On 02/16/2017 10:32 AM, Sreekanth Reddy wrote:
> On Tue, Jan 31, 2017 at 2:55 PM, Hannes Reinecke <hare@xxxxxxx> wrote:
>> Cleanup the MSI-X handling allowing us to use
>> the PCI-layer provided vector allocation.
>>
>> Signed-off-by: Hannes Reinecke <hare@xxxxxxxx>
>> ---
>> drivers/scsi/mpt3sas/mpt3sas_base.c | 100 +++++++++++++++++-------------------
>> drivers/scsi/mpt3sas/mpt3sas_base.h | 2 -
>> 2 files changed, 46 insertions(+), 56 deletions(-)
>>
[ .. ]
>>
>> ioc->msix_enable = 1;
>> - for (i = 0, a = entries; i < ioc->reply_queue_count; i++, a++) {
>> - r = _base_request_irq(ioc, i, a->vector);
>> + ioc->msix_vector_count = r;
>
> Here we should also reinitialize ioc->reply_queue_count if
> pci_alloc_irq_vectors returns with value less than requested MSIx
> vector count (i.e. ioc->reply_queue_count).
>
Sigh. The usual confusion between 'reply_queue_count' and
'msix_vector_count'.
I have _tried_ to make 'reply_queue_count' the number of _allocated_
reply queues, and 'msix_vector_count' the number of MSI-x vectors allocated.
As pci_alloc_irq_vectors might be returning less vectors than specified
(but we don't modify the allocation for reply_queue_count in that case)
I considered it an error to modify reply_queue_count, too.
But I'd _very_ much appreciate if we could clear up the confusion and
fix the meaning of 'reply_queue_count' and 'msix_vector_count'.
ATM I don't think we have a consistent usage here.
Cheers,
Hannes
--
Dr. Hannes Reinecke Teamlead Storage & Networking
hare@xxxxxxx +49 911 74053 688
SUSE LINUX GmbH, Maxfeldstr. 5, 90409 Nürnberg
GF: F. Imendörffer, J. Smithard, J. Guild, D. Upmanyu, G. Norton
HRB 21284 (AG Nürnberg)
