Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > linux.kernel > #1727695 > unrolled thread

[PATCH] netfilter: xt_hashlimit: avoid 64-bit division

Started byArnd Bergmann <arnd@arndb.de>
First post2017-09-06 22:00 +0200
Last post2017-09-07 16:20 +0200
Articles 6 — 4 participants

Back to article view | Back to linux.kernel


Contents

  [PATCH] netfilter: xt_hashlimit: avoid 64-bit division Arnd Bergmann <arnd@arndb.de> - 2017-09-06 22:00 +0200
    Re: [PATCH] netfilter: xt_hashlimit: avoid 64-bit division Vishwanath Pai <vpai@akamai.com> - 2017-09-06 22:30 +0200
      Re: [PATCH] netfilter: xt_hashlimit: avoid 64-bit division Arnd Bergmann <arnd@arndb.de> - 2017-09-06 22:50 +0200
        Re: [PATCH] netfilter: xt_hashlimit: avoid 64-bit division Pablo Neira Ayuso <pablo@netfilter.org> - 2017-09-07 12:30 +0200
          Re: [PATCH] netfilter: xt_hashlimit: avoid 64-bit division Arnd Bergmann <arnd@arndb.de> - 2017-09-07 13:20 +0200
    Re: [PATCH] netfilter: xt_hashlimit: avoid 64-bit division Geert Uytterhoeven <geert@linux-m68k.org> - 2017-09-07 16:20 +0200

#1727695 — [PATCH] netfilter: xt_hashlimit: avoid 64-bit division

FromArnd Bergmann <arnd@arndb.de>
Date2017-09-06 22:00 +0200
Subject[PATCH] netfilter: xt_hashlimit: avoid 64-bit division
Message-ID<umOEq-6jL-11@gated-at.bofh.it>
64-bit division is expensive on 32-bit architectures, and
requires a special function call to avoid a link error like:

net/netfilter/xt_hashlimit.o: In function `hashlimit_mt_common':
xt_hashlimit.c:(.text+0x1328): undefined reference to `__aeabi_uldivmod'

In the case of hashlimit_mt_common, we don't actually need a
64-bit operation, we can simply rewrite the function slightly
to make that clear to the compiler.

Fixes: bea74641e378 ("netfilter: xt_hashlimit: add rate match mode")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
---
 net/netfilter/xt_hashlimit.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/net/netfilter/xt_hashlimit.c b/net/netfilter/xt_hashlimit.c
index 10d48234f5f4..50b53d86eef5 100644
--- a/net/netfilter/xt_hashlimit.c
+++ b/net/netfilter/xt_hashlimit.c
@@ -531,7 +531,10 @@ static u64 user2rate_bytes(u64 user)
 {
 	u64 r;
 
-	r = user ? 0xFFFFFFFFULL / user : 0xFFFFFFFFULL;
+	if (user > 0xFFFFFFFFULL)
+		return 0;
+
+	r = user ? 0xFFFFFFFFULL / (u32)user : 0xFFFFFFFFULL;
 	r = (r - 1) << 4;
 	return r;
 }
-- 
2.9.0

[toc] | [next] | [standalone]


#1727715

FromVishwanath Pai <vpai@akamai.com>
Date2017-09-06 22:30 +0200
Message-ID<umP7s-6KN-33@gated-at.bofh.it>
In reply to#1727695
On 09/06/2017 03:57 PM, Arnd Bergmann wrote:
> 64-bit division is expensive on 32-bit architectures, and
> requires a special function call to avoid a link error like:
> 
> net/netfilter/xt_hashlimit.o: In function `hashlimit_mt_common':
> xt_hashlimit.c:(.text+0x1328): undefined reference to `__aeabi_uldivmod'
> 
> In the case of hashlimit_mt_common, we don't actually need a
> 64-bit operation, we can simply rewrite the function slightly
> to make that clear to the compiler.
> 
> Fixes: bea74641e378 ("netfilter: xt_hashlimit: add rate match mode")
> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
> ---
>  net/netfilter/xt_hashlimit.c | 5 ++++-
>  1 file changed, 4 insertions(+), 1 deletion(-)
> 
> diff --git a/net/netfilter/xt_hashlimit.c b/net/netfilter/xt_hashlimit.c
> index 10d48234f5f4..50b53d86eef5 100644
> --- a/net/netfilter/xt_hashlimit.c
> +++ b/net/netfilter/xt_hashlimit.c
> @@ -531,7 +531,10 @@ static u64 user2rate_bytes(u64 user)
>  {
>  	u64 r;
>  
> -	r = user ? 0xFFFFFFFFULL / user : 0xFFFFFFFFULL;
> +	if (user > 0xFFFFFFFFULL)
> +		return 0;
> +
> +	r = user ? 0xFFFFFFFFULL / (u32)user : 0xFFFFFFFFULL;
>  	r = (r - 1) << 4;
>  	return r;
>  }
> 

I have submitted another patch to fix this:
https://patchwork.ozlabs.org/patch/809881/

We have seen this problem before, I was careful not to introduce this
again in the new patch but clearly I overlooked this particular line :(

In the other cases we fixed it by replacing division with div64_u64().

-Vishwanath

[toc] | [prev] | [next] | [standalone]


#1727725

FromArnd Bergmann <arnd@arndb.de>
Date2017-09-06 22:50 +0200
Message-ID<umPqN-6Rp-17@gated-at.bofh.it>
In reply to#1727715
On Wed, Sep 6, 2017 at 10:22 PM, Vishwanath Pai <vpai@akamai.com> wrote:
> On 09/06/2017 03:57 PM, Arnd Bergmann wrote:
>> 64-bit division is expensive on 32-bit architectures, and
>> requires a special function call to avoid a link error like:
>>
>> net/netfilter/xt_hashlimit.o: In function `hashlimit_mt_common':
>> xt_hashlimit.c:(.text+0x1328): undefined reference to `__aeabi_uldivmod'
>>
>> In the case of hashlimit_mt_common, we don't actually need a
>> 64-bit operation, we can simply rewrite the function slightly
>> to make that clear to the compiler.
>>
>> Fixes: bea74641e378 ("netfilter: xt_hashlimit: add rate match mode")
>> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
>> ---
>>  net/netfilter/xt_hashlimit.c | 5 ++++-
>>  1 file changed, 4 insertions(+), 1 deletion(-)
>>
>> diff --git a/net/netfilter/xt_hashlimit.c b/net/netfilter/xt_hashlimit.c
>> index 10d48234f5f4..50b53d86eef5 100644
>> --- a/net/netfilter/xt_hashlimit.c
>> +++ b/net/netfilter/xt_hashlimit.c
>> @@ -531,7 +531,10 @@ static u64 user2rate_bytes(u64 user)
>>  {
>>       u64 r;
>>
>> -     r = user ? 0xFFFFFFFFULL / user : 0xFFFFFFFFULL;
>> +     if (user > 0xFFFFFFFFULL)
>> +             return 0;
>> +
>> +     r = user ? 0xFFFFFFFFULL / (u32)user : 0xFFFFFFFFULL;
>>       r = (r - 1) << 4;
>>       return r;
>>  }
>>
>
> I have submitted another patch to fix this:
> https://patchwork.ozlabs.org/patch/809881/
>
> We have seen this problem before, I was careful not to introduce this
> again in the new patch but clearly I overlooked this particular line :(
>
> In the other cases we fixed it by replacing division with div64_u64().

div64_u64() seems needlessly expensive here since the dividend
is known to be a 32-bit number. I guess the function is not called
frequently though, so it doesn't matter much.

      Arnd

[toc] | [prev] | [next] | [standalone]


#1728127

FromPablo Neira Ayuso <pablo@netfilter.org>
Date2017-09-07 12:30 +0200
Message-ID<un2em-74v-3@gated-at.bofh.it>
In reply to#1727725
On Wed, Sep 06, 2017 at 10:48:22PM +0200, Arnd Bergmann wrote:
> On Wed, Sep 6, 2017 at 10:22 PM, Vishwanath Pai <vpai@akamai.com> wrote:
> > On 09/06/2017 03:57 PM, Arnd Bergmann wrote:
> >> 64-bit division is expensive on 32-bit architectures, and
> >> requires a special function call to avoid a link error like:
> >>
> >> net/netfilter/xt_hashlimit.o: In function `hashlimit_mt_common':
> >> xt_hashlimit.c:(.text+0x1328): undefined reference to `__aeabi_uldivmod'
> >>
> >> In the case of hashlimit_mt_common, we don't actually need a
> >> 64-bit operation, we can simply rewrite the function slightly
> >> to make that clear to the compiler.
> >>
> >> Fixes: bea74641e378 ("netfilter: xt_hashlimit: add rate match mode")
> >> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
> >> ---
> >>  net/netfilter/xt_hashlimit.c | 5 ++++-
> >>  1 file changed, 4 insertions(+), 1 deletion(-)
> >>
> >> diff --git a/net/netfilter/xt_hashlimit.c b/net/netfilter/xt_hashlimit.c
> >> index 10d48234f5f4..50b53d86eef5 100644
> >> --- a/net/netfilter/xt_hashlimit.c
> >> +++ b/net/netfilter/xt_hashlimit.c
> >> @@ -531,7 +531,10 @@ static u64 user2rate_bytes(u64 user)
> >>  {
> >>       u64 r;
> >>
> >> -     r = user ? 0xFFFFFFFFULL / user : 0xFFFFFFFFULL;
> >> +     if (user > 0xFFFFFFFFULL)
> >> +             return 0;
> >> +
> >> +     r = user ? 0xFFFFFFFFULL / (u32)user : 0xFFFFFFFFULL;
> >>       r = (r - 1) << 4;
> >>       return r;
> >>  }
> >>
> >
> > I have submitted another patch to fix this:
> > https://patchwork.ozlabs.org/patch/809881/
> >
> > We have seen this problem before, I was careful not to introduce this
> > again in the new patch but clearly I overlooked this particular line :(
> >
> > In the other cases we fixed it by replacing division with div64_u64().
> 
> div64_u64() seems needlessly expensive here since the dividend
> is known to be a 32-bit number. I guess the function is not called
> frequently though, so it doesn't matter much.

This is called from the packet path, only for the first packet for
each new destination IP entry in the hashtable, still from the
datapath. So if we can take something faster (for 32 bit arches) that
is correct, I think it's sensible to take.

Let me know in any case.

[toc] | [prev] | [next] | [standalone]


#1728153

FromArnd Bergmann <arnd@arndb.de>
Date2017-09-07 13:20 +0200
Message-ID<un30K-7Eu-19@gated-at.bofh.it>
In reply to#1728127
On Thu, Sep 7, 2017 at 12:19 PM, Pablo Neira Ayuso <pablo@netfilter.org> wrote:
> On Wed, Sep 06, 2017 at 10:48:22PM +0200, Arnd Bergmann wrote:
>> On Wed, Sep 6, 2017 at 10:22 PM, Vishwanath Pai <vpai@akamai.com> wrote:
>> > On 09/06/2017 03:57 PM, Arnd Bergmann wrote:
>> >> 64-bit division is expensive on 32-bit architectures, and
>> >> requires a special function call to avoid a link error like:
>> >>
>> >> net/netfilter/xt_hashlimit.o: In function `hashlimit_mt_common':
>> >> xt_hashlimit.c:(.text+0x1328): undefined reference to `__aeabi_uldivmod'
>> >>
>> >> In the case of hashlimit_mt_common, we don't actually need a
>> >> 64-bit operation, we can simply rewrite the function slightly
>> >> to make that clear to the compiler.
>> >>
>> >> Fixes: bea74641e378 ("netfilter: xt_hashlimit: add rate match mode")
>> >> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
>> >> ---
>> >>  net/netfilter/xt_hashlimit.c | 5 ++++-
>> >>  1 file changed, 4 insertions(+), 1 deletion(-)
>> >>
>> >> diff --git a/net/netfilter/xt_hashlimit.c b/net/netfilter/xt_hashlimit.c
>> >> index 10d48234f5f4..50b53d86eef5 100644
>> >> --- a/net/netfilter/xt_hashlimit.c
>> >> +++ b/net/netfilter/xt_hashlimit.c
>> >> @@ -531,7 +531,10 @@ static u64 user2rate_bytes(u64 user)
>> >>  {
>> >>       u64 r;
>> >>
>> >> -     r = user ? 0xFFFFFFFFULL / user : 0xFFFFFFFFULL;
>> >> +     if (user > 0xFFFFFFFFULL)
>> >> +             return 0;
>> >> +
>> >> +     r = user ? 0xFFFFFFFFULL / (u32)user : 0xFFFFFFFFULL;
>> >>       r = (r - 1) << 4;
>> >>       return r;
>> >>  }
>> >>
>> >
>> > I have submitted another patch to fix this:
>> > https://patchwork.ozlabs.org/patch/809881/
>> >
>> > We have seen this problem before, I was careful not to introduce this
>> > again in the new patch but clearly I overlooked this particular line :(
>> >
>> > In the other cases we fixed it by replacing division with div64_u64().
>>
>> div64_u64() seems needlessly expensive here since the dividend
>> is known to be a 32-bit number. I guess the function is not called
>> frequently though, so it doesn't matter much.
>
> This is called from the packet path, only for the first packet for
> each new destination IP entry in the hashtable, still from the
> datapath. So if we can take something faster (for 32 bit arches) that
> is correct, I think it's sensible to take.
>
> Let me know in any case.

I think my version should be slightly better then, unless someone
finds something wrong with it.

       Arnd

[toc] | [prev] | [next] | [standalone]


#1728237

FromGeert Uytterhoeven <geert@linux-m68k.org>
Date2017-09-07 16:20 +0200
Message-ID<un5OW-Y6-15@gated-at.bofh.it>
In reply to#1727695
Hi Arnd,

On Wed, Sep 6, 2017 at 9:57 PM, Arnd Bergmann <arnd@arndb.de> wrote:
> 64-bit division is expensive on 32-bit architectures, and
> requires a special function call to avoid a link error like:
>
> net/netfilter/xt_hashlimit.o: In function `hashlimit_mt_common':
> xt_hashlimit.c:(.text+0x1328): undefined reference to `__aeabi_uldivmod'
>
> In the case of hashlimit_mt_common, we don't actually need a
> 64-bit operation, we can simply rewrite the function slightly
> to make that clear to the compiler.
>
> Fixes: bea74641e378 ("netfilter: xt_hashlimit: add rate match mode")
> Signed-off-by: Arnd Bergmann <arnd@arndb.de>

Thanks, this fixes a similar issue (__udivdi3 undefined) on m68k.

Acked-by: Geert Uytterhoeven <geert@linux-m68k.org>

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

[toc] | [prev] | [standalone]


Back to top | Article view | linux.kernel


csiph-web