xdiff/xhistogram: move index allocation into find_lcs

This fixes a memory issue when recursing a lot, which can be reproduced as

    seq 1   100000 >one
    seq 1 4 100000 >two
    git diff --no-index --histogram one two

Before this patch, histogram_diff would call itself recursively before
calling free_index, which would mean a lot of memory is allocated during
the recursion and only freed afterwards. By moving the memory allocation
(and its free call) into find_lcs, the memory is free'd before we recurse,
such that memory is reused in the next step of the recursion instead of
using new memory.

This addresses only the memory pressure, not the run time complexity,
that is also awful for the corner case outlined above.

Helpful in understanding the code (in addition to the sparse history of
this file), was https://stackoverflow.com/a/32367597 which reproduces
most of the code comments of the JGit implementation.

Signed-off-by: Stefan Beller <sbeller@google.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

xdiff/xhistogram.c

@@ -251,44 +251,13 @@ static inline void free_index(struct histindex *index)
 	xdl_cha_free(&index->rcha);
 }
 
-static int find_lcs(struct histindex *index, struct region *lcs,
+static int find_lcs(xpparam_t const *xpp, xdfenv_t *env,
-	int line1, int count1, int line2, int count2) {
+		    struct region *lcs,
-	int b_ptr;
+		    int line1, int count1, int line2, int count2)
-
-	if (scanA(index, line1, count1))
-		return -1;
-
-	index->cnt = index->max_chain_length + 1;
-
-	for (b_ptr = line2; b_ptr <= LINE_END(2); )
-		b_ptr = try_lcs(index, lcs, b_ptr, line1, count1, line2, count2);
-
-	return index->has_common && index->max_chain_length < index->cnt;
-}
-
-static int histogram_diff(xpparam_t const *xpp, xdfenv_t *env,
-	int line1, int count1, int line2, int count2)
 {
+	int b_ptr;
+	int sz, ret = -1;
 	struct histindex index;
-	struct region lcs;
-	int sz;
-	int result = -1;
-
-	if (count1 <= 0 && count2 <= 0)
-		return 0;
-
-	if (LINE_END(1) >= MAX_PTR)
-		return -1;
-
-	if (!count1) {
-		while(count2--)
-			env->xdf2.rchg[line2++ - 1] = 1;
-		return 0;
-	} else if (!count2) {
-		while(count1--)
-			env->xdf1.rchg[line1++ - 1] = 1;
-		return 0;
-	}
 
 	memset(&index, 0, sizeof(index));
 
@@ -326,8 +295,52 @@ static int histogram_diff(xpparam_t const *xpp, xdfenv_t *env,
 	index.ptr_shift = line1;
 	index.max_chain_length = 64;
 
+	if (scanA(&index, line1, count1))
+		goto cleanup;
+
+	index.cnt = index.max_chain_length + 1;
+
+	for (b_ptr = line2; b_ptr <= LINE_END(2); )
+		b_ptr = try_lcs(&index, lcs, b_ptr, line1, count1, line2, count2);
+
+	if (index.has_common && index.max_chain_length < index.cnt)
+		ret = 1;
+	else
+		ret = 0;
+
+cleanup:
+	free_index(&index);
+	return ret;
+}
+
+static int histogram_diff(xpparam_t const *xpp, xdfenv_t *env,
+	int line1, int count1, int line2, int count2)
+{
+	struct region lcs;
+	int lcs_found;
+	int result = -1;
+
+	if (count1 <= 0 && count2 <= 0)
+		return 0;
+
+	if (LINE_END(1) >= MAX_PTR)
+		return -1;
+
+	if (!count1) {
+		while(count2--)
+			env->xdf2.rchg[line2++ - 1] = 1;
+		return 0;
+	} else if (!count2) {
+		while(count1--)
+			env->xdf1.rchg[line1++ - 1] = 1;
+		return 0;
+	}
+
 	memset(&lcs, 0, sizeof(lcs));
-	if (find_lcs(&index, &lcs, line1, count1, line2, count2))
+	lcs_found = find_lcs(xpp, env, &lcs, line1, count1, line2, count2);
+	if (lcs_found < 0)
+		goto out;
+	else if (lcs_found)
 		result = fall_back_to_classic_diff(xpp, env, line1, count1, line2, count2);
 	else {
 		if (lcs.begin1 == 0 && lcs.begin2 == 0) {
@@ -341,18 +354,15 @@ static int histogram_diff(xpparam_t const *xpp, xdfenv_t *env,
 						line1, lcs.begin1 - line1,
 						line2, lcs.begin2 - line2);
 			if (result)
-				goto cleanup;
+				goto out;
 			result = histogram_diff(xpp, env,
 						lcs.end1 + 1, LINE_END(1) - lcs.end1,
 						lcs.end2 + 1, LINE_END(2) - lcs.end2);
 			if (result)
-				goto cleanup;
+				goto out;
 		}
 	}
-
+out:
-cleanup:
-	free_index(&index);
-
 	return result;
 }