facebook · terrelln · Jan 5, 2021 · Jan 4, 2021 · Jan 4, 2021 · Jan 4, 2021
diff --git a/lib/compress/zstd_compress.c b/lib/compress/zstd_compress.c
@@ -1188,15 +1188,26 @@ ZSTD_adjustCParams_internal(ZSTD_compressionParameters cPar,
     const U64 maxWindowResize = 1ULL << (ZSTD_WINDOWLOG_MAX-1);
     assert(ZSTD_checkCParams(cPar)==0);
 
-    if (dictSize && srcSize == ZSTD_CONTENTSIZE_UNKNOWN)
-        srcSize = minSrcSize;
-
     switch (mode) {
-    case ZSTD_cpm_noAttachDict:
     case ZSTD_cpm_unknown:
+    case ZSTD_cpm_noAttachDict:
+        /* If we don't know the source size, don't make any
+         * assumptions about it. We will already have selected
+         * smaller parameters if a dictionary is in use.
+         */
+        break;
     case ZSTD_cpm_createCDict:
+        /* Assume a small source size when creating a dictionary
+         * with an unkown source size.
+         */
+        if (dictSize && srcSize == ZSTD_CONTENTSIZE_UNKNOWN)
+            srcSize = minSrcSize;
         break;
     case ZSTD_cpm_attachDict:
+        /* Dictionary has its own dedicated parameters which have
+         * already been selected. We are selecting parameters
+         * for only the source.
+         */
         dictSize = 0;
         break;
     default:
@@ -1213,7 +1224,8 @@ ZSTD_adjustCParams_internal(ZSTD_compressionParameters cPar,
                             ZSTD_highbit32(tSize-1) + 1;
         if (cPar.windowLog > srcLog) cPar.windowLog = srcLog;
     }
-    {   U32 const dictAndWindowLog = ZSTD_dictAndWindowLog(cPar.windowLog, (U64)srcSize, (U64)dictSize);
+    if (srcSize != ZSTD_CONTENTSIZE_UNKNOWN) {
+        U32 const dictAndWindowLog = ZSTD_dictAndWindowLog(cPar.windowLog, (U64)srcSize, (U64)dictSize);
         U32 const cycleLog = ZSTD_cycleLog(cPar.chainLog, cPar.strategy);
         if (cPar.hashLog > dictAndWindowLog+1) cPar.hashLog = dictAndWindowLog+1;
         if (cycleLog > dictAndWindowLog)
@@ -2955,9 +2967,9 @@ static size_t ZSTD_writeFrameHeader(void* dst, size_t dstCapacity,
 }
 
 /* ZSTD_writeSkippableFrame_advanced() :
- * Writes out a skippable frame with the specified magic number variant (16 are supported), 
+ * Writes out a skippable frame with the specified magic number variant (16 are supported),
  * from ZSTD_MAGIC_SKIPPABLE_START to ZSTD_MAGIC_SKIPPABLE_START+15, and the desired source data.
- * 
+ *
  * Returns the total number of bytes written, or a ZSTD error code.
  */
 size_t ZSTD_writeSkippableFrame(void* dst, size_t dstCapacity,

diff --git a/tests/fuzzer.c b/tests/fuzzer.c
@@ -3051,6 +3051,32 @@ static int basicUnitTests(U32 const seed, double compressibility)
         free(dict);
     }
     DISPLAYLEVEL(3, "OK \n");
+
+    DISPLAYLEVEL(3, "test%3i : ZSTD_getCParams() + dictionary ", testNb++);
+    {
+        ZSTD_compressionParameters const medium = ZSTD_getCParams(1, 16*1024-1, 0);
+        ZSTD_compressionParameters const large = ZSTD_getCParams(1, 128*1024-1, 0);
+        ZSTD_compressionParameters const smallDict = ZSTD_getCParams(1, 0, 400);
+        ZSTD_compressionParameters const mediumDict = ZSTD_getCParams(1, 0, 10000);
+        ZSTD_compressionParameters const largeDict = ZSTD_getCParams(1, 0, 100000);
+
+        assert(!memcmp(&smallDict, &mediumDict, sizeof(smallDict)));
+        assert(!memcmp(&medium, &mediumDict, sizeof(medium)));
+        assert(!memcmp(&large, &largeDict, sizeof(large)));
+    }
+    DISPLAYLEVEL(3, "OK \n");
+
+    DISPLAYLEVEL(3, "test%3i : ZSTD_adjustCParams() + dictionary ", testNb++);
+    {
+        ZSTD_compressionParameters const cParams = ZSTD_getCParams(1, 0, 0);
+        ZSTD_compressionParameters const smallDict = ZSTD_adjustCParams(cParams, 0, 400);
+        ZSTD_compressionParameters const smallSrcAndDict = ZSTD_adjustCParams(cParams, 500, 400);
+
+        assert(smallSrcAndDict.windowLog == 10);
+        assert(!memcmp(&cParams, &smallDict, sizeof(cParams)));
+    }
+    DISPLAYLEVEL(3, "OK \n");
+
 #endif
 
 _end:

diff --git a/tests/regression/config.c b/tests/regression/config.c
@@ -59,6 +59,14 @@ static config_t no_pledged_src_size = {
     .no_pledged_src_size = 1,
 };
 
+static config_t no_pledged_src_size_with_dict = {
+    .name = "no source size with dict",
+    .cli_args = "",
+    .param_values = PARAM_VALUES(level_0_param_values),
+    .no_pledged_src_size = 1,
+    .use_dictionary = 1,
+};
+
 static param_value_t const ldm_param_values[] = {
     {.param = ZSTD_c_enableLongDistanceMatching, .value = 1},
 };
@@ -192,6 +200,7 @@ static config_t const* g_configs[] = {
 #undef FAST_LEVEL
 
     &no_pledged_src_size,
+    &no_pledged_src_size_with_dict,
     &ldm,
     &mt,
     &mt_ldm,

diff --git a/tests/regression/data.c b/tests/regression/data.c
@@ -67,10 +67,27 @@ data_t github = {
         },
 };
 
+data_t github_tar = {
+    .name = "github.tar",
+    .type = data_type_file,
+    .data =
+        {
+            .url = REGRESSION_RELEASE("github.tar.zst"),
+            .xxhash64 = 0xa9b1b44b020df292LL,
+        },
+    .dict =
+        {
+            .url = REGRESSION_RELEASE("github.dict.zst"),
+            .xxhash64 = 0x1eddc6f737d3cb53LL,
+
+        },
+};
+
 static data_t* g_data[] = {
     &silesia,
     &silesia_tar,
     &github,
+    &github_tar,
     NULL,
 };