[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <CAJfuBxwubk2DatnVrHU1C6WiYNKW1CcuQZe29vToMov4Ch0dHA@mail.gmail.com>
Date: Sat, 1 May 2021 20:49:35 -0600
From: jim.cromie@...il.com
To: Jason Baron <jbaron@...mai.com>
Cc: Greg KH <gregkh@...uxfoundation.org>,
LKML <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH 1/2] dyndbg: avoid calling dyndbg_emit_prefix when it has
no work
On Thu, Apr 29, 2021 at 3:03 PM Jason Baron <jbaron@...mai.com> wrote:
>
> Hi Jim,
>
> On 4/27/21 9:00 PM, Jim Cromie wrote:
> > Wrap function in a static-inline one, which checks flags to avoid
> > calling the function unnecessarily.
> >
> > Signed-off-by: Jim Cromie <jim.cromie@...il.com>
> > ---
> > include/linux/dynamic_debug.h | 9 +++++++++
> > lib/dynamic_debug.c | 9 ++++++++-
> > 2 files changed, 17 insertions(+), 1 deletion(-)
> >
> > diff --git a/include/linux/dynamic_debug.h b/include/linux/dynamic_debug.h
> > index a57ee75342cf..173535e725f7 100644
> > --- a/include/linux/dynamic_debug.h
> > +++ b/include/linux/dynamic_debug.h
> > @@ -32,6 +32,15 @@ struct _ddebug {
> > #define _DPRINTK_FLAGS_INCL_FUNCNAME (1<<2)
> > #define _DPRINTK_FLAGS_INCL_LINENO (1<<3)
> > #define _DPRINTK_FLAGS_INCL_TID (1<<4)
> > +
> > +#define _DPRINTK_FLAGS_INCL_ANYSITE \
> > + (_DPRINTK_FLAGS_INCL_MODNAME \
> > + | _DPRINTK_FLAGS_INCL_FUNCNAME \
> > + | _DPRINTK_FLAGS_INCL_LINENO)
> > +#define _DPRINTK_FLAGS_INCL_ANY \
> > + (_DPRINTK_FLAGS_INCL_ANYSITE \
> > + | _DPRINTK_FLAGS_INCL_TID)
> > +
>
> I'm not sure it's worth having an unused define here by dynamic_debug.c.
>
> I would prefer to just have _DPRINTK_FLAGS_INCL_ANY that has all the flags
> in a single define.
I have another patch to emit_prefix that uses !ANYSITE to return early
if only TID is wanted
but I can split that out,
or maybe I can pull the other patch forward out of the dd-diet-plan
set Im working.
>
> hmmm - looking at __dynamic_emit_prefix() it starts by doing:
>
>
> 589 static char *dynamic_emit_prefix(const struct _ddebug *desc, char *buf)
> 590 {
> 591 int pos_after_tid;
> 592 int pos = 0;
> 593
> 594 *buf = '\0';
>
>
> So now we are missing the string termination if no flags are set...
yes thats wrong. looks like I can hoist that init into the caller
> Thanks,
>
> -Jason
>
thanks
Jim
Powered by blists - more mailing lists