IRC logs for #openrisc Thursday, 2016-06-23

--- Log opened Thu Jun 23 00:00:06 2016
olofkHmm... we really should get this adv_debug_sys uart working. Got a board now without any easy way to break out two uart pins04:16
wallentoor you can switch to Open SoC Debug ;)04:23
wallentogetting the run-control (aka adv_debug_sys) running there is still on top of my wishlist only unfortunately..04:24
wallentobut it has a "virtual" UART04:24
wallento;)04:24
stekernolofk: isn't it almost working?05:12
olofkstekern: It works in polling mode, but fails in Linux when interrupts are enabled06:35
stekernah, right06:35
olofkwallento: Program loading and UART are what I need. Is program loading considered part of run-control?06:36
olofkBut yes. I'm interested in Open SoC Debug, but I got a setup for a client now and can't justify changing debug system atm06:46
olofkWhy is it always so hard to remember how to space-extend a string with verilog $display06:57
olofkaha! It's just %0s. For strings and decimal values that trims the 0's, but for hex and binary it insert zeros07:00
olofkVeri logical :)07:00
ZipCPUolofk: Did I hear you mention the word "client"?  You mean there's actual money to be made in this business?07:01
olofkZipCPU: Well yes. If the client asks for some FPGA functionality and it makes sense to put in a CPU, I much rather use an OpenRISC than MicroBlaze or NIOS07:03
olofkThat makes it more portable, I can simulate it properly and I have less proprietary crap that I have to get working07:04
ZipCPUOkay, that makes sense.  A lot of sense in fact.  Keep going, then, and don't let me slow you down.  :)07:05
olofk:)07:05
olofkAnd the good thing here is that I could reuse much of the code that I wrote for another client some time ago. That client explicitly asked for having as much as possible Open Source07:05
olofkEverybody wins! :)07:06
wallentoolofk: loading is another part of Open SoC Debug, I hope we can release the OpenRISC stuff over summer07:50
shorneok, another newlib patch sent, I think that fixes the duplicate environ thing.  but its really simple, not sure why it was done like that in the first place08:29
shornehttps://github.com/openrisc/newlib/pull/1008:30
SMDwrkCan anyone tell me how system registers work for mor1kx? Let's say I'm implementing PCU, so I need to create my own module, connect it to spr bus and then check if addr/data on spr bus relates to PCU, make something I want, right?08:55
-!- heroux_ is now known as heroux09:36
olofkSMDwrk: I realized now that I'm not sure what a PCU is09:56
SMDwrkperfomance counters unit09:56
olofkah right09:57
olofkYou've talked about that before09:57
SMDwrkI'm finally have some time at work while tests are running)09:58
olofkI haven't really looked into that part of the or1k spec09:59
olofkdoing it now09:59
olofkBut yes, that sounds about right09:59
SMDwrkOk then I'll start implementing it that way, thanks10:00
olofkI would assume that there is a SPR address mux somewhere in the code and some code to handle the other SPR accesses that you can look at10:01
olofkI have only touched the SPR stuff for hacking on the version register... and that could have been the or1200... like five years ago10:02
SMDwrkI think I'll take PIC as example10:03
olofkshorne: The travis build failed with your patch, but it looks like it was because binutils was unavailable10:04
olofkBut your patch is probably fine. Don't know if it breaks using environ from bare-metal apps, because no one has probably ever tried10:06
SMDwrkAnd what should happen if you try to read sysreg from user mode while this register is available only in system? Should you read just 0 or get an exception?10:13
-!- heroux_ is now known as heroux10:15
shorneolofk: having a look at the failure. Thats what I wasnt sure about would you compile bare-metal app expecting libc and not link libc?10:37
shorneI mean libgloss has the main and handlers, but you wouldnt want the environ without libc10:37
shornenot sure if that makes sense10:37
shorneI guess we would need to rerun the test10:46
olofkshorne: I think that you could theoretically set up environ to be loaded from a specified memory address or via a register value (like we do for the Linux device tree)11:50
--- Log closed Thu Jun 23 16:27:31 2016
--- Log opened Thu Jun 23 16:28:16 2016
-!- Irssi: #openrisc: Total of 51 nicks [0 ops, 0 halfops, 0 voices, 51 normal]16:28
-!- Irssi: Join to #openrisc was synced in 14 secs16:28
olofkI think I'm getting ready to release FuseSoC 1.5. Any known issues I should fix first?16:37
olofkGot a lot of big stuff lined up that I want to add after the release16:37
olofkAhh.. just remembered one thing16:38
--- Log closed Fri Jun 24 00:00:07 2016

Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!