#qi-hardware IRC log for Thursday, 2013-05-23

mthlarsc: do you think there is an advantage to switching to COMMON_CLK instead of implementing clk_enable etc in the platform code?15:36
larscmth: yes, already did that15:37
larscsave quite a few lines15:37
mthoh, then I'm looking at the wrong branch15:38
larscI haven't been able to integrate your pll changes into it15:38
mthcould I have a look at what you have so far? I want to do the same switch to the JZ4770 code15:41
larscyes, I'll push it later15:43
mthok, thanks15:43
mthlarsc: in the mmc driver, if no card detect irq is specified in the platform data, host->card_detect_irq is not initialized17:52
mthwhich means it would be zero because of kzmalloc, I think17:53
mthhowever, 0 is a valid irq number, so it will freed at module removal17:53
larscno zero is not a valid irq number, although the check in remove() might be wrong18:07
larscbut I reworked that codepath as well18:07
larscand it's gone now18:07
wpwraklarsc: are you sure that 0 is not valid ? i think that's just what i got in the old atusb code (that politely asked for the first unassigned interrupt)19:11
larscmth: I just pushed a jz-3.9-clk branch20:37
apeleteoh boy, I just spent a couple of hours tonight trying to solder a serial output to my NN, and only succeeded soldering 1 wire out of 320:39
apeletewhat a waste :-(20:39
apeletethe surface under the battery looks messy now, I hope I didn't melted the pads off...20:41
mthlarsc: thanks20:42
mththat's a lot of changes, fortunately most are deletes :)20:42
larscyes the diffstat looks pretty good20:44
larscI'll probably send this upstream for 3.1220:44
pcercueijz4740 support finally upstream?20:52
mthlarsc: you could use devm_request_irq in the last DMA patch, now that DMA is a device20:52
mthpcercuei: jz4740 support has been upstream for ages20:52
larsctwo years or so20:53
mthwe were talking about using the common clock framework instead of the custom clock implementation20:53
larscapelete: were you using candles to melt the tin?20:54
paul_boddieapelete: Oh well! I managed to spend a good hour soldering four connections to a board recently, and I'm sure people would be unhappy with the joints, but they did work.20:54
pcercueiis jz4770 upstream as well?20:54
mthno, only 474020:54
mthfrantony and d_smagin are looking at making a clean 4755 kernel though20:54
mthso maybe at some point we can combine all 47xx kernels20:55
mthI'm currently cleaning up the GPIO usage in our 4770 kernel20:55
apeletelarsc: what do you mean by candles ? I used a sharp soldering iron tip to melt the tin (not the smallest but relatively small)21:01
apeleteapparently the pads are stll in place (cleaned with a desoldering pump), will try again tomorrow21:02
larscapelete: I'm just kidding, because it took so long21:04
apeletelarsc: not funny at all :D21:10
apeleteanyway, I'll try again tomorrow, making sure I tin the pads before try to solder the wires21:11
paul_boddieapelete, larsc: My problem seems to be getting the iron to be hot enough at the right point: the tip! That and flux-cored solder not apparently being enough to get the solder to behave.21:12
apeletepaul_boddie: working with cheap tools & materials here: 35W soldering iron and not-flux-cored solder. but I won"t be too fast to blame it on the tools or materials, my problem seems to be between the chair and the soldering station :-)21:21
paul_boddieapelete: I'm no soldering expert. I managed to solder connections to the UBB precisely once successfully, but it has been downhill from there ever since then. :-/21:21
paul_boddieI do actually blame the tools. ;-)21:22
mthI always prefer to solder by daylight, it's much easier to see what I'm doing then21:22
apeletepaul_boddie: ha ha :D21:22
paul_boddieBut I'm happy anyway because I got my experiment to work. I'll try and add a wiki page about it soon.21:22
paul_boddiemth: Daylight isn't a problem where I am for most of the day, so bad light is not an excuse I can use at the moment. :-)21:23
--- Fri May 24 201300:00

Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!