Can the “BOOLEAN wait” parameter to AcquireForLazyWrite and
AcquireForReadAhead ever come in as FALSE. Since, these callbacks are
always invoked from the top of a system thread, is it safe to assume that
they can always block (for resources). Or can under certain conditions (like
low resources), the system expect TRY semantics in these 2 callbacks.
Thanks,
Vipul.
FastFat handles the case where Wait is FALSE, therefore so should you. I
have personally seen callbacks of other varieties come down as can’t
wait, so it’s a safe assumption that these may be invoked likewise.
Vipul wrote:
Can the “BOOLEAN wait” parameter to AcquireForLazyWrite and
AcquireForReadAhead ever come in as FALSE. Since, these callbacks are
always invoked from the top of a system thread, is it safe to assume that
they can always block (for resources). Or can under certain conditions (like
low resources), the system expect TRY semantics in these 2 callbacks.
Thanks,
Vipul.
–
Nick Ryan (MVP for DDK)