IRC logs for #openrisc Friday, 2014-12-26

--- Log opened Fri Dec 26 00:00:10 2014
-!- Netsplit *.net <-> *.split quits: poke53281, kiwichri_, Amadiro, heroux02:49
-!- heroux_ is now known as heroux02:49
stekernpoke53282: what do you mean that gcc doesn't use l.div?07:25
stekernI don't think bug 107 is still present07:39
olofkpoke53282: The printk in bug 106 is still there in the upstream kernel. Was this problem fixed while still allowing the printk to stay?08:48
olofkstekern: I remember there was some talk about divsi and friends. The logs hint that it was done some work on that in early August08:50
olofkblueCmd: Do you remember any of this?08:50
olofkThis seems relevant https://github.com/openrisc/or1k-gcc/commit/bdd3ad496930c61218ea683b9fd3dbcc093b9a1408:51
olofkhaha. bug 4 might be a hint why it looks like part of the JTAG ID code was shifted a bit08:58
poke53282stekern: I haven't checked it thouroughly. But for me it looks like, that my compiler use a libgcc function for all divs09:06
poke53282l.mul works.09:06
poke53282olofk: http://git.openrisc.net/cgit.cgi/jonas/linux/commit/?id=e43f1288e64639e9f0d46ff1d8a3dea1ff9cb2b209:11
poke53282If this is upstream, then the bug is fixed09:11
olofkpoke53282: Thanks. It's there, so I'm closing the bug09:25
blueCmdolofk: sorry no09:30
olofkI'm closing bug 27 now as worksforme. If anyone has deeper knowledge about why this works/didn't work, please add a comment09:33
olofkHow's the dwarf2 support nowadays?09:51
stekernpoke53282: I at least have l.div in my vmlinux13:24
stekernolofk: yes, that commit should make 107 not relevant anymore13:25
olofkstekern: Thanks. 107 is closed now14:05
olofkWhat the fuck should we do with the swapped PICP/PMP bits?14:13
olofkmor1kx is aligned with the spec, but the kernel and or1200 isn't14:15
olofkI'm all for pretending this isn't a problem and just push fixes for linux and or120014:16
olofkI'm not even sure it is a problem in practice14:16
olofkThe slightly nicer thing to do would be to do something like stekern did with the device tree for the edge/level weirdness in or1200. Not sure if something similar can be done here though14:17
olofkOr can we use the version regs?14:18
olofkAnd should spr_defs.h in the kernel be regenerated from or1k-headers14:19
poke53282olofk: I guess, at the moment I am the only one who is using the PMP.14:29
poke53282and in stekerns SMP branch is is already corrected.14:30
-!- atgreen` is now known as atgreen14:58
olofkpoke53282: That makes things easier. I'll just wait a week for comments, if any, and apply the RTL fix15:08
olofkAnd we should see if we could get jonibo to take on stekern as co-maintainer for the kernel15:09
poke53282yes, there are a lot of patches waiting for the kernel. Mine and stekerns.15:35
poke53282I have a few ptrace patches too.15:36
poke53282Even I would do the mainting.15:37
poke53282.oO  ( Hopefully stekern don't want to upstream the current ompic )15:39
poke53282:)15:39
stekernyeah, no15:46
olofkpoke53282, stekern: It would be great if any of you could send Jonas a mail or something and see if we can work things out.17:29
olofkI have gotten the feeling that he's been pretty tied up by other stuff, but if I understood things correctly, the upstream port isn't really usable, which is pretty bad in that case17:30
poke53282as far as I know the upstream version works, but you shoudln't do anything with signals.18:38
--- Log closed Sat Dec 27 00:00:11 2014

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