| --- Log opened Thu Apr 16 00:00:09 2015 | ||
| -!- Hesham1 is now known as hesham | 11:43 | |
| olofk | Hmm.. or1ksim doesn't want to load my assembled file. Says it's not an elf | 14:24 |
|---|---|---|
| olofk | Wonder if this is the same problem as with OpenOCD | 14:24 |
| dalias | stekern, can you proofread some musl or1k asm for me? :) | 14:56 |
| stekern | dalias: hit me | 17:52 |
| dalias | http://sprunge.us/DTJc <-- new sigsetjmp for all archs; some untested so far | 17:57 |
| dalias | what it needs to do is save lr and one call-saved register inside the unused space of the jmp_buf, backup the argument (jmpbuf) ptr inside the call-saved reg it just got... | 17:57 |
| dalias | then call setjmp | 17:57 |
| dalias | then restore lr and the call-saved reg it used and tail-call to __sigsetjmp_tail(jb, retval_of_setjmp) | 17:58 |
| dalias | this is the magic by which i'm making signal mask get restored _after_ siglongjmp jumps back to the saved context rather than before | 17:59 |
| dalias | (the way most systems do it restores the signal mask first, but then under signal barrage you can get unbounded stack usage) | 17:59 |
| stekern | dalias: looks good, I think. The only comment I have is that r16 happens to be our GOT register. | 18:20 |
| dalias | oh. does that matter? | 18:40 |
| stekern | I don't think so, but I stopped an extra second to think if it had any importance that you used the GOT reg there. | 18:43 |
| dalias | bbl | 18:43 |
| --- Log closed Fri Apr 17 00:00:10 2015 | ||
Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!