cache: move duplicated condition to a function

While subtly duplicating such checks happens a lot in this file, this
particular case can be easily factored into a function.
This commit is contained in:
wm4 2017-05-15 15:48:31 +02:00
parent 3f17b101ee
commit 5f2c4d4799
1 changed files with 12 additions and 6 deletions

View File

@ -211,15 +211,21 @@ static size_t read_buffer(struct priv *s, unsigned char *dst,
return read;
}
// Whether a seek will be needed to get to the position. This honors seek_limit,
// which is a heuristic to prevent dropping the cache with small forward seeks.
// This helps in situations where waiting for network a bit longer would quickly
// reach the target position. Especially if the demuxer seeks back and forth,
// not dropping the backwards cache will be a major performance win.
static bool needs_seek(struct priv *s, int64_t pos)
{
return pos < s->min_filepos || pos > s->max_filepos + s->seek_limit;
}
static bool cache_update_stream_position(struct priv *s)
{
int64_t read = s->read_filepos;
// drop cache contents only if seeking backward or too much fwd.
// This is also done for on-disk files, since it loses the backseek cache.
// That in turn can cause major bandwidth increase and performance
// issues with e.g. mov or badly interleaved files
if (read < s->min_filepos || read > s->max_filepos + s->seek_limit) {
if (needs_seek(s, read)) {
MP_VERBOSE(s, "Dropping cache at pos %"PRId64", "
"cached range: %"PRId64"-%"PRId64".\n", read,
s->min_filepos, s->max_filepos);
@ -607,7 +613,7 @@ static int cache_seek(stream_t *cache, int64_t pos)
// This check is not quite exact - if the reader thread is blocked in
// a read, the read might advance file position enough that a seek
// forward is no longer needed.
if (pos < s->min_filepos || pos > s->max_filepos + s->seek_limit) {
if (needs_seek(s, pos)) {
s->eof = false;
s->control = CACHE_CTRL_SEEK;
s->control_res = 0;