13:45 karolherbst: heh
13:45 karolherbst: prog->hdr[2] |= (((info->maxCFDepth * 2) + 47) / 48) * 0x200;
13:46 karolherbst: that one is called ShaderLocalMemoryHighSize
13:46 karolherbst: _OHHHHH_
13:46 karolherbst: imirkin:
13:46 karolherbst: ^^ we have this code in nvc0
13:46 karolherbst: but it's disabled
13:47 karolherbst: _but_
13:47 karolherbst: I think it's wrong
13:47 karolherbst: it has to be [3]
13:47 karolherbst: then it's the off-chip crs thing
13:47 karolherbst: there is also this "prog->need_tls = true;" thing
13:48 karolherbst: 6d1cdec3ba15
13:48 karolherbst: mhh, unrelated commit
13:50 karolherbst: heh
13:50 karolherbst: it was added back then still
13:50 karolherbst: RSpliet: ^^
13:55 ItsMarlin: hi
13:56 ItsMarlin: quick question, what kind of stuff should i learn to get into reverse engineering firmware blobs for graphics cards?
14:00 karolherbst: ItsMarlin: learn the ISA
14:00 ItsMarlin: i was thinking about the new radeon cards
14:01 karolherbst: but it's kind of pointless as on newer GPUs you also need nvidias key to be able to run them
14:01 ItsMarlin: they only lack free firmware
14:01 karolherbst: ohh, wrong channel then, we do nvidia stuff here :p
14:01 ItsMarlin: Yeah, i just didn't know where else to ask about the topic :P
14:02 ItsMarlin: don't think there is a movement for reverse engineering radeon firmware
14:02 karolherbst: #radeon maybe
14:02 karolherbst: uhhh huh?
14:03 ItsMarlin: i'll check if this topic is accepted there
14:03 ItsMarlin: thanks
16:19 karolherbst: imirkin: would you mind if I remove the support for the compute symbol table from codegen for now? That way I wouldn't feel bad for ignoring it when caching our shaders and we don't use it for anything anyway
16:20 imirkin: compute symbol table?
16:20 karolherbst: nv50_prog_info.syms
16:20 karolherbst: uhm
16:20 karolherbst: nv50_ir_prog_info.syms
16:20 imirkin: eh, wtvr
16:20 karolherbst: yeah, that's what I thought, but I just wanted to check back before removing it :)