Re: [PATCH 1/2] staging: vme_user: Replace printk() with pr_*(),dev_*()

[Soumya Negi]

On Thu, Oct 19, 2023 at 09:42:26PM +0200, Greg Kroah-Hartman wrote:
> On Thu, Oct 19, 2023 at 12:06:18PM -0700, Soumya Negi wrote:
> > On Thu, Oct 19, 2023 at 05:34:01PM +0200, Greg Kroah-Hartman wrote:
> > > On Wed, Oct 18, 2023 at 12:38:56PM -0700, Soumya Negi wrote:
> > > > Hi Greg,
> > > > 
> > > > On Wed, Oct 18, 2023 at 03:26:07PM +0200, Greg Kroah-Hartman wrote:
> > > > > On Tue, Oct 17, 2023 at 09:36:32PM -0700, Soumya Negi wrote:
> > > > > > vme.c uses printk() to log messages. To improve and standardize message
> > > > > > formatting, use logging mechanisms pr_err()/pr_warn() and
> > > > > > dev_err()/dev_warn() instead. Retain the printk log levels of the
> > > > > > messages during replacement.
> > > > > > 
> > > > > > Issue found by checkpatch.pl
> > > > > > 
> > > > > > Signed-off-by: Soumya Negi <soumya.negi97@xxxxxxxxx>
> > > > > > ---
> > > > > >  drivers/staging/vme_user/vme.c | 175 ++++++++++++++++++---------------
> > > > > >  1 file changed, 94 insertions(+), 81 deletions(-)
> > > > > > 
> > > > > > diff --git a/drivers/staging/vme_user/vme.c b/drivers/staging/vme_user/vme.c
> > > > > > index 6519a7c994a0..e8c2c1e77b7d 100644
> > > > > > --- a/drivers/staging/vme_user/vme.c
> > > > > > +++ b/drivers/staging/vme_user/vme.c
> > > > > > @@ -9,6 +9,8 @@
> > > > > >   * Copyright 2004 Motorola Inc.
> > > > > >   */
> > > > > >  
> > > > > > +#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
> > > > > 
> > > > > No, this is a driver, as others have pointed out, always use dev_*()
> > > > > calls instead.
> > > > 
> > > > Some of the pr_ fns can be dev_, but I don't think all can.
> > > > e.g. device NULL-check error messages
> > > 
> > > I would argue that those are pointless and can be removed and also the
> > > check is probably not needed either.
> > 
> > Got it. The pr_() in find_bridge() can't be converted to dev_ so I'll remove 
> > the message entirely in another patch.
> > 
> > I understand that the device-NULL checks should be done on the caller's side. 
> > Since empty devices would mean something went wrong, would it be better to
> > put in an assertion(..WARN_ON) when removing the check? 
> 
> WARN_ON() means "I have no idea what can happen here so I give up",
> which is not a good idea in kernel development.  If that every hits,
> then your machine will reboot as the huge majority of all Linux systems
> in the world run with panic-on-warn enabled.
> 
> If it is impossible for something to happen (i.e. you control all
> callers) then just do not check for it.  If it happens, you will get a
> NULL-dereference which is the same as a WARN_ON() in a way.
> 
> No new WARN_ON() should ever be added to the kernel, especially in a
> driver.  Handle the condition if it is possible to be hit.  If it can
> never be hit, don't even check it.
> 
> thanks,
> 
> greg k-h

Hi Greg,

Thank you for explaining in detail. I'll remove the device NULL-checks
completely.

Regards,
Soumya