Date:   Tue, 28 Dec 2021 10:55:55 -0800
From:   Nick Desaulniers <ndesaulniers@...gle.com>
To:     Tom Rix <trix@...hat.com>
Cc:     johannes@...solutions.net, davem@...emloft.net, kuba@...nel.org,
        nathan@...nel.org, linville@...driver.com,
        linux-wireless@...r.kernel.org, netdev@...r.kernel.org,
        linux-kernel@...r.kernel.org, llvm@...ts.linux.dev,
        Linus Torvalds <torvalds@...ux-foundation.org>
Subject: Re: [PATCH] mac80211: initialize variable have_higher_than_11mbit

On Fri, Dec 24, 2021 at 6:01 AM Tom Rix <trix@...hat.com> wrote:
>
>
> On 12/23/21 12:30 PM, Nick Desaulniers wrote:
> > On Thu, Dec 23, 2021 at 8:29 AM <trix@...hat.com> wrote:
> >> From: Tom Rix <trix@...hat.com>
> >>
> >> Clang static analysis reports this warnings
> >>
> >> mlme.c:5332:7: warning: Branch condition evaluates to a
> >>    garbage value
> >>      have_higher_than_11mbit)
> >>      ^~~~~~~~~~~~~~~~~~~~~~~
> >>
> >> have_higher_than_11mbit is only set to true some of the time in
> >> ieee80211_get_rates() but is checked all of the time.  So
> >> have_higher_than_11mbit needs to be initialized to false.
> > LGTM. There's only one caller of ieee80211_get_rates() today; if there
> > were others, they could make a similar mistake in the future. An
> > alternate approach: ieee80211_get_rates() could unconditionally write
> > false before the loop that could later write true. Then call sites
> > don't need to worry about this conditional assignment. Perhaps that
> > would be preferable? If not:
>
> The have_higher_than_11mbit variable had previously be initialized to false.
>
> The commit 5d6a1b069b7f moved the variable without initializing.

I'm not disagreeing with that.

My point is that these sometimes uninitialized warnings you're
finding+fixing with clang static analyzer are demonstrating a
recurring pattern with code.

When _not_ using the static analyzer, -Wuninitialized and
-Wsometimes-uninitialized work in Clang by building a control flow
graph, but they only analyze a function locally.

For example, consider the following code:
```
_Bool is_thursday(void);
void hello(int);

void init (int* x) {
  if (is_thursday())
    *x = 1;
}

void foo (void) {
  int x;
  init(&x);
  hello(x);
}
```
(Clang+GCC today will warn on the above; x is considered to "escape"
the scope of foo as init could write the address of x to a global.
Instead clang's static analyzer will take the additional time to
analyze the callee.  But here's a spooky question: what happens when
init is in another translation unit? IIRC, the static analyzer doesn't
do cross TU analysis; I could be wrong though, I haven't run it in a
while.)

My point is that you're sending patches initializing x, when I think
it might be nicer to instead have functions like init always write a
value (unconditionally, rather than conditionally).  That way other
callers of init don't have to worry about sometimes initialized
variables.

>
> Tom
>
> >
> > Reviewed-by: Nick Desaulniers <ndesaulniers@...gle.com>
> >
> >> Fixes: 5d6a1b069b7f ("mac80211: set basic rates earlier")
> >> Signed-off-by: Tom Rix <trix@...hat.com>
> >> ---
> >>   net/mac80211/mlme.c | 2 +-
> >>   1 file changed, 1 insertion(+), 1 deletion(-)
> >>
> >> diff --git a/net/mac80211/mlme.c b/net/mac80211/mlme.c
> >> index 51f55c4ee3c6e..766cbbc9c3a72 100644
> >> --- a/net/mac80211/mlme.c
> >> +++ b/net/mac80211/mlme.c
> >> @@ -5279,7 +5279,7 @@ static int ieee80211_prep_connection(struct ieee80211_sub_if_data *sdata,
> >>           */
> >>          if (new_sta) {
> >>                  u32 rates = 0, basic_rates = 0;
> >> -               bool have_higher_than_11mbit;
> >> +               bool have_higher_than_11mbit = false;
> >>                  int min_rate = INT_MAX, min_rate_index = -1;
> >>                  const struct cfg80211_bss_ies *ies;
> >>                  int shift = ieee80211_vif_get_shift(&sdata->vif);
> >> --
> >> 2.26.3
> >>
> >
>


-- 
Thanks,
~Nick Desaulniers

Powered by blists - more mailing lists