From 7c9ff5f1667a16733163b75dfd4b509662c387f4 Mon Sep 17 00:00:00 2001 From: Jia Tan Date: Wed, 21 Dec 2022 21:12:03 +0800 Subject: [PATCH] Tests: Adds lzip decoder tests --- .gitignore | 1 + tests/Makefile.am | 2 + tests/test_lzip_decoder.c | 471 ++++++++++++++++++++++++++++++++++++++ 3 files changed, 474 insertions(+) create mode 100644 tests/test_lzip_decoder.c diff --git a/.gitignore b/.gitignore index f0b11aa9..44f269cc 100644 --- a/.gitignore +++ b/.gitignore @@ -62,6 +62,7 @@ coverage /tests/test_filter_flags /tests/test_hardware /tests/test_index +/test/test_lzip_decoder /tests/test_memlimit /tests/test_stream_flags /tests/test_vli diff --git a/tests/Makefile.am b/tests/Makefile.am index e612aa68..f84c1749 100644 --- a/tests/Makefile.am +++ b/tests/Makefile.am @@ -46,6 +46,7 @@ check_PROGRAMS = \ test_index \ test_bcj_exact_size \ test_memlimit \ + test_lzip_decoder \ test_vli TESTS = \ @@ -57,6 +58,7 @@ TESTS = \ test_index \ test_bcj_exact_size \ test_memlimit \ + test_lzip_decoder \ test_vli \ test_files.sh \ test_compress_prepared_bcj_sparc \ diff --git a/tests/test_lzip_decoder.c b/tests/test_lzip_decoder.c new file mode 100644 index 00000000..a5495bf9 --- /dev/null +++ b/tests/test_lzip_decoder.c @@ -0,0 +1,471 @@ +/////////////////////////////////////////////////////////////////////////////// +// +/// \file test_lzip_decoder.c +/// \brief Tests decoding lzip data +// +// Author: Jia Tan +// +// This file has been put into the public domain. +// You can do whatever you want with this file. +// +/////////////////////////////////////////////////////////////////////////////// + +#include "tests.h" + +#ifdef HAVE_LZIP_DECODER + +// Memlimit large enough to pass all of the test files +#define MEMLIMIT (1U << 20) +#define DECODE_CHUNK_SIZE 1024 + + +// Avoiding using data buffers so we don't have to store the data buffers +// as large hex strings. Instead, store the CRC32 value of the expected data. +// CRC32 value of "Hello\nWorld\n" +static const uint32_t hello_world_crc = 0x15A2A343; + +// CRC32 value of "Trailing garbage\n" +static const uint32_t trailing_garbage_crc = 0x87081A60; + + +// Helper function to decode a good file with no flags and plenty high memlimit +static void +basic_lzip_decode(const char *src, const uint32_t expected_crc) { + size_t file_size; + uint8_t *data = tuktest_file_from_srcdir(src, &file_size); + uint32_t checksum = 0; + + lzma_stream strm = LZMA_STREAM_INIT; + assert_lzma_ret(lzma_lzip_decoder(&strm, MEMLIMIT, 0), LZMA_OK); + + uint8_t *output_buffer = tuktest_malloc(DECODE_CHUNK_SIZE); + + strm.next_in = data; + strm.next_out = output_buffer; + strm.avail_out = DECODE_CHUNK_SIZE; + + // Feed 1 byte at a time to the decoder to look for any bugs + // when switching between decoding sequences + lzma_ret ret = LZMA_OK; + while (ret == LZMA_OK) { + strm.avail_in = 1; + ret = lzma_code(&strm, LZMA_RUN); + if (strm.avail_out == 0) { + checksum = lzma_crc32(output_buffer, + strm.next_out - output_buffer, + checksum); + // No need to free output_buffer because it will + // automatically be freed at the end of the test by + // tuktest. + output_buffer = tuktest_malloc(DECODE_CHUNK_SIZE); + strm.next_out = output_buffer; + strm.avail_out = DECODE_CHUNK_SIZE; + } + } + + assert_lzma_ret(ret, LZMA_STREAM_END); + assert_uint_eq(strm.total_in, file_size); + + checksum = lzma_crc32(output_buffer, strm.next_out - output_buffer, + checksum); + assert_uint_eq(checksum, expected_crc); + + lzma_end(&strm); +} + + +static void +test_options(void) +{ + // Test NULL stream + assert_lzma_ret(lzma_lzip_decoder(NULL, MEMLIMIT, 0), + LZMA_PROG_ERROR); + + // Test invalid flags + lzma_stream strm = LZMA_STREAM_INIT; + assert_lzma_ret(lzma_lzip_decoder(&strm, MEMLIMIT, UINT32_MAX), + LZMA_OPTIONS_ERROR); + // Memlimit tests are done elsewhere +} + + +static void +test_v0_decode(void) { + // This tests if liblzma can decode lzip version 0 files. + // lzip 1.17 and older can decompress this, but lzip 1.18 + // and newer can no longer decode these files. + basic_lzip_decode("files/good-1-v0.lz", hello_world_crc); +} + + +static void +test_v1_decode(void) { + // This tests decoding a basic lzip v1 file + basic_lzip_decode("files/good-1-v1.lz", hello_world_crc); +} + + +// Helper function to decode a good file with trailing bytes after +// the lzip stream +static void +trailing_helper(const char *src, const uint32_t expected_data_checksum, + const uint32_t expected_trailing_checksum) { + size_t file_size; + uint32_t checksum = 0; + uint8_t *data = tuktest_file_from_srcdir(src, &file_size); + lzma_stream strm = LZMA_STREAM_INIT; + assert_lzma_ret(lzma_lzip_decoder(&strm, MEMLIMIT, + LZMA_CONCATENATED), LZMA_OK); + + uint8_t *output_buffer = tuktest_malloc(DECODE_CHUNK_SIZE); + + strm.next_in = data; + strm.next_out = output_buffer; + strm.avail_in = file_size; + strm.avail_out = DECODE_CHUNK_SIZE; + + lzma_ret ret = LZMA_OK; + while (ret == LZMA_OK) { + ret = lzma_code(&strm, LZMA_RUN); + if (strm.avail_out == 0) { + checksum = lzma_crc32(output_buffer, + strm.next_out - output_buffer, + checksum); + // No need to free output_buffer because it will + // automatically be freed at the end of the test by + // tuktest. + output_buffer = tuktest_malloc(DECODE_CHUNK_SIZE); + strm.next_out = output_buffer; + strm.avail_out = DECODE_CHUNK_SIZE; + } + } + + assert_lzma_ret(ret, LZMA_STREAM_END); + assert_uint(strm.total_in, <, file_size); + + checksum = lzma_crc32(output_buffer, + strm.next_out - output_buffer, + checksum); + + assert_uint_eq(checksum, expected_data_checksum); + + // Trailing data should be readable from strm.next_in + checksum = lzma_crc32(strm.next_in, strm.avail_in, 0); + assert_uint_eq(checksum, expected_trailing_checksum); + + lzma_end(&strm); +} + + +// Helper function to decode a bad file and compare to returned error to +// what the caller expects +static void +decode_expect_error(const char *src, lzma_ret expected_error) +{ + lzma_stream strm = LZMA_STREAM_INIT; + size_t file_size; + uint8_t *data = tuktest_file_from_srcdir(src, &file_size); + + assert_lzma_ret(lzma_lzip_decoder(&strm, MEMLIMIT, + LZMA_CONCATENATED), LZMA_OK); + + uint8_t output_buffer[DECODE_CHUNK_SIZE]; + + strm.avail_in = file_size; + strm.next_in = data; + strm.avail_out = DECODE_CHUNK_SIZE; + strm.next_out = output_buffer; + + lzma_ret ret = LZMA_OK; + + while (ret == LZMA_OK) { + // Discard output since we are only looking for errors + strm.next_out = output_buffer; + strm.avail_out = DECODE_CHUNK_SIZE; + if (strm.avail_in == 0) + ret = lzma_code(&strm, LZMA_FINISH); + else + ret = lzma_code(&strm, LZMA_RUN); + } + + assert_lzma_ret(ret, expected_error); + lzma_end(&strm); +} + + +static void +test_v0_trailing(void) { + trailing_helper("files/good-1-v0-trailing-1.lz", hello_world_crc, + trailing_garbage_crc); +} + + +static void +test_v1_trailing(void) { + trailing_helper("files/good-1-v1-trailing-1.lz", hello_world_crc, + trailing_garbage_crc); + + // The second files/good-1-v1-trailing-2.lz will have the same + // expected output and trailing output as + // files/good-1-v1-trailing-1.lz, but this tests if the prefix + // to the trailing data contains lzip magic bytes. + // When this happens, the expected behavior is to silently ignore + // the magic byte prefix and consume it from the input file. + trailing_helper("files/good-1-v1-trailing-2.lz", hello_world_crc, + trailing_garbage_crc); + + // Expect LZMA_BUF error if a file ends with the lzip magic bytes + // but does not contain any data after + decode_expect_error("files/bad-1-v1-trailing-magic.lz", + LZMA_BUF_ERROR); +} + + +static void +test_concatentated(void) +{ + // First test a file with one v0 member and one v1 member + // The first member should contain "Hello\n" and + // the second member should contain "World!\n" + + lzma_stream strm = LZMA_STREAM_INIT; + size_t file_size; + uint8_t *v0_v1 = tuktest_file_from_srcdir("files/good-2-v0-v1.lz", + &file_size); + + assert_lzma_ret(lzma_lzip_decoder(&strm, MEMLIMIT, + LZMA_CONCATENATED), LZMA_OK); + + uint8_t output_buffer[DECODE_CHUNK_SIZE]; + + strm.avail_in = file_size; + strm.next_in = v0_v1; + strm.avail_out = DECODE_CHUNK_SIZE; + strm.next_out = output_buffer; + + assert_lzma_ret(lzma_code(&strm, LZMA_FINISH), LZMA_STREAM_END); + + assert_uint_eq(strm.total_in, file_size); + + uint32_t checksum = lzma_crc32(output_buffer, strm.total_out, 0); + assert_uint_eq(checksum, hello_world_crc); + + // The second file contains one v1 member and one v2 member + uint8_t *v1_v0 = tuktest_file_from_srcdir("files/good-2-v1-v0.lz", + &file_size); + + assert_lzma_ret(lzma_lzip_decoder(&strm, MEMLIMIT, + LZMA_CONCATENATED), LZMA_OK); + + strm.avail_in = file_size; + strm.next_in = v1_v0; + strm.avail_out = DECODE_CHUNK_SIZE; + strm.next_out = output_buffer; + + assert_lzma_ret(lzma_code(&strm, LZMA_FINISH), LZMA_STREAM_END); + + assert_uint_eq(strm.total_in, file_size); + checksum = lzma_crc32(output_buffer, strm.total_out, 0); + assert_uint_eq(checksum, hello_world_crc); + + // The third file contains 2 v1 members + uint8_t *v1_v1 = tuktest_file_from_srcdir("files/good-2-v1-v1.lz", + &file_size); + + assert_lzma_ret(lzma_lzip_decoder(&strm, MEMLIMIT, + LZMA_CONCATENATED), LZMA_OK); + + strm.avail_in = file_size; + strm.next_in = v1_v1; + strm.avail_out = DECODE_CHUNK_SIZE; + strm.next_out = output_buffer; + + assert_lzma_ret(lzma_code(&strm, LZMA_FINISH), LZMA_STREAM_END); + + assert_uint_eq(strm.total_in, file_size); + checksum = lzma_crc32(output_buffer, strm.total_out, 0); + assert_uint_eq(checksum, hello_world_crc); + + lzma_end(&strm); +} + + +static void +test_crc(void) { + // Test invalid checksum + lzma_stream strm = LZMA_STREAM_INIT; + size_t file_size; + uint8_t *data = tuktest_file_from_srcdir("files/bad-1-v1-crc32.lz", + &file_size); + + assert_lzma_ret(lzma_lzip_decoder(&strm, MEMLIMIT, + LZMA_CONCATENATED), LZMA_OK); + + uint8_t output_buffer[DECODE_CHUNK_SIZE]; + + strm.avail_in = file_size; + strm.next_in = data; + strm.avail_out = DECODE_CHUNK_SIZE; + strm.next_out = output_buffer; + + assert_lzma_ret(lzma_code(&strm, LZMA_FINISH), LZMA_DATA_ERROR); + + // Test ignoring the checksum value - should decode successfully + assert_lzma_ret(lzma_lzip_decoder(&strm, MEMLIMIT, + LZMA_CONCATENATED | LZMA_IGNORE_CHECK), LZMA_OK); + + strm.avail_in = file_size; + strm.next_in = data; + strm.avail_out = DECODE_CHUNK_SIZE; + strm.next_out = output_buffer; + + assert_lzma_ret(lzma_code(&strm, LZMA_FINISH), LZMA_STREAM_END); + assert_uint_eq(strm.total_in, file_size); + + // Test tell check + assert_lzma_ret(lzma_lzip_decoder(&strm, MEMLIMIT, + LZMA_CONCATENATED | LZMA_TELL_ANY_CHECK), LZMA_OK); + + strm.avail_in = file_size; + strm.next_in = data; + strm.avail_out = DECODE_CHUNK_SIZE; + strm.next_out = output_buffer; + + assert_lzma_ret(lzma_code(&strm, LZMA_FINISH), LZMA_GET_CHECK); + assert_uint_eq(lzma_get_check(&strm), LZMA_CHECK_CRC32); + assert_lzma_ret(lzma_code(&strm, LZMA_FINISH), LZMA_DATA_ERROR); + lzma_end(&strm); +} + + +static void +test_invalid_magic_bytes(void) { + uint8_t lzip_id_string[] = { 0x4C, 0x5A, 0x49, 0x50 }; + lzma_stream strm = LZMA_STREAM_INIT; + + for (uint32_t i = 0; i < ARRAY_SIZE(lzip_id_string); i++) { + // Corrupt magic bytes + lzip_id_string[i] ^= 1; + uint8_t output_buffer[DECODE_CHUNK_SIZE]; + + assert_lzma_ret(lzma_lzip_decoder(&strm, MEMLIMIT, 0), + LZMA_OK); + + strm.next_in = lzip_id_string; + strm.avail_in = sizeof(lzip_id_string); + strm.next_out = output_buffer; + strm.avail_out = DECODE_CHUNK_SIZE; + + assert_lzma_ret(lzma_code(&strm, LZMA_RUN), + LZMA_FORMAT_ERROR); + + // Reset magic bytes + lzip_id_string[i] ^= 1; + } + + lzma_end(&strm); +} + + +static void +test_invalid_version(void) +{ + // The file contains a version number that is not 0 or 1, + // so it should cause an error + decode_expect_error("files/unsupported-1-v234.lz", + LZMA_OPTIONS_ERROR); +} + + +static void +test_invalid_dictionary_size(void) { + // First file has too small dictionary size field + decode_expect_error("files/bad-1-v1-dict-1.lz", LZMA_DATA_ERROR); + + // Second file has too large dictionary size field + decode_expect_error("files/bad-1-v1-dict-2.lz", LZMA_DATA_ERROR); +} + + +static void +test_invalid_uncomp_size(void) { + // Test invalid v0 lzip file uncomp size + decode_expect_error("files/bad-1-v0-uncomp-size.lz", + LZMA_DATA_ERROR); + + // Test invalid v1 lzip file uncomp size + decode_expect_error("files/bad-1-v1-uncomp-size.lz", + LZMA_DATA_ERROR); +} + + +static void +test_invalid_member_size(void) { + decode_expect_error("files/bad-1-v1-member-size.lz", + LZMA_DATA_ERROR); +} + + +static void +test_invalid_memlimit(void) { + // A very low memlimit should prevent decoding. + // Should be able to update the memlimit after failing + size_t file_size; + uint8_t *data = tuktest_file_from_srcdir("files/good-1-v1.lz", + &file_size); + + uint8_t output_buffer[DECODE_CHUNK_SIZE]; + + lzma_stream strm = LZMA_STREAM_INIT; + + assert_lzma_ret(lzma_lzip_decoder(&strm, 1, 0), LZMA_OK); + + strm.next_in = data; + strm.avail_in = file_size; + strm.next_out = output_buffer; + strm.avail_out = DECODE_CHUNK_SIZE; + + assert_lzma_ret(lzma_code(&strm, LZMA_FINISH), LZMA_MEMLIMIT_ERROR); + + // Up the memlimit so decoding can continue. + // First only increase by a small amount and expect an error + assert_lzma_ret(lzma_memlimit_set(&strm, 100), LZMA_MEMLIMIT_ERROR); + assert_lzma_ret(lzma_memlimit_set(&strm, MEMLIMIT), LZMA_OK); + + // Finish decoding + assert_lzma_ret(lzma_code(&strm, LZMA_FINISH), LZMA_STREAM_END); + + assert_uint_eq(strm.total_in, file_size); + uint32_t checksum = lzma_crc32(output_buffer, strm.total_out, 0); + assert_uint_eq(checksum, hello_world_crc); + + lzma_end(&strm); +} +#endif + + +extern int +main(int argc, char **argv) +{ + tuktest_start(argc, argv); + +#ifndef HAVE_LZIP_DECODER + tuktest_early_skip("lzip decoder disabled"); +#else + tuktest_run(test_options); + tuktest_run(test_v0_decode); + tuktest_run(test_v1_decode); + tuktest_run(test_v0_trailing); + tuktest_run(test_v1_trailing); + tuktest_run(test_concatentated); + tuktest_run(test_crc); + tuktest_run(test_invalid_magic_bytes); + tuktest_run(test_invalid_version); + tuktest_run(test_invalid_dictionary_size); + tuktest_run(test_invalid_uncomp_size); + tuktest_run(test_invalid_member_size); + tuktest_run(test_invalid_memlimit); + return tuktest_end(); +#endif + +}