stekern | poke53281: wow, way cool! | 04:11 |
---|---|---|
stekern | regarding libgmp and libmpfr, when reading up on the libs, my guess is that they are used to evaluate arithmetics on constants | 04:58 |
-!- Netsplit *.net <-> *.split quits: O01eg, Xark | 12:03 | |
-!- Netsplit over, joins: O01eg | 12:12 | |
blueCmd | stekern: do your know anything about relro support in GCC? | 13:02 |
stekern | blueCmd: nope, I'm familiar with what it means, but not at all how it is implemented | 13:05 |
stekern | but I'd assume it's in the linker, not in the compiler? | 13:06 |
blueCmd | stekern: I would think that yes. I'm just for the fun of it trying to start porting glibc, and it apparenly wants relro | 13:08 |
blueCmd | reading up on it atm | 13:08 |
blueCmd | yeah, it looks for a GNU_RELRO section | 13:18 |
blueCmd | stekern: do you have any idea why that section isn't added with or1k-gcc? I tried grepping on RELRO in gcc and it seems to be pretty much arch independent | 13:27 |
blueCmd | ah right, the linker! stupid me | 13:28 |
blueCmd | hm, adding "COMMONPAGESIZE="CONSTANT (COMMONPAGESIZE)"" to ld/emulparams/elf32or1k.sh seems to enable relro but I have no idea what I just did :) | 15:07 |
blueCmd | also enabling CFI doesn't seem too hard, is there a reason why this hasn't been done? | 17:02 |
stekern | blueCmd: probably just because no one have had time to look into it ;) | 17:05 |
stekern | i.e. the slot is open for you to step in! | 17:07 |
blueCmd | stekern: great | 17:15 |
blueCmd | I have it working, but I guessed some numbers so it will need some testing before I submit the patch :) | 17:15 |
blueCmd | or, it will need some research into what the numbers really should be | 17:16 |
stekern | cool | 17:21 |
stekern | a glibc port is pretty high on the wishlist too, so if you sit down with that, it'd be great! | 17:21 |
blueCmd | I already started :) | 17:22 |
stekern | great :) | 17:22 |
blueCmd | eglibc-port really | 17:22 |
blueCmd | Not saying that it won't take a couple of months or so though - it's quite massive | 17:23 |
blueCmd | but i'm sure you know that | 17:23 |
ams | Shouldn't take that long to port glibc. | 17:27 |
blueCmd | ams: perhaps, it's not that much code (mean seems to be around 6000 per arch), but testing and making sure it works will take some time I figure | 17:40 |
blueCmd | stekern: I submitted a pull request for the relro fix, it's not big ;) | 17:41 |
Erant | Hmm, minsoc's code could do with a bit of tidying up. | 18:07 |
-!- LoneTech_ is now known as LoneTech | 19:04 | |
poke53281 | @blueCmd: glibc support would be great | 19:20 |
ams | blueCmd: Nah, not much work that ... "couple of months" is quite alot of time :-) | 20:06 |
blueCmd | ams: I hope yoy're right :) | 20:28 |
blueCmd | stekern: have you hacked anything with eh_frames? When I added the CFI support GCC started to sort of pad .eh_frame with 32 bit zeros, making ld believe the table is empty. | 20:30 |
blueCmd | really weird. it only does this on crtstuff.c and not on a simple file with only an "int main" function. | 20:31 |
blueCmd | oh well, time for a break anyway | 20:31 |
stekern | blueCmd: can't say I really have touched eh_frames, no | 21:54 |
stekern | as you might notice, I'm not really a gnu toolchain expert ;) | 21:56 |
stekern | I know the stuff I've touched, the PIC stuff, but most of the other stuff is muddy waters for me | 21:56 |
blueCmd | stekern: I didn't expect you to be but since I never hacked on gcc or binutils in the past you're still an expert compared to me ;) | 22:20 |
stekern | oh, you'll catch up quickly ;) | 22:21 |
stekern | but ask on, I don't mind learning from the questions | 22:22 |
blueCmd | yeah, I think I'm getting the hang on it - I'll instrument gcc and see if I can hunt this down. | 22:22 |
blueCmd | great! | 22:22 |
blueCmd | stekern: if you're interested; http://pastebin.com/mjgbF7St - the first case is crtendS.c from GCC after preprocessor - notice the first 0:d 32 bits in .eh_frame. | 22:28 |
blueCmd | it's not there in my conftest.c which is a simple function | 22:28 |
blueCmd | so atm I'm doing a binary removal to find the bad guy :) | 22:29 |
blueCmd | hm that wasn't hard.. there was an asm that actually prepended 0 to the eh_frame | 22:30 |
* blueCmd will do silent work and stop spaming :) | 22:31 | |
stekern | heh, nono, keep using irc as your private notepad | 22:31 |
stekern | I do that all the time | 22:32 |
stekern | ;) | 22:32 |
stekern | time to go to sleep here, early morning tomorrow | 22:32 |
stekern | see you | 22:33 |
blueCmd | stekern: take care! | 22:33 |
Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!