| --- Log opened Sun Jul 28 00:00:46 2013 | ||
| stekern | olofk: we basically just settled the date and place for orconf'13 | 02:13 |
|---|---|---|
| stekern | and I have asked the same question about the rams several times, no-one has ever responded | 02:14 |
| poke53281 | stekern: You are my hero. Thanks for the patches. We have to include the tlb hardware reload in or1ksim too. | 05:45 |
| poke53281 | I am still a little bit depressed that X is not stable. | 05:50 |
| poke53281 | So I though I play a little bit | 05:51 |
| poke53281 | simulationcorner.net/monkey.png | 05:51 |
| poke53281 | :) | 05:51 |
| stekern | hah! that's the coolest thing I've ever seen being run on an openrisc implementation | 06:08 |
| stekern | I'm a bit depressed that you beat me to it though =P | 06:09 |
| stekern | regarding hw tlb reload in or1ksim, we could probably just about copy what I have in qemu: http://pastie.org/8182913 | 06:13 |
| poke53281 | Yes, looks doable. I should also work on my or1ksim stuff. There is so much to do. | 06:31 |
| poke53281 | The one thing I don't like in the tlb reload is the definition that the MMUCR registers must be zero for software dtlb refill. | 06:32 |
| poke53281 | A flag bit would be better | 06:32 |
| poke53281 | There are still 10 bits reserved in in the SR register. And for something important like this I would take one of those bits. | 06:35 |
| stekern | perhaps, but what should be done when it's zero? it'll generate a pagefault in the end, so falling back to tlb miss exception seems like a sensible thing to do | 06:38 |
| stekern | even if there would be a föag for it | 06:38 |
| poke53281 | Flag=0: always software tlb refill. Flag=1: Always Hardware TLB Refill. If the register mmucr is zero then it will try to read the pagetable from zero, which will probably lead in the end to a page fault. But almost all implementation that I have seen so far behave like this. | 06:51 |
| stekern | yeah, I don't really have strong feelings either way on that matter | 06:55 |
| poke53281 | For or1ksim you have to change the refill to extend for the different tlb sizes. Unfortunately just copy and paste will not do it. | 06:57 |
| stekern | you still have to write the mmucr register to enable/disable hw refill, so I don't see the huge benefit though | 06:57 |
| stekern | you can have the pagetables at address 0, but that's about it | 06:59 |
| stekern | (or1ksim) ah, but that's a minor thing to fix | 07:00 |
| poke53281 | Well, call it symmetry :). Otherwise it will break the consistency of the architecture. | 07:00 |
| poke53281 | Why everything important can be enabled and disables in the SR register instead of the hardware refill? | 07:01 |
| poke53281 | There are already flags for the MMU in it. | 07:01 |
| stekern | aha, you are speaking sbout *SR*, i thought you meant the reserved bits in mmucr | 07:03 |
| poke53281 | Yes, I wrote *SR* :) | 07:03 |
| poke53281 | I am pretty sure the original writers of the specification would have done it that way. | 07:04 |
| poke53281 | original authors | 07:04 |
| stekern | yes, I'm just a sloppy reader... ;) | 07:08 |
| stekern | I guess that would be fine, I'll think about it some more | 07:10 |
| poke53281 | Ok, think about it. I will write a short reply wit the suggestion tomorrow. Now I have to go to bed. | 07:12 |
| stekern | sleep well, and thanks | 07:28 |
| stekern | immediate problem in mor1kx is that SR is only 16 bit wide, but that's not a good reason against it ;) | 07:34 |
| olofk | poke53281: Have to agree with stekern. That's the coolest and most useful thing I've ever seen the OpenRISC being used for | 10:10 |
| olofk | stekern: Great. I was mostly interested in the date so I can book plane tickets. When is it? Wiki is not updated | 10:12 |
| stekern | olofk: october 12th I believe was the verdict, jeremybennett or juliusb might want to confirm | 15:14 |
| hno | where? | 17:00 |
| juliusb | http://opencores.org/or1k/ORCONF | 17:19 |
| juliusb | finally got around to it today | 17:19 |
| stekern | juliusb: is it much harder to get to cambridge from gatwick than stansted? | 17:43 |
| stekern | looking at flights, they are offering me 22 h travel in the range 180 € - 1268 € helsinki - stansted | 17:44 |
| stekern | helsinki - gatwick is 3 hours straight flight for 157 € | 17:45 |
| juliusb | Stansted is a lot closer than Gatwick, but it's not the end of the world | 18:18 |
| juliusb | it's just another couple of hours of travel all up | 18:19 |
| juliusb | stekern: are you thinking of coming Friday or Saturday? | 18:21 |
| juliusb | yes it looks like Helsink is not well connected to Stansted. It is a big Ryan Air hub | 18:30 |
| juliusb | so if ther's a Ryan Air portal of pain nearby, then I'd look at flights there | 18:30 |
| hno | Ryan Air do have a quite reasonable flight from Stockholm, 438:- (51€). Friday night - Monday morning. | 19:18 |
| --- Log closed Mon Jul 29 00:00:47 2013 | ||
Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!