[Intel-gfx] [PATCH] detect hdmi monitor by reading edid
ling.ma at intel.com
Wed Jul 1 18:42:13 PDT 2009
>From: Eric Anholt [mailto:eric at anholt.net]
>Sent: 2009年7月2日 9:25
>To: Ma, Ling
>Cc: Keith Packard; intel-gfx at lists.freedesktop.org
>Subject: Re: [Intel-gfx] [PATCH] detect hdmi monitor by reading edid
>On Tue, 2009-06-30 at 13:44 +0800, Ma, Ling wrote:
>> >-----Original Message-----
>> >From: intel-gfx-bounces at lists.freedesktop.org
>> >[mailto:intel-gfx-bounces at lists.freedesktop.org] On Behalf Of Keith Packard
>> >Sent: 2009年6月29日 23:07
>> >To: Ma, Ling
>> >Cc: intel-gfx at lists.freedesktop.org
>> >Subject: Re: [Intel-gfx] [PATCH] detect hdmi monitor by reading edid
>> >On Mon, 2009-06-29 at 14:41 +0800, ling.ma at intel.com wrote:
>> >> We detect HDMI output status by setting interrupt enable bit,
>> >> then read interrupt status, which is effective for most cases.
>> >> However G4x platform will indicate incorrect detection result
>> >> even if current output is HDMI monitor. The patch intends
>> >> to read edid, then determin connect status by EDID type.
>> >Can you describe the precise errors in detection you're seeing here?
>> >From my experience, as the hotplug bits are shared among multiple
>> >outputs, they will never provide a false-positive reading.
>> We run into two related bugs #21322 and #21084
>> For #21322, The G43 can not find dvi monitor by hot plug approach.
>> For #21084 There are 2 root causes, one is in KMS Q45( DQ45EK in our
>> lab) connected by VGA through DVI-I can't fetch EDID info, another is
>> the it will mistake VGA as HDMI monitor.
>> Our original approach is to set Hot Plug Interrupt Detect Enable, then
>> read corresponding live state bit. However According to the HW team,
>> live status detection need special board design, and won't work on all
>> Meanwhile I ever have seen bios code for G4X always read first 4 bytes
>> of EDID to detect HDMI monitor and not use hot plug approach.
>So, if we're not doing hotplug interrupt based detection any more,
>doesn't that mean that the PEG_BAND_GAP_DATA workaround doesn't need to
>stay any more?
Yes, but in order to avoid regression, whether should we keep hot plug, hdmi edid detection become one additional approach?
>eric at anholt.net eric.anholt at intel.com
More information about the Intel-gfx