Performance impact of workaround for ERR009598 (SATA: PRD not flushed ...)

[piotrfigiel]

Hi,

 we are now porting kernel tree linux-fslc-imx 4.1 for one of our boards, upgrading from kernel 3.10. Upon switch we observed noticeable performance degradation on SATA interface. It's been quickly traced to be caused by commit implementing workaround for ERR009598.

 Errata description suggests that only ATAPI and ATA devices implementing streaming feature are affected. I wonder is that correct (complete)?

 We haven't observed the issue with devices used here (which don't implement stream commands) on kernel 3.10 and we consider reverting the workaround for that errata in our tree.

[igorpadykov]

Hi Piotr

please check

https://community.nxp.com/message/859275

Best regards
igor

================================================================
Best regards
igor
-----------------------------------------------------------------------------------------------------------------------
Note: If this post answers your question, please click the Correct Answer button. Thank you!
-----------------------------------------------------------------------------------------------------------------------

[piotrfigiel]

Hi igor, thanks for reply.

 I read referenced thread though I'm more interested in conditions which can trigger the workaround to be relevant. Basing only on the errata description, would it be correct to enable the workaround conditionally - i.e. only when the connected non-ATAPI ATA drive supports streaming feature (as indicated by IDENTIFY DEVICE) or at the moment the feature becomes enabled? Do you foresee any changes in the workaround for errata to improve performance in official IMX tree?

Best regards,

Piotr.