ALSA: hda/hdmi: packet buffer index must be set before reading value
PATCH 1/1 ALSA: hda/hdmi: packet buffer index must be set before reading value
The check for infoframe transmit status in hdmi_infoframe_uptodate() makes the assumption that packet buffer index is set to zero.
Align code with specification and explicitly set the index before AC_VERB_GET_HDMI_DIP_XMIT. The packet index setting affects both DIP-Data and DIP-XmitCtrl verbs.
There are no known cases where the old implementation has caused driver to work incorrectly. This change is purely based on code review against the specification (HDA spec rev1.0a).
--- sound/pci/hda/patch_hdmi.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
sound/pci/hda/patch_hdmi.c +1 -1
@@ -638,11 +638,11 @@ static bool hdmi_infoframe_uptodate(struct hda_codec *codec, hda_nid_t pin_nid,u8 val;int i; +hdmi_set_dip_index (codec, pin_nid,0x0 ,0x0 );if (snd_hda_codec_read (codec, pin_nid,0 ,AC_VERB_GET_HDMI_DIP_XMIT ,0 ) !=AC_DIPXMIT_BEST )return false; -hdmi_set_dip_index (codec, pin_nid,0x0 ,0x0 );for (i =0 ; i < size; i++) { val =snd_hda_codec_read (codec, pin_nid,0 ,AC_VERB_GET_HDMI_DIP_DATA ,0 );
Signature · 1 line
2.29.2
Quoted 8 lines · Kai Vehmanen
The check for infoframe transmit status in
hdmi_infoframe_uptodate()makes the assumption that packet buffer index is set to zero.Align code with specification and explicitly set the index before
AC_VERB_GET_HDMI_DIP_XMIT. The packet index setting affects both DIP-Data and DIP-XmitCtrl verbs.
Quoted 5 lines
There are no known cases where the old implementation has caused driver to work incorrectly. This change is purely based on code review against the specification (HDA spec rev1.0a).
Signed-off-by Kai Vehmanen <[email protected]>
Currently there is no other place to change DIP index, i.e. it's always zero, and the bug wouldn't hit, but it's better to be addressed beforehand, indeed.
Applied now. Thanks.
Takashi
Hi,
Quoted 8 lines · Takashi Iwai
Quoted 3 lines · Kai Vehmanen
The check for infoframe transmit status in
hdmi_infoframe_uptodate()makes the assumption that packet buffer index is set to zero.Currently there is no other place to change DIP index, i.e. it's always zero, and the bug wouldn't hit, but it's better to be addressed beforehand, indeed.
indeed, agreed on that. Most likely scenario where this could actually cause harm if someone later adds new functionality to this driver that uses other DIP indices for something. That's pretty unlikely as well, but yeah, better align with the spec.
Br, Kai