[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <871pksk3cg.fsf@yellow.woof>
Date: Thu, 18 Dec 2025 08:04:15 +0700
From: Nam Cao <namcao@...utronix.de>
To: Gabriele Monaco <gmonaco@...hat.com>, linux-kernel@...r.kernel.org,
Steven Rostedt <rostedt@...dmis.org>, Gabriele Monaco
<gmonaco@...hat.com>, linux-trace-kernel@...r.kernel.org
Cc: Tomas Glozar <tglozar@...hat.com>, Juri Lelli <jlelli@...hat.com>, Clark
Williams <williams@...hat.com>, John Kacur <jkacur@...hat.com>
Subject: Re: [PATCH v3 01/13] rv: Unify DA event handling functions across
monitor types
Gabriele Monaco <gmonaco@...hat.com> writes:
> The DA event handling functions are mostly duplicated because the
> per-task monitors need to propagate the task to use the pid in the trace
> events. This is a maintenance burden for a little advantage.
> The task can be obtained with some pointer arithmetic from the da_mon,
> hence only the function tracing events really need to differ.
>
> Unify all code handling the events, create da_trace_event() and
> da_trace_error() that only call the tracepoint function.
> Propagate the monitor id through the calls, the do_trace_ functions use
> the id (pid) in case of per-task monitors but ignore it for implicit
> monitors.
I think this overexplains. I would just explain the "WHY", then the
general idea how the patch solves the problem. Let the diff speaks for
the rest.
Perhaps something like:
The DA event handling functions are mostly duplicated for
differerent monitor types, because per-task monitors' functions
require a task_struct parameter while the others do not.
Unify the functions, handle the difference by always passing a
da_id_type which is the task's pid for per-task monitor but is
ignored for the other types.
Regardless:
Reviewed-by: Nam Cao <namcao@...utronix.de>
Nam
Powered by blists - more mailing lists