[PATCH] bpf: Update sysctl documentation to list all supported architectures

[PATCH] bpf: Update sysctl documentation to list all supported architectures

[Michael Ellerman]

The sysctl documentation states that the JIT is only available on
x86_64, which is no longer correct.

Update the list to include all architectures that enable HAVE_CBPF_JIT
or HAVE_EBPF_JIT under some configuration.

Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
---
 Documentation/sysctl/net.txt | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/Documentation/sysctl/net.txt b/Documentation/sysctl/net.txt
index 14db18c970b1..f68356024d09 100644
--- a/Documentation/sysctl/net.txt
+++ b/Documentation/sysctl/net.txt
@@ -36,8 +36,9 @@ bpf_jit_enable
 --------------
 
 This enables Berkeley Packet Filter Just in Time compiler.
-Currently supported on x86_64 architecture, bpf_jit provides a framework
-to speed packet filtering, the one used by tcpdump/libpcap for example.
+Currently supported on arm, arm64, mips, powerpc, s390, sparc and x86_64
+architectures, bpf_jit provides a framework to speed packet filtering, the one
+used by tcpdump/libpcap for example.
 Values :
 	0 - disable the JIT (default value)
 	1 - enable the JIT
-- 
2.7.4

Re: [PATCH] bpf: Update sysctl documentation to list all supported architectures

[Daniel Borkmann]

Hi Michael,

On 08/16/2017 07:15 AM, Michael Ellerman wrote:
> The sysctl documentation states that the JIT is only available on
> x86_64, which is no longer correct.
>
> Update the list to include all architectures that enable HAVE_CBPF_JIT
> or HAVE_EBPF_JIT under some configuration.
>
> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

Thanks for the patch!

>   Documentation/sysctl/net.txt | 5 +++--
>   1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/Documentation/sysctl/net.txt b/Documentation/sysctl/net.txt
> index 14db18c970b1..f68356024d09 100644
> --- a/Documentation/sysctl/net.txt
> +++ b/Documentation/sysctl/net.txt
> @@ -36,8 +36,9 @@ bpf_jit_enable
>   --------------
>
>   This enables Berkeley Packet Filter Just in Time compiler.
> -Currently supported on x86_64 architecture, bpf_jit provides a framework
> -to speed packet filtering, the one used by tcpdump/libpcap for example.
> +Currently supported on arm, arm64, mips, powerpc, s390, sparc and x86_64
> +architectures, bpf_jit provides a framework to speed packet filtering, the one
> +used by tcpdump/libpcap for example.

Good point, could we actually make that as a bullet list and
differentiate between cBPF and eBPF JITs, so that a user doesn't
need to run git grep HAVE_{E,C}BPF_JIT to figure it out what the
switch enables on the arch used? That would be great.

So for eBPF JITs, we have covered:

  * x86_64
  * arm64
  * ppc64
  * sparc64
  * mips64

For old cBPF, there is:

  * arm
  * mips
  * ppc
  * sparc

Thanks,
Daniel

>   Values :
>   	0 - disable the JIT (default value)
>   	1 - enable the JIT
>

Re: [PATCH] bpf: Update sysctl documentation to list all supported architectures

[Michael Ellerman]

Daniel Borkmann <daniel@iogearbox.net> writes:

> Hi Michael,
>
> On 08/16/2017 07:15 AM, Michael Ellerman wrote:
>> The sysctl documentation states that the JIT is only available on
>> x86_64, which is no longer correct.
>>
>> Update the list to include all architectures that enable HAVE_CBPF_JIT
>> or HAVE_EBPF_JIT under some configuration.
>>
>> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
>
> Thanks for the patch!
>
>>   Documentation/sysctl/net.txt | 5 +++--
>>   1 file changed, 3 insertions(+), 2 deletions(-)
>>
>> diff --git a/Documentation/sysctl/net.txt b/Documentation/sysctl/net.txt
>> index 14db18c970b1..f68356024d09 100644
>> --- a/Documentation/sysctl/net.txt
>> +++ b/Documentation/sysctl/net.txt
>> @@ -36,8 +36,9 @@ bpf_jit_enable
>>   --------------
>>
>>   This enables Berkeley Packet Filter Just in Time compiler.
>> -Currently supported on x86_64 architecture, bpf_jit provides a framework
>> -to speed packet filtering, the one used by tcpdump/libpcap for example.
>> +Currently supported on arm, arm64, mips, powerpc, s390, sparc and x86_64
>> +architectures, bpf_jit provides a framework to speed packet filtering, the one
>> +used by tcpdump/libpcap for example.
>
> Good point, could we actually make that as a bullet list and
> differentiate between cBPF and eBPF JITs, so that a user doesn't
> need to run git grep HAVE_{E,C}BPF_JIT to figure it out what the
> switch enables on the arch used? That would be great.

We could.

Does a user of the sysctl want/need to know the difference though? Or do
they just want to turn on "the JIT"?

cheers

Re: [PATCH] bpf: Update sysctl documentation to list all supported architectures

[Daniel Borkmann]

On 08/16/2017 01:10 PM, Michael Ellerman wrote:
> Daniel Borkmann <daniel@iogearbox.net> writes:
>> On 08/16/2017 07:15 AM, Michael Ellerman wrote:
>>> The sysctl documentation states that the JIT is only available on
>>> x86_64, which is no longer correct.
>>>
>>> Update the list to include all architectures that enable HAVE_CBPF_JIT
>>> or HAVE_EBPF_JIT under some configuration.
>>>
>>> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
>>
>> Thanks for the patch!
>>
>>>    Documentation/sysctl/net.txt | 5 +++--
>>>    1 file changed, 3 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/Documentation/sysctl/net.txt b/Documentation/sysctl/net.txt
>>> index 14db18c970b1..f68356024d09 100644
>>> --- a/Documentation/sysctl/net.txt
>>> +++ b/Documentation/sysctl/net.txt
>>> @@ -36,8 +36,9 @@ bpf_jit_enable
>>>    --------------
>>>
>>>    This enables Berkeley Packet Filter Just in Time compiler.
>>> -Currently supported on x86_64 architecture, bpf_jit provides a framework
>>> -to speed packet filtering, the one used by tcpdump/libpcap for example.
>>> +Currently supported on arm, arm64, mips, powerpc, s390, sparc and x86_64
>>> +architectures, bpf_jit provides a framework to speed packet filtering, the one
>>> +used by tcpdump/libpcap for example.
>>
>> Good point, could we actually make that as a bullet list and
>> differentiate between cBPF and eBPF JITs, so that a user doesn't
>> need to run git grep HAVE_{E,C}BPF_JIT to figure it out what the
>> switch enables on the arch used? That would be great.
>
> We could.
>
> Does a user of the sysctl want/need to know the difference though? Or do
> they just want to turn on "the JIT"?

They would just turn it on, but I think it would be nice to inform
them which archs support eBPF (which is a superset of cBPF in term
of what can be jited), so in case they have some native eBPF programs
they would see whether these can also be jited.

Re: [PATCH] bpf: Update sysctl documentation to list all supported architectures

[Michael Ellerman]

Daniel Borkmann <daniel@iogearbox.net> writes:
> On 08/16/2017 01:10 PM, Michael Ellerman wrote:
>> Daniel Borkmann <daniel@iogearbox.net> writes:
>>> On 08/16/2017 07:15 AM, Michael Ellerman wrote:
>>>> diff --git a/Documentation/sysctl/net.txt b/Documentation/sysctl/net.txt
>>>> index 14db18c970b1..f68356024d09 100644
>>>> --- a/Documentation/sysctl/net.txt
>>>> +++ b/Documentation/sysctl/net.txt
>>>> @@ -36,8 +36,9 @@ bpf_jit_enable
>>>>    --------------
>>>>
>>>>    This enables Berkeley Packet Filter Just in Time compiler.
>>>> -Currently supported on x86_64 architecture, bpf_jit provides a framework
>>>> -to speed packet filtering, the one used by tcpdump/libpcap for example.
>>>> +Currently supported on arm, arm64, mips, powerpc, s390, sparc and x86_64
>>>> +architectures, bpf_jit provides a framework to speed packet filtering, the one
>>>> +used by tcpdump/libpcap for example.
>>>
>>> Good point, could we actually make that as a bullet list and
>>> differentiate between cBPF and eBPF JITs, so that a user doesn't
>>> need to run git grep HAVE_{E,C}BPF_JIT to figure it out what the
>>> switch enables on the arch used? That would be great.
>>
>> We could.
>>
>> Does a user of the sysctl want/need to know the difference though? Or do
>> they just want to turn on "the JIT"?
>
> They would just turn it on, but I think it would be nice to inform
> them which archs support eBPF (which is a superset of cBPF in term
> of what can be jited), so in case they have some native eBPF programs
> they would see whether these can also be jited.

OK. v2 just sent.

The text could probably use some more tweaking to mention the other
things BPF is used for these days, but I didn't really feel qualified to
do that.

cheers