parse-options: add OPT_STRING_LIST helper
[gitweb.git] / diff.c
diff --git a/diff.c b/diff.c
index 19b5bf63ed37aa49c00e63f7aec621ce601b6daa..559bf574a88f97402f32e1bd4ab51f6f0c9d5026 100644 (file)
--- a/diff.c
+++ b/diff.c
@@ -23,7 +23,7 @@
 #endif
 
 static int diff_detect_rename_default;
-static int diff_rename_limit_default = 200;
+static int diff_rename_limit_default = 400;
 static int diff_suppress_blank_empty;
 int diff_use_color_default = -1;
 static const char *diff_word_regex_cfg;
@@ -31,6 +31,7 @@ static const char *external_diff_cmd_cfg;
 int diff_auto_refresh_index = 1;
 static int diff_mnemonic_prefix;
 static int diff_no_prefix;
+static struct diff_options default_diff_options;
 
 static char diff_colors[][COLOR_MAXLEN] = {
        GIT_COLOR_RESET,
@@ -107,6 +108,9 @@ int git_diff_ui_config(const char *var, const char *value, void *cb)
        if (!strcmp(var, "diff.wordregex"))
                return git_config_string(&diff_word_regex_cfg, var, value);
 
+       if (!strcmp(var, "diff.ignoresubmodules"))
+               handle_ignore_submodules_arg(&default_diff_options, value);
+
        return git_diff_basic_config(var, value, cb);
 }
 
@@ -141,6 +145,9 @@ int git_diff_basic_config(const char *var, const char *value, void *cb)
                return 0;
        }
 
+       if (!prefixcmp(var, "submodule."))
+               return parse_submodule_config_option(var, value);
+
        return git_color_default_config(var, value, cb);
 }
 
@@ -238,6 +245,15 @@ static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
        return 0;
 }
 
+/* like fill_mmfile, but only for size, so we can avoid retrieving blob */
+static unsigned long diff_filespec_size(struct diff_filespec *one)
+{
+       if (!DIFF_FILE_VALID(one))
+               return 0;
+       diff_populate_filespec(one, 1);
+       return one->size;
+}
+
 static int count_trailing_blank(mmfile_t *mf, unsigned ws_rule)
 {
        char *ptr = mf->ptr;
@@ -599,22 +615,20 @@ static void diff_words_append(char *line, unsigned long len,
        buffer->text.ptr[buffer->text.size] = '\0';
 }
 
-struct diff_words_style_elem
-{
+struct diff_words_style_elem {
        const char *prefix;
        const char *suffix;
        const char *color; /* NULL; filled in by the setup code if
                            * color is enabled */
 };
 
-struct diff_words_style
-{
+struct diff_words_style {
        enum diff_words_type type;
        struct diff_words_style_elem new, old, ctx;
        const char *newline;
 };
 
-struct diff_words_style diff_words_styles[] = {
+static struct diff_words_style diff_words_styles[] = {
        { DIFF_WORDS_PORCELAIN, {"+", "\n"}, {"-", "\n"}, {" ", "\n"}, "~\n" },
        { DIFF_WORDS_PLAIN, {"{+", "+}"}, {"[-", "-]"}, {"", ""}, "\n" },
        { DIFF_WORDS_COLOR, {"", ""}, {"", ""}, {"", ""}, "\n" }
@@ -1036,8 +1050,16 @@ static void fn_out_consume(void *priv, char *line, unsigned long len)
                        emit_line(ecbdata->opt, plain, reset, line, len);
                        fputs("~\n", ecbdata->opt->file);
                } else {
-                       /* don't print the prefix character */
-                       emit_line(ecbdata->opt, plain, reset, line+1, len-1);
+                       /*
+                        * Skip the prefix character, if any.  With
+                        * diff_suppress_blank_empty, there may be
+                        * none.
+                        */
+                       if (line[0] != '\n') {
+                             line++;
+                             len--;
+                       }
+                       emit_line(ecbdata->opt, plain, reset, line, len);
                }
                return;
        }
@@ -1228,7 +1250,7 @@ static void show_stats(struct diffstat_t *data, struct diff_options *options)
        uintmax_t max_change = 0, max_len = 0;
        int total_files = data->nr;
        int width, name_width;
-       const char *reset, *set, *add_c, *del_c;
+       const char *reset, *add_c, *del_c;
        const char *line_prefix = "";
        struct strbuf *msg = NULL;
 
@@ -1255,7 +1277,6 @@ static void show_stats(struct diffstat_t *data, struct diff_options *options)
 
        /* Find the longest filename and max number of changes */
        reset = diff_get_color_opt(options, DIFF_RESET);
-       set   = diff_get_color_opt(options, DIFF_PLAIN);
        add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
        del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
 
@@ -1525,8 +1546,36 @@ static void show_dirstat(struct diff_options *options)
                struct diff_filepair *p = q->queue[i];
                const char *name;
                unsigned long copied, added, damage;
+               int content_changed;
+
+               name = p->two->path ? p->two->path : p->one->path;
 
-               name = p->one->path ? p->one->path : p->two->path;
+               if (p->one->sha1_valid && p->two->sha1_valid)
+                       content_changed = hashcmp(p->one->sha1, p->two->sha1);
+               else
+                       content_changed = 1;
+
+               if (!content_changed) {
+                       /*
+                        * The SHA1 has not changed, so pre-/post-content is
+                        * identical. We can therefore skip looking at the
+                        * file contents altogether.
+                        */
+                       damage = 0;
+                       goto found_damage;
+               }
+
+               if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE)) {
+                       /*
+                        * In --dirstat-by-file mode, we don't really need to
+                        * look at the actual file contents at all.
+                        * The fact that the SHA1 changed is enough for us to
+                        * add this file to the list of results
+                        * (with each file contributing equal damage).
+                        */
+                       damage = 1;
+                       goto found_damage;
+               }
 
                if (DIFF_FILE_VALID(p->one) && DIFF_FILE_VALID(p->two)) {
                        diff_populate_filespec(p->one, 0);
@@ -1550,14 +1599,18 @@ static void show_dirstat(struct diff_options *options)
                /*
                 * Original minus copied is the removed material,
                 * added is the new material.  They are both damages
-                * made to the preimage. In --dirstat-by-file mode, count
-                * damaged files, not damaged lines. This is done by
-                * counting only a single damaged line per file.
+                * made to the preimage.
+                * If the resulting damage is zero, we know that
+                * diffcore_count_changes() considers the two entries to
+                * be identical, but since content_changed is true, we
+                * know that there must have been _some_ kind of change,
+                * so we force all entries to have damage > 0.
                 */
                damage = (p->one->size - copied) + added;
-               if (DIFF_OPT_TST(options, DIRSTAT_BY_FILE) && damage > 0)
+               if (!damage)
                        damage = 1;
 
+found_damage:
                ALLOC_GROW(dir.files, dir.nr + 1, dir.alloc);
                dir.files[dir.nr].name = name;
                dir.files[dir.nr].changed = damage;
@@ -1764,8 +1817,14 @@ static void emit_binary_diff(FILE *file, mmfile_t *one, mmfile_t *two, char *pre
 
 static void diff_filespec_load_driver(struct diff_filespec *one)
 {
-       if (!one->driver)
+       /* Use already-loaded driver */
+       if (one->driver)
+               return;
+
+       if (S_ISREG(one->mode))
                one->driver = userdiff_find_by_path(one->path);
+
+       /* Fallback to default settings */
        if (!one->driver)
                one->driver = userdiff_find_by_name("default");
 }
@@ -1813,8 +1872,7 @@ struct userdiff_driver *get_textconv(struct diff_filespec *one)
 {
        if (!DIFF_FILE_VALID(one))
                return NULL;
-       if (!S_ISREG(one->mode))
-               return NULL;
+
        diff_filespec_load_driver(one);
        if (!one->driver->textconv)
                return NULL;
@@ -2067,25 +2125,28 @@ static void builtin_diffstat(const char *name_a, const char *name_b,
                data->is_unmerged = 1;
                return;
        }
-       if (complete_rewrite) {
+
+       if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
+               data->is_binary = 1;
+               data->added = diff_filespec_size(two);
+               data->deleted = diff_filespec_size(one);
+       }
+
+       else if (complete_rewrite) {
                diff_populate_filespec(one, 0);
                diff_populate_filespec(two, 0);
                data->deleted = count_lines(one->data, one->size);
                data->added = count_lines(two->data, two->size);
-               goto free_and_return;
        }
-       if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
-               die("unable to read files to diff");
 
-       if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
-               data->is_binary = 1;
-               data->added = mf2.size;
-               data->deleted = mf1.size;
-       } else {
+       else {
                /* Crazy xdl interfaces.. */
                xpparam_t xpp;
                xdemitconf_t xecfg;
 
+               if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
+                       die("unable to read files to diff");
+
                memset(&xpp, 0, sizeof(xpp));
                memset(&xecfg, 0, sizeof(xecfg));
                xpp.flags = o->xdl_opts;
@@ -2093,7 +2154,6 @@ static void builtin_diffstat(const char *name_a, const char *name_b,
                              &xpp, &xecfg);
        }
 
- free_and_return:
        diff_free_filespec_data(one);
        diff_free_filespec_data(two);
 }
@@ -2146,7 +2206,7 @@ static void builtin_checkdiff(const char *name_a, const char *name_b,
 
                        ecbdata.ws_rule = data.ws_rule;
                        check_blank_at_eof(&mf1, &mf2, &ecbdata);
-                       blank_at_eof = ecbdata.blank_at_eof_in_preimage;
+                       blank_at_eof = ecbdata.blank_at_eof_in_postimage;
 
                        if (blank_at_eof) {
                                static char *err;
@@ -2379,10 +2439,14 @@ int diff_populate_filespec(struct diff_filespec *s, int size_only)
        }
        else {
                enum object_type type;
-               if (size_only)
+               if (size_only) {
                        type = sha1_object_info(s->sha1, &s->size);
-               else {
+                       if (type < 0)
+                               die("unable to read %s", sha1_to_hex(s->sha1));
+               } else {
                        s->data = read_sha1_file(s->sha1, &type, &s->size);
+                       if (!s->data)
+                               die("unable to read %s", sha1_to_hex(s->sha1));
                        s->should_free = 1;
                }
        }
@@ -2822,7 +2886,7 @@ static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
 
 void diff_setup(struct diff_options *options)
 {
-       memset(options, 0, sizeof(*options));
+       memcpy(options, &default_diff_options, sizeof(*options));
 
        options->file = stdout;
 
@@ -2998,9 +3062,100 @@ static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *va
 
 static int diff_scoreopt_parse(const char *opt);
 
+static inline int short_opt(char opt, const char **argv,
+                           const char **optarg)
+{
+       const char *arg = argv[0];
+       if (arg[0] != '-' || arg[1] != opt)
+               return 0;
+       if (arg[2] != '\0') {
+               *optarg = arg + 2;
+               return 1;
+       }
+       if (!argv[1])
+               die("Option '%c' requires a value", opt);
+       *optarg = argv[1];
+       return 2;
+}
+
+int parse_long_opt(const char *opt, const char **argv,
+                  const char **optarg)
+{
+       const char *arg = argv[0];
+       if (arg[0] != '-' || arg[1] != '-')
+               return 0;
+       arg += strlen("--");
+       if (prefixcmp(arg, opt))
+               return 0;
+       arg += strlen(opt);
+       if (*arg == '=') { /* sticked form: --option=value */
+               *optarg = arg + 1;
+               return 1;
+       }
+       if (*arg != '\0')
+               return 0;
+       /* separate form: --option value */
+       if (!argv[1])
+               die("Option '--%s' requires a value", opt);
+       *optarg = argv[1];
+       return 2;
+}
+
+static int stat_opt(struct diff_options *options, const char **av)
+{
+       const char *arg = av[0];
+       char *end;
+       int width = options->stat_width;
+       int name_width = options->stat_name_width;
+       int argcount = 1;
+
+       arg += strlen("--stat");
+       end = (char *)arg;
+
+       switch (*arg) {
+       case '-':
+               if (!prefixcmp(arg, "-width")) {
+                       arg += strlen("-width");
+                       if (*arg == '=')
+                               width = strtoul(arg + 1, &end, 10);
+                       else if (!*arg && !av[1])
+                               die("Option '--stat-width' requires a value");
+                       else if (!*arg) {
+                               width = strtoul(av[1], &end, 10);
+                               argcount = 2;
+                       }
+               } else if (!prefixcmp(arg, "-name-width")) {
+                       arg += strlen("-name-width");
+                       if (*arg == '=')
+                               name_width = strtoul(arg + 1, &end, 10);
+                       else if (!*arg && !av[1])
+                               die("Option '--stat-name-width' requires a value");
+                       else if (!*arg) {
+                               name_width = strtoul(av[1], &end, 10);
+                               argcount = 2;
+                       }
+               }
+               break;
+       case '=':
+               width = strtoul(arg+1, &end, 10);
+               if (*end == ',')
+                       name_width = strtoul(end+1, &end, 10);
+       }
+
+       /* Important! This checks all the error cases! */
+       if (*end)
+               return 0;
+       options->output_format |= DIFF_FORMAT_DIFFSTAT;
+       options->stat_name_width = name_width;
+       options->stat_width = width;
+       return argcount;
+}
+
 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
 {
        const char *arg = av[0];
+       const char *optarg;
+       int argcount;
 
        /* Output format options */
        if (!strcmp(arg, "-p") || !strcmp(arg, "-u") || !strcmp(arg, "--patch"))
@@ -3037,49 +3192,28 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
                options->output_format |= DIFF_FORMAT_NAME_STATUS;
        else if (!strcmp(arg, "-s"))
                options->output_format |= DIFF_FORMAT_NO_OUTPUT;
-       else if (!prefixcmp(arg, "--stat")) {
-               char *end;
-               int width = options->stat_width;
-               int name_width = options->stat_name_width;
-               arg += 6;
-               end = (char *)arg;
-
-               switch (*arg) {
-               case '-':
-                       if (!prefixcmp(arg, "-width="))
-                               width = strtoul(arg + 7, &end, 10);
-                       else if (!prefixcmp(arg, "-name-width="))
-                               name_width = strtoul(arg + 12, &end, 10);
-                       break;
-               case '=':
-                       width = strtoul(arg+1, &end, 10);
-                       if (*end == ',')
-                               name_width = strtoul(end+1, &end, 10);
-               }
-
-               /* Important! This checks all the error cases! */
-               if (*end)
-                       return 0;
-               options->output_format |= DIFF_FORMAT_DIFFSTAT;
-               options->stat_name_width = name_width;
-               options->stat_width = width;
-       }
+       else if (!prefixcmp(arg, "--stat"))
+               /* --stat, --stat-width, or --stat-name-width */
+               return stat_opt(options, av);
 
        /* renames options */
-       else if (!prefixcmp(arg, "-B")) {
+       else if (!prefixcmp(arg, "-B") || !prefixcmp(arg, "--break-rewrites=") ||
+                !strcmp(arg, "--break-rewrites")) {
                if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
-                       return -1;
+                       return error("invalid argument to -B: %s", arg+2);
        }
-       else if (!prefixcmp(arg, "-M")) {
+       else if (!prefixcmp(arg, "-M") || !prefixcmp(arg, "--find-renames=") ||
+                !strcmp(arg, "--find-renames")) {
                if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
-                       return -1;
+                       return error("invalid argument to -M: %s", arg+2);
                options->detect_rename = DIFF_DETECT_RENAME;
        }
-       else if (!prefixcmp(arg, "-C")) {
+       else if (!prefixcmp(arg, "-C") || !prefixcmp(arg, "--find-copies=") ||
+                !strcmp(arg, "--find-copies")) {
                if (options->detect_rename == DIFF_DETECT_COPY)
                        DIFF_OPT_SET(options, FIND_COPIES_HARDER);
                if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
-                       return -1;
+                       return error("invalid argument to -C: %s", arg+2);
                options->detect_rename = DIFF_DETECT_COPY;
        }
        else if (!strcmp(arg, "--no-renames"))
@@ -3157,10 +3291,11 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
                else
                        die("bad --word-diff argument: %s", type);
        }
-       else if (!prefixcmp(arg, "--word-diff-regex=")) {
+       else if ((argcount = parse_long_opt("word-diff-regex", av, &optarg))) {
                if (options->word_diff == DIFF_WORDS_NONE)
                        options->word_diff = DIFF_WORDS_PLAIN;
-               options->word_regex = arg + 18;
+               options->word_regex = optarg;
+               return argcount;
        }
        else if (!strcmp(arg, "--exit-code"))
                DIFF_OPT_SET(options, EXIT_WITH_STATUS);
@@ -3174,11 +3309,13 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
                DIFF_OPT_SET(options, ALLOW_TEXTCONV);
        else if (!strcmp(arg, "--no-textconv"))
                DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
-       else if (!strcmp(arg, "--ignore-submodules"))
+       else if (!strcmp(arg, "--ignore-submodules")) {
+               DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
                handle_ignore_submodules_arg(options, "all");
-       else if (!prefixcmp(arg, "--ignore-submodules="))
+       } else if (!prefixcmp(arg, "--ignore-submodules=")) {
+               DIFF_OPT_SET(options, OVERRIDE_SUBMODULE_CONFIG);
                handle_ignore_submodules_arg(options, arg + 20);
-       else if (!strcmp(arg, "--submodule"))
+       else if (!strcmp(arg, "--submodule"))
                DIFF_OPT_SET(options, SUBMODULE_LOG);
        else if (!prefixcmp(arg, "--submodule=")) {
                if (!strcmp(arg + 12, "log"))
@@ -3188,18 +3325,31 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
        /* misc options */
        else if (!strcmp(arg, "-z"))
                options->line_termination = 0;
-       else if (!prefixcmp(arg, "-l"))
-               options->rename_limit = strtoul(arg+2, NULL, 10);
-       else if (!prefixcmp(arg, "-S"))
-               options->pickaxe = arg + 2;
+       else if ((argcount = short_opt('l', av, &optarg))) {
+               options->rename_limit = strtoul(optarg, NULL, 10);
+               return argcount;
+       }
+       else if ((argcount = short_opt('S', av, &optarg))) {
+               options->pickaxe = optarg;
+               options->pickaxe_opts |= DIFF_PICKAXE_KIND_S;
+               return argcount;
+       } else if ((argcount = short_opt('G', av, &optarg))) {
+               options->pickaxe = optarg;
+               options->pickaxe_opts |= DIFF_PICKAXE_KIND_G;
+               return argcount;
+       }
        else if (!strcmp(arg, "--pickaxe-all"))
-               options->pickaxe_opts = DIFF_PICKAXE_ALL;
+               options->pickaxe_opts |= DIFF_PICKAXE_ALL;
        else if (!strcmp(arg, "--pickaxe-regex"))
-               options->pickaxe_opts = DIFF_PICKAXE_REGEX;
-       else if (!prefixcmp(arg, "-O"))
-               options->orderfile = arg + 2;
-       else if (!prefixcmp(arg, "--diff-filter="))
-               options->filter = arg + 14;
+               options->pickaxe_opts |= DIFF_PICKAXE_REGEX;
+       else if ((argcount = short_opt('O', av, &optarg))) {
+               options->orderfile = optarg;
+               return argcount;
+       }
+       else if ((argcount = parse_long_opt("diff-filter", av, &optarg))) {
+               options->filter = optarg;
+               return argcount;
+       }
        else if (!strcmp(arg, "--abbrev"))
                options->abbrev = DEFAULT_ABBREV;
        else if (!prefixcmp(arg, "--abbrev=")) {
@@ -3209,26 +3359,31 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
                else if (40 < options->abbrev)
                        options->abbrev = 40;
        }
-       else if (!prefixcmp(arg, "--src-prefix="))
-               options->a_prefix = arg + 13;
-       else if (!prefixcmp(arg, "--dst-prefix="))
-               options->b_prefix = arg + 13;
+       else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
+               options->a_prefix = optarg;
+               return argcount;
+       }
+       else if ((argcount = parse_long_opt("dst-prefix", av, &optarg))) {
+               options->b_prefix = optarg;
+               return argcount;
+       }
        else if (!strcmp(arg, "--no-prefix"))
                options->a_prefix = options->b_prefix = "";
        else if (opt_arg(arg, '\0', "inter-hunk-context",
                         &options->interhunkcontext))
                ;
-       else if (!prefixcmp(arg, "--output=")) {
-               options->file = fopen(arg + strlen("--output="), "w");
+       else if ((argcount = parse_long_opt("output", av, &optarg))) {
+               options->file = fopen(optarg, "w");
                if (!options->file)
-                       die_errno("Could not open '%s'", arg + strlen("--output="));
+                       die_errno("Could not open '%s'", optarg);
                options->close_file = 1;
+               return argcount;
        } else
                return 0;
        return 1;
 }
 
-static int parse_num(const char **cp_p)
+int parse_rename_score(const char **cp_p)
 {
        unsigned long num, scale;
        int ch, dot;
@@ -3271,10 +3426,26 @@ static int diff_scoreopt_parse(const char *opt)
        if (*opt++ != '-')
                return -1;
        cmd = *opt++;
+       if (cmd == '-') {
+               /* convert the long-form arguments into short-form versions */
+               if (!prefixcmp(opt, "break-rewrites")) {
+                       opt += strlen("break-rewrites");
+                       if (*opt == 0 || *opt++ == '=')
+                               cmd = 'B';
+               } else if (!prefixcmp(opt, "find-copies")) {
+                       opt += strlen("find-copies");
+                       if (*opt == 0 || *opt++ == '=')
+                               cmd = 'C';
+               } else if (!prefixcmp(opt, "find-renames")) {
+                       opt += strlen("find-renames");
+                       if (*opt == 0 || *opt++ == '=')
+                               cmd = 'M';
+               }
+       }
        if (cmd != 'M' && cmd != 'C' && cmd != 'B')
                return -1; /* that is not a -M, -C nor -B option */
 
-       opt1 = parse_num(&opt);
+       opt1 = parse_rename_score(&opt);
        if (cmd != 'B')
                opt2 = 0;
        else {
@@ -3284,7 +3455,7 @@ static int diff_scoreopt_parse(const char *opt)
                        return -1; /* we expect -B80/99 or -B80 */
                else {
                        opt++;
-                       opt2 = parse_num(&opt);
+                       opt2 = parse_rename_score(&opt);
                }
        }
        if (*opt != 0)
@@ -3437,7 +3608,7 @@ static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
 
        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
-               return; /* no tree diffs in patch format */
+               return; /* no useful stat for tree diffs */
 
        run_diffstat(p, o, diffstat);
 }
@@ -3450,7 +3621,7 @@ static void diff_flush_checkdiff(struct diff_filepair *p,
 
        if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
-               return; /* no tree diffs in patch format */
+               return; /* nothing to check in tree diffs */
 
        run_checkdiff(p, o);
 }
@@ -3766,9 +3937,16 @@ static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
                                        len2, p->two->path);
                git_SHA1_Update(&ctx, buffer, len1);
 
+               if (diff_filespec_is_binary(p->one) ||
+                   diff_filespec_is_binary(p->two)) {
+                       git_SHA1_Update(&ctx, sha1_to_hex(p->one->sha1), 40);
+                       git_SHA1_Update(&ctx, sha1_to_hex(p->two->sha1), 40);
+                       continue;
+               }
+
                xpp.flags = 0;
                xecfg.ctxlen = 3;
-               xecfg.flags = XDL_EMIT_FUNCNAMES;
+               xecfg.flags = 0;
                xdi_diff_outf(&mf1, &mf2, patch_id_consume, &data,
                              &xpp, &xecfg);
        }
@@ -3817,6 +3995,28 @@ static int is_summary_empty(const struct diff_queue_struct *q)
        return 1;
 }
 
+static const char rename_limit_warning[] =
+"inexact rename detection was skipped due to too many files.";
+
+static const char degrade_cc_to_c_warning[] =
+"only found copies from modified paths due to too many files.";
+
+static const char rename_limit_advice[] =
+"you may want to set your %s variable to at least "
+"%d and retry the command.";
+
+void diff_warn_rename_limit(const char *varname, int needed, int degraded_cc)
+{
+       if (degraded_cc)
+               warning(degrade_cc_to_c_warning);
+       else if (needed)
+               warning(rename_limit_warning);
+       else
+               return;
+       if (0 < needed && needed < 32767)
+               warning(rename_limit_advice, varname, needed);
+}
+
 void diff_flush(struct diff_options *options)
 {
        struct diff_queue_struct *q = &diff_queued_diff;
@@ -4079,7 +4279,7 @@ void diffcore_std(struct diff_options *options)
                        diffcore_merge_broken();
        }
        if (options->pickaxe)
-               diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
+               diffcore_pickaxe(options);
        if (options->orderfile)
                diffcore_order(options->orderfile);
        if (!options->found_follow)
@@ -4098,6 +4298,10 @@ void diffcore_std(struct diff_options *options)
 int diff_result_code(struct diff_options *opt, int status)
 {
        int result = 0;
+
+       diff_warn_rename_limit("diff.renamelimit",
+                              opt->needed_rename_limit,
+                              opt->degraded_cc_to_c);
        if (!DIFF_OPT_TST(opt, EXIT_WITH_STATUS) &&
            !(opt->output_format & DIFF_FORMAT_CHECKDIFF))
                return status;
@@ -4110,6 +4314,24 @@ int diff_result_code(struct diff_options *opt, int status)
        return result;
 }
 
+/*
+ * Shall changes to this submodule be ignored?
+ *
+ * Submodule changes can be configured to be ignored separately for each path,
+ * but that configuration can be overridden from the command line.
+ */
+static int is_submodule_ignored(const char *path, struct diff_options *options)
+{
+       int ignored = 0;
+       unsigned orig_flags = options->flags;
+       if (!DIFF_OPT_TST(options, OVERRIDE_SUBMODULE_CONFIG))
+               set_diffopt_flags_from_submodule_config(options, path);
+       if (DIFF_OPT_TST(options, IGNORE_SUBMODULES))
+               ignored = 1;
+       options->flags = orig_flags;
+       return ignored;
+}
+
 void diff_addremove(struct diff_options *options,
                    int addremove, unsigned mode,
                    const unsigned char *sha1,
@@ -4117,7 +4339,7 @@ void diff_addremove(struct diff_options *options,
 {
        struct diff_filespec *one, *two;
 
-       if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(mode))
+       if (S_ISGITLINK(mode) && is_submodule_ignored(concatpath, options))
                return;
 
        /* This may look odd, but it is a preparation for
@@ -4164,8 +4386,8 @@ void diff_change(struct diff_options *options,
 {
        struct diff_filespec *one, *two;
 
-       if (DIFF_OPT_TST(options, IGNORE_SUBMODULES) && S_ISGITLINK(old_mode)
-                       && S_ISGITLINK(new_mode))
+       if (S_ISGITLINK(old_mode) && S_ISGITLINK(new_mode) &&
+           is_submodule_ignored(concatpath, options))
                return;
 
        if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
@@ -4193,20 +4415,20 @@ void diff_change(struct diff_options *options,
                DIFF_OPT_SET(options, HAS_CHANGES);
 }
 
-void diff_unmerge(struct diff_options *options,
-                 const char *path,
-                 unsigned mode, const unsigned char *sha1)
+struct diff_filepair *diff_unmerge(struct diff_options *options, const char *path)
 {
+       struct diff_filepair *pair;
        struct diff_filespec *one, *two;
 
        if (options->prefix &&
            strncmp(path, options->prefix, options->prefix_length))
-               return;
+               return NULL;
 
        one = alloc_filespec(path);
        two = alloc_filespec(path);
-       fill_filespec(one, sha1, mode);
-       diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
+       pair = diff_queue(&diff_queued_diff, one, two);
+       pair->is_unmerged = 1;
+       return pair;
 }
 
 static char *run_textconv(const char *pgm, struct diff_filespec *spec,
@@ -4264,7 +4486,7 @@ size_t fill_textconv(struct userdiff_driver *driver,
                return df->size;
        }
 
-       if (driver->textconv_cache) {
+       if (driver->textconv_cache && df->sha1_valid) {
                *outbuf = notes_cache_get(driver->textconv_cache, df->sha1,
                                          &size);
                if (*outbuf)
@@ -4275,7 +4497,7 @@ size_t fill_textconv(struct userdiff_driver *driver,
        if (!*outbuf)
                die("unable to read files to diff");
 
-       if (driver->textconv_cache) {
+       if (driver->textconv_cache && df->sha1_valid) {
                /* ignore errors, as we might be in a readonly repository */
                notes_cache_put(driver->textconv_cache, df->sha1, *outbuf,
                                size);