2adbd7ee43
write a new test to exercise the hardlink strategies used by different archive formats (tar, old cpio, new cpio). This uncovered two problems, both fixed by this commit: 1) Enforce file size when writing files to disk. 2) When restoring hardlink entries, if they have data associated, go ahead and open the file so we can write the data. In particular, this fixes bsdtar/bsdcpio extraction of new cpio formats where the "original" is empty and the subsequent "hardlink" entry actually carries the data. It also provides correct behavior for old cpio archives where hardlinked entries have their bodies stored multiple times in the archive; the last body should always be the one that ends up in the final file. The new pax format also permits (but does not require) hardlinks to carry file data; again, the last contents should always win. Note that with any of these, a size of zero on a hardlink simply means that the hardlink carries no data; it does not mean that the file has zero size. A non-zero size on a hardlink does provide the file size. Thanks to: John Baldwin, for reminding me about this long-standing bug and sending me a simple example archive that prompted this test case |
||
---|---|---|
.. | ||
main.c | ||
Makefile | ||
read_open_memory.c | ||
README | ||
test_acl_basic.c | ||
test_acl_pax.c | ||
test_archive_api_feature.c | ||
test_bad_fd.c | ||
test_compat_gtar_1.tgz.uu | ||
test_compat_gtar.c | ||
test_compat_zip_1.zip.uu | ||
test_compat_zip.c | ||
test_empty_write.c | ||
test_entry_strmode.c | ||
test_entry.c | ||
test_read_compress_program.c | ||
test_read_data_large.c | ||
test_read_extract.c | ||
test_read_format_ar.c | ||
test_read_format_cpio_bin_bz2.c | ||
test_read_format_cpio_bin_gz.c | ||
test_read_format_cpio_bin_Z.c | ||
test_read_format_cpio_bin.c | ||
test_read_format_cpio_odc.c | ||
test_read_format_cpio_svr4_gzip.c | ||
test_read_format_cpio_svr4c_Z.c | ||
test_read_format_empty.c | ||
test_read_format_gtar_gz.c | ||
test_read_format_gtar_sparse_1_13.tgz.uu | ||
test_read_format_gtar_sparse_1_17_posix00.tgz.uu | ||
test_read_format_gtar_sparse_1_17_posix01.tgz.uu | ||
test_read_format_gtar_sparse_1_17_posix10_modified.tar.uu | ||
test_read_format_gtar_sparse_1_17_posix10.tgz.uu | ||
test_read_format_gtar_sparse_1_17.tgz.uu | ||
test_read_format_gtar_sparse.c | ||
test_read_format_iso_gz.c | ||
test_read_format_isorr_bz2.c | ||
test_read_format_mtree.c | ||
test_read_format_pax_bz2.c | ||
test_read_format_tar.c | ||
test_read_format_tbz.c | ||
test_read_format_tgz.c | ||
test_read_format_tz.c | ||
test_read_format_zip.c | ||
test_read_large.c | ||
test_read_pax_truncated.c | ||
test_read_position.c | ||
test_read_truncated.c | ||
test_tar_filenames.c | ||
test_tar_large.c | ||
test_write_compress_program.c | ||
test_write_disk_hardlink.c | ||
test_write_disk_perms.c | ||
test_write_disk_secure.c | ||
test_write_disk.c | ||
test_write_format_ar.c | ||
test_write_format_cpio_empty.c | ||
test_write_format_cpio_newc.c | ||
test_write_format_cpio_odc.c | ||
test_write_format_cpio.c | ||
test_write_format_shar_empty.c | ||
test_write_format_tar_empty.c | ||
test_write_format_tar.c | ||
test_write_open_memory.c | ||
test.h |
$FreeBSD$ This is the test harness for libarchive. It compiles into a single program "libarchive_test" that is intended to exercise as much of the library as possible. It is, of course, very much a work in progress. Each test is a function named test_foo in a file named test_foo.c. Note that the file name is the same as the function name. Each file must start with this line: #include "test.h" The test function must be declared with a line of this form DEFINE_TEST(test_foo) Nothing else should appear on that line. When you add a test, please update the Makefile to add your file to the list of tests. The Makefile and main.c use various macro trickery to automatically collect a list of test functions to be invoked. Each test function can rely on the following: * The current directory will be a freshly-created empty directory suitable for that test. (The top-level main() creates a directory for each separate test and chdir()s to that directory before running the test.) * The test function should use assert(), assertA() and similar macros defined in test.h. If you need to add new macros of this form, feel free to do so. The current macro set includes assertEqualInt() and assertEqualString() that print out additional detail about their arguments if the assertion does fail. 'A' versions also accept a struct archive * and display any error message from there on failure. * You are encouraged to document each assertion with a failure() call just before the assert. The failure() function is a printf-like function whose text is displayed only if the assertion fails. It can be used to display additional information relevant to the failure: failure("The data read from file %s did not match the data written to that file.", filename); assert(strcmp(buff1, buff2) == 0); * Tests are encouraged to be economical with their memory and disk usage, though this is not essential. The test is occasionally run under a memory debugger to try to locate memory leaks in the library; as a result, tests should be careful to release any memory they allocate. * Disable tests on specific platforms as necessary. Please don't use config.h to adjust feature requirements, as I want the tests to also serve as a check on the configure process. The following form is appropriate: #if !defined(__PLATFORM) && !defined(__Platform2__) assert(xxxx) #endif