From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jason Gunthorpe Subject: Re: [PATCH rdma-next] RDMA/srpt: Fix RCU debug build error Date: Tue, 23 Jan 2018 08:21:42 -0700 Message-ID: <20180123152142.GB30619@mellanox.com> References: <20180123115013.12213-1-leonro@mellanox.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: <20180123115013.12213-1-leonro-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org> Sender: linux-rdma-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org To: Leon Romanovsky , "Paul E. McKenney" Cc: Bart Van Assche , Doug Ledford , linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org List-Id: linux-rdma@vger.kernel.org On Tue, Jan 23, 2018 at 01:50:13PM +0200, Leon Romanovsky wrote: > Combination of CONFIG_DEBUG_OBJECTS_RCU_HEAD=y and > CONFIG_INFINIBAND_SRPT=m produces the following build error. > > ERROR: "init_rcu_head" [drivers/infiniband/ulp/srpt/ib_srpt.ko] undefined! > make[1]: *** [scripts/Makefile.modpost:92: __modpost] Error 1 > make: *** [Makefile:1216: modules] Error 2 > > The reason to it that init_rcu_head() is not exported and not supposed > to be used in modules. The correct function to use is init_rcu_head_on_stack(). > > Fixes: 795bc112cd5a ("IB/srpt: Make it safe to use RCU for srpt_device.rch_list") > Fixes: a11253142e6d ("IB/srpt: Rework multi-channel support") > Signed-off-by: Leon Romanovsky > drivers/infiniband/ulp/srpt/ib_srpt.c | 4 ++-- > 1 file changed, 2 insertions(+), 2 deletions(-) > > diff --git a/drivers/infiniband/ulp/srpt/ib_srpt.c b/drivers/infiniband/ulp/srpt/ib_srpt.c > index bf37816a1b12..65bbc3dbf0fd 100644 > +++ b/drivers/infiniband/ulp/srpt/ib_srpt.c > @@ -1950,7 +1950,7 @@ static struct srpt_nexus *srpt_get_nexus(struct srpt_port *sport, > nexus = ERR_PTR(-ENOMEM); > break; > } > - init_rcu_head(&tmp_nexus->rcu); > + init_rcu_head_on_stack(&tmp_nexus->rcu); The comments for init_rcu_head_on_stack say it should only be used for stack allocations and tmp_nexus is not a stack allocation. Removing the init_rcu_head entirely seems like the right thing to do, and maybe that function is poorly named? /* * init_rcu_head_on_stack()/destroy_rcu_head_on_stack() are needed for dynamic * initialization and destruction of rcu_head on the stack. rcu_head structures * allocated dynamically in the heap or defined statically don't need any * initialization. */ The commit that introduced init_rcu_head is 546a9d8519ed137b2804a3f5a3659003039dd49c which suggests it exists only to help the slab allocators, although curiously the allocators never call it these days.. Maybe Paul can confirm? Jason -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org More majordomo info at http://vger.kernel.org/majordomo-info.html