15:02 swiftgeek: any idea why monitor is exposing only 0x30 and not 0x50?
15:03 swiftgeek: I'm not sure how to make newer monitors expose edid
15:06 RSpliet: swiftgeek: no idea what 0x30 and 0x50 means, nor where you conclude that the monitor isn't exposing it
15:07 swiftgeek: addresses over i2c bus
15:07 RSpliet: randr filters out modes that aren't supported by the GPU, for example because the pixelclock doesn't go up high enough to drive the monitor at said mode
15:07 ajax: "newer"?
15:07 swiftgeek: from what i'm understanding anything in past 10-15 years
15:08 swiftgeek: i always assumed that 24c02 would always appear right away and it looks like i'm wrong
15:08 RSpliet: Oh, this is about how EDIDs can be split up into further and further extensions I guess? I'm going to shut up then, I know too little about all that :-D
15:09 ajax: i'm going to need some more context here. monitors still support EDID.
15:09 RSpliet: I think the monitor can detect the cable type/version and hide incompatible modes/parts of the EDID too
15:09 ajax: you mean "monitor" as in the gpu engine
15:09 ajax: ?
15:09 swiftgeek: i mean as in external panel with scaler that has DVI/HDMI inputs
15:10 ajax: and it's not like there's a way to "make" it expose edid if it doesn't.
15:10 ajax: it's basically just a rom you can read at an i2c address
15:11 ajax: so maybe the connector to that little rom in the monitor is dying, or the ddc pins in the monitor cable, or their connection on the gpu side
15:11 swiftgeek: yep and now i have ran into a wall of not seeing it immediately and monitor actively gating it
15:11 swiftgeek: and i have no idea what the hell it expects
15:11 ajax: yeah that part doesn't make any sense to me, why would it gate anything, how would it even know
15:12 swiftgeek: well it doesn't make sense to me either but that's how it appears to be
15:15 swiftgeek: I would guess it's some combination of HPD/GND and I2C lines that are used for this handshake, and the only consistent portion i observed so far is 0x30 appearing in i2cdetect
15:18 swiftgeek: I thought i had broken something until I made this happen https://www.youtube.com/watch?v=SNDrinyFtSQ
15:20 hell__: I've seen monitors which don't have an actual I2C EEPROM, instead the scaler chip emulates the EEPROM
15:21 swiftgeek: yeah that would be an evolution of that, which is even harder to make sense of
15:22 hell__: and I once soldered half a dozen EEPROMs to fix three monitors (2 EEPROMs per monitor) which had troubles providing an EDID
15:22 karolherbst: swiftgeek: tried with nvidias driver?
15:22 karolherbst: if the behavior is different with it, we might be able to tell what they are doing differently
15:24 swiftgeek: will need to check that later then
15:45 swiftgeek: ok figured it out, at the very least DVI needs shield to be GND when EDID isn't always exposed
15:45 swiftgeek: otherwise hotplug detection of DVI fails
15:45 swiftgeek: which in turn makes EDID disappear
15:48 swiftgeek: why it isn't using HPD for HPD is beyond me
15:49 karolherbst: swiftgeek: some displays are just weird
15:49 swiftgeek: consistent across many vendors
15:49 swiftgeek: though all mstar based
15:50 karolherbst: I know about displays which expose the same EDID over multiple connectors, but changing some bits without updating the checksum :)
15:50 karolherbst: somebody trying to be smart and failing at it
15:51 swiftgeek: but why use GND pin for HPD xD
15:51 swiftgeek: OH
15:51 swiftgeek: so you can have that fancy "cable disconnected" message
15:51 swiftgeek: without PC powered
15:51 swiftgeek: ....
15:51 karolherbst: :D
15:52 karolherbst: I am sure there are more expensive but spec compliant ways of figuring that out
15:53 swiftgeek: yeah just OR detection with HPD
15:53 swiftgeek: you know , the signal it's meant to handle it xD
15:53 karolherbst: anyway, some displays are just weird and we need workarounds for those
15:54 karolherbst: I was asking nvidia once for sharing their list of workarounds, but maybe now with their driver open sourced, we could scrap that
15:54 swiftgeek: right now i'm using an aligator clips to connect shell to GND
15:54 karolherbst: I am sure nvidia does have something for this case
15:54 swiftgeek: gonna grab dremel later and solder a jumper
15:55 swiftgeek: more like nvidia cards happen to have shell at GND
15:56 karolherbst: ahh