Bug fixes and cleanups

This contains a few bug fixes and some header and API cleanups.

The main API change is that filters should now use GIT_PASSTHROUGH
to indicate that they wish to skip processing a file instead of
GIT_ENOTFOUND.

The bug fixes include a possible out-of-range buffer access in
the ident filter, a filter ordering problem I introduced into the
custom filter tests on Windows, and a filter buf NUL termination
issue that was coming up on Linux.

5 files changed, 33 insertions, 28 deletions

diff --git a/src/array.h b/src/array.h
index b82079bd8..d7272d78c 100644
--- a/src/array.h
+++ b/src/array.h
@@ -59,7 +59,7 @@ GIT_INLINE(void *) git_array_grow(void *_a, size_t item_size)
 #define git_array_alloc(a) \
 	((a).size >= (a).asize) ? \
 	git_array_grow(&(a), sizeof(*(a).ptr)) : \
-	(a).ptr ? &(a).ptr[(a).size++] : NULL
+	((a).ptr ? &(a).ptr[(a).size++] : NULL)
 
 #define git_array_last(a) ((a).size ? &(a).ptr[(a).size - 1] : NULL)
 
diff --git a/src/crlf.c b/src/crlf.c
index 6b1fe46a3..b4eda267b 100644
--- a/src/crlf.c
+++ b/src/crlf.c
@@ -143,7 +143,7 @@ static int crlf_apply_to_odb(
 		 * stuff?
 		 */
 		if (stats.cr != stats.crlf)
-			return GIT_ENOTFOUND;
+			return GIT_PASSTHROUGH;
 
 		if (ca->crlf_action == GIT_CRLF_GUESS) {
 			/*
@@ -151,11 +151,11 @@ static int crlf_apply_to_odb(
 			 * This is the new safer autocrlf handling.
 			 */
 			if (has_cr_in_index(src))
-				return GIT_ENOTFOUND;
+				return GIT_PASSTHROUGH;
 		}
 
 		if (!stats.cr)
-			return GIT_ENOTFOUND;
+			return GIT_PASSTHROUGH;
 	}
 
 	/* Actually drop the carriage returns */
@@ -211,7 +211,7 @@ static int crlf_apply_to_workdir(
 
 	/* Don't filter binary files */
 	if (git_buf_text_is_binary(from))
-		return GIT_ENOTFOUND;
+		return GIT_PASSTHROUGH;
 
 	/* Determine proper line ending */
 	workdir_ending = line_ending(ca);
@@ -220,10 +220,10 @@ static int crlf_apply_to_workdir(
 
 	if (!strcmp("\n", workdir_ending)) {
 		if (ca->crlf_action == GIT_CRLF_GUESS && ca->auto_crlf)
-			return GIT_ENOTFOUND;
+			return GIT_PASSTHROUGH;
 
 		if (git_buf_find(from, '\r') < 0)
-			return GIT_ENOTFOUND;
+			return GIT_PASSTHROUGH;
 
 		if (git_buf_text_crlf_to_lf(to, from) < 0)
 			return -1;
@@ -267,7 +267,7 @@ static int crlf_check(
 	ca.crlf_action = crlf_input_action(&ca);
 
 	if (ca.crlf_action == GIT_CRLF_BINARY)
-		return GIT_ENOTFOUND;
+		return GIT_PASSTHROUGH;
 
 	if (ca.crlf_action == GIT_CRLF_GUESS) {
 		error = git_repository__cvar(
@@ -276,7 +276,7 @@ static int crlf_check(
 			return error;
 
 		if (ca.auto_crlf == GIT_AUTO_CRLF_FALSE)
-			return GIT_ENOTFOUND;
+			return GIT_PASSTHROUGH;
 	}
 
 	*payload = git__malloc(sizeof(ca));
@@ -296,7 +296,7 @@ static int crlf_apply(
 	/* initialize payload in case `check` was bypassed */
 	if (!*payload) {
 		int error = crlf_check(self, payload, src, NULL);
-		if (error < 0 && error != GIT_ENOTFOUND)
+		if (error < 0 && error != GIT_PASSTHROUGH)
 			return error;
 	}
 
diff --git a/src/filter.c b/src/filter.c
index 378209800..503f18555 100644
--- a/src/filter.c
+++ b/src/filter.c
@@ -235,7 +235,7 @@ int git_filter_register(
 	if (!filter_registry_find(NULL, name)) {
 		giterr_set(
 			GITERR_FILTER, "Attempt to reregister existing filter '%s'", name);
-		return -1;
+		return GIT_EEXISTS;
 	}
 
 	if (filter_def_scan_attrs(&attrs, &nattr, &nmatch, filter->attributes) < 0)
@@ -270,7 +270,7 @@ int git_filter_unregister(const char *name)
 	git_filter_def *fdef;
 
 	/* cannot unregister default filters */
-	if (!strcmp(GIT_FILTER_CRLF, name)) {
+	if (!strcmp(GIT_FILTER_CRLF, name) || !strcmp(GIT_FILTER_IDENT, name)) {
 		giterr_set(GITERR_FILTER, "Cannot unregister filter '%s'", name);
 		return -1;
 	}
@@ -476,7 +476,7 @@ int git_filter_list_load(
 
 		git__free((void *)values);
 
-		if (error == GIT_ENOTFOUND)
+		if (error == GIT_PASSTHROUGH)
 			error = 0;
 		else if (error < 0)
 			break;
@@ -609,11 +609,13 @@ int git_filter_list_apply_to_data(
 		error = fe->filter->apply(
 			fe->filter, &fe->payload, dbuffer[di], dbuffer[si], &fl->source);
 
-		if (error == GIT_ENOTFOUND)
+		if (error == GIT_PASSTHROUGH) {
+			/* PASSTHROUGH means filter decided not to process the buffer */
 			error = 0;
-		else if (!error)
+		} else if (!error) {
+			git_buf_shorten(dbuffer[di], 0); /* force NUL termination */
 			si = di; /* swap buffers */
-		else {
+		} else {
 			tgt->size = 0;
 			return error;
 		}
diff --git a/src/ident.c b/src/ident.c
index 23c407f16..51630879d 100644
--- a/src/ident.c
+++ b/src/ident.c
@@ -13,23 +13,25 @@
 static int ident_find_id(
 	const char **id_start, const char **id_end, const char *start, size_t len)
 {
-	const char *found;
+	const char *end = start + len, *found = NULL;
 
-	while (len > 0 && (found = memchr(start, '$', len)) != NULL) {
-		size_t remaining = len - (size_t)(found - start);
+	while (len > 3 && (found = memchr(start, '$', len)) != NULL) {
+		size_t remaining = (size_t)(end - found) - 1;
 		if (remaining < 3)
 			return GIT_ENOTFOUND;
-		if (found[1] == 'I' && found[2] == 'd')
-			break;
+
 		start = found + 1;
-		len = remaining - 1;
+		len   = remaining;
+
+		if (start[0] == 'I' && start[1] == 'd')
+			break;
 	}
 
-	if (!found || len < 3)
+	if (len < 3 || !found)
 		return GIT_ENOTFOUND;
 	*id_start = found;
 
-	if ((found = memchr(found + 3, '$', len - 3)) == NULL)
+	if ((found = memchr(start + 2, '$', len - 2)) == NULL)
 		return GIT_ENOTFOUND;
 
 	*id_end = found + 1;
@@ -46,12 +48,12 @@ static int ident_insert_id(
 	/* replace $Id$ with blob id */
 
 	if (!git_filter_source_id(src))
-		return GIT_ENOTFOUND;
+		return GIT_PASSTHROUGH;
 
 	git_oid_tostr(oid, sizeof(oid), git_filter_source_id(src));
 
 	if (ident_find_id(&id_start, &id_end, from->ptr, from->size) < 0)
-		return GIT_ENOTFOUND;
+		return GIT_PASSTHROUGH;
 
 	need_size = (size_t)(id_start - from->ptr) +
 		5 /* "$Id: " */ + GIT_OID_HEXSZ + 1 /* "$" */ +
@@ -76,7 +78,7 @@ static int ident_remove_id(
 	size_t need_size;
 
 	if (ident_find_id(&id_start, &id_end, from->ptr, from->size) < 0)
-		return GIT_ENOTFOUND;
+		return GIT_PASSTHROUGH;
 
 	need_size = (size_t)(id_start - from->ptr) +
 		4 /* "$Id$" */ + (size_t)(from_end - id_end);
@@ -102,7 +104,7 @@ static int ident_apply(
 
 	/* Don't filter binary files */
 	if (git_buf_text_is_binary(from))
-		return GIT_ENOTFOUND;
+		return GIT_PASSTHROUGH;
 
 	if (git_filter_source_mode(src) == GIT_FILTER_SMUDGE)
 		return ident_insert_id(to, from, src);
diff --git a/src/win32/pthread.c b/src/win32/pthread.c
index 8c7ef2856..db8927471 100644
--- a/src/win32/pthread.c
+++ b/src/win32/pthread.c
@@ -6,6 +6,7 @@
  */
 
 #include "pthread.h"
+#include "../global.h"
 
 int pthread_create(
 	pthread_t *GIT_RESTRICT thread,