aboutsummaryrefslogtreecommitdiff
path: root/src/xz
diff options
context:
space:
mode:
authorLasse Collin <lasse.collin@tukaani.org>2008-12-27 19:30:19 +0200
committerLasse Collin <lasse.collin@tukaani.org>2008-12-27 19:30:19 +0200
commit4820f10d0f173864f6a2ea7479663b509ac53358 (patch)
treea4ba58d89bbfccd333a721b02523742b05d3e62f /src/xz
parentBunch of liblzma tweaks, including some API changes. (diff)
downloadxz-4820f10d0f173864f6a2ea7479663b509ac53358.tar.xz
Some xz command line tool improvements.
Diffstat (limited to 'src/xz')
-rw-r--r--src/xz/args.c23
-rw-r--r--src/xz/message.c4
-rw-r--r--src/xz/options.c2
-rw-r--r--src/xz/process.c133
-rw-r--r--src/xz/process.h3
5 files changed, 121 insertions, 44 deletions
diff --git a/src/xz/args.c b/src/xz/args.c
index a2efb277..c9d1dc14 100644
--- a/src/xz/args.c
+++ b/src/xz/args.c
@@ -53,7 +53,7 @@ parse_real(args_info *args, int argc, char **argv)
OPT_FILES0,
};
- static const char short_opts[] = "cC:dfF:hHlLkM:p:qrS:tT:vVz123456789";
+ static const char short_opts[] = "cC:defF:hHlkM:qrS:tT:vVz0123456789";
static const struct option long_opts[] = {
// Operation mode
@@ -77,11 +77,11 @@ parse_real(args_info *args, int argc, char **argv)
// Basic compression settings
{ "format", required_argument, NULL, 'F' },
{ "check", required_argument, NULL, 'C' },
- { "preset", required_argument, NULL, 'p' },
{ "memory", required_argument, NULL, 'M' },
{ "threads", required_argument, NULL, 'T' },
- { "fast", no_argument, NULL, '1' },
+ { "extreme", no_argument, NULL, 'e' },
+ { "fast", no_argument, NULL, '0' },
{ "best", no_argument, NULL, '9' },
// Filters
@@ -114,20 +114,12 @@ parse_real(args_info *args, int argc, char **argv)
while ((c = getopt_long(argc, argv, short_opts, long_opts, NULL))
!= -1) {
switch (c) {
- // gzip-like options
-
- case '1': case '2': case '3': case '4':
+ // Compression preset (also for decompression if --format=raw)
+ case '0': case '1': case '2': case '3': case '4':
case '5': case '6': case '7': case '8': case '9':
coder_set_preset(c - '0');
break;
- case 'p': {
- const uint64_t preset = str_to_uint64(
- "preset", optarg, 1, 9);
- coder_set_preset(preset);
- break;
- }
-
// --memory
case 'M':
// On 32-bit systems, SIZE_MAX would make more sense
@@ -162,6 +154,11 @@ parse_real(args_info *args, int argc, char **argv)
opt_mode = MODE_DECOMPRESS;
break;
+ // --extreme
+ case 'e':
+ coder_set_extreme();
+ break;
+
// --force
case 'f':
opt_force = true;
diff --git a/src/xz/message.c b/src/xz/message.c
index d9733742..c2e35e8e 100644
--- a/src/xz/message.c
+++ b/src/xz/message.c
@@ -891,8 +891,8 @@ message_help(bool long_help)
}
puts(_(
-" -p, --preset=NUM compression preset: 1-2 fast compression, 3-6 good\n"
-" compression, 7-9 excellent compression; default is 7"));
+" -0 .. -9 compression preset; 0-2 fast compression, 3-5 good\n"
+" compression, 6-9 excellent compression; default is 6"));
puts(_(
" -M, --memory=NUM use roughly NUM bytes of memory at maximum; 0 indicates\n"
diff --git a/src/xz/options.c b/src/xz/options.c
index e754c266..efceba9e 100644
--- a/src/xz/options.c
+++ b/src/xz/options.c
@@ -315,7 +315,7 @@ options_lzma(const char *str)
};
static const option_map opts[] = {
- { "preset", NULL, 1, 9 },
+ { "preset", NULL, 0, 9 },
{ "dict", NULL, LZMA_DICT_SIZE_MIN,
(UINT32_C(1) << 30) + (UINT32_C(1) << 29) },
{ "lc", NULL, LZMA_LCLP_MIN, LZMA_LCLP_MAX },
diff --git a/src/xz/process.c b/src/xz/process.c
index 191dfc09..e6870f81 100644
--- a/src/xz/process.c
+++ b/src/xz/process.c
@@ -34,14 +34,22 @@ static lzma_filter filters[LZMA_FILTERS_MAX + 1];
/// Number of filters. Zero indicates that we are using a preset.
static size_t filters_count = 0;
-/// Number of the preset (1-9)
-static size_t preset_number = 7;
+/// Number of the preset (0-9)
+static size_t preset_number = 6;
-/// Indicate if no preset has been given. In that case, we will auto-adjust
-/// the compression preset so that it doesn't use too much RAM.
-// FIXME
+/// True if we should auto-adjust the compression settings to use less memory
+/// if memory usage limit is too low for the original settings.
+static bool auto_adjust = true;
+
+/// Indicate if no preset has been explicitly given. In that case, if we need
+/// to auto-adjust for lower memory usage, we won't print a warning.
static bool preset_default = true;
+/// If a preset is used (no custom filter chain) and preset_extreme is true,
+/// a significantly slower compression is used to achieve slightly better
+/// compression ratio.
+static bool preset_extreme = false;
+
/// Integrity check type
static lzma_check check = LZMA_CHECK_CRC64;
@@ -64,6 +72,14 @@ coder_set_preset(size_t new_preset)
extern void
+coder_set_extreme(void)
+{
+ preset_extreme = true;
+ return;
+}
+
+
+extern void
coder_add_filter(lzma_vli id, void *options)
{
if (filters_count == LZMA_FILTERS_MAX)
@@ -77,6 +93,15 @@ coder_add_filter(lzma_vli id, void *options)
}
+static void lzma_attribute((noreturn))
+memlimit_too_small(uint64_t memory_usage, uint64_t memory_limit)
+{
+ message_fatal(_("Memory usage limit (%" PRIu64 " MiB) is too small "
+ "for the given filter setup (%" PRIu64 " MiB)"),
+ memory_limit >> 20, memory_usage >> 20);
+}
+
+
extern void
coder_set_compression_settings(void)
{
@@ -99,6 +124,9 @@ coder_set_compression_settings(void)
}
// Get the preset for LZMA1 or LZMA2.
+ if (preset_extreme)
+ preset_number |= LZMA_PRESET_EXTREME;
+
if (lzma_lzma_preset(&opt_lzma, preset_number))
message_bug();
@@ -107,6 +135,8 @@ coder_set_compression_settings(void)
? LZMA_FILTER_LZMA1 : LZMA_FILTER_LZMA2;
filters[0].options = &opt_lzma;
filters_count = 1;
+ } else {
+ preset_default = false;
}
// Terminate the filter options array.
@@ -139,30 +169,77 @@ coder_set_compression_settings(void)
message_fatal("Unsupported filter chain or filter options");
// Print memory usage info.
- message(V_DEBUG, _("%" PRIu64 " MiB of memory is required per thread, "
- "limit is %" PRIu64 " MiB"),
- memory_usage / (1024 * 1024),
- memory_limit / (1024 * 1024));
-
- if (preset_default) {
- // When no preset was explicitly requested, we use the default
- // preset only if the memory usage limit allows. Otherwise we
- // select a lower preset automatically.
- while (memory_usage > memory_limit) {
- if (preset_number == 1)
- message_fatal(_("Memory usage limit is too "
- "small for any internal "
- "filter preset"));
-
- if (lzma_lzma_preset(&opt_lzma, --preset_number))
- message_bug();
+ message(V_DEBUG, _("%'" PRIu64 " MiB (%'" PRIu64 " B) of memory is "
+ "required per thread, "
+ "limit is %'" PRIu64 " MiB (%'" PRIu64 " B)"),
+ memory_usage >> 20, memory_usage,
+ memory_limit >> 20, memory_limit);
+
+ if (memory_usage > memory_limit) {
+ // If --no-auto-adjust was used or we didn't find LZMA1 or
+ // LZMA2 as the last filter, give an error immediatelly.
+ // --format=raw implies --no-auto-adjust.
+ if (!auto_adjust || opt_format == FORMAT_RAW)
+ memlimit_too_small(memory_usage, memory_limit);
+
+ assert(opt_mode == MODE_COMPRESS);
+
+ // Look for the last filter if it is LZMA2 or LZMA1, so
+ // we can make it use less RAM. With other filters we don't
+ // know what to do.
+ size_t i = 0;
+ while (filters[i].id != LZMA_FILTER_LZMA2
+ && filters[i].id != LZMA_FILTER_LZMA1) {
+ if (filters[i].id == LZMA_VLI_UNKNOWN)
+ memlimit_too_small(memory_usage, memory_limit);
+
+ ++i;
+ }
+
+ // Decrease the dictionary size until we meet the memory
+ // usage limit. First round down to full mebibytes.
+ lzma_options_lzma *opt = filters[i].options;
+ const uint32_t orig_dict_size = opt->dict_size;
+ opt->dict_size &= ~((UINT32_C(1) << 20) - 1);
+ while (true) {
+ // If it is below 1 MiB, auto-adjusting failed. We
+ // could be more sophisticated and scale it down even
+ // more, but let's see if many complain about this
+ // version.
+ //
+ // FIXME: Displays the scaled memory usage instead
+ // of the original.
+ if (opt->dict_size < (UINT32_C(1) << 20))
+ memlimit_too_small(memory_usage, memory_limit);
memory_usage = lzma_memusage_encoder(filters);
+ if (memory_usage == UINT64_MAX)
+ message_bug();
+
+ // Accept it if it is low enough.
+ if (memory_usage <= memory_limit)
+ break;
+
+ // Otherwise 1 MiB down and try again. I hope this
+ // isn't too slow method for cases where the original
+ // dict_size is very big.
+ opt->dict_size -= UINT32_C(1) << 20;
}
- } else {
- if (memory_usage > memory_limit)
- message_fatal(_("Memory usage limit is too small "
- "for the given filter setup"));
+
+ // Tell the user that we decreased the dictionary size.
+ // However, omit the message if no preset or custom chain
+ // was given. FIXME: Always warn?
+ if (!preset_default)
+ message(V_WARNING, "Adjusted LZMA%c dictionary size "
+ "from %'" PRIu32 " MiB to "
+ "%'" PRIu32 " MiB to not exceed "
+ "the memory usage limit of "
+ "%'" PRIu64 " MiB",
+ filters[i].id == LZMA_FILTER_LZMA2
+ ? '2' : '1',
+ orig_dict_size >> 20,
+ opt->dict_size >> 20,
+ memory_limit >> 20);
}
// Limit the number of worker threads so that memory usage
@@ -224,8 +301,8 @@ coder_init(void)
break;
case FORMAT_RAW:
- // Memory usage has already been checked in args.c.
- // FIXME Comment
+ // Memory usage has already been checked in
+ // coder_set_compression_settings().
ret = lzma_raw_decoder(&strm, filters);
break;
}
diff --git a/src/xz/process.h b/src/xz/process.h
index de23eacb..38485285 100644
--- a/src/xz/process.h
+++ b/src/xz/process.h
@@ -57,6 +57,9 @@ extern void coder_set_check(lzma_check check);
/// Set preset number
extern void coder_set_preset(size_t new_preset);
+/// Enable extreme mode
+extern void coder_set_extreme(void);
+
/// Add a filter to the custom filter chain
extern void coder_add_filter(lzma_vli id, void *options);