Re: [Qemu-devel] [PATCH] ahci.c: mask the interrupt on complete flag to allow ahci.c to read the correct size for the PRDT

[Alexander Graf <agraf@suse.de>]

On 27.06.14 17:23, John Snow wrote:
>
> On 06/26/2014 07:28 PM, Reza Jelveh wrote:
>> +#define AHCI_PRDT_SIZE_MASK                0x3fffff
>>
> out of rampant curiosity, is there ever a case where the lower bits 
> might be set and the mask 0x3fffc is not desirable, or can we always 
> trust those bits to simply be off anyway?

We can't really trust anything from an OS :). But the reason for this 
patch is that PRDT.I was set on some entries to enable notification of 
the OS when the entry has been successfully processed.

We currently don't emulate the I bit correctly, but get away without 
doing so. However, we have to make sure we mask it out when interpreting 
the values. Hence the mask.

I do agree that this should have been in the patch description. Reza, 
could you please repost this with a proper patch description and as a 
checkpatch.pl compliant patch? Also please CC me on the next iteration :).


Alex

PS: Is your name really John Snow? That is so cool!