diff --git a/.travis.yml b/.travis.yml
index bece49300..0967f60e0 100644
--- a/.travis.yml
+++ b/.travis.yml
@@ -31,9 +31,7 @@ matrix:
     - env: Cmd='make ppcinstall && make ppc64fuzz'
     - env: Cmd='make -j uasanregressiontest && make clean && make -j msanregressiontest'
 
-    - env: Cmd='make lz4install && make -C tests test-lz4'
-
-    - env: Cmd='bash tests/libzstd_partial_builds.sh'
+    - env: Cmd='make lz4install && make -C tests test-lz4 test-pool && make clean && bash tests/libzstd_partial_builds.sh'
 
     # tag-specific test
     - if: tag =~ ^v[0-9]\.[0-9]
diff --git a/programs/zstd.1 b/programs/zstd.1
index 507933c97..b136bb33e 100644
--- a/programs/zstd.1
+++ b/programs/zstd.1
@@ -1,5 +1,5 @@
 .
-.TH "ZSTD" "1" "2018-05-12" "zstd 1.3.4" "User Commands"
+.TH "ZSTD" "1" "2018-06-27" "zstd 1.3.5" "User Commands"
 .
 .SH "NAME"
 \fBzstd\fR \- zstd, zstdmt, unzstd, zstdcat \- Compress or decompress \.zst files
@@ -160,7 +160,7 @@ operate recursively on dictionaries
 .
 .TP
 \fB\-\-format=FORMAT\fR
-compress and decompress in other formats\. If compiled with support, zstd can compress to or decompress from other compression algorithm formats\. Possibly available options are \fBgzip\fR, \fBxz\fR, \fBlzma\fR, and \fBlz4\fR\.
+compress and decompress in other formats\. If compiled with support, zstd can compress to or decompress from other compression algorithm formats\. Possibly available options are \fBzstd\fR, \fBgzip\fR, \fBxz\fR, \fBlzma\fR, and \fBlz4\fR\. If no such format is provided, \fBzstd\fR is the default\.
 .
 .TP
 \fB\-h\fR/\fB\-H\fR, \fB\-\-help\fR
@@ -349,13 +349,13 @@ The impact of this field vary depending on selected strategy\.
 For ZSTD_btopt and ZSTD_btultra, it specifies the minimum match length that causes match finder to stop searching for better matches\. A larger \fBtargetLen\fR usually improves compression ratio but decreases compression speed\.
 .
 .IP
-For ZSTD_fast, it specifies the amount of data skipped between match sampling\. Impact is reversed : a larger \fBtargetLen\fR increases compression speed but decreases compression ratio\.
+For ZSTD_fast, it triggers ultra\-fast mode when > 0\. The value represents the amount of data skipped between match sampling\. Impact is reversed : a larger \fBtargetLen\fR increases compression speed but decreases compression ratio\.
 .
 .IP
 For all other strategies, this field has no impact\.
 .
 .IP
-The minimum \fItlen\fR is 1 and the maximum is 999\.
+The minimum \fItlen\fR is 0 and the maximum is 999\.
 .
 .TP
 \fBoverlapLog\fR=\fIovlog\fR, \fBovlog\fR=\fIovlog\fR
diff --git a/programs/zstd.1.md b/programs/zstd.1.md
index 4b3818141..c8b8d8d72 100644
--- a/programs/zstd.1.md
+++ b/programs/zstd.1.md
@@ -355,14 +355,14 @@ The list of available _options_:
     A larger `targetLen` usually improves compression ratio
     but decreases compression speed.
 
-    For ZSTD\_fast, it specifies
-    the amount of data skipped between match sampling.
+    For ZSTD\_fast, it triggers ultra-fast mode when > 0.
+    The value represents the amount of data skipped between match sampling.
     Impact is reversed : a larger `targetLen` increases compression speed
     but decreases compression ratio.
 
     For all other strategies, this field has no impact.
 
-    The minimum _tlen_ is 1 and the maximum is 999.
+    The minimum _tlen_ is 0 and the maximum is 999.
 
 - `overlapLog`=_ovlog_,  `ovlog`=_ovlog_:
     Determine `overlapSize`, amount of data reloaded from previous job.
diff --git a/programs/zstdcli.c b/programs/zstdcli.c
index 1af5f26fa..be6204667 100644
--- a/programs/zstdcli.c
+++ b/programs/zstdcli.c
@@ -561,13 +561,15 @@ int main(int argCount, const char* argv[])
                         continue;
                     }
                     if (longCommandWArg(&argument, "--fast")) {
-                        /* Parse optional window log */
+                        /* Parse optional acceleration factor */
                         if (*argument == '=') {
                             U32 fastLevel;
                             ++argument;
                             fastLevel = readU32FromChar(&argument);
                             if (fastLevel) {
                               dictCLevel = cLevel = -(int)fastLevel;
+                            } else {
+                              CLEAN_RETURN(badusage(programName));
                             }
                         } else if (*argument != 0) {
                             /* Invalid character following --fast */
diff --git a/tests/playTests.sh b/tests/playTests.sh
index 52d4dd0b9..09a7377f2 100755
--- a/tests/playTests.sh
+++ b/tests/playTests.sh
@@ -108,6 +108,7 @@ $ECHO "test : --fast aka negative compression levels"
 $ZSTD --fast -f tmp  # == -1
 $ZSTD --fast=3 -f tmp  # == -3
 $ZSTD --fast=200000 -f tmp  # == no compression
+! $ZSTD -c --fast=0 tmp > $INTOVOID # should fail
 $ECHO "test : too large numeric argument"
 $ZSTD --fast=9999999999 -f tmp  && die "should have refused numeric value"
 $ECHO "test : compress to stdout"
@@ -532,7 +533,7 @@ fi
 if [ $GZIPMODE -eq 1 ]; then
     ./datagen > tmp
     rm -f tmp.zst
-    $ZSTD --format=gzip --format=zstd -f tmp 
+    $ZSTD --format=gzip --format=zstd -f tmp
     test -f tmp.zst
 fi
 
@@ -637,11 +638,11 @@ $ECHO "\n===> suffix list test"
 
 ! $ZSTD -d tmp.abc 2> tmplg
 
-if [ $GZIPMODE -ne 1 ]; then 
+if [ $GZIPMODE -ne 1 ]; then
     grep ".gz" tmplg > $INTOVOID && die "Unsupported suffix listed"
 fi
 
-if [ $LZMAMODE -ne 1 ]; then 
+if [ $LZMAMODE -ne 1 ]; then
     grep ".lzma" tmplg > $INTOVOID && die "Unsupported suffix listed"
     grep ".xz" tmplg > $INTOVOID && die "Unsupported suffix listed"
 fi
diff --git a/tests/poolTests.c b/tests/poolTests.c
index 6a058a5a3..9661b5299 100644
--- a/tests/poolTests.c
+++ b/tests/poolTests.c
@@ -105,7 +105,8 @@ void waitLongFn(void *opaque) {
 static int testThreadReduction_internal(POOL_ctx* ctx, poolTest_t test)
 {
     int const nbWaits = 16;
-    UTIL_time_t startTime, time4threads, time2threads;
+    UTIL_time_t startTime;
+    U64 time4threads, time2threads;
 
     test.val = 0;
     test.max = nbWaits;