From: Yann Collet <cyan@fb.com>
Date: Mon, 23 Jan 2017 07:49:52 +0000 (-0800)
Subject: ZSTDMT_compressStream() becomes blocking when required to ensure forward progresses
X-Git-Tag: v1.1.3^2~19^2~13
X-Git-Url: http://git.ipfire.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=1a2547f6540c8f208fe116072cd935960acfe62a;p=thirdparty%2Fzstd.git

ZSTDMT_compressStream() becomes blocking when required to ensure forward progresses

In some (rare) cases, job list could be blocked by a first job still being processed,
while all following ones are completed, waiting to be flushed.
In such case, the current job-table implementation is unable to accept new job.
As a consequence, a call to ZSTDMT_compressStream() can be useless (nothing read, nothing flushed),
with the risk to trigger a busy-wait on the caller side
(needlessly loop over ZSTDMT_compressStream() ).

In such a case, ZSTDMT_compressStream() will block until the first job is completed and ready to flush.
It ensures some forward progress by guaranteeing it will flush at least a part of the completed job.
Energy-wasting busy-wait is avoided.
---

diff --git a/lib/compress/zstdmt_compress.c b/lib/compress/zstdmt_compress.c
index 283f76841..176f940c8 100644
--- a/lib/compress/zstdmt_compress.c
+++ b/lib/compress/zstdmt_compress.c
@@ -565,6 +565,10 @@ size_t ZSTDMT_compressStream(ZSTDMT_CCtx* zcs, ZSTD_outBuffer* output, ZSTD_inBu
     {   unsigned const jobID = zcs->doneJobID & zcs->jobIDMask;
         unsigned jobCompleted;
         pthread_mutex_lock(&zcs->jobCompleted_mutex);
+        while (zcs->jobs[jobID].jobCompleted == 0 && zcs->inBuff.filled == zcs->inBuffSize) {
+            /* when no new job could be started, block until there is something to flush, ensuring forward progress */
+            pthread_cond_wait(&zcs->jobCompleted_cond, &zcs->jobCompleted_mutex);
+        }
         jobCompleted = zcs->jobs[jobID].jobCompleted;
         pthread_mutex_unlock(&zcs->jobCompleted_mutex);
         if (jobCompleted) {