IRC logs for #openrisc Tuesday, 2013-07-23

--- Log opened Tue Jul 23 00:00:39 2013
poke53281olofk: Patches send to the mailing list. Let's see03:16
stekernpoke53281: how did you generate those patches?06:48
stekernit looks like git format-patch, but with the e-mail information stripped, so it doesn't apply with git am06:50
poke53281well, the email was wrong, so I removed this part. I thought the interesting part is below "---"06:53
poke53281no problem to send the patches again ... tomorrow.06:55
stekernyeah, np, I can apply them for testing as is, the commit info get lost though06:56
stekernbut that's not so important, jeremybennett is the one that should commit them06:57
poke53281just changing the email address in the document seems not problem. I think the first line includes a sha1 hash07:04
poke53281....seems to be a problem ...07:05
stekern'git send-email' works pretty well too07:09
stekernbut that would of course not have fixed your first problem, that the e-mail was wrong07:11
jeremybennettpoke53281: Thanks for your patch. I've asked for a couple of extra things, then it can be committed to both branches08:25
-!- Netsplit *.net <-> *.split quits: enghong, jakob_10:59
-!- Netsplit over, joins: enghong10:59
* stekern got to Linux prompt with hw itlb reload15:10
stekernin the verilated model at least15:10
stekernsimple stuff like 'cat /proc/cpuinfo', 'top', 'ls' and 'uname -a' works too at least15:11
stekerncoremark too15:41
stekernthis seems stable enough, let's move onto dtlb reloading15:41
stekernthat should be a walk in the park, no annoying fetch corner cases to give you headaches15:43
juliusbawesome stuff16:27
juliusbhow much faster do you estimate this is stekern?16:27
poke53281stekern: Perfect16:36
poke53281Did someone test the eglibc port so far. I have realized that it moved to the official site16:40
stekernjuliusb: extremely rough estimate, from just comparing the time a random software dtlb exceptions take and a random hardware itlb reload in the waveforms I get 1500ns/200ns16:40
juliusb... you mean 1500ns down to 200ns?16:43
stekernyeah, 1500 vs 200 for the two cases16:45
juliusboh nice16:45
juliusbwe usually simulate at 50MHz (20ns clock period), so about 75cycles versus 10?16:46
juliusbthat's a real performance win :)16:46
stekernand with the stats from poke53281s jor1k, we saw that dtlb misses can go up to about 25% (do I remember correct?) during a 'gcc hello_world.c'16:46
olofkFarewell sweet uptime16:46
stekernI mean, 25% of the time spent on handling dtlb misses16:47
stekernpoke53281: I think only blueCmd (who's done the work on it) have tested it, afaik it's not complete, but somewhat working16:50
juliusboh jebus!16:52
juliusbthat's a high proportion of time spent doing that16:52
stekernI just forked it over to the openrisc site to show that there exist an effort of a eglibc port =P16:52
stekernjebus, the prophet of openrisc?16:53
juliusbMaybe that's Lampret16:54
poke53281stekern: Yes, you are right. The dtlb misses took a lot of performance. 25% at least16:55
juliusbhe came, gave us the word of open source CPU development, and has then gone16:55
juliusband we hold out hope for his return16:55
juliusb(well, not actually)16:56
juliusboh yeah btw I'm going to talk at OHS 2013 in Boston this year17:04
juliusbI'm getting 6 minutes17:04
poke532816 minutes sound more like Hello and Good bye. I guess 1 minute is for questions. :)17:08
stekernwell, think about it this way, most internet meme videos on youtube are less than 6 minutes, you can make an impression in that time at least =P17:15
poke53281There is a very annoying bug in the kernel which I think started around 3.9. There is a unaligned access in this loop:
poke53281The kbd_struct uses a lot of bit variables.
poke53281So I am not sure, either change the gcc compiler or patch the Linux kernel or include a proper unaligned access handler in the kernel.18:06
poke53281Indeed that is the only unaligned access I could find. All other user-space progs I have  compiled sp far don't use unaligned accesses.18:07
stekernah, I think I've seen that too, but iirc in 3.8 already18:14
stekernI had to disable the ps/2 interface on my atlys board, to be able to continue with what I was doing right then18:16
stekernwithout looking closer though, sounds odd that bitfields in chars would generate unaligned accesses18:18
stekernbut I think our struct padding in gcc is buggy, so could be related to that18:18
poke53281in the evening I will take a closer look and program a small test program.18:19
stekerndoing sizeof(struct eth_hdr) does not return 14 for instance18:19
stekernand I *think* one of the testsuite fails is related to that18:19
stekernthe first fail, that is18:19
poke53281bitfields should be always be padded to word alignment I think18:19
stekernhmm, yeah, that might be. That would be a problem then, since the struct starts with 2 chars18:21
stekernhave to take a look in the standard18:21
stekern...but right now it's time for bed18:22
poke53281good night18:22
olofkIt's insane that there are only five minute slots at OHS.21:34
poke53281How many talks are there in howmany days?. And how many people are visiting this place?22:16
--- Log closed Wed Jul 24 00:00:40 2013

Generated by 2.15.2 by Marius Gedminas - find it at!