Wednesday, 2016-11-09

*** tpb has joined #timvideos00:00
mithroshenki: \o/00:04
CarlFKmithro: this cable works with the opsis:
*** CarlFK has quit IRC00:45
*** CarlFK has joined #timvideos01:30
*** ChanServ sets mode: +v CarlFK01:30
*** wanig has joined #timvideos01:47
*** wanig has quit IRC01:59
*** wanig has joined #timvideos02:23
*** wanig has quit IRC02:27
*** wanig has joined #timvideos02:27
*** sb0 has quit IRC03:53
*** rohitksingh_work has joined #timvideos04:02
*** Bertl_oO is now known as Bertl_zZ04:53
*** sb0 has joined #timvideos06:02
*** fumblehool has joined #timvideos06:05
*** CarlFK has quit IRC07:01
*** CarlFK has joined #timvideos07:23
*** ChanServ sets mode: +v CarlFK07:23
*** cr1901_modern has quit IRC07:37
*** fumblehool has quit IRC08:09
*** sb0 has quit IRC08:18
shenkimithro: progress:08:54
shenki#0  calibrate_delay_converge () at init/calibrate.c:19608:54
shenki#1  calibrate_delay () at init/calibrate.c:30308:54
shenki#2  0x40000b2c in start_kernel () at init/main.c:61808:54
mithroshenki: I'm not sure what that means, but yay!?08:54
shenkimithro: progress over last night :)08:55
shenkistuck calibrating the delay loop08:55
shenkiit also means the uart is working08:56
mithroshenki: is this the upstream qemu uart or key2's qemu uart?08:56
shenkimithro: upstream08:57
mithroshenki: Okay, I think that isn't compatible with the misoc uart?08:57
shenkino idea. it's the milkymist uart08:57
shenkii updated it to modern linux, but otherwise it's unchanged08:57
shenkimithro: where was key2's qemu?08:57
shenkimithro: my plan was to get the milkymist port of the kernel working, and then move it over to our hardware08:58
shenkias that lets us do the core arch support08:58
tpbTitle: key2 / qemu Bitbucket (at
mithroshenki: Yeah, that seems pretty reasonable08:58
shenkithat said, now that we're up to the delay loop, perhaps i should move to our hardware08:58
mithroHow does one calibrate a delay loop?08:58
shenkiit uses a clocksource driver to time how fast things are moving08:58
shenkiso my time might be better spent getting it working with our timer peripheral, instead of the milkymist one08:59
mithroDo we have a timer peripheral? :-P08:59
shenkithat was my next question :)08:59
shenkiwe need one08:59
shenkisomething that can produce a periodic irq09:00
mithroLooks like09:00
mithro        "timer0":         4,  # provided by default (optional)09:00
tpbTitle: litex/ at master · enjoy-digital/litex · GitHub (at
tpbTitle: litex/ at master · enjoy-digital/litex · GitHub (at
shenkiand key2's qemu implements them?09:03
mithroLooking now09:04
mithroLooks like that might be the case09:04
tpbTitle: key2 / qemu / commit / 012c29ad5966 Bitbucket (at
shenkihis code is one big patch added to the tip of qemu09:04
shenkiwe should find out if he's interested in upstreaming, and if not, do that ourselves09:04
mithroshenki: BTW, our lm32 implementation has an experimental MMU09:09
tpbTitle: GitHub - m-labs/lm32: LatticeMico32 soft processor (at
tpbTitle: lm32/mmu.rst at master · m-labs/lm32 · GitHub (at
shenkimithro: can you remind me why we wanted to have linux running?09:11
mithroshenki: Mainly so we have access to DRI, possible network stack and a user space09:12
mithroshenki: IE just use readline for tab completion09:12
shenkiok. i will keep hacking at this until it works09:12
shenkibut in order to upstream it i think i will have to throw most of it out09:13
shenkiand start over09:13
shenkiat least we have the current code as a reference09:13
mithroI'm not sure if it makes sense to upstream lm32 arch - it does make sense to upstream the drivers if they can be used with or1k/riscv/jcore archs too?09:14
mithroshenki: How much "code" is actually an arch?09:15
shenkidunno. never done it before09:22
shenki 76 files changed, 4968 insertions(+)09:22
shenkithat's the current tree i have09:23
shenkimithro: have we decided out of those three other options what we're going to go with?09:23
mithroshenki: If we can choose not to decide for as long as possible, I recommend we do that09:24
mithroThey said like 90% of Linux is the drivers, right? :-P09:28
shenkiyeah, it is09:31
mithroSo we /really/ care about things like the Ethernet and video input/output stuff09:32
mithroWhich is all driver side, right?09:32
mithroSo, if we can share these drivers between multiple arch, then whichever one actually wins we can end up using09:35
mithroshenki: Also, if we are in a situation like now, were we have a known working lm32 arch but unknown others we can work on that first, get the parts working and then have less effort on the other archs?09:45
shenkifor sure09:56
shenkimithro: trying to get it booting on the litex model atm. the address map is quite different09:56
mithroAddress map? You mean the memory layout?09:57
shenkimy kernel is trying to execute from 0x00008000 and is crashing qemu10:02
mithroshenki: on key2's qemu?10:25
tpbTitle: key2 / qemu / source / include / generated / mem.h Bitbucket (at
tpbTitle: key2 / qemu / source / include / generated / regions.ld Bitbucket (at
mithroshenki: Which branch of linux-lm32 are you using?10:30
mithroshenki: Looking at
tpbTitle: linux-lm32/Kconfig.cpu at ng · shenki/linux-lm32 · GitHub (at
*** Bertl_zZ is now known as Bertl11:04
*** sb0 has joined #timvideos12:55
*** rohitksingh_work has quit IRC13:04
*** cr1901_modern has joined #timvideos13:08
shenkimithro: im using the lm32 branch13:28
mithroshenki: How does that compare to the ng branch?13:28
shenkimithro: ng is the latest that was on m-labs13:29
mithrolm32 branch is your rework onto latest?13:29
shenkimithro: which is based on 3.15. my tree is based on 4.813:29
mithroshenki: Have you pushed the latest?13:29
shenkimithro: yep13:30
mithroshenki: Did you figure out the 0x8000 thing?13:30
shenkimithro: nah, i left hackerspace soon after i sent that message13:30
shenkithe milkymist board maps a flash chip at 0x00000000, whereas the litex one maps a memory region there13:31
shenki0x8000 is the last address after the memory region/flash13:31
shenkii'm not sure what opcode 'zero' turns out to be, but the cpu could be fetching from 0 all the way upto 0x8000, when it crashes qemu13:32
shenkibut if that's the case, i don't know why my kernel works on the milkymist soc13:32
shenkiwill keep investigating tomorrow13:32
mithroshenki: Did you see the mem map I linked?13:33
mithroI should also go home myself13:35
shenkimithro: i wrote some notes:
tpbTitle: Home · shenki/linux-lm32 Wiki · GitHub (at
shenkimostly just in case i lose my bash history :)13:45
*** rohitksingh has joined #timvideos13:59
*** Bertl is now known as Bertl_oO14:48
*** aps has quit IRC16:32
*** hyadez has quit IRC16:32
*** aps has joined #timvideos18:07
*** aps is now known as Guest918818:07
*** hyadez has joined #timvideos18:14
*** rohitksingh has quit IRC19:29
*** CarlFK has quit IRC22:04

Generated by 2.13.1 by Marius Gedminas - find it at!