Message ID | 20190331161024.31035-1-shamir.rabinovitch@oracle.com (mailing list archive) |
---|---|
Headers | show |
Series | IB/{hw,sw}: remove 'uobject->context' dependency APIs | expand |
On Sun, Mar 31, 2019 at 07:10:02PM +0300, Shamir Rabinovitch wrote: > This patch set continue the cleanup started with prior series. This > patch set clean the ib_x destroy APIs. > > The final goal of this cleanup is to remove the dependency in the IB > code in the ib_x->uobject pointer as step toward shared ib_x objects. > > This series also remove ib_ucontext from functions signature where > ib_udata is available. > > Changelog: > > v2 -> v3: > * Fix HNS compile > > v1 -> v2: > * rebase to latest rdma/for-next > * Patch 1/5: > Replace the ufile with attrs->ufile > ufd_get_read - pass attrs not NULL > * Patch 2/5: > uverbs_close_fd now pass valid uverbs_attr_bundle > __uverbs_cleanup_ufile loop optimize as Jason suggested > * Patch 3/5: > Use static inline instead of macros > * Patch 5/5: > New patch to clean-up the functions prototypes that > had duplicate ib_ucontext & ib_udata > > Shamir Rabinovitch (5): > IB/{rdma,core}: ucontext should be set properly for all cmd & ioctl > paths > IB/{core,hw}: pass uverbs_attr_bundle down uobject destroy path > IB/{core,hw,sw}: pass uverbs_attr_bundle down ib_x destroy path > IB:{core,hw}: remove 'uobject->context' dependency in object destroy > APIs > IB/{core,hw,sw}: pass only ib_udata in function prototypes I fixed the things I noted, did some clang-format and applied it to for-next, thanks Please send a folowup patch fixing the attr.context not being set during get_context Jason
On Mon, Apr 01, 2019 at 04:32:18PM -0300, Jason Gunthorpe wrote: > On Sun, Mar 31, 2019 at 07:10:02PM +0300, Shamir Rabinovitch wrote: > > This patch set continue the cleanup started with prior series. This > > patch set clean the ib_x destroy APIs. > > > > The final goal of this cleanup is to remove the dependency in the IB > > code in the ib_x->uobject pointer as step toward shared ib_x objects. > > > > This series also remove ib_ucontext from functions signature where > > ib_udata is available. > > > > Changelog: > > > > v2 -> v3: > > * Fix HNS compile > > > > v1 -> v2: > > * rebase to latest rdma/for-next > > * Patch 1/5: > > Replace the ufile with attrs->ufile > > ufd_get_read - pass attrs not NULL > > * Patch 2/5: > > uverbs_close_fd now pass valid uverbs_attr_bundle > > __uverbs_cleanup_ufile loop optimize as Jason suggested > > * Patch 3/5: > > Use static inline instead of macros > > * Patch 5/5: > > New patch to clean-up the functions prototypes that > > had duplicate ib_ucontext & ib_udata > > > > Shamir Rabinovitch (5): > > IB/{rdma,core}: ucontext should be set properly for all cmd & ioctl > > paths > > IB/{core,hw}: pass uverbs_attr_bundle down uobject destroy path > > IB/{core,hw,sw}: pass uverbs_attr_bundle down ib_x destroy path > > IB:{core,hw}: remove 'uobject->context' dependency in object destroy > > APIs > > IB/{core,hw,sw}: pass only ib_udata in function prototypes > > I fixed the things I noted, did some clang-format and applied it to > for-next, thanks > > Please send a folowup patch fixing the attr.context not being set > during get_context > > Jason Thanks Jason! Will send the follow up patch.