Hi, >> > As mentioned, the flush requires immediately after reading the CI. >> > Otherwise, there is still an chance that the command is completed and >> > the OS notified the upper layer while the data is still in flight. For >> > the initial version, I will remove the flush (IRQ wrapper) and submit >> > separate patch. >> >> The function is called with ap->lock held. Nothing can happen >> inbetween. > > Oh, or do you mean that upper layer may get data which hasn't finish > transferring without the flush? > In the ISR, the AHCI library code reads the CI register and then performs XOR to determine which commands are completed. Then it goes and processes the completed command(s). I am worry that the process of processing the completed command(s), the upper layer may act on the data before the data arrived at the DDR/cache. In answering your question, yes. -Loc. -- To unsubscribe from this list: send the line "unsubscribe devicetree" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
