[PATCH v7 8/9] media: Avoid parsing quantization and huffman tables
From: Mirela Rabulea <hidden>
Date: 2021-01-11 19:32:16
Also in:
linux-media, lkml
Subsystem:
media input infrastructure (v4l/dvb), the rest · Maintainers:
Mauro Carvalho Chehab, Linus Torvalds
From: Mirela Rabulea <mirela.rabulea@nxp.com> These are optional in struct v4l2_jpeg_header, so skip DHT/DQT segment parsing if huffman_tables/quantization_tables were not requested by user, to save time. However, do count them (num_dht/num_dqt). Signed-off-by: Mirela Rabulea <mirela.rabulea@nxp.com> Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de> --- Changes in v7: Added Reviewed-by: Philipp Zabel [off-list ref] drivers/media/v4l2-core/v4l2-jpeg.c | 8 ++++++++ 1 file changed, 8 insertions(+)
diff --git a/drivers/media/v4l2-core/v4l2-jpeg.c b/drivers/media/v4l2-core/v4l2-jpeg.c
index 42aaabc71b7a..4b008d630252 100644
--- a/drivers/media/v4l2-core/v4l2-jpeg.c
+++ b/drivers/media/v4l2-core/v4l2-jpeg.c@@ -540,6 +540,10 @@ int v4l2_jpeg_parse_header(void *buf, size_t len, struct v4l2_jpeg_header *out) &out->dht[out->num_dht++ % 4]); if (ret < 0) return ret; + if (!out->huffman_tables) { + ret = jpeg_skip_segment(&stream); + break; + } ret = jpeg_parse_huffman_tables(&stream, out->huffman_tables); break;
@@ -548,6 +552,10 @@ int v4l2_jpeg_parse_header(void *buf, size_t len, struct v4l2_jpeg_header *out) &out->dqt[out->num_dqt++ % 4]); if (ret < 0) return ret; + if (!out->quantization_tables) { + ret = jpeg_skip_segment(&stream); + break; + } ret = jpeg_parse_quantization_tables(&stream, out->frame.precision, out->quantization_tables);
--
2.17.1