Monday, 2018-09-10

*** tpb has joined #timvideos00:00
cr1901_modernI'm going to deliberately increase the SRAM size to force a compile, but why don't we see which beefy symbols/variables can be trimmed out if we are tight on SRAM (using a new preprocessor define as well)?00:12
mithrocr1901_modern: just use a smaller firmware to start with?00:13
cr1901_modernI suppose that works too as long as you don't have a problem with something like a "stub" firmware directory (analogous to what _florent_ has for tinyfpga-soc)00:16
cr1901_modernmithro: ^^00:26
mithroLets get it in ASAP00:29
*** micolous[m] has quit IRC00:49
*** CarlFK[m] has quit IRC00:49
*** micolous[m] has joined #timvideos00:51
*** CarlFK[m] has joined #timvideos00:51
*** mauz555 has quit IRC01:06
fitzsimmithro: no, I've never built any RISC-V toolchain stuff before01:57
fitzsimmithro: for your purposes, I would just copy that recipe, and remove everything01:57
fitzsimmithro: then change all the relevant fields to point to an HX8K example bitfile01:58
fitzsimmithro: its "make" target should work, and generate a bit file in a recipe01:58
fitzsimmithro: since you're already at the point where you have all the iCE40 tools built, and staged in the sysroots01:58
shornemithro: I was asking about qemu devicetree on #qemu no response yet.  I looked more at the code, it seems there is not much 'configuration' done with device tree03:13
shornethe support in qemu is mostly for passing information to the kernel03:13
shorneie. qemu generates device tree details and passes that down to the kernel03:14
shorneMyabe we can do that? i.e. pass device tree details to kernel or litex bios? when running in qemu03:15
*** CarlFK has quit IRC03:16
*** thaytan has quit IRC03:22
* cr1901_modern shudders at the idea of litex-bios becoming device-tree aware03:23
*** thaytan has joined #timvideos03:23
*** ChanServ sets mode: +v thaytan03:23
cr1901_modern(nothing against you/you work shorne)03:23
cr1901_modernJust device-tree + cr1901 = Match made in hell (though I understand why it exists)03:23
*** CarlFK has joined #timvideos03:43
*** ChanServ sets mode: +v CarlFK03:43
*** rohitksingh has joined #timvideos05:52
cr1901_modernmithro (If you're still awake): My sleep schedule is inverted (I've been working mostly at night the past few days). Chances are I'll make a PR for tinyfpga in litex-buildenv while you're asleep tonight.06:01
cr1901_modernmithro: Btw, perhaps we should consider moving some of the litex-buildenv firmare into its own support library that applications (such as stub firmware) can pick and choose to use. Just throwing darts and seeing what'll stick tho :P06:02
*** rohitksingh has quit IRC06:49
*** rohitksingh has joined #timvideos06:53
CarlFKmithro: I finely got around to "git fetch upstream; git rebase -i upstream/master; git push origin mybranch --force"06:59
xobsI seem to remember that one trick to quicker peripheral debugging is to remove the CPU.  However, I've changed BaseSoC.init() to specify cpu_type=None, and EtherboneSoC claims that "object has no attribute 'cpu_or_bridge'".  How do I have BaseSoC create a bridge?07:08
cr1901_modernxobs BaseSoC.add_cpu_or_bridge07:47
cr1901_modernthis could be documented better (<jest>PRs welcome</jest>)07:48
xobscr1901_modern: Should I add the Etherbone bridge?  Or, why isn't a bridge added when BaseSoC.__init__() is called?07:49
cr1901_modernxobs: TLDR; Yes to the first part, probably b/c you can add any WB peripheral you like to the bus as why __init__ doesn't create it.07:52
cr1901_modernWhile clearly, this level of internal impl shouldn't be necessary... The Etherbone bridge should have a "bus" attribute, and anything with the wishbone "bus" attribute can be passed on07:52
xobsThanks, I'll try self.add_cpu_or_bridge(self.etherbone.bridge) and see if that builds.07:55
xobsLiteEthEtherbone has no attribute 'bis' or 'bridge'07:57
xobs'bus' or 'bridge', my fault.07:57
cr1901_modernTry UARTWishboneBridge07:58
xobsThat makes it happy.  Why do I need to add a UARTWishboneBridge to it in order to get Ethernet working?  Does Ethernet go over UART?07:59
cr1901_modernYou don't07:59
cr1901_modernthat makes the UART work :)07:59
cr1901_modernI just wanted to see if it still worked07:59
*** Shari2 has joined #timvideos07:59
cr1901_modernI've never used LiteEthEtherbone07:59
*** rohitksingh has quit IRC08:17
*** rohitksingh has joined #timvideos08:37
*** rohitksingh has quit IRC09:29
*** mauz555 has joined #timvideos09:31
*** CarlFK has quit IRC09:32
*** mauz555 has quit IRC09:36
*** rohitksingh has joined #timvideos09:37
xobsI'll probably need to rename the repo at some point, but I've cleaned up the wishbone stuff and refined litex-devmem2:
tpbTitle: GitHub - xobs/wishbone-adapter: An adapter for the Wishbone network, and in particular the Ethernet interface (at
xobsIt now supports using litex_server and communicating directly over Ethernet with "--target [addr] --direct"09:43
xobsIn litex, how do I designate a signal as "inout"?  I'm still working on the GPIO stuff.  At least now I know Wishbone is getting memory mapped properly (Though I'm not sure why I need the UART).10:06
xobsIn top.v, the pads are showing up as "input".  How do I designate something as "inout", like the mdio or dq pads?10:06
xobsAh, okay, I need to add another parameter to the Tristate module.  Let's see if this synthesizes.10:09
*** rohitksingh has quit IRC10:09
*** rohitksingh has joined #timvideos10:10
*** rohitksingh has quit IRC10:12
xobsIt does, but it doesn't work.  It seems like the "io_" prefix for an Instance enables "inout".  Let's see if I can drop down an "IOBUF".10:32
*** rohitksingh has joined #timvideos10:37
*** rohitksingh1 has joined #timvideos11:01
*** rohitksingh has quit IRC11:03
*** rohitksingh has joined #timvideos11:25
*** rohitksingh1 has quit IRC11:28
*** Kripton has quit IRC11:29
*** Kripton has joined #timvideos11:29
cr1901_modernxobs: I'm sure the Ethernet stuff works, I just don't know how to use it12:11
cr1901_modernand re: the io_ i_ p_ o_ stuff... oops, sorry I wasn't paying attention to answer your q :P12:12
cr1901_modernFor once I'm trying to correct my poor sleeping habits12:12
*** rohitksingh has quit IRC12:34
xobsSleep is very good.12:57
cr1901_modernI should consider learning how to powernap13:02
xobsOr travel to the correct time zone.13:06
*** CarlFK has joined #timvideos13:07
*** ChanServ sets mode: +v CarlFK13:07
cr1901_modernIf only we weren't required by physical limitations to lose 1/3 of our lives13:24
cr1901_modernto involuntarily doing nothing13:25
*** Shari2 has quit IRC13:45
mithrocr1901_modern: Morning?17:08
*** tac-tics has joined #timvideos17:15
CarlFKmithro: can you review my PR soon so I can see how smooth the LCA howto is17:53
CarlFKshorne:  is your qemu/litex ready for public consumption?  (which is about 2 people total)17:56
mithroCarlFK: How do you keep screwing up indenting?17:56
CarlFKmithro: tabs.17:56
*** rohitksingh has joined #timvideos18:39
CarlFKmithro: I tried to find a shell script linter that would warn me before I commited messed up scripts.  best I could find was  and it didnt' compain about mixed tabs and intending problems18:40
tpbTitle: GitHub - koalaman/shellcheck: ShellCheck, a static analysis tool for shell scripts (at
mithroCarlFK: Do you actually look at the pull request before you send it?18:51
CarlFKmithro: the file looks ok in vim, and I do a git diff which also doesn't show a problem18:53
CarlFKreally it is ok as long as your vim settings are right, it is github's version that looks funny :p18:54
*** TheAssassin has quit IRC20:09
*** TheAssassin has joined #timvideos20:14
*** rohitksingh has quit IRC21:18
*** twoolie has joined #timvideos23:13
*** twoolie has quit IRC23:40
*** CarlFK has quit IRC23:48
mithrocr1901_modern: Soooo... pull request?23:58

Generated by 2.13.1 by Marius Gedminas - find it at!