[libre-riscv-dev] [Bug 178] first coriolis2 tutorial, workflow and "test project" page

[bugzilla-daemon at libre-riscv.org]

http://bugs.libre-riscv.org/show_bug.cgi?id=178

--- Comment #88 from Luke Kenneth Casson Leighton <lkcl at lkcl.net> ---
(In reply to Jean-Paul.Chaput from comment #87)
> (In reply to Luke Kenneth Casson Leighton from comment #86)
> > (In reply to Jean-Paul.Chaput from comment #80)
> > > (In reply to Luke Kenneth Casson Leighton from comment #78)
> > > > yuk! :)
> > > > 
> > > > i am using yosys as of yesterday, now.  can you do yosys -V and let me know?
> > > > 
> > > > i can then do binary search compiling different versions, looking for the
> > > > commit that fails.  i have ccache installed and a stupidly high resource
> > > > laptop.  64GB DDR4 RAM, 8 core i9, 2TB NVMe 2500mbytes/sec muhahaha
> > > > 
> > > > if you let me know the yosys version you use it sets a lower bound on the
> > > > git version to start from.
> > > 
> > > I did use this one, less than a week ago.
> > > 
> > > https://github.com/cliffordwolf/yosys/archive/yosys-0.9.tar.gz
> > 
> > ok that appears to be git tagged:
> > 
> > commit 1979e0b1f2482dbf0562f5116ab444280a377773
> > Author: Clifford Wolf <clifford at clifford.at>
> > Date:   Mon Aug 26 10:37:53 2019 +0200
> > 
> >     Yosys 0.9
> >     
> >     Signed-off-by: Clifford Wolf <clifford at clifford.at>
> > 
> > 
> > although it is easier to track down / confirm with "yosys -V".  hmmm, i
> > don't think it will report the version correctly though if built from a .tgz
> > 
> > i'll *assume* the git tag yosys-0.9 was where that tarball came from.
> 
> Hello Luke,
> 
> Yes, that's the right one and, yes when built from a tar.gz it do not
> report it's version correctly. I've build both versions.

just tried it: no repro yet, because *sigh* i've moved on, on both
ieee754fpu and the experiments/* trying to randomly "fix" things over
the past 2 days.

> BUT, I was half wrong about the bug. If I persist to think that the
> blif file is incorrect, this was not causing the lvx error.

ah interesting.

> I'm correcting the Coriolis blif parser to account for the righ one.

ok.  and that's from comment #77


> Strange thing is that now the P&R always fail, whatever the version.
> I must have made something wrong in my previous tests (some cached
> file perhaps). 

i have found that sometimes even "make clean" and rm * and git reset --hard
will result, about 1/40 times, the router or something else going awry.

> Sorry for the wrong lead.

if you did a "git pull" on either soclayout or ieee754fpu then there
is *another* error causing build failures (here as well).

i must apologise that the designs, even the "simple" ones, are large
enough that creating even a small repro case is itself very tricky.

-- 
You are receiving this mail because:
You are on the CC list for the bug.