Re: [Qemu-block] [Qemu-devel] [PATCH 4/5] atapi: call ide_set_irq before ide_transfer_start

[Paolo Bonzini]

On 21/03/2018 01:35, John Snow wrote:
>> The ATAPI_INT_REASON_IO interrupt is raised when I/O starts, but in the
>> AHCI case ide_set_irq was actually called at the end of a mutual recursion.
>> Move it early, with the side effect that ide_transfer_start becomes a tail
>> call in ide_atapi_cmd_reply_end.
>>
> Do you know in which spec it specifies that we should interrupt?
> 
> Seems okay, but I wanted to double check the wording in the spec (was
> looking in scsi MMC and ATA8) and couldn't find it quickly.

I have an "Interrupt Reason" field in my copy of the ATA/ATAPI spec:

---
6.4.3 Input/Output (I/O) bit
The Input/Output bit shall be cleared to zero if the transfer is to the
device. The Input/Output bit shall be set to one if the transfer is to
the host. In ATA/ATAPI-7 this bit was documented as the I/O bit.
---

I suppose the commit message needs some improvement---the point is that
since this is a READ, the I/O bit must be set before we start the
transfer.  It can be done after the disk I/O starts, but it must be done
before the PIO transfer starts; in other words, the bug is only there
for AHCI.

Paolo