io: move csumming away from bstream to gunzip

in future we want to checksum on gzip boundary basis, not the
full file.

7 files changed, 131 insertions, 75 deletions

diff --git a/src/apk_io.h b/src/apk_io.h
index b5e984a..629729a 100644
--- a/src/apk_io.h
+++ b/src/apk_io.h
@@ -36,7 +36,7 @@ struct apk_istream {
 
 struct apk_bstream {
 	size_t (*read)(void *stream, void **ptr);
-	void (*close)(void *stream, csum_p csum, size_t *size);
+	void (*close)(void *stream, size_t *size);
 };
 
 struct apk_ostream {
@@ -44,7 +44,20 @@ struct apk_ostream {
 	void (*close)(void *stream);
 };
 
-struct apk_istream *apk_bstream_gunzip(struct apk_bstream *, int);
+#define APK_MPART_BEGIN		0
+#define APK_MPART_BOUNDARY	1
+#define APK_MPART_END		2
+
+typedef int (*apk_multipart_cb)(void *ctx, EVP_MD_CTX *mdctx, int part);
+
+struct apk_istream *apk_bstream_gunzip_mpart(struct apk_bstream *, int,
+					     apk_multipart_cb cb, void *ctx);
+static inline struct apk_istream *apk_bstream_gunzip(struct apk_bstream *bs,
+						     int autoclose)
+{
+	return apk_bstream_gunzip_mpart(bs, autoclose, NULL, NULL);
+}
+
 struct apk_ostream *apk_ostream_gzip(struct apk_ostream *);
 
 struct apk_istream *apk_istream_from_fd(int fd);
diff --git a/src/archive.c b/src/archive.c
index a05aee2..6681b61 100644
--- a/src/archive.c
+++ b/src/archive.c
@@ -4,7 +4,7 @@
  * Copyright (C) 2008 Timo Teräs <timo.teras@iki.fi>
  * All rights reserved.
  *
- * This program is free software; you can redistribute it and/or modify it 
+ * This program is free software; you can redistribute it and/or modify it
  * under the terms of the GNU General Public License version 2 as published
  * by the Free Software Foundation. See http://www.gnu.org/ for details.
  */
@@ -91,7 +91,7 @@ int apk_parse_tar(struct apk_istream *is, apk_archive_entry_parser parser,
 		if (buf.name[0] == '\0') {
 			if (end) {
 				r = 0;
-				break;
+				//break;
 			}
 			end++;
 			continue;
diff --git a/src/audit.c b/src/audit.c
index 9ea0f25..3732aac 100644
--- a/src/audit.c
+++ b/src/audit.c
@@ -27,12 +27,10 @@ static int audit_directory(apk_hash_item item, void *ctx)
 	struct apk_db_file *dbf;
 	struct apk_database *db = (struct apk_database *) ctx;
 	struct dirent *de;
-	struct stat st;
-	struct apk_bstream *bs;
+	struct apk_file_info fi;
+	apk_blob_t bdir = APK_BLOB_STR(dbd->dirname);
 	char tmp[512], reason;
 	DIR *dir;
-	apk_blob_t bdir = APK_BLOB_STR(dbd->dirname);
-	csum_t csum;
 
 	if (!(dbd->flags & APK_DBDIRF_PROTECTED))
 		return 0;
@@ -49,10 +47,10 @@ static int audit_directory(apk_hash_item item, void *ctx)
 		snprintf(tmp, sizeof(tmp), "%s/%s",
 			 dbd->dirname, de->d_name);
 
-		if (stat(tmp, &st) < 0)
+		if (apk_file_get_info(tmp, &fi) < 0)
 			continue;
 
-		if (S_ISDIR(st.st_mode)) {
+		if (S_ISDIR(fi.mode)) {
 			if (apk_db_dir_query(db, APK_BLOB_STR(tmp)) != NULL)
 				continue;
 
@@ -60,13 +58,7 @@ static int audit_directory(apk_hash_item item, void *ctx)
 		} else {
 			dbf = apk_db_file_query(db, bdir, APK_BLOB_STR(de->d_name));
 			if (dbf != NULL) {
-				bs = apk_bstream_from_file(tmp);
-				if (bs == NULL)
-					continue;
-
-				bs->close(bs, csum, NULL);
-
-				if (apk_blob_compare(APK_BLOB_BUF(csum),
+				if (apk_blob_compare(APK_BLOB_BUF(fi.csum),
 						     APK_BLOB_BUF(dbf->csum)) == 0)
 					continue;
 
diff --git a/src/database.c b/src/database.c
index 293af4c..10c0b5e 100644
--- a/src/database.c
+++ b/src/database.c
@@ -36,6 +36,7 @@ struct install_ctx {
 
 	int script;
 	struct apk_db_dir_instance *diri;
+	csum_t data_csum;
 
 	apk_progress_cb cb;
 	void *cb_ctx;
@@ -1271,13 +1272,28 @@ static void apk_db_purge_pkg(struct apk_database *db,
 	apk_pkg_set_state(db, pkg, APK_PKG_NOT_INSTALLED);
 }
 
+static int apk_db_gzip_part(void *pctx, EVP_MD_CTX *mdctx, int part)
+{
+	struct install_ctx *ctx = (struct install_ctx *) pctx;
+
+	switch (part) {
+	case APK_MPART_BEGIN:
+		EVP_DigestInit_ex(mdctx, EVP_md5(), NULL);
+		break;
+	case APK_MPART_END:
+		EVP_DigestFinal_ex(mdctx, ctx->data_csum, NULL);
+		break;
+	}
+	return 0;
+}
+
 static int apk_db_unpack_pkg(struct apk_database *db,
 			     struct apk_package *newpkg,
-			     int upgrade, csum_t csum,
-			     apk_progress_cb cb, void *cb_ctx)
+			     int upgrade, apk_progress_cb cb, void *cb_ctx)
 {
 	struct install_ctx ctx;
 	struct apk_bstream *bs = NULL;
+	struct apk_istream *tar;
 	char pkgname[256], file[256];
 	int i, need_copy = FALSE;
 	size_t length;
@@ -1334,10 +1350,17 @@ static int apk_db_unpack_pkg(struct apk_database *db,
 		.cb = cb,
 		.cb_ctx = cb_ctx,
 	};
-	if (apk_parse_tar_gz(bs, apk_db_install_archive_entry, &ctx) != 0)
+
+	tar = apk_bstream_gunzip_mpart(bs, FALSE, apk_db_gzip_part, &ctx);
+	if (apk_parse_tar(tar, apk_db_install_archive_entry, &ctx) != 0)
 		goto err_close;
+	bs->close(bs, &length);
+
+	/* Check the package checksum */
+	if (memcmp(ctx.data_csum, newpkg->csum, sizeof(csum_t)) != 0)
+		apk_warning("%s-%s: checksum does not match",
+			    newpkg->name->name, newpkg->version);
 
-	bs->close(bs, csum, &length);
 	if (need_copy) {
 		if (length == newpkg->size) {
 			char file2[256];
@@ -1351,7 +1374,7 @@ static int apk_db_unpack_pkg(struct apk_database *db,
 
 	return 0;
 err_close:
-	bs->close(bs, NULL, NULL);
+	bs->close(bs, NULL);
 	return -1;
 }
 
@@ -1360,7 +1383,6 @@ int apk_db_install_pkg(struct apk_database *db,
 		       struct apk_package *newpkg,
 		       apk_progress_cb cb, void *cb_ctx)
 {
-	csum_t csum;
 	int r;
 
 	if (fchdir(db->root_fd) < 0)
@@ -1382,19 +1404,13 @@ int apk_db_install_pkg(struct apk_database *db,
 
 	/* Install the new stuff */
 	if (!(newpkg->name->flags & APK_NAME_VIRTUAL)) {
-		r = apk_db_unpack_pkg(db, newpkg, (oldpkg != NULL), csum,
-				      cb, cb_ctx);
+		r = apk_db_unpack_pkg(db, newpkg, (oldpkg != NULL), cb, cb_ctx);
 		if (r != 0)
 			return r;
 	}
 
 	apk_pkg_set_state(db, newpkg, APK_PKG_INSTALLED);
 
-	if (!(newpkg->name->flags & APK_NAME_VIRTUAL) &&
-	    memcmp(csum, newpkg->csum, sizeof(csum)) != 0)
-		apk_warning("%s-%s: checksum does not match",
-			    newpkg->name->name, newpkg->version);
-
 	if (oldpkg != NULL)
 		apk_db_purge_pkg(db, oldpkg);
 
diff --git a/src/gunzip.c b/src/gunzip.c
index 2c7ffc2..1d2881c 100644
--- a/src/gunzip.c
+++ b/src/gunzip.c
@@ -23,13 +23,18 @@ struct apk_gzip_istream {
 	z_stream zs;
 	int z_err;
 	int autoclose;
+
+	EVP_MD_CTX mdctx;
+	void *mdblock;
+
+	apk_multipart_cb cb;
+	void *cbctx;
 };
 
 static size_t gz_read(void *stream, void *ptr, size_t size)
 {
 	struct apk_gzip_istream *gis =
 		container_of(stream, struct apk_gzip_istream, is);
-	int restart_count = 0;
 
 	if (gis->z_err == Z_DATA_ERROR || gis->z_err == Z_ERRNO)
 		return -1;
@@ -44,22 +49,40 @@ static size_t gz_read(void *stream, void *ptr, size_t size)
 
 	while (gis->zs.avail_out != 0 && gis->z_err == Z_OK) {
 		if (gis->zs.avail_in == 0) {
-			gis->zs.avail_in = gis->bs->read(gis->bs, (void **) &gis->zs.next_in);
+			if (gis->cb != NULL && gis->mdblock != NULL) {
+				/* Digest the inflated bytes */
+				EVP_DigestUpdate(&gis->mdctx, gis->mdblock,
+						 (void *)gis->zs.next_in - gis->mdblock);
+			}
+			gis->zs.avail_in = gis->bs->read(gis->bs, &gis->mdblock);
+			gis->zs.next_in = (void *) gis->mdblock;
 			if (gis->zs.avail_in < 0) {
 				gis->z_err = Z_DATA_ERROR;
 				return size - gis->zs.avail_out;
+			} else if (gis->zs.avail_in == 0) {
+				if (gis->cb != NULL)
+					gis->cb(gis->cbctx, &gis->mdctx,
+						APK_MPART_END);
+				gis->z_err = Z_STREAM_END;
+				return size - gis->zs.avail_out;
 			}
 		}
 
 		gis->z_err = inflate(&gis->zs, Z_NO_FLUSH);
-		if (restart_count == 0 && gis->z_err == Z_STREAM_END) {
+		if (gis->z_err == Z_STREAM_END) {
+			/* Digest the inflated bytes */
+			if (gis->cb != NULL) {
+				EVP_DigestUpdate(&gis->mdctx, gis->mdblock,
+						 (void *)gis->zs.next_in - gis->mdblock);
+				gis->mdblock = gis->zs.next_in;
+				gis->cb(gis->cbctx, &gis->mdctx,
+					APK_MPART_BOUNDARY);
+			}
 			inflateEnd(&gis->zs);
+
 			if (inflateInit2(&gis->zs, 15+32) != Z_OK)
 				return -1;
 			gis->z_err = Z_OK;
-			restart_count++;
-		} else {
-			restart_count = 0;
 		}
 	}
 
@@ -74,13 +97,16 @@ static void gz_close(void *stream)
 	struct apk_gzip_istream *gis =
 		container_of(stream, struct apk_gzip_istream, is);
 
+	if (gis->cb != NULL)
+		EVP_MD_CTX_cleanup(&gis->mdctx);
 	inflateEnd(&gis->zs);
 	if (gis->autoclose)
-		gis->bs->close(gis->bs, NULL, NULL);
+		gis->bs->close(gis->bs, NULL);
 	free(gis);
 }
 
-struct apk_istream *apk_bstream_gunzip(struct apk_bstream *bs, int autoclose)
+struct apk_istream *apk_bstream_gunzip_mpart(struct apk_bstream *bs, int autoclose,
+					     apk_multipart_cb cb, void *ctx)
 {
 	struct apk_gzip_istream *gis;
 
@@ -97,6 +123,8 @@ struct apk_istream *apk_bstream_gunzip(struct apk_bstream *bs, int autoclose)
 		.bs = bs,
 		.z_err = 0,
 		.autoclose = autoclose,
+		.cb = cb,
+		.cbctx = ctx,
 	};
 
 	if (inflateInit2(&gis->zs, 15+32) != Z_OK) {
@@ -104,6 +132,11 @@ struct apk_istream *apk_bstream_gunzip(struct apk_bstream *bs, int autoclose)
 		return NULL;
 	}
 
+	if (gis->cb != NULL) {
+		EVP_MD_CTX_init(&gis->mdctx);
+		cb(ctx, &gis->mdctx, APK_MPART_BEGIN);
+	}
+
 	return &gis->is;
 }
 
diff --git a/src/io.c b/src/io.c
index 722bccb..a250b27 100644
--- a/src/io.c
+++ b/src/io.c
@@ -150,7 +150,6 @@ err:
 struct apk_istream_bstream {
 	struct apk_bstream bs;
 	struct apk_istream *is;
-	csum_ctx_t csum_ctx;
 	unsigned char buffer[8*1024];
 	size_t size;
 };
@@ -165,31 +164,17 @@ static size_t is_bs_read(void *stream, void **ptr)
 	if (size <= 0)
 		return size;
 
-	csum_process(&isbs->csum_ctx, isbs->buffer, size);
 	isbs->size += size;
 
 	*ptr = isbs->buffer;
 	return size;
 }
 
-static void is_bs_close(void *stream, csum_t csum, size_t *size)
+static void is_bs_close(void *stream, size_t *size)
 {
 	struct apk_istream_bstream *isbs =
 		container_of(stream, struct apk_istream_bstream, bs);
 
-	if (csum != NULL) {
-		size_t size;
-
-		do {
-			size = isbs->is->read(isbs->is, isbs->buffer,
-					      sizeof(isbs->buffer));
-			csum_process(&isbs->csum_ctx, isbs->buffer, size);
-			isbs->size += size;
-		} while (size == sizeof(isbs->buffer));
-
-		csum_finish(&isbs->csum_ctx, csum);
-	}
-
 	if (size != NULL)
 		*size = isbs->size;
 
@@ -210,14 +195,12 @@ struct apk_bstream *apk_bstream_from_istream(struct apk_istream *istream)
 		.close = is_bs_close,
 	};
 	isbs->is = istream;
-	csum_init(&isbs->csum_ctx);
 
 	return &isbs->bs;
 }
 
 struct apk_mmap_bstream {
 	struct apk_bstream bs;
-	csum_ctx_t csum_ctx;
 	int fd;
 	size_t size;
 	unsigned char *ptr;
@@ -235,24 +218,16 @@ static size_t mmap_read(void *stream, void **ptr)
 		size = 1024*1024;
 
 	*ptr = (void *) &mbs->ptr[mbs->pos];
-	csum_process(&mbs->csum_ctx, &mbs->ptr[mbs->pos], size);
 	mbs->pos += size;
 
 	return size;
 }
 
-static void mmap_close(void *stream, csum_t csum, size_t *size)
+static void mmap_close(void *stream, size_t *size)
 {
 	struct apk_mmap_bstream *mbs =
 		container_of(stream, struct apk_mmap_bstream, bs);
 
-	if (csum != NULL) {
-		if (mbs->pos != mbs->size)
-			csum_process(&mbs->csum_ctx, &mbs->ptr[mbs->pos],
-				     mbs->size - mbs->pos);
-		csum_finish(&mbs->csum_ctx, csum);
-	}
-
 	if (size != NULL)
 		*size = mbs->size;
 
@@ -288,7 +263,6 @@ static struct apk_bstream *apk_mmap_bstream_from_fd(int fd)
 	mbs->size = st.st_size;
 	mbs->ptr = ptr;
 	mbs->pos = 0;
-	csum_init(&mbs->csum_ctx);
 
 	return &mbs->bs;
 }
@@ -340,12 +314,12 @@ static size_t tee_read(void *stream, void **ptr)
 	return size;
 }
 
-static void tee_close(void *stream, csum_t csum, size_t *size)
+static void tee_close(void *stream, size_t *size)
 {
 	struct apk_tee_bstream *tbs =
 		container_of(stream, struct apk_tee_bstream, bs);
 
-	tbs->inner_bs->close(tbs->inner_bs, csum, NULL);
+	tbs->inner_bs->close(tbs->inner_bs, NULL);
 	if (size != NULL)
 		*size = tbs->size;
 	close(tbs->fd);
@@ -431,6 +405,7 @@ int apk_file_get_info(const char *filename, struct apk_file_info *fi)
 {
 	struct stat st;
 	struct apk_bstream *bs;
+	csum_ctx_t ctx;
 
 	if (stat(filename, &st) != 0)
 		return -1;
@@ -445,8 +420,17 @@ int apk_file_get_info(const char *filename, struct apk_file_info *fi)
 	};
 
 	bs = apk_bstream_from_file(filename);
-	if (bs != NULL)
-		bs->close(bs, fi->csum, NULL);
+	if (bs != NULL) {
+		ssize_t size;
+		void *ptr;
+
+		csum_init(&ctx);
+		while ((size = bs->read(bs, &ptr)) > 0)
+			csum_process(&ctx, ptr, size);
+		csum_finish(&ctx, fi->csum);
+
+		bs->close(bs, NULL);
+	}
 
 	return 0;
 }
diff --git a/src/package.c b/src/package.c
index 6c51924..30b09b1 100644
--- a/src/package.c
+++ b/src/package.c
@@ -4,7 +4,7 @@
  * Copyright (C) 2008 Timo Teräs <timo.teras@iki.fi>
  * All rights reserved.
  *
- * This program is free software; you can redistribute it and/or modify it 
+ * This program is free software; you can redistribute it and/or modify it
  * under the terms of the GNU General Public License version 2 as published
  * by the Free Software Foundation. See http://www.gnu.org/ for details.
  */
@@ -415,10 +415,26 @@ static int read_info_entry(void *ctx, const struct apk_file_info *ae,
 	return 0;
 }
 
+static int apk_pkg_gzip_part(void *ctx, EVP_MD_CTX *mdctx, int part)
+{
+	struct read_info_ctx *ri = (struct read_info_ctx *) ctx;
+
+	switch (part) {
+	case APK_MPART_BEGIN:
+		EVP_DigestInit_ex(mdctx, EVP_md5(), NULL);
+		break;
+	case APK_MPART_END:
+		EVP_DigestFinal_ex(mdctx, ri->pkg->csum, NULL);
+		break;
+	}
+	return 0;
+}
+
 struct apk_package *apk_pkg_read(struct apk_database *db, const char *file)
 {
 	struct read_info_ctx ctx;
 	struct apk_bstream *bs;
+	struct apk_istream *tar;
 	char realfile[PATH_MAX];
 
 	if (realpath(file, realfile) < 0)
@@ -434,12 +450,14 @@ struct apk_package *apk_pkg_read(struct apk_database *db, const char *file)
 
 	ctx.db = db;
 	ctx.has_install = 0;
-	if (apk_parse_tar_gz(bs, read_info_entry, &ctx) < 0) {
+
+	tar = apk_bstream_gunzip_mpart(bs, FALSE, apk_pkg_gzip_part, &ctx);
+	if (apk_parse_tar(tar, read_info_entry, &ctx) < 0) {
 		apk_error("File %s is not an APK archive", file);
-		bs->close(bs, NULL, NULL);
+		bs->close(bs, NULL);
 		goto err;
 	}
-	bs->close(bs, ctx.pkg->csum, &ctx.pkg->size);
+	bs->close(bs, &ctx.pkg->size);
 
 	if (ctx.pkg->name == NULL) {
 		apk_error("File %s is corrupted", file);
@@ -682,7 +700,7 @@ struct apk_dependency apk_dep_from_str(struct apk_database *db,
 		mask = apk_version_result_mask(v++);
 		if (*v == '=')
 			v++;
-	} 
+	}
 	return (struct apk_dependency) {
 		.name = apk_db_get_name(db, name),
 		.version = v,