[Libre-soc-dev] daily kan-ban update 14jun2021

[Luke Kenneth Casson Leighton]

tobias you need to be very caregul here, the microwatt cache assumes a
1 cycle delay between address presented and data presented.

we are NOT going to alter the microwatt cache code so please do not
try, simply make sure that all unit tests and loadstore.py take that
correctly into account.

l.



On 6/14/21, Tobias Platen <libre-soc at platen-software.de> wrote:
> On Mon, 2021-06-14 at 19:43 +0200, Tobias Platen wrote:
>> today: continuing to work on mmu/dcache.
>>
>> Recently I found a bug in src/soc/fu/ldst/loadstore.py that causes
>> wrong data to be written into the cache.
> Fixed the test case, now I am looking for a proper solution for
> set_wr_data which needs to latch the data in, as it is written
> to the cache later. Just using a comb assignment will fail here.
>
> def set_wr_data(self, m, data, wen):
>         # do the "blip" on write data
>         m.d.comb += self.d_valid.eq(1)
>         # put data into comb which is picked up in main elaborate()
>         m.d.comb += self.d_w_valid.eq(1)
>         # FIXME latch data in
>         m.d.comb += self.store_data.eq(data)
>         m.d.sync += self.store_data_sync.eq(data)
>
>         st_ok = self.done # TODO indicates write data is valid
>         return st_ok
>>
>> _______________________________________________
>> Libre-soc-dev mailing list
>> Libre-soc-dev at lists.libre-soc.org
>> http://lists.libre-soc.org/mailman/listinfo/libre-soc-dev
>
>
> _______________________________________________
> Libre-soc-dev mailing list
> Libre-soc-dev at lists.libre-soc.org
> http://lists.libre-soc.org/mailman/listinfo/libre-soc-dev
>