| -!- Netsplit *.net <-> *.split quits: bradfa | 03:04 | |
| stekern | LoneTech: yup, that's what I found out from googling too. I ignored it for know, and just ran 'boot disk'. Everything else seems to work fine, but I need to reset the root password and I don't have a sunos cd. Seems like there is no write-support for UFS under linux. | 05:14 |
|---|---|---|
| stekern | (sorry about being so highly off-topic) | 05:15 |
| stekern | blueCmd: just FYI, I'm running the regression tests now | 06:23 |
| stekern | I started them yesterday, but it seems that or1ksim (or linux) locked up before it was done | 06:24 |
| stekern | so I restart it now | 06:24 |
| stekern | looked like it was about the same result as before your patches up to the point where it crashed though | 06:25 |
| -!- Kevin` is now known as Kevin`_ | 09:59 | |
| blueCmd | stekern: cool! | 10:53 |
| _franck_ | stekern: do you know how to run the regression test on the or32 toolchain ? | 10:58 |
| jeremybennett | _franck_: Saw your email - I was offline yesterday. | 11:25 |
| jeremybennett | Should just work as "make check-gcc" etc. | 11:25 |
| jeremybennett | You need to set up DEJAGNU to point to your site.exp file, and that might need some hard-coded details changed - I can't remember exactly. | 11:26 |
| _franck_ | ok thanks, I'll try to look in this direction :) | 11:38 |
| LoneTech | huh. gcc 4.5.1-or32-1.0rc4 __ffssi2 is surprisingly horrible (used for ffs()). I guess it hasn't been told about l.f[fl]1 | 14:51 |
| stekern | LoneTech: the find first instructions are optional... | 15:13 |
| stekern | I ignored that fact in LLVM though | 15:13 |
| LoneTech | ok | 15:14 |
| stekern | fwiw, the or1k gcc (4.8) expand __builtin_ffs to __ffssi2 as well | 15:26 |
| stekern | IIRC, newlib has a l.ff1/l.fl1 in it's startup code, so it's not so "optional" after all | 15:28 |
| stekern | http://pastie.org/5688303 | 15:34 |
| LoneTech | yep, in newlib or1k_interrupt_handler l.ff1 appears | 15:39 |
| jeremybennett | stekern: That would appear to be a bug. | 15:40 |
| stekern | the appearing in newlib? | 15:40 |
| jeremybennett | Yes | 15:41 |
| stekern | I agree | 15:41 |
| stekern | me neglecting the fact in LLVM is also a bug | 15:41 |
| stekern | I should put a switch to turn it on and off with in there | 15:41 |
| LoneTech | it's in gnu-stable/newlib-1.18.0/newlib/libc/machine/or32/or1k-support-asm.S | 15:45 |
| -!- bradfa_ is now known as bradfa | 18:12 | |
| _franck_ | stekern: as you ran the testsuite on or1k, could you send me your test log ? | 23:38 |
| _franck_ | running make check RUNTESTFLAGS="--target_board=or1k-sim" in the gdb directory and I get a lot of | 23:39 |
| _franck_ | ERROR: Couldn't set target for simulator. | 23:39 |
| _franck_ | during the tests | 23:39 |
| _franck_ | do yo also have this ? | 23:39 |
| _franck_ | s/yo/you | 23:39 |
Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!