-
-
Notifications
You must be signed in to change notification settings - Fork 78
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Segmentation fault in tests on 32-bit architectures (i686 and arm) #223
Comments
For reference, the Sundials build is: https://github.com/JuliaDiffEq/SundialsBuilder/releases/tag/sundials-3.1.1.6 . I assume that segfault means we need to fix that binary builder repo build script. |
Is there a minimal example I can try? It could even be Julia itself crashing. |
using Sundials
function lorenz(du,u,p,t)
du[1] = 10.0(u[2]-u[1])
du[2] = u[1]*(28.0-u[3]) - u[2]
du[3] = u[1]*u[2] - (8/3)*u[3]
end
u0 = [1.0;0.0;0.0]
tspan = (0.0,100.0)
prob = ODEProblem(lorenz,u0,tspan)
sol = solve(prob,CVODE_Adams())
using Plots; plot(sol,vars=(1,2,3)) is what I'd try. |
That example works fine. It's crashing here (which I can verify manually and is what the failing test above showed as well): |
Isn't it the Yggdrasil one? https://github.com/JuliaPackaging/Yggdrasil/tree/master/S/Sundials |
Off topic, but we should just build KLU as part of the regular SuiteSparse build, and simplify the Sundials build. |
This Sundials build was the first BinaryBuilder setup, and it hasn't really been modified since. So anything like Yggdrasil isn't apart of this unless that happened outside of these repos. No BLAS here, that's an open issue. And getting KLU out of this build would indeed be nice. |
I feel like we should get the build upgraded first. KLU would be next. But not sure if any of this would help fix the crash unless it is happening due to some bad combination of things being done in the old way and it is only triggered on arm. |
We might as well update to Sundials 5.0 at the same time too. It's an easy enough API change. It mostly just needs someone who wants to make the binaries build |
@ChrisRackauckas Should I try this again now? Will it pull Sundials 5 and the new changes? I don't have any use for this myself, but just figured I can close htis out if if it works. |
Yes, it should just test it on Sundials 5. |
Looks like an error in the same place. Using Julia 1.4 and Sundials 5 through artifacts.
|
This fails now.
|
I get a failure using 32-bit i686 Julia as well, which suggests that this failure is not arm specific:
|
@jd-lara Any thoughts? |
It seems that the I would test the basic tests first to see if the problem is in the DiffEqCommon interface set-up of the problem |
I just tried running the Sundials tests on Raspberry Pi and they failed with 1.3. Just reporting here for the record.
The text was updated successfully, but these errors were encountered: