You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Commit 219be4f
Browse filesBrowse the repository at this point in the historyBrowse files
test(cuda.core): cover the forkserver start method for VMM IPC buffers
The forkserver is a fresh interpreter and children fork from it, so it
is safe after CUDA initialization, and it is the Linux default since
Python 3.14. Its Popen collects file descriptors during pickling and
sends them afterwards, like spawn, so the Process-argument path in
VirtualMemoryBuffer.__reduce__ has to work there too. Add a Process
argument test and a Queue round trip under mp.get_context("forkserver"),
skipped where that start method is unavailable.
Closes#3111
Co-Authored-By: Claude Fable 5.1 <[email protected]>
0 commit comments