* tree in the index -- this means that the index must be in
* sync with the head commit. The strategies are responsible
* to ensure this.
*/
if (use_strategies_nr == 1 ||
/*
* Stash away the local changes so that we can try more than one.
*/
save_state(&stash))
oidclr(&stash);
for (i = 0; !merge_was_ok && i < use_strategies_nr; i++) {
int ret, cnt;
if (i) {
printf(_("Rewinding the tree to pristine...\n"));
restore_state(&head_commit->object.oid, &stash);
}
if (use_strategies_nr != 1)
printf(_("Trying merge strategy %s...\n"),
use_strategies[i]->name);
/*
* Remember which strategy left the state in the working
* tree.
*/
wt_strategy = use_strategies[i]->name;
ret = try_merge_strategy(use_strategies[i]->name,
common, remoteheads,
head_commit);
/*
* The backend exits with 1 when conflicts are
* left to be resolved, with 2 when it does not
* handle the given merge at all.
*/
if (ret < 2) {
if (!ret) {
if (option_commit) {
/* Automerge succeeded. */
automerge_was_ok = 1;
break;
}
merge_was_ok = 1;
}
cnt = (use_strategies_nr > 1) ? evaluate_result() : 0;
if (best_cnt <= 0 || cnt <= best_cnt) {
best_strategy = use_strategies[i]->name;
best_cnt = cnt;
}
}
}
/*
* If we have a resulting tree, that means the strategy module
* auto resolved the merge cleanly.
*/
if (automerge_was_ok) {
ret = finish_automerge(head_commit, head_subsumed,
common, remoteheads,
&result_tree, wt_strategy);
goto done;
}
/*
* Pick the result from the best strategy and have the user fix
* it up.
*/
if (!best_strategy) {
restore_state(&head_commit->object.oid, &stash);
if (use_strategies_nr > 1)
fprintf(stderr,
_("No merge strategy handled the merge.\n"));
else
fprintf(stderr, _("Merge with strategy %s failed.\n"),
use_strategies[0]->name);
apply_autostash(git_path_merge_autostash(the_repository));
ret = 2;
goto done;
} else if (best_strategy == wt_strategy)
; /* We already have its result in the working tree. */
else {
printf(_("Rewinding the tree to pristine...\n"));
restore_state(&head_commit->object.oid, &stash);
printf(_("Using the %s strategy to prepare resolving by hand.\n"),
best_strategy);
try_merge_strategy(best_strategy, common, remoteheads,
head_commit);
}
if (squash) {
finish(head_commit, remoteheads, NULL, NULL);
git_test_write_commit_graph_or_die();
} else
write_merge_state(remoteheads);
if (merge_was_ok)
fprintf(stderr, _("Automatic merge went well; "
"stopped before committing as requested\n"));
else
ret = suggest_conflicts();
done:
merge: fix memory leaks in cmd_merge() There were two commit_lists created in cmd_merge() that were only conditionally free()'d. Add a quick conditional call to free_commit_list() for each of them at the end of the function. Testing this commit against t6404 under valgrind shows that this patch fixes the following two leaks: 16 bytes in 1 blocks are definitely lost in loss record 16 of 126 at 0x484086F: malloc (vg_replace_malloc.c:380) by 0x69FFEB: do_xmalloc (wrapper.c:41) by 0x6A0073: xmalloc (wrapper.c:62) by 0x52A72D: commit_list_insert (commit.c:556) by 0x47FC93: reduce_parents (merge.c:1114) by 0x4801EE: collect_parents (merge.c:1214) by 0x480B56: cmd_merge (merge.c:1465) by 0x40686E: run_builtin (git.c:464) by 0x406C51: handle_builtin (git.c:716) by 0x406E96: run_argv (git.c:783) by 0x40730A: cmd_main (git.c:914) by 0x4E7DFA: main (common-main.c:56) 8 (16 direct, 32 indirect) bytes in 1 blocks are definitely lost in \ loss record 61 of 126 at 0x484086F: malloc (vg_replace_malloc.c:380) by 0x69FFEB: do_xmalloc (wrapper.c:41) by 0x6A0073: xmalloc (wrapper.c:62) by 0x52A72D: commit_list_insert (commit.c:556) by 0x52A8F2: commit_list_insert_by_date (commit.c:620) by 0x5270AC: get_merge_bases_many_0 (commit-reach.c:413) by 0x52716C: repo_get_merge_bases (commit-reach.c:438) by 0x480E5A: cmd_merge (merge.c:1520) by 0x40686E: run_builtin (git.c:464) by 0x406C51: handle_builtin (git.c:716) by 0x406E96: run_argv (git.c:783) by 0x40730A: cmd_main (git.c:914) There are still 3 leaks in chdir_notify_register() after this, but chdir_notify_register() has been brought up on the list before and folks were not a fan of fixing those, so I'm not touching them. Signed-off-by: Elijah Newren <newren@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-01-20 10:47:15 +03:00
if (!automerge_was_ok) {
free_commit_list(common);
free_commit_list(remoteheads);
}
strbuf_release(&buf);
free(branch_to_free);
return ret;
}