| --- Log opened Sat Nov 14 00:00:09 2015 | ||
| andrzejr_ | so, I've added support for 8/16b access in wb_bfm and it works well. Too well, in fact - a wb slave which is broken in hw (8/16b access) passes the test just fine. :-/ | 02:16 |
|---|---|---|
| andrzejr_ | made a pull request https://github.com/openrisc/orpsoc-cores/pull/90 | 02:50 |
| andrzejr_ | would be good if someone could test it with an existing and working memory slave. | 02:52 |
| andrzejr_ | the code assumes 8/16b accesses are aligned to 32b and bytes/half-words are selected with wb_sel. | 02:53 |
| -!- Netsplit *.net <-> *.split quits: ysionneau, yang, Empyrium, heroux, jkent, sung | 05:14 | |
| -!- heroux_ is now known as heroux | 05:14 | |
| -!- Empyrium_ is now known as Empyrium | 05:30 | |
| -!- jkent_ is now known as jkent | 05:42 | |
| --- Log closed Sat Nov 14 05:49:17 2015 | ||
| --- Log opened Sat Nov 14 05:49:29 2015 | ||
| -!- Irssi: #openrisc: Total of 45 nicks [0 ops, 0 halfops, 0 voices, 45 normal] | 05:49 | |
| -!- Irssi: Join to #openrisc was synced in 8 secs | 05:49 | |
| -!- wallento_ is now known as wallento | 06:20 | |
| olofk | andrzejr_: Awesome. I'll take a look at it | 09:26 |
| -!- ysionnea1 is now known as ysionneau | 15:30 | |
| -!- yang_ is now known as yang | 17:05 | |
| andrzejr_ | olofk, I've redesigned wb_data_resizer again. In 8bit mode it behaves as your implementation - only allows 8b access, although it now also signals an error if it is accessed in 16 or 32bit mode. | 22:39 |
| olofk | andrzejr_: Ahh. that's a good feature | 22:40 |
| andrzejr_ | In 16bit mode it currently allows only 16b access, and signals an error for 32b or 8b accesses. | 22:40 |
| andrzejr_ | We could in theory allow 8b accesses for those slaves that support it but then wb_data_resize would have to have a wbs_sel_o signal. So for now, I didn't bother with that. | 22:41 |
| olofk | andrzejr_: Did you do your changes in orpsoc-cores? | 22:42 |
| andrzejr_ | yes, but I haven't pushed them yet | 22:42 |
| andrzejr_ | still testing | 22:42 |
| olofk | Because latest versions of wb_bfm and wb_intercon are in separate repositories now | 22:42 |
| olofk | wb_bfm-1.0 and wb_intercon-1.0 | 22:42 |
| olofk | I don't think there are many changes to wb_interon, but wb_bfm is quite different from the old version in orpsoc-cores | 22:43 |
| andrzejr_ | I've put it here for now: http://pastebin.com/Xa14hhdp | 22:44 |
| andrzejr_ | hmm.. I assumed version 1.0 was an older core "before upstreaming". atlys used the ones from orpsoc-cores | 22:48 |
| andrzejr_ | are there any significant changes in 1.0? | 22:48 |
| olofk | Yeah. Sorry about that. I'm trying to change all the cores to use a version number, but I haven't really communicated that | 22:50 |
| olofk | For wb_intercon, there are only small changes, and wb_data_resize is identical, so the patch should apply there too | 22:51 |
| olofk | But if you're doing any work on wb_bfm, please use the one in https://github.com/olofk/wb_bfm | 22:52 |
| andrzejr_ | Thanks, I'll switch both cores to version 1.0 | 22:53 |
| olofk | Cool. And sorry about not being more clear on that | 22:54 |
| andrzejr_ | olofk, wb_data_resize works fine here, maybe you could check on your side if it doesn't break anything that was working before | 23:28 |
| olofk | I'll try to take a look tomorrow | 23:37 |
| olofk | Good night | 23:37 |
| --- Log closed Sun Nov 15 00:00:10 2015 | ||
Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!