| *** tpb has joined #openrisc | 00:00 | |
| shorne | ysionneau: I am making progress on TLS | 02:01 |
|---|---|---|
| shorne | The fix I made allows the linker to work and I confirmed access works for LE and IE type access | 02:02 |
| shorne | but GD is having problem, this test mixes GD, LE, IE and LD access to the same variable | 02:02 |
| shorne | it seems that mixing IE and LD is causing the .got table entries to be messed up | 02:03 |
| shorne | I mean mixing IE and GD | 02:03 |
| shorne | i.e. if first writes the entry IE, i.e. just the tp_offset to the .got | 02:04 |
| shorne | then it tries to write GD (mod + tpoffset) to the .got at the same place, it skips this because it protects from doing 2 writes | 02:04 |
| shorne | so, IE access works fine, but GD access fails, because the .got entry is wrong | 02:05 |
| shorne | (Thats what it looks like at least) | 02:06 |
| shorne | I will have to fix the linker | 02:06 |
| shorne | again :( | 02:06 |
| *** tpb has joined #openrisc | 03:49 | |
| shorne | I think I fixed it, made a patch | 08:12 |
| *** rohitksingh has quit IRC | 10:28 | |
| *** o01eg has joined #openrisc | 11:55 | |
| *** ZipCPU has quit IRC | 17:25 | |
| *** ZipCPU has joined #openrisc | 17:48 | |
| *** rohitksingh has joined #openrisc | 17:53 | |
| *** rohitksingh has quit IRC | 18:09 | |
| *** rohitksingh has joined #openrisc | 18:12 | |
| ysionneau | 04:01 < shorne> ysionneau: I am making progress on TLS < great! | 22:07 |
| ysionneau | on our k1c port, only 12 non-passing uClibc tests so far regarding nptl/pthread/tls | 22:08 |
| ysionneau | but those involve pthread_cancel stuff ... (exceptions...) | 22:08 |
| ysionneau | that topic seems to be some kind of nightmare | 22:08 |
| *** rohitksingh has quit IRC | 22:46 | |
| *** rohitksingh has joined #openrisc | 22:53 | |
| *** rohitksingh has quit IRC | 23:06 | |
| *** rohitksingh has joined #openrisc | 23:10 | |
| shorne | ysionneau: I had to work on one pthread_cancel bug, I might have the mailthread. It was interesting | 23:24 |
| shorne | ah, it was this patch: https://patchwork.ozlabs.org/patch/590640/ | 23:26 |
| tpb | Title: [RFC] linux-user: Use SIGRTMAX-1 for guest SIGRTMIN+1 to avoid conflict with host libc - Patchwork (at patchwork.ozlabs.org) | 23:26 |
| shorne | that patch fixed an issue running pthread_cancel on qemu-user* mode | 23:27 |
| shorne | It might not explain your issue | 23:27 |
| shorne | ~~~ | 23:27 |
| shorne | My TLS linker patch doesnt fix the issue | 23:28 |
Generated by irclog2html.py 2.13.1 by Marius Gedminas - find it at mg.pov.lt!