RE: [RFC v2 00/11] QLogic RDMA Driver (qedr) RFC
> > The module parameters is no-go. > We'll remove it.
Re: [RFC v2 00/11] QLogic RDMA Driver (qedr) RFC
On Tue, Sep 20, 2016 at 01:33:25PM +, Elior, Ariel wrote: > > On Tue, Sep 20, 2016 at 01:35:52PM +0300, Ram Amrani wrote: > ... > > > The series adds on top of RFC v1: > > > * a check for all drivers that IB_ACCESS_MW_BIND isn't set for > > > ib_get_dma_mr > > > * relocation of qedr user API to include/rdma/uapi/providers/ > > > * removal of qedr_devlist_local > > > * fixed error handling in qedr_alloc_resources() > > > * configuration of PBL in ib_map_mr_sg() driver implementation, > > > rather than post_send's IB_WR_REG_MR > > > * misc.: placed code in proper patch, fixed a few comments, > > > removed extra parentheses > > > > > > Thanks for everyone which pointed out problems in the driver. > > > > > > Any review/comment is appreciated. > > > > Very nice, > > Any reason why didn't you drop debug module parameter and decided to > > mimic already available kernel core functionality? > > > > You got technical explanations why it is bad idea to use it. If you need > > additional > > voices > > to support my claims, you will find them in thread about VERBOSE flag and > > responses > > from > > Doug, and Dennis. > > > > Thanks > Hi Leon, > The RFC cover letter lists what has been addressed. Debug printouts are not > addressed in V2 as the discussion on that topic is not concluded (more > thoughts from us on debug printouts incoming on the thread). There were many > comments to V1 which are not relevant to debug printouts which are addressed > by V2. We are requesting further comment, hence RFC V2. Rest assured, if it > is the final opinion in the relevant discussion that pr_debug is the way to > go, that's what we'll do. The module parameters is no-go. > Thanks, > Ariel signature.asc Description: PGP signature
RE: [RFC v2 00/11] QLogic RDMA Driver (qedr) RFC
> On Tue, Sep 20, 2016 at 01:35:52PM +0300, Ram Amrani wrote: ... > > The series adds on top of RFC v1: > > * a check for all drivers that IB_ACCESS_MW_BIND isn't set for ib_get_dma_mr > > * relocation of qedr user API to include/rdma/uapi/providers/ > > * removal of qedr_devlist_local > > * fixed error handling in qedr_alloc_resources() > > * configuration of PBL in ib_map_mr_sg() driver implementation, > > rather than post_send's IB_WR_REG_MR > > * misc.: placed code in proper patch, fixed a few comments, > > removed extra parentheses > > > > Thanks for everyone which pointed out problems in the driver. > > > > Any review/comment is appreciated. > > Very nice, > Any reason why didn't you drop debug module parameter and decided to > mimic already available kernel core functionality? > > You got technical explanations why it is bad idea to use it. If you need > additional > voices > to support my claims, you will find them in thread about VERBOSE flag and > responses > from > Doug, and Dennis. > > Thanks Hi Leon, The RFC cover letter lists what has been addressed. Debug printouts are not addressed in V2 as the discussion on that topic is not concluded (more thoughts from us on debug printouts incoming on the thread). There were many comments to V1 which are not relevant to debug printouts which are addressed by V2. We are requesting further comment, hence RFC V2. Rest assured, if it is the final opinion in the relevant discussion that pr_debug is the way to go, that's what we'll do. Thanks, Ariel
Re: [RFC v2 00/11] QLogic RDMA Driver (qedr) RFC
On Tue, Sep 20, 2016 at 01:35:52PM +0300, Ram Amrani wrote: > From: Ram amrani > > This series introduces RoCE RDMA driver for the 579xx RDMA products by Qlogic. > The RDMA support is added as an additional loadable module (qedr) over the > Ethernet qede driver. > The qedr module will support both RoCE and iWarp, although this series only > adds RoCE support. > The qed and qede drivers are enhanced with functionality required for RDMA > support. > > Based on net-next 4c73195edbe3a5d7e14ea549bb261cf35c29f0cc. > > The series adds on top of RFC v1: > * a check for all drivers that IB_ACCESS_MW_BIND isn't set for ib_get_dma_mr > * relocation of qedr user API to include/rdma/uapi/providers/ > * removal of qedr_devlist_local > * fixed error handling in qedr_alloc_resources() > * configuration of PBL in ib_map_mr_sg() driver implementation, > rather than post_send's IB_WR_REG_MR > * misc.: placed code in proper patch, fixed a few comments, > removed extra parentheses > > Thanks for everyone which pointed out problems in the driver. > > Any review/comment is appreciated. Very nice, Any reason why didn't you drop debug module parameter and decided to mimic already available kernel core functionality? You got technical explanations why it is bad idea to use it. If you need additional voices to support my claims, you will find them in thread about VERBOSE flag and responses from Doug, and Dennis. Thanks signature.asc Description: PGP signature