16:08 cdo256: I need to know im not going crazy.
16:09 cdo256: To me it looks like an MMIO write fault at one address is being triggered by a read at a completely different address
16:11 cdo256: commenting out the read produces no mmio fault. and replacing the read with a write the address that's shown on the fault message, produces no fault.
16:12 cdo256: can i assume that there's a delay on the interupt and that it's just reporting the last accessed address at the time the interrupt fires?
18:45 imirkin: cdo256: yes, there's a slight delay in reporting the error
18:46 imirkin: cdo256: but the address is the erroring address
18:46 imirkin: (if you trigger 100 errors by the time the interrupt runs, it's probably the last error)
18:49 cdo256: if removing the read removes the error then it has to be the only error right?
18:49 imirkin: what are you reading?
18:49 cdo256: address 0x022554
18:49 imirkin: and the error you get?
18:51 cdo256: MMIO write of 80000061 FAULT at 10eb14 [ PRIVRING ]
18:51 imirkin: what's 22554 again? it's some capabilities reg, no?
18:52 cdo256: <reg32 offset="0x154" name="HW_MC_DISABLE_MASK" variants="GF100:GM107"/>
18:52 imirkin: right yea
18:52 imirkin: and you're on a?
18:52 cdo256: GM204
18:53 imirkin: so ... that envytools snippset says that reg is there up to but not including GM107 ...
18:53 imirkin: so why are you reading that reg?
18:54 cdo256: because the driver is using ramgk104.c
18:54 imirkin: ok well
18:54 imirkin: it shouldn't be reading that.
18:55 imirkin: i have no real clue why you're getting a seemingly unrelated address in the error though
18:56 imirkin: also the source is "PRIVRING" which is ... odd
18:56 imirkin: (mostly coz i dunno wtf that means)
18:56 imirkin: and i've been around the proverbial block, so ... it's definitely an odd thing
18:56 cdo256: I was just about to ask you what that meant haha
18:58 cdo256: I guess the next step is to flesh out ramgm200.c
18:59 cdo256: I'm going to read some more code, familiarise myself with the mmiotrace and come back to it
18:59 imirkin: you have your own fan connected right?
19:00 cdo256: you mean on my case? I thought I did but it's not spinning. Do you think something overheated?
19:01 imirkin: i mean on your GPU
19:01 imirkin: the whole thing with GM20x reclocking is we can't control the fan without high-secure firmware
19:02 imirkin: so you have to have some fan connected that's not controlled internally by the GPU
19:02 cdo256: ah I see
19:03 imirkin: (otherwise sadness will ensue)
19:04 cdo256: thanks ilia I'll look into that
19:06 imirkin: if it's a laptop it's probably fine
19:06 imirkin: since the fan is generally controlled by the laptop's EC rather than the GPU itself
23:28 mangix: imirkin: that patchset never made it upstream.