<zyp>
it's making my build fail with ValueError: Multiple submodules with local clock domains cannot be anonymous
AndrewD has joined #litex
<AndrewD>
tnt: thanks, I assumed there must be a reason for the pattern but using litex is my first reason to take python seriously. I've discovered I quite like it!
C-Man has joined #litex
AndrewD has quit [Ping timeout: 250 seconds]
tucanae47_ has quit [Quit: Connection closed for inactivity]
lexano has joined #litex
<_florent_>
zyp: ok, I'll see if I can do this differently
<tnt>
_florent_: btw, tried again with updated litex/litepci still same issue in x8 :/
<tnt>
So I if you have the litescope code to probe the inside, I can try that next.
<_florent_>
and trigger on m_axis_rc.valid before starting the dma_test
<_florent_>
to see if you the DMA Reader receives the first completions. And once stuck, if you can do another capture to see if that's rc channel that is stuck (valid=1, ready=0)
<tnt>
Ack.
<_florent_>
zyp: not very clean, but this should fix your build. I'll probably have another look at it.
<Melkhior>
hello; what would be the 'recommended' way to clear (set all words to 0) a Memory() object? I find myself in need of clearing the tag memory of a wishbone cache to "flush" it but not sure how to do that (no lines need eviction as I converted it from write-back to write-through/write-no-allocate)
<Melkhior>
thanks
<Melkhior>
(I tried a ResetInserter() but those don't seem to work on Memory ?)
<_florent_>
tnt: ok, thanks. So it seems to start correctly and nothing indicates that a channel is blocked when it's hanging.
<_florent_>
tnt: it would be interesting to add these signals:
<_florent_>
Melkhior: A Block RAM does not have a reset pin, you'll have to implement your own logic to do clear the sections you want
<_florent_>
Melkhior: You could share one of the port of the Memory for this, or add another specific. But if you already have 2 ports, adding another will prevent the use of BlockRAM.
<Melkhior>
@_florent_ OK thanks! it should only have one port (it's basically the tag_mem from the wishbone cache) so I could add a loop to reset all tag bits to zero
<Melkhior>
(or I could try and do some snooping to only invalidates lines that have been write-touched by the host... but that's probably harder)
sebo has joined #litex
sebo has quit [Ping timeout: 252 seconds]
Martoni has joined #litex
Martoni has quit [Ping timeout: 260 seconds]
_franck_ has quit [Quit: Ping timeout (120 seconds)]