IRC logs for #openrisc Monday, 2016-12-19

--- Log opened Mon Dec 19 00:00:34 2016
mithroblueCmd: Hi! You about at all?05:54
mithroblueCmd: I'm interested in knowing the status of your wb<->axi bridge at
olofk_mithro: Talk to ZipCPU too. He's done one as well06:19
mithroolofk_: Okay06:27
blueCmdmithro: I don't remember, was pretty long ago - IIRC it "should work" for the thing I wanted it to work for06:37
blueCmdshould have tests that you can derrive what functionallity is there06:37
mithroblueCmd: Any chance you are coming to 33C3?06:48
blueCmdmithro: no, it's during the time of year I spend with my parents07:19
blueCmdthey kind of have dibs on those dates07:19
shornestekern: I think probably the cache flush stuff could in next as well. But the commit message need some cleanup.  It seems that came via a github PR, do you know if Jan is around lately?07:22
stekernI don't, but yeah, those commits needs cleaning up before sending upstream07:41
ZipCPU|Laptopmithro: works.  I use it regularly.  It's wb4/pipeline compliant.07:56
stekernfifo_neck and fifo_torso ;)08:03
ZipCPU|LaptopI didn't say the AXI -> WB converter worked, just the WB->AXI converter.  The other's been written, just not yet tested.08:15
ZipCPU|LaptopBesides ... if you need two pointers between the head and the tail ... what names would you pick?08:16
ZipCPU|Laptopshoulders and haunches?08:16
stekernno, I thought the naming was clever08:25
stekernmuch better than "almost_full" and similar that I've seen in the past08:28
ZipCPU|Laptopstekern: Ok.  Do you have a need for an AXI -> WB converter?  (Since you were looking, and I need testing ...)08:48
stekernnaw, I was just quickly browsing through the code and chuckled when I saw the names ;)08:55
ZipCPUThat's fine.  I'm chuckling here too ... while trying to talk all serious-like ;)09:11
kc5tjaZipCPU: Didn't you say you needed four pointers to make your FIFO?  So that would be head, torso, hips, and tail.  ;)12:43
ZipCPUWhat, not head, shoulders, knees, and toes?13:33
kc5tjaOf course, is it really still a FIFO when you have four pointers?  Doesn't it become a gap buffer at that point?  ;)13:49
ZipCPUkc5tja: No ... I think I FIFO is more appropriate, since the head and tail match the original definition of a FIFO.15:10
ZipCPUThe difference being, I need to keep track of two points within the FIFO.15:10
ZipCPUI mean ... basically, there are four operations that must be done in order, starting with 1) filling out the request list into the fifo at the head, 2) issueing the requests to the wishbone, 3) gathering the requests back, and then 4) returning the acks to AXI.15:11
ZipCPUNormally I'd combine 3 and 4, but AXI allows for stall on the ACK side of things, whereas wishbone does not.  So ... four pointers are needed.15:12
kc5tjaI was being somewhat cheeky.15:13
ZipCPUWho ... you???  I would've never known!15:15
--- Log closed Tue Dec 20 00:00:35 2016

Generated by 2.15.2 by Marius Gedminas - find it at!