(In reply to Kershaw Chang [:kershaw] from comment #17) > The `PR_POLL_NVAL` error appears to come from this [check](https://searchfox.org/mozilla-central/rev/5aa2a9f1de4e513c0c979e1ee019c014fc43a24c/nsprpub/pr/src/md/unix/unix.c#3420) that treats a socket as invalid if its osfd is greater than `FD_SETSIZE`. In Firefox, FD_SETSIZE is defined as 4096. What is the indication of that line being reached? Before it even gets to that point, the fd is skipped by [an `osfd >= FD_SETSIZE` check](https://searchfox.org/mozilla-central/rev/5aa2a9f1de4e513c0c979e1ee019c014fc43a24c/nsprpub/pr/src/md/unix/unix.c#3375-3377,3420). > From the system trace Rob provided, all the failed sockets had file descriptors greater than 4096. Regardless of whether the code pointer is accurate, I think that you may be on something.
Bug 1980171 Comment 19 Edit History
Note: The actual edited comment in the bug view page will always show the original commenter’s name and original timestamp.
(In reply to Kershaw Chang [:kershaw] from comment #17) > The `PR_POLL_NVAL` error appears to come from this [check](https://searchfox.org/mozilla-central/rev/5aa2a9f1de4e513c0c979e1ee019c014fc43a24c/nsprpub/pr/src/md/unix/unix.c#3420) that treats a socket as invalid if its osfd is greater than `FD_SETSIZE`. In Firefox, FD_SETSIZE is defined as 4096. What is the indication of that line being reached? Before it even gets to that point, the fd is skipped by [an `osfd >= FD_SETSIZE` check](https://searchfox.org/mozilla-central/rev/5aa2a9f1de4e513c0c979e1ee019c014fc43a24c/nsprpub/pr/src/md/unix/unix.c#3375-3377,3420). EDIT: Those are two separate loops, it does indeed look plausible. > From the system trace Rob provided, all the failed sockets had file descriptors greater than 4096. Regardless of whether the code pointer is accurate, I think that you may be on something.