[Libre-soc-dev] daily kan-ban update 06jan2021

Luke Kenneth Casson Leighton lkcl at lkcl.net
Thu Jan 6 17:29:22 GMT 2022


On Thu, Jan 6, 2022 at 4:43 PM Tobias Platen
<libre-soc at platen-software.de> wrote:
>
> On Thu, 2022-01-06 at 14:20 +0000, Luke Kenneth Casson Leighton wrote:
> >  it looks like this will have to be added: it is quite
> > a lot of work, to get LoadStore1 to support double-LD/STs.
> I can do part of that work

that'd be great.  deep breath: i really *really* do not like branches,
but in this case it's probably a good justification to use one, because
now the loadstore is crucial to be working and i can see this being
quite a bit of work.

it involves adding SECOND_REQ from loadstore1.vhdl, but that in
turn also means messing with the data, dealing with byte-reversal,
sign-extension and so on.

so can you start a branch of soc.git and go from there?

On Thu, Jan 6, 2022 at 4:49 PM Jacob Lifshay <programmerjake at gmail.com> wrote:

> I came up with a possible workaround on irc, disable microwatt's support
> for unaligned load/store (maybe 1 line change), then see if linux correctly
> sw emulates unaligned memory accesses, that way the instruction trace from
> libre-soc will match microwatt and testing can proceed.
>
> https://libre-soc.org/irclog/%23libre-soc.2022-01-06.log.html#t2022-01-06T16:07:07

yes i like this idea, it would save a lot of work in the diff comparisons.
for now i am just going to trust that the linux kernel 0x600 exception
handler is capable of doing its job, until i see some evidence to the
contrary [in the massive 13,000,000 line-long log files]

l.



More information about the Libre-soc-dev mailing list