James Bottomley writes: > On Mon, 2005-05-16 at 10:06 +0200, Arjan van de Ven wrote: >> > + spin_lock( instance->host_lock ); >> > + cmd->scmd->scsi_done( cmd->scmd ); >> > + spin_unlock( instance->host_lock ); >> >> are you really sure you don't want to use spin_lock_irqsave() here ? > >Actually, don't bother with the lock at all. scsi_done() is designed to >be called locklessly. Could I get an historical (2.4 & Distribution) perspective on this. At which point, or what code/include/manifest/version delineating it, would you say the driver is no longer, if ever, required to place a lock (host_lock or io_request_lock) around the scsi_done call? I expect (or hope) the answer to be: always needs io_request_lock in 2.4, never needed the host_lock in 2.5+. -- Mark Salyzyn - : send the line "unsubscribe linux-scsi" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
