[v1] libbpf: Fix is_pow_of_2

[PATCH] libbpf: Fix is_pow_of_2

Posted by Ian Rogers 3 years, 10 months ago

From: Yuze Chi <chiyuze@google.com>

There is a missing not. Consider a power of 2 number like 4096:

x && (x & (x - 1))
4096 && (4096 & (4096 - 1))
4096 && (4096 & 4095)
4096 && 0
0

with the not this is:
x && !(x & (x - 1))
4096 && !(4096 & (4096 - 1))
4096 && !(4096 & 4095)
4096 && !0
4096 && 1
1

Reported-by: Yuze Chi <chiyuze@google.com>
Signed-off-by: Yuze Chi <chiyuze@google.com>
Signed-off-by: Ian Rogers <irogers@google.com>
---
 tools/lib/bpf/libbpf.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
index 3f4f18684bd3..fd0414ea00df 100644
--- a/tools/lib/bpf/libbpf.c
+++ b/tools/lib/bpf/libbpf.c
@@ -4956,7 +4956,7 @@ static void bpf_map__destroy(struct bpf_map *map);
 
 static bool is_pow_of_2(size_t x)
 {
-	return x && (x & (x - 1));
+	return x && !(x & (x - 1));
 }
 
 static size_t adjust_ringbuf_sz(size_t sz)
-- 
2.36.1.255.ge46751e96f-goog

Re: [PATCH] libbpf: Fix is_pow_of_2

Posted by Pavel Machek 3 years, 10 months ago

Hi!

> From: Yuze Chi <chiyuze@google.com>
> 
> +++ b/tools/lib/bpf/libbpf.c
> @@ -4956,7 +4956,7 @@ static void bpf_map__destroy(struct bpf_map *map);
>  
>  static bool is_pow_of_2(size_t x)
>  {
> -	return x && (x & (x - 1));
> +	return x && !(x & (x - 1));
>  }

I'm pretty sure we have this test in macro in includes somewhere... should we use
that instead?
									Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Re: [PATCH] libbpf: Fix is_pow_of_2

Posted by Zvi Effron 3 years, 10 months ago

On Sun, Jun 19, 2022 at 12:13 PM Pavel Machek <pavel@ucw.cz> wrote:
>
> Hi!
>
> > From: Yuze Chi <chiyuze@google.com>
> >
> > +++ b/tools/lib/bpf/libbpf.c
> > @@ -4956,7 +4956,7 @@ static void bpf_map__destroy(struct bpf_map *map);
> >
> > static bool is_pow_of_2(size_t x)
> > {
> > - return x && (x & (x - 1));
> > + return x && !(x & (x - 1));
> > }
>
> I'm pretty sure we have this test in macro in includes somewhere... should we use
> that instead?

I went looking for a macro that provided this check and could not find one. I
did find the inlined static function is_power_of_2 in log2.h, though, that we
could use.

> Pavel
> --
> (english) http://www.livejournal.com/~pavelmachek
> (cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Re: [PATCH] libbpf: Fix is_pow_of_2

Posted by Tiezhu Yang 3 years, 10 months ago


On 06/22/2022 07:03 AM, Zvi Effron wrote:
> On Sun, Jun 19, 2022 at 12:13 PM Pavel Machek <pavel@ucw.cz> wrote:
>>
>> Hi!
>>
>>> From: Yuze Chi <chiyuze@google.com>
>>>
>>> +++ b/tools/lib/bpf/libbpf.c
>>> @@ -4956,7 +4956,7 @@ static void bpf_map__destroy(struct bpf_map *map);
>>>
>>> static bool is_pow_of_2(size_t x)
>>> {
>>> - return x && (x & (x - 1));
>>> + return x && !(x & (x - 1));
>>> }
>>
>> I'm pretty sure we have this test in macro in includes somewhere... should we use
>> that instead?
>
> I went looking for a macro that provided this check and could not find one. I

arch/microblaze/mm/pgtable.c

#define is_power_of_2(x)	((x) != 0 && (((x) & ((x) - 1)) == 0))

> did find the inlined static function is_power_of_2 in log2.h, though, that we
> could use.

Here is a patch, but it seems that this is not worth the extra pain.

https://lore.kernel.org/bpf/8e5291b7-bd89-6fea-bfb7-954cacdb8523@iogearbox.net/

Thanks,
Tiezhu

Re: [PATCH] libbpf: Fix is_pow_of_2

Posted by Zvi Effron 3 years, 10 months ago

On Thu, Jun 2, 2022 at 9:17 PM Ian Rogers <irogers@google.com> wrote:
>
> From: Yuze Chi <chiyuze@google.com>
>
> There is a missing not. Consider a power of 2 number like 4096:
>
> x && (x & (x - 1))
> 4096 && (4096 & (4096 - 1))
> 4096 && (4096 & 4095)
> 4096 && 0
> 0
>
> with the not this is:
> x && !(x & (x - 1))
> 4096 && !(4096 & (4096 - 1))
> 4096 && !(4096 & 4095)
> 4096 && !0
> 4096 && 1
> 1
>
> Reported-by: Yuze Chi <chiyuze@google.com>
> Signed-off-by: Yuze Chi <chiyuze@google.com>
> Signed-off-by: Ian Rogers <irogers@google.com>
> ---
>  tools/lib/bpf/libbpf.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
> index 3f4f18684bd3..fd0414ea00df 100644
> --- a/tools/lib/bpf/libbpf.c
> +++ b/tools/lib/bpf/libbpf.c
> @@ -4956,7 +4956,7 @@ static void bpf_map__destroy(struct bpf_map *map);
>
>  static bool is_pow_of_2(size_t x)
>  {
> -       return x && (x & (x - 1));
> +       return x && !(x & (x - 1));

No idea if anyone cares about the consistency, but in linker.c (same directory)
the same static function is defined using == 0 at the end instead of using the
not operator.

Aside from the consistency issue, personally I find the == 0 version a little
bit easier to read and understand because it's a bit less dense (and a "!" next
to a "(" is an easy character to overlook).

>  }
>
>  static size_t adjust_ringbuf_sz(size_t sz)
> --
> 2.36.1.255.ge46751e96f-goog
>

Re: [PATCH] libbpf: Fix is_pow_of_2

Posted by Andrii Nakryiko 3 years, 10 months ago

On Thu, Jun 2, 2022 at 9:31 PM Zvi Effron <zeffron@riotgames.com> wrote:
>
> On Thu, Jun 2, 2022 at 9:17 PM Ian Rogers <irogers@google.com> wrote:
> >
> > From: Yuze Chi <chiyuze@google.com>
> >
> > There is a missing not. Consider a power of 2 number like 4096:
> >
> > x && (x & (x - 1))
> > 4096 && (4096 & (4096 - 1))
> > 4096 && (4096 & 4095)
> > 4096 && 0
> > 0
> >
> > with the not this is:
> > x && !(x & (x - 1))
> > 4096 && !(4096 & (4096 - 1))
> > 4096 && !(4096 & 4095)
> > 4096 && !0
> > 4096 && 1
> > 1
> >
> > Reported-by: Yuze Chi <chiyuze@google.com>
> > Signed-off-by: Yuze Chi <chiyuze@google.com>
> > Signed-off-by: Ian Rogers <irogers@google.com>
> > ---
> >  tools/lib/bpf/libbpf.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
> > index 3f4f18684bd3..fd0414ea00df 100644
> > --- a/tools/lib/bpf/libbpf.c
> > +++ b/tools/lib/bpf/libbpf.c
> > @@ -4956,7 +4956,7 @@ static void bpf_map__destroy(struct bpf_map *map);
> >
> >  static bool is_pow_of_2(size_t x)
> >  {
> > -       return x && (x & (x - 1));
> > +       return x && !(x & (x - 1));

ugh... *facepalm*

>
> No idea if anyone cares about the consistency, but in linker.c (same directory)
> the same static function is defined using == 0 at the end instead of using the
> not operator.
>
> Aside from the consistency issue, personally I find the == 0 version a little
> bit easier to read and understand because it's a bit less dense (and a "!" next
> to a "(" is an easy character to overlook).
>

I agree, even more so, logical not used with arbitrary integer (not a
pointer or bool) is a mental stumbling block for me, so much so that I
avoid doing !strcmp(), for example.

But in this case, I'm not sure why I copy/pasted is_pow_of_2() instead
of moving the one from linker.c into libbpf_internal.h as static
inline. Let's do that instead?

> >  }
> >
> >  static size_t adjust_ringbuf_sz(size_t sz)
> > --
> > 2.36.1.255.ge46751e96f-goog
> >

Re: [PATCH] libbpf: Fix is_pow_of_2

Posted by Andrii Nakryiko 3 years, 10 months ago

On Thu, Jun 2, 2022 at 10:33 PM Andrii Nakryiko
<andrii.nakryiko@gmail.com> wrote:
>
> On Thu, Jun 2, 2022 at 9:31 PM Zvi Effron <zeffron@riotgames.com> wrote:
> >
> > On Thu, Jun 2, 2022 at 9:17 PM Ian Rogers <irogers@google.com> wrote:
> > >
> > > From: Yuze Chi <chiyuze@google.com>
> > >
> > > There is a missing not. Consider a power of 2 number like 4096:
> > >
> > > x && (x & (x - 1))
> > > 4096 && (4096 & (4096 - 1))
> > > 4096 && (4096 & 4095)
> > > 4096 && 0
> > > 0
> > >
> > > with the not this is:
> > > x && !(x & (x - 1))
> > > 4096 && !(4096 & (4096 - 1))
> > > 4096 && !(4096 & 4095)
> > > 4096 && !0
> > > 4096 && 1
> > > 1
> > >
> > > Reported-by: Yuze Chi <chiyuze@google.com>
> > > Signed-off-by: Yuze Chi <chiyuze@google.com>
> > > Signed-off-by: Ian Rogers <irogers@google.com>

also can you please add Fixes: tag?

> > > ---
> > >  tools/lib/bpf/libbpf.c | 2 +-
> > >  1 file changed, 1 insertion(+), 1 deletion(-)
> > >
> > > diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
> > > index 3f4f18684bd3..fd0414ea00df 100644
> > > --- a/tools/lib/bpf/libbpf.c
> > > +++ b/tools/lib/bpf/libbpf.c
> > > @@ -4956,7 +4956,7 @@ static void bpf_map__destroy(struct bpf_map *map);
> > >
> > >  static bool is_pow_of_2(size_t x)
> > >  {
> > > -       return x && (x & (x - 1));
> > > +       return x && !(x & (x - 1));
>
> ugh... *facepalm*
>
> >
> > No idea if anyone cares about the consistency, but in linker.c (same directory)
> > the same static function is defined using == 0 at the end instead of using the
> > not operator.
> >
> > Aside from the consistency issue, personally I find the == 0 version a little
> > bit easier to read and understand because it's a bit less dense (and a "!" next
> > to a "(" is an easy character to overlook).
> >
>
> I agree, even more so, logical not used with arbitrary integer (not a
> pointer or bool) is a mental stumbling block for me, so much so that I
> avoid doing !strcmp(), for example.
>
> But in this case, I'm not sure why I copy/pasted is_pow_of_2() instead
> of moving the one from linker.c into libbpf_internal.h as static
> inline. Let's do that instead?
>
> > >  }
> > >
> > >  static size_t adjust_ringbuf_sz(size_t sz)
> > > --
> > > 2.36.1.255.ge46751e96f-goog
> > >