On Wed, Apr 17, 2019 at 09:41:11AM +0300, Gal Pressman wrote: > On 16-Apr-19 19:49, Dennis Dalessandro wrote: > > On 4/15/2019 8:20 AM, Gal Pressman wrote: > >> diff --git a/lib/dynamic_debug.c b/lib/dynamic_debug.c > >> index 7bdf98c37e91..dfcf6cfa1c70 100644 > >> +++ b/lib/dynamic_debug.c > >> @@ -37,6 +37,8 @@ > >> #include <linux/device.h> > >> #include <linux/netdevice.h> > >> +#include <rdma/ib_verbs.h> > >> + > >> extern struct _ddebug __start___verbose[]; > >> extern struct _ddebug __stop___verbose[]; > >> @@ -636,6 +638,44 @@ EXPORT_SYMBOL(__dynamic_netdev_dbg); > >> #endif > >> +#if IS_ENABLED(CONFIG_INFINIBAND) > >> + > >> +void __dynamic_ibdev_dbg(struct _ddebug *descriptor, > >> + const struct ib_device *ibdev, const char *fmt, ...) > >> +{ > >> + struct va_format vaf; > >> + va_list args; > >> + > >> + BUG_ON(!descriptor); > >> + BUG_ON(!fmt); > > > > I don't think BUG_ON is going to be acceptable. > > These are the exact same BUG_ON's that are present in __dynamic_pr_dbg, > __dynamic_dev_dbg and __dynamic_netdev_dbg. > I prefer the ibdev variation to be consistent with its counterparts. > Jason, do you prefer to remove them? Linus has yelled at RDMA for adding them, so if this goes through the rdma tree (which I prefer) then it must be removed, or someone who owns that file must stand up to say they are needed.. Jason