pub fn check_write(
limits: WriteLimits,
already_written: u64,
bytes: u64,
available: Option<u64>,
) -> WriteVerdictExpand description
Whether a write of bytes may proceed.
available is what the filesystem reports, or None when it could not be
measured. An unmeasurable filesystem allows the write: a guard that
cannot see has nothing to say, and refusing on it would block every write on
any filesystem the probe cannot read. The other two ceilings still apply.
already_written counts the run so far, excluding this call.
Checked in that order on purpose. Running out of disk is the only one that harms anything outside this run, so it is reported first when more than one applies - a user reading “over the per-call limit” would go raise the limit, which is exactly wrong when the real problem is a full disk.