Wednesday, 2019-07-03

*** tpb has joined #litex00:00
futarisIRCcloud - Bordeaux, France - September, 27th to 29th, 201900:42
tpbTitle: ORConf 2019 (at
*** tweakoz has quit IRC00:43
*** tweakoz has joined #litex04:43
*** rohitksingh_work has joined #litex05:17
tweakoz@ _florent_ : is there a reason not to have an assert like this: (line 202 in the post)
tpbTitle: Update · tweakoz/[email protected] · GitHub (at
tweakozI switched to a CSR like you said up above to get my re signal. it did not work correctly due to the CSR in question being 32 bits. 32 bit CSR's allocated 4 slots of the address decoder space, but they only incremented i by 1 in the CSRBank "for i, c in enumerate(self.simple_csrs):" loop. this caused a csr select failure.08:24
tweakozso the assert I added would have caught that. Though maybe I should try incrementing i by c.size and see if that works also.08:26
tweakozIm guessing incrementing by c.size wouldn't work due to bus size. duh..08:45
tweakozI'd rather do 32 bit bus size anyway. I think it was the SDRAM system preventing it...08:46
*** futarisIRCcloud has quit IRC09:38
_florent_tweakoz: i'll have a look, but yes we should probably add an assert when using CSR (vs CSRStorage/CSRStatus that can be > bus.data_width)09:59
_florent_tweakoz: the SDRAM calibration logic does not handle csr_data_width of 32, it should be rewritten to be more generic, so yes, if you use SDRAM, you have to use csr_data_width = 810:00
*** futarisIRCcloud has joined #litex10:01
*** tweakoz has quit IRC11:26
*** rohitksingh_work has quit IRC11:47
*** somlo has quit IRC11:53
*** somlo has joined #litex11:56
*** somlo has quit IRC12:21
*** somlo has joined #litex12:29
*** futarisIRCcloud has quit IRC13:38
*** ambro718 has joined #litex17:37
*** tweakoz has joined #litex17:38
*** ambro718 has quit IRC19:13
*** Dolu has quit IRC20:40
*** Dolu has joined #litex20:52
*** futarisIRCcloud has joined #litex22:04

Generated by 2.13.1 by Marius Gedminas - find it at!