On Sun, Nov 26, 2023 at 10:26:41AM +0000, Greg Kroah-Hartman wrote:
> On Thu, Nov 23, 2023 at 08:00:22PM +0200, Laurent Pinchart wrote:
> > On Thu, Nov 23, 2023 at 05:31:32PM +0000, Greg Kroah-Hartman wrote:
> > > On Thu, Nov 23, 2023 at 07:28:25PM +0200, Laurent Pinchart wrote:
> > > > On Thu, Nov 23, 2023 at 01:53:42PM +0000, Greg Kroah-Hartman wrote:
> > > > > On Thu, Nov 23, 2023 at 03:49:41PM +0200, Laurent Pinchart wrote:
> > > > > > On Thu, Nov 23, 2023 at 01:02:45PM +0000, Greg Kroah-Hartman wrote:
> > > > > > > On Tue, Nov 07, 2023 at 04:51:53AM -0500, Umang Jain wrote:
> > > > > > > > Drop vchiq_log_error() macro which wraps dev_dbg(). Introduce the usage
> > > > > > > > of dev_dbg() directly.
> > > > > > > >
> > > > > > > > Add a new enum vchiq_log_type and log_type() helper to faciliate the
> > > > > > > > type of logging in dev_dbg(). This will help to determine the type of
> > > > > > > > logging("error", "warning", "debug", "trace") to dynamic debug.
> > > > > > > >
> > > > > > > > Signed-off-by: Umang Jain <umang.jain@xxxxxxxxxxxxxxxx>
> > > > > > > > ---
> > > > > > > > .../interface/vchiq_arm/vchiq_arm.c | 54 ++++----
> > > > > > > > .../interface/vchiq_arm/vchiq_connected.c | 6 +-
> > > > > > > > .../interface/vchiq_arm/vchiq_core.c | 126 ++++++++++--------
> > > > > > > > .../interface/vchiq_arm/vchiq_core.h | 23 +++-
> > > > > > > > .../interface/vchiq_arm/vchiq_dev.c | 47 ++++---
> > > > > > > > 5 files changed, 143 insertions(+), 113 deletions(-)
> > > > > > > >
> > > > > > > > diff --git a/drivers/staging/vc04_services/interface/vchiq_arm/vchiq_arm.c b/drivers/staging/vc04_services/interface/vchiq_arm/vchiq_arm.c
> > > > > > > > index 9fb3e240d9de..2cb2a6503058 100644
> > > > > > > > --- a/drivers/staging/vc04_services/interface/vchiq_arm/vchiq_arm.c
> > > > > > > > +++ b/drivers/staging/vc04_services/interface/vchiq_arm/vchiq_arm.c
> > > > > > > > @@ -696,8 +696,8 @@ int vchiq_initialise(struct vchiq_instance **instance_out)
> > > > > > > >
> > > > > > > > instance = kzalloc(sizeof(*instance), GFP_KERNEL);
> > > > > > > > if (!instance) {
> > > > > > > > - vchiq_log_error(state->dev, VCHIQ_CORE,
> > > > > > > > - "%s: error allocating vchiq instance\n", __func__);
> > > > > > > > + dev_dbg(state->dev, "%s: %s: %s: error allocating vchiq instance\n",
> > > > > > > > + log_cat(VCHIQ_CORE), log_type(ERROR), __func__);
> > > > > > >
> > > > > > > All dev_dbg() calls have __func__ in them automatically, you never need
> > > > > > > to duplicate it again as that's redundant :(
> > > > > >
> > > > > > Oh ? I didn't know that, and can't find it in the code. I may be missing
> > > > > > something though. Are you referring to the +f flag for dynamic debug
> > > > > > entries ? It won't work if dynamic debug isn't enabled though, but maybe
> > > > > > we don't care about that ?
> > > > >
> > > > > Yes, the "f" flag is what controls this, and if dynamic debug isn't
> > > > > enabled, you don't get any message here and we don't care about it :)
> > > >
> > > > You do if you #define DEBUG, that's one of the three options for
> > > > dev_dbg() (dynamic debug and no_printk() being the other two). Maybe
> > > > __func__ should be added to the dev_printk() version of dev_dbg() to
> > > > have a consistent behaviour.
> > >
> > > Drivers should NOT be defining DEBUG for anything in the tree, just use
> > > the normal interfaces, as no one will be selecting debug options from
> > > Kconfig. DEBUG is really only good for out-of-tree work.
> >
> > I didn't know that either. Of course '#define DEBUG' shouldn't be merged
> > upstream, but I thought it was supported by the kernel to make that
> > possible during development, as an alternative to dynamic debug. Does it
> > mean we should drop '#define DEBUG' support from dev_dbg() eventually ?
>
> Probably, once all of the "define DEBUG" lines are dropped from the
> kernel itself, which might take a while :)
Would a patch that simply drops them be acceptable ? If not, what's
missing to remove them ?
--
Regards,
Laurent Pinchart
