| stekern | blueCmd: the got is actually more linked with PIC than dynamic linking, needing PIC for dynamic linking libraries is just a result of that | 08:32 |
|---|---|---|
| blueCmd | stekern: yes, a lot of things cleared by reading the document, drawing some things and venting thoughts - maybe I should get a rubber duck instead of the irc channel? ;-) | 11:17 |
| blueCmd | stekern: btw, feel free to merge my latest pull request (I don't have time to clean up my potential mess if I try :P) | 12:00 |
| blueCmd | it will fix the last (what I can see anyway) regressions from my dw2 patch series | 12:01 |
| stekern | blueCmd: will do | 14:46 |
| stekern | yay, I'm starting to get console output from Linux on mor1kx | 16:01 |
| stekern | in simulation at least | 16:01 |
| stekern | woho, in hardware as well! | 16:14 |
| stekern | it crashes after a while, but still | 16:15 |
| stekern | http://pastie.org/5868447 | 16:20 |
| stekern | looks like it's a dtlb miss that turns into bus error after the tlb has been updating and the exception is rfeing back to the load | 18:24 |
| stekern | i.e. probably some rfe->dmmu enable glitch | 18:25 |
| blueCmd | I can now link to my non-hacked eglibc (only static), and run C programs \o/ | 19:59 |
| blueCmd | there is something not quite right in the linker. | 19:59 |
| blueCmd | stekern: do you remember that we hunted a bug related to weak refs? I got multiple definitions | 20:00 |
| blueCmd | now I have the problem that I have to explicitly specificy to link with certain .o files - if I just just libc.a (which contains the .o files I need) the final output will only contain weak references, not the real symbols | 20:01 |
| stekern | hmm, no, can't say I remember | 20:02 |
| stekern | I know that attr-weakref-1.c have been failing in the testsuite | 20:02 |
| stekern | it has been failing upstream as well | 20:02 |
| stekern | but the problem there is that a weakref to a weakref isn't emitted | 20:03 |
| blueCmd | hm, might be related | 20:03 |
| blueCmd | I just checked now - g++ emits weak refs as well so perhaps the linker does the right thing | 20:04 |
| stekern | yeah, from what I have seen, it does weak refs right | 20:04 |
| stekern | doesn't mean there might be corner cases I haven't seen | 20:04 |
| stekern | I need more blockram in my cyclone iv... | 20:06 |
| stekern | hmm, this is odd... seems like sr just get's overwritten for no apperant reason | 20:13 |
| stekern | +a | 20:13 |
| blueCmd | stekern: the issue I'm having is that _nl_current_LC_* (and what I can see, only those) are called with relocation addr 0 | 20:30 |
| blueCmd | which makes it hard for me to relocate them, since I don't know where they are | 20:30 |
| blueCmd | relocate to them* | 20:44 |
| _franck_ | jeremybennett: finally got some results: http://pastebin.com/GYJTy7LU | 22:32 |
| _franck_ | do we have a reference point to say ir is good or bad ? Does it make sens to compare this to our gdb 7.2 results ? | 22:34 |
| _franck_ | I also need your help to identify which FAIL is target dependant | 22:37 |
Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!