diff options
| author | Robin Watts <robin.watts@artifex.com> | 2012-02-29 18:39:16 +0000 |
|---|---|---|
| committer | Robin Watts <robin.watts@artifex.com> | 2012-02-29 18:39:16 +0000 |
| commit | 46ddb6ea2c7fa8c56006d39dc41836fa6c59f40d (patch) | |
| tree | c776f573fdef6bbd2c9eec6deecfe7511267007b | |
| parent | bbfe635555dce16858403706e2031dd3bfa1a9f1 (diff) | |
| download | mupdf-46ddb6ea2c7fa8c56006d39dc41836fa6c59f40d.tar.xz | |
Fix typo in hash table code.
When detecting a clash when inserting into a hash table (which
should only ever happen in multithreaded cases), we should give
a warning, and return the existing item from the table.
The current code doesn't do this due to a stupid typo. Fixed here.
CLUSTER_UNTESTED as we don't test multithreaded operation in the
cluster.
| -rw-r--r-- | fitz/base_hash.c | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/fitz/base_hash.c b/fitz/base_hash.c index 0fda86cd..76c95304 100644 --- a/fitz/base_hash.c +++ b/fitz/base_hash.c @@ -127,7 +127,7 @@ do_hash_insert(fz_context *ctx, fz_hash_table *table, void *key, void *val) if (memcmp(key, ents[pos].key, table->keylen) == 0) { fz_warn(ctx, "assert: overwrite hash slot"); - return ents[pos].val = val; + return ents[pos].val; } pos = (pos + 1) % size; |