00:54 <GitHub113> [migen] sbourdeauducq pushed 1 new commit to master: http://git.io/E6WnuA

00:54 <GitHub113> migen/master 8baa957 Sebastien Bourdeauducq: genlib/fifo: use synchronous memory read instead of additional register...

13:18 <sb0> rjo, do you know this group? http://www.phy.cuhk.edu.hk/~djwang/index.html

13:26 <GitHub161> [migen] sbourdeauducq pushed 1 new commit to master: http://git.io/Y9BO2Q

13:26 <GitHub161> migen/master a0d0742 Florent Kermarrec: mibuild/generic_platform: add recursive parameter to add_source_dir

14:05 <_florent_> hi

14:06 <_florent_> the misoc-kc705 port has 128-bits registers for each DFI data phase

14:06 <_florent_> which raise the NotImplementedError("Register too large") error

14:07 <_florent_> ln my case the 64 MSB bits ot these registers are not used by the LM32 software

14:07 <_florent_> but my python script for protoryping use the 128 bits

14:07 <_florent_> I'm using this hack to be able to generate the design:

14:07 <_florent_> http://pastebin.com/xASL4ft8

14:08 <_florent_> What would be an acceptable solution for this?

14:09 <_florent_> is a truncation with a warning something acceptable? (it will be then the first warning of Migen... )

14:10 <_florent_> or do you have others ideas?

14:16 <sb0> _florent_, I'd say, just not generate access functions in that case, and use the CSR_*_ADDR CSR_*_SIZE defines

14:16 <sb0> are you writing a patch for the sdrrd/sdrwr commands?

14:17 <sb0> they could use the _SIZE define to automatically get the correct sdram data length

14:18 <_florent_> not I was just trying to use upstream version of MiSoC

14:18 <_florent_> no

14:19 <_florent_> but these 128 bits registers are probably indeed only used by sdrrd/sdrwr commands

14:20 <sb0> yes

14:20 <sb0> and the sddrd/sdrwr commands currently hardcode the length and should be fixed.

14:23 <_florent_> ok, I'll try to fix that by the way, thanks

14:39 <_florent_> would this be ok (only for cpuif.py) ? : http://pastebin.com/fM5casL5

14:40 <_florent_> if size > 64 we only generate CSR_*_ADDR & CSR_*_SIZE

14:46 <sb0> yeah, looks good

14:52 <_florent_> thanks, I will also fix sdrrd/sdrwr commands and send you a patch