Merge pull request #2686 from stoeckmann/lseek_win_regression

Fix Windows off_t handling

Makefile.am

@@ -132,6 +132,7 @@ libarchive_la_SOURCES= \
 	libarchive/archive_pathmatch.h \
 	libarchive/archive_platform.h \
 	libarchive/archive_platform_acl.h \
+	libarchive/archive_platform_stat.h \
 	libarchive/archive_platform_xattr.h \
 	libarchive/archive_ppmd_private.h \
 	libarchive/archive_ppmd7.c \

libarchive/CMakeLists.txt

@@ -52,6 +52,7 @@ SET(libarchive_SOURCES
   archive_pathmatch.h
   archive_platform.h
   archive_platform_acl.h
+  archive_platform_stat.h
   archive_platform_xattr.h
   archive_ppmd_private.h
   archive_ppmd8.c

libarchive/archive_platform_stat.h

@@ -0,0 +1,45 @@
+/*-
+ * SPDX-License-Identifier: BSD-2-Clause
+ *
+ * Copyright (c) 2025 Tobias Stoeckmann
+ * All rights reserved.
+ */
+
+/* !!ONLY FOR USE INTERNALLY TO LIBARCHIVE!! */
+
+#ifndef ARCHIVE_PLATFORM_STAT_H_INCLUDED
+#define ARCHIVE_PLATFORM_STAT_H_INCLUDED
+
+#ifndef __LIBARCHIVE_BUILD
+#error This header is only to be used internally to libarchive.
+#endif
+
+#if defined(_WIN32) && !defined(__CYGWIN__)
+/* We use _lseeki64() on Windows. */
+typedef int64_t la_seek_t;
+
+struct la_seek_stat {
+	int64_t		st_mtime;
+	ino_t		st_ino;
+	unsigned short	st_mode;
+	uint32_t	st_nlink;
+	gid_t		st_gid;
+	la_seek_t	st_size;
+	uid_t		st_uid;
+	dev_t		st_dev;
+	dev_t		st_rdev;
+};
+typedef struct la_seek_stat la_seek_stat_t;
+
+#define la_seek_fstat(fd, st)	__la_seek_fstat((fd), (st))
+#define la_seek_stat(fd, st)	__la_seek_stat((fd), (st))
+
+#else
+typedef off_t la_seek_t;
+typedef struct stat la_seek_stat_t;
+
+#define la_seek_fstat(fd, st)	fstat((fd), (st))
+#define la_seek_stat(fd, st)	stat((fd), (st))
+#endif
+
+#endif	/* !ARCHIVE_PLATFORM_STAT_H_INCLUDED */

libarchive/archive_read_open_fd.c

@@ -48,6 +48,7 @@
 #endif
 
 #include "archive.h"
+#include "archive_platform_stat.h"
 
 struct read_fd_data {
 	int	 fd;
@@ -65,12 +66,12 @@ static int64_t	file_skip(struct archive *, void *, int64_t request);
 int
 archive_read_open_fd(struct archive *a, int fd, size_t block_size)
 {
-	struct stat st;
+	la_seek_stat_t st;
 	struct read_fd_data *mine;
 	void *b;
 
 	archive_clear_error(a);
-	if (fstat(fd, &st) != 0) {
+	if (la_seek_fstat(fd, &st) != 0) {
 		archive_set_error(a, errno, "Can't stat fd %d", fd);
 		return (ARCHIVE_FATAL);
 	}
@@ -133,7 +134,7 @@ static int64_t
 file_skip(struct archive *a, void *client_data, int64_t request)
 {
 	struct read_fd_data *mine = (struct read_fd_data *)client_data;
-	off_t skip = (off_t)request;
+	la_seek_t skip = (la_seek_t)request;
 	int64_t old_offset, new_offset;
 	int skip_bits = sizeof(skip) * 8 - 1;  /* off_t is a signed type. */
 
@@ -149,7 +150,8 @@ file_skip(struct archive *a, void *client_data, int64_t request)
 	}
 
 	/* Reduce 'skip' to the next smallest multiple of block_size */
-	skip = (off_t)(((int64_t)skip / mine->block_size) * mine->block_size);
+	skip = (la_seek_t)(((int64_t)skip / mine->block_size) * mine->block_size);
+
 	if (skip == 0)
 		return (0);
 
@@ -185,27 +187,28 @@ static int64_t
 file_seek(struct archive *a, void *client_data, int64_t request, int whence)
 {
 	struct read_fd_data *mine = (struct read_fd_data *)client_data;
-	off_t seek = (off_t)request;
+	la_seek_t seek = (la_seek_t)request;
 	int64_t r;
 	int seek_bits = sizeof(seek) * 8 - 1;  /* off_t is a signed type. */
 
 	/* We use off_t here because lseek() is declared that way. */
 
-	/* Reduce a request that would overflow the 'seek' variable. */
+	/* Do not perform a seek which cannot be fulfilled. */
 	if (sizeof(request) > sizeof(seek)) {
 		const int64_t max_seek =
 		    (((int64_t)1 << (seek_bits - 1)) - 1) * 2 + 1;
 		const int64_t min_seek = ~max_seek;
-		if (request > max_seek)
-			seek = (off_t)max_seek;
-		else if (request < min_seek)
-			seek = (off_t)min_seek;
+		if (request < min_seek || request > max_seek) {
+			errno = EOVERFLOW;
+			goto err;
+		}
 	}
 
 	r = lseek(mine->fd, seek, whence);
 	if (r >= 0)
 		return r;
 
+err:
 	if (errno == ESPIPE) {
 		archive_set_error(a, errno,
 		    "A file descriptor(%d) is not seekable(PIPE)", mine->fd);

libarchive/archive_read_open_file.c

@@ -48,6 +48,7 @@
 #endif
 
 #include "archive.h"
+#include "archive_platform_stat.h"
 
 struct read_FILE_data {
 	FILE    *f;
@@ -65,7 +66,7 @@ static int64_t	FILE_skip(struct archive *, void *, int64_t);
 int
 archive_read_open_FILE(struct archive *a, FILE *f)
 {
-	struct stat st;
+	la_seek_stat_t st;
 	struct read_FILE_data *mine;
 	size_t block_size = 128 * 1024;
 	void *b;
@@ -88,7 +89,7 @@ archive_read_open_FILE(struct archive *a, FILE *f)
 	 * streams that don't support fileno()).  As a result, fileno()
 	 * should be used cautiously.)
 	 */
-	if (fstat(fileno(mine->f), &st) == 0 && S_ISREG(st.st_mode)) {
+	if (la_seek_fstat(fileno(mine->f), &st) == 0 && S_ISREG(st.st_mode)) {
 		archive_read_extract_set_skip_file(a, st.st_dev, st.st_ino);
 		/* Enable the seek optimization only for regular files. */
 		mine->can_skip = 1;
@@ -205,15 +206,15 @@ FILE_seek(struct archive *a, void *client_data, int64_t request, int whence)
 	int seek_bits = sizeof(seek) * 8 - 1;
 	(void)a; /* UNUSED */
 
-	/* Reduce a request that would overflow the 'seek' variable. */
+	/* Do not perform a seek which cannot be fulfilled. */
 	if (sizeof(request) > sizeof(seek)) {
 		const int64_t max_seek =
 		    (((int64_t)1 << (seek_bits - 1)) - 1) * 2 + 1;
 		const int64_t min_seek = ~max_seek;
-		if (request > max_seek)
-			seek = max_seek;
-		else if (request < min_seek)
-			seek = min_seek;
+		if (request < min_seek || request > max_seek) {
+			errno = EOVERFLOW;
+			goto err;
+		}
 	}
 
 #ifdef __ANDROID__
@@ -236,6 +237,7 @@ FILE_seek(struct archive *a, void *client_data, int64_t request, int whence)
 	}
 #endif
 	/* If we arrive here, the input is corrupted or truncated so fail. */
+err:
 	archive_set_error(a, errno, "Error seeking in FILE* pointer");
 	return (ARCHIVE_FATAL);
 }

libarchive/archive_read_open_filename.c

@@ -59,6 +59,7 @@
 #endif
 
 #include "archive.h"
+#include "archive_platform_stat.h"
 #include "archive_private.h"
 #include "archive_string.h"
 
@@ -252,7 +253,7 @@ archive_read_open_filename_w(struct archive *a, const wchar_t *wfilename,
 static int
 file_open(struct archive *a, void *client_data)
 {
-	struct stat st;
+	la_seek_stat_t st;
 	struct read_file_data *mine = (struct read_file_data *)client_data;
 	void *buffer;
 	const char *filename = NULL;
@@ -317,7 +318,7 @@ file_open(struct archive *a, void *client_data)
 		goto fail;
 #endif
 	}
-	if (fstat(fd, &st) != 0) {
+	if (la_seek_fstat(fd, &st) != 0) {
 #if defined(_WIN32) && !defined(__CYGWIN__)
 		if (mine->filename_type == FNT_WCS)
 			archive_set_error(a, errno, "Can't stat '%ls'",
@@ -486,10 +487,11 @@ file_skip_lseek(struct archive *a, void *client_data, int64_t request)
 	struct read_file_data *mine = (struct read_file_data *)client_data;
 #if defined(_WIN32) && !defined(__CYGWIN__)
 	/* We use _lseeki64() on Windows. */
-	int64_t old_offset, new_offset, skip = request;
+	int64_t old_offset, new_offset;
 #else
-	off_t old_offset, new_offset, skip = (off_t)request;
+	off_t old_offset, new_offset;
 #endif
+	la_seek_t skip = (la_seek_t)request;
 	int skip_bits = sizeof(skip) * 8 - 1;
 
 	/* We use off_t here because lseek() is declared that way. */
@@ -556,21 +558,21 @@ static int64_t
 file_seek(struct archive *a, void *client_data, int64_t request, int whence)
 {
 	struct read_file_data *mine = (struct read_file_data *)client_data;
-	off_t seek = (off_t)request;
+	la_seek_t seek = (la_seek_t)request;
 	int64_t r;
 	int seek_bits = sizeof(seek) * 8 - 1;
 
 	/* We use off_t here because lseek() is declared that way. */
 
-	/* Reduce a request that would overflow the 'seek' variable. */
+	/* Do not perform a seek which cannot be fulfilled. */
 	if (sizeof(request) > sizeof(seek)) {
 		const int64_t max_seek =
 		    (((int64_t)1 << (seek_bits - 1)) - 1) * 2 + 1;
 		const int64_t min_seek = ~max_seek;
-		if (request > max_seek)
-			seek = (off_t)max_seek;
-		else if (request < min_seek)
-			seek = (off_t)min_seek;
+		if (request < min_seek || request > max_seek) {
+			errno = EOVERFLOW;
+			goto err;
+		}
 	}
 
 	r = lseek(mine->fd, seek, whence);
@@ -578,6 +580,7 @@ file_seek(struct archive *a, void *client_data, int64_t request, int whence)
 		return r;
 
 	/* If the input is corrupted or truncated, fail. */
+err:
 	if (mine->filename_type == FNT_STDIN)
 		archive_set_error(a, errno, "Error seeking in stdin");
 	else if (mine->filename_type == FNT_MBS)

libarchive/archive_read_support_format_mtree.c

@@ -51,6 +51,7 @@
 #include "archive.h"
 #include "archive_entry.h"
 #include "archive_entry_private.h"
+#include "archive_platform_stat.h"
 #include "archive_private.h"
 #include "archive_rb.h"
 #include "archive_read_private.h"
@@ -1177,7 +1178,7 @@ parse_file(struct archive_read *a, struct archive_entry *entry,
     struct mtree *mtree, struct mtree_entry *mentry, int *use_next)
 {
 	const char *path;
-	struct stat st_storage, *st;
+	la_seek_stat_t st_storage, *st;
 	struct mtree_entry *mp;
 	struct archive_entry *sparse_entry;
 	int r = ARCHIVE_OK, r1, parsed_kws;
@@ -1272,7 +1273,7 @@ parse_file(struct archive_read *a, struct archive_entry *entry,
 
 		st = &st_storage;
 		if (mtree->fd >= 0) {
-			if (fstat(mtree->fd, st) == -1) {
+			if (la_seek_fstat(mtree->fd, st) == -1) {
 				archive_set_error(&a->archive, errno,
 						"Could not fstat %s", path);
 				r = ARCHIVE_WARN;
@@ -1285,7 +1286,7 @@ parse_file(struct archive_read *a, struct archive_entry *entry,
 #ifdef HAVE_LSTAT
 		else if (lstat(path, st) == -1)
 #else
-		else if (la_stat(path, st) == -1)
+		else if (la_seek_stat(path, st) == -1)
 #endif
 		{
 			st = NULL;

libarchive/archive_windows.c

@@ -45,6 +45,7 @@
 #if defined(_WIN32) && !defined(__CYGWIN__)
 
 #include "archive_platform.h"
+#include "archive_platform_stat.h"
 #include "archive_private.h"
 #include "archive_entry.h"
 #include "archive_time_private.h"
@@ -723,6 +724,53 @@ __la_stat(const char *path, struct stat *st)
 	return (ret);
 }
 
+static void
+copy_seek_stat(la_seek_stat_t *st, struct ustat *us)
+{
+	st->st_mtime = us->st_mtime;
+	st->st_gid = us->st_gid;
+	st->st_ino = getino(us);
+	st->st_mode = us->st_mode;
+	st->st_nlink = us->st_nlink;
+	st->st_size = (la_seek_t)us->st_size;
+	if (st->st_size < 0 || (uint64_t)st->st_size != us->st_size)
+		st->st_size = -1;
+	st->st_uid = us->st_uid;
+	st->st_dev = us->st_dev;
+	st->st_rdev = us->st_rdev;
+}
+
+int
+__la_seek_fstat(int fd, la_seek_stat_t *st)
+{
+	struct ustat u;
+	int ret;
+
+	ret = __hstat((HANDLE)_get_osfhandle(fd), &u);
+	copy_seek_stat(st, &u);
+	return (ret);
+}
+
+int
+__la_seek_stat(const char *path, la_seek_stat_t *st)
+{
+	HANDLE handle;
+	struct ustat u;
+	int ret;
+
+	handle = la_CreateFile(path, 0, FILE_SHARE_READ, NULL, OPEN_EXISTING,
+		FILE_FLAG_BACKUP_SEMANTICS,
+		NULL);
+	if (handle == INVALID_HANDLE_VALUE) {
+		la_dosmaperr(GetLastError());
+		return (-1);
+	}
+	ret = __hstat(handle, &u);
+	CloseHandle(handle);
+	copy_seek_stat(st, &u);
+	return (ret);
+}
+
 /*
  * This waitpid is limited implementation.
  */

libarchive/archive_windows.h

@@ -72,6 +72,8 @@
 #include <windows.h>
 //#define	EFTYPE 7
 
+#include "archive_platform_stat.h"
+
 #if defined(__BORLANDC__)
 #pragma warn -8068	/* Constant out of range in comparison. */
 #pragma warn -8072	/* Suspicious pointer arithmetic. */
@@ -262,6 +264,9 @@
     #define	F_OK    0       /*  Test for existence of file  */
 #endif
 
+/* Functions to circumvent off_t limitations */
+int __la_seek_fstat(int fd, la_seek_stat_t *st);
+int __la_seek_stat(const char *path, la_seek_stat_t *st);
 
 /* Replacement POSIX function */
 extern int	 __la_fstat(int fd, struct stat *st);