wroclaw-git-backups/nix
1
0
Fork 0
mirror of https://github.com/NixOS/nix synced 2025-06-25 14:51:16 +02:00
Code Activity

MonitorFdHup: Don't sleep anymore

Browse source 
After the previous commit it should not be necessary. Furthermore, if we
*do* sleep, we'll exacerbate a race condition (in conjunction with
getting rid of the thread cancellation) that will cause test failures.
This commit is contained in:
John Ericson 2025-03-23 19:11:17 -04:00
parent 9b3352c3c8
commit 49f486d8e0
1 changed files with 18 additions and 5 deletions
Download patch file Download diff file Expand all files Collapse all files

23
src/libutil/unix/monitor-fd.hh
View file

@ -98,11 +98,24 @@ public:
if (fds[1].revents & POLLHUP) { if (fds[1].revents & POLLHUP) {
break; break;
} }
// On macOS, it is possible (although not observed on macOS // On macOS, (jade thinks that) it is possible (although not
// 14.5) that in some limited cases on buggy kernel versions, // observed on macOS 14.5) that in some limited cases on buggy
// all the non-POLLHUP events for the socket get delivered. // kernel versions, all the non-POLLHUP events for the socket
// Sleeping avoids pointlessly spinning a thread on those. // get delivered.
sleep(1); //
// We could sleep to avoid pointlessly spinning a thread on
// those, but this opens up a different problem, which is that
// if do sleep, it will be longer before the daemon fork for a
// client exits. Imagine a sequential shell script, running Nix
// commands, each of which talk to the daemon. If the previous
// command registered a temp root, exits, and then the next
// command issues a delete request before the temp root is
// cleaned up, that delete request might fail.
//
// Not sleeping doesn't actually fix the race condition --- we
// would need to block on the old connections' tempt roots being
// cleaned up in in the new connection --- but it does make it
// much less likely.
} }
}); });
}; };