linux-kernel - Re: [PATCH v2 3/5] tracing: Fix operator precedence for hist triggers expression

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <CAC_TJvfsF9BF2wfGck1icX_Ya7dLWO+hOBA5cR56PPr0Dn9D9Q@mail.gmail.com>
Date:   Wed, 20 Oct 2021 09:11:27 -0700
From:   Kalesh Singh <kaleshsingh@...gle.com>
To:     Steven Rostedt <rostedt@...dmis.org>
Cc:     Suren Baghdasaryan <surenb@...gle.com>,
        Hridya Valsaraju <hridya@...gle.com>,
        Namhyung Kim <namhyung@...nel.org>,
        "Cc: Android Kernel" <kernel-team@...roid.com>,
        Jonathan Corbet <corbet@....net>,
        Ingo Molnar <mingo@...hat.com>, Shuah Khan <shuah@...nel.org>,
        Masami Hiramatsu <mhiramat@...nel.org>,
        Tom Zanussi <zanussi@...nel.org>,
        "open list:DOCUMENTATION" <linux-doc@...r.kernel.org>,
        LKML <linux-kernel@...r.kernel.org>,
        "open list:KERNEL SELFTEST FRAMEWORK" 
        <linux-kselftest@...r.kernel.org>
Subject: Re: [PATCH v2 3/5] tracing: Fix operator precedence for hist triggers expression

On Wed, Oct 20, 2021 at 8:48 AM Steven Rostedt <rostedt@...dmis.org> wrote:
>
> On Tue, 19 Oct 2021 18:31:40 -0700
> Kalesh Singh <kaleshsingh@...gle.com> wrote:
>
> > @@ -2391,60 +2460,61 @@ static int check_expr_operands(struct trace_array *tr,
> >  static struct hist_field *parse_expr(struct hist_trigger_data *hist_data,
> >                                    struct trace_event_file *file,
> >                                    char *str, unsigned long flags,
> > -                                  char *var_name, unsigned int level)
> > +                                  char *var_name, unsigned int *n_subexprs)
> >  {
> >       struct hist_field *operand1 = NULL, *operand2 = NULL, *expr = NULL;
> >       unsigned long operand_flags;
> >       int field_op, ret = -EINVAL;
> >       char *sep, *operand1_str;
> >
> > -     if (level > 3) {
> > +     if (*n_subexprs > 3) {
>
> Why limit the sub expressions, and not just keep the limit of the level of
> recursion. We allow 3 levels of recursion, but we could have more than 3
> sub expressions.

The main reason for this is that it's predictable behavior form the
user's perspective. Before this patch the recursion always walked down
a single branch so limiting by level worked out the same as limiting
by sub expressions and is in line with the error the user would see
("Too many sub-expressions (3 max)"). Now that we take multiple paths
in the recursion, using the level to reflect the number of
sub-expressions would lead to only seeing the error in some of the
cases (Sometimes we allow 4, 5, 6 sub-expressions depending on how
balanced the tree is, and sometimes we error out on 4 - when the tree
is list-like). Limiting by sub-expression keeps this consistent
(always error out if we have more than 3 sub-expressions) and is in
line with the previous behavior.

- Kalesh

>
>
> If we have:  a * b + c / d - e * f / h
>
> It would break down into:
>               -
>        +            /
>    *       /     *     h
>  a   b   c  d  e  f
>
>
> Which I believe is 6 "sub expressions", but never goes more than three deep
> in recursion:
>
>    "a * b + c / d - e * f / h"
>
> Step 1:
>
>   op = "-"
>   operand1 = "a * b + c / d"
>   operand2 = "e * f / h"
>
> Process operand1: (recursion level 1)
>
>   op = "+"
>   operand1a = "a * b"
>   operand2a = "c / d"
>
> Process operand1a: (recursion level 2)
>
>   op = "*"
>   operand1b = "a"
>   operand2b = "b"
>
> return;
>
> Process operand1b: (recursion level 2)
>
>   op = "/"
>   operand1b = "c"
>   operand2b = "d"
>
> return;
>
> return;
>
> Process operand2: (recursion level 1)
>
>   op = "/"
>   operand1c = "e * f"
>   operand2c = "h"
>
> Process operand1c: (recursion level 2)
>
>   op = "*"
>   operand1c = "e"
>   operand2c = "f"
>
> return;
>
> return;
>
>
>
> > +
> > +     /* LHS of string is an expression e.g. a+b in a+b+c */
> > +     operand1 = parse_expr(hist_data, file, operand1_str, operand_flags, NULL, n_subexprs);
> >       if (IS_ERR(operand1)) {
> >               ret = PTR_ERR(operand1);
> >               operand1 = NULL;
>
> I wonder if we should look for optimizations, in case of operand1 and
> operand2 are both constants?
>
> Just perform the function, and convert it into a constant as well.
>
> -- Steve