sysctl.conf: Enable Loose Reverse Path Filter according to RFC 3704

  For historical reasons, we were always reluctant to reverse path
filtering, since configuration changes were tricky to evaluate for a
larger userbase, IPFire permits a number of complex scenarios, and due
to limited resources.

As a compromise, this patch suggests to enable Loose Reverse Path
Filtering, as specified in RFC 3704 (section 2.4), to gain at least some
security achievement on this end.

To quote from that:

   Loose Reverse Path Forwarding (Loose RPF) is algorithmically similar
   to strict RPF, but differs in that it checks only for the existence
   of a route (even a default route, if applicable), not where the route
   points to.  Practically, this could be considered as a "route
   presence check" ("loose RPF is a misnomer in a sense because there is
   no "reverse path" check in the first place).

   The questionable benefit of Loose RPF is found in asymmetric routing
   situations: a packet is dropped if there is no route at all, such as
   to "Martian addresses" or addresses that are not currently routed,
   but is not dropped if a route exists.

There is no legitimate reason why we cannot enable this: If IPFire
receives a packet on some interface it cannot route on _any_ interface
at all, there is no sense in processing it.

While testing this change, I was unable to produce a situation where it
actually causes any harm. In theory, it shouldn't do so anyways.

In the future, we will hopefully be able to set these sysctl's to "1",
using Strict Reverse Path Filtering, as specified in RFC 3704 (section
2.2). Doing so was found to work fine in my testing environment as well,
but there is no asymmetric routing in place there.

Signed-off-by: Peter Müller <peter.mueller@ipfire.org>
---
 config/etc/sysctl.conf | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
  

Hello,

How are we going to test this with a wider audience?

I do not expect this to break anything, but I would like to make sure that this assumption holds true.

-Michael

> On 16 Jan 2022, at 14:47, Peter Müller <peter.mueller@ipfire.org> wrote:
> 
> For historical reasons, we were always reluctant to reverse path
> filtering, since configuration changes were tricky to evaluate for a
> larger userbase, IPFire permits a number of complex scenarios, and due
> to limited resources.
> 
> As a compromise, this patch suggests to enable Loose Reverse Path
> Filtering, as specified in RFC 3704 (section 2.4), to gain at least some
> security achievement on this end.
> 
> To quote from that:
> 
>   Loose Reverse Path Forwarding (Loose RPF) is algorithmically similar
>   to strict RPF, but differs in that it checks only for the existence
>   of a route (even a default route, if applicable), not where the route
>   points to.  Practically, this could be considered as a "route
>   presence check" ("loose RPF is a misnomer in a sense because there is
>   no "reverse path" check in the first place).
> 
>   The questionable benefit of Loose RPF is found in asymmetric routing
>   situations: a packet is dropped if there is no route at all, such as
>   to "Martian addresses" or addresses that are not currently routed,
>   but is not dropped if a route exists.
> 
> There is no legitimate reason why we cannot enable this: If IPFire
> receives a packet on some interface it cannot route on _any_ interface
> at all, there is no sense in processing it.
> 
> While testing this change, I was unable to produce a situation where it
> actually causes any harm. In theory, it shouldn't do so anyways.
> 
> In the future, we will hopefully be able to set these sysctl's to "1",
> using Strict Reverse Path Filtering, as specified in RFC 3704 (section
> 2.2). Doing so was found to work fine in my testing environment as well,
> but there is no asymmetric routing in place there.
> 
> Signed-off-by: Peter Müller <peter.mueller@ipfire.org>
> ---
> config/etc/sysctl.conf | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/config/etc/sysctl.conf b/config/etc/sysctl.conf
> index bc2d21c93..c8c775d13 100644
> --- a/config/etc/sysctl.conf
> +++ b/config/etc/sysctl.conf
> @@ -12,13 +12,13 @@ net.ipv4.tcp_syn_retries = 3
> net.ipv4.tcp_synack_retries = 3
> 
> net.ipv4.conf.default.arp_filter = 1
> -net.ipv4.conf.default.rp_filter = 0
> +net.ipv4.conf.default.rp_filter = 2
> net.ipv4.conf.default.accept_redirects = 0
> net.ipv4.conf.default.accept_source_route = 0
> net.ipv4.conf.default.log_martians = 1
> 
> net.ipv4.conf.all.arp_filter = 1
> -net.ipv4.conf.all.rp_filter = 0
> +net.ipv4.conf.all.rp_filter = 2
> net.ipv4.conf.all.accept_redirects = 0
> net.ipv4.conf.all.accept_source_route = 0
> net.ipv4.conf.all.log_martians = 1
> -- 
> 2.31.1
  

Hello Michael,

thanks for your reply.

Well, besides amending the patch for Core Update 164 (or beyond), mention it in the changelog,
and encourage people running special setups to test this update, I have no idea.

Since the vast majority of IPFire installations will have a default route set, Loose Reverse
Path Filtering cannot break anything for them. Therefore, I am willing to risk the procedure
mentioned above.

Thanks, and best regards,
Peter Müller


> Hello,
> 
> How are we going to test this with a wider audience?
> 
> I do not expect this to break anything, but I would like to make sure that this assumption holds true.
> 
> -Michael
> 
>> On 16 Jan 2022, at 14:47, Peter Müller <peter.mueller@ipfire.org> wrote:
>>
>> For historical reasons, we were always reluctant to reverse path
>> filtering, since configuration changes were tricky to evaluate for a
>> larger userbase, IPFire permits a number of complex scenarios, and due
>> to limited resources.
>>
>> As a compromise, this patch suggests to enable Loose Reverse Path
>> Filtering, as specified in RFC 3704 (section 2.4), to gain at least some
>> security achievement on this end.
>>
>> To quote from that:
>>
>>   Loose Reverse Path Forwarding (Loose RPF) is algorithmically similar
>>   to strict RPF, but differs in that it checks only for the existence
>>   of a route (even a default route, if applicable), not where the route
>>   points to.  Practically, this could be considered as a "route
>>   presence check" ("loose RPF is a misnomer in a sense because there is
>>   no "reverse path" check in the first place).
>>
>>   The questionable benefit of Loose RPF is found in asymmetric routing
>>   situations: a packet is dropped if there is no route at all, such as
>>   to "Martian addresses" or addresses that are not currently routed,
>>   but is not dropped if a route exists.
>>
>> There is no legitimate reason why we cannot enable this: If IPFire
>> receives a packet on some interface it cannot route on _any_ interface
>> at all, there is no sense in processing it.
>>
>> While testing this change, I was unable to produce a situation where it
>> actually causes any harm. In theory, it shouldn't do so anyways.
>>
>> In the future, we will hopefully be able to set these sysctl's to "1",
>> using Strict Reverse Path Filtering, as specified in RFC 3704 (section
>> 2.2). Doing so was found to work fine in my testing environment as well,
>> but there is no asymmetric routing in place there.
>>
>> Signed-off-by: Peter Müller <peter.mueller@ipfire.org>
>> ---
>> config/etc/sysctl.conf | 4 ++--
>> 1 file changed, 2 insertions(+), 2 deletions(-)
>>
>> diff --git a/config/etc/sysctl.conf b/config/etc/sysctl.conf
>> index bc2d21c93..c8c775d13 100644
>> --- a/config/etc/sysctl.conf
>> +++ b/config/etc/sysctl.conf
>> @@ -12,13 +12,13 @@ net.ipv4.tcp_syn_retries = 3
>> net.ipv4.tcp_synack_retries = 3
>>
>> net.ipv4.conf.default.arp_filter = 1
>> -net.ipv4.conf.default.rp_filter = 0
>> +net.ipv4.conf.default.rp_filter = 2
>> net.ipv4.conf.default.accept_redirects = 0
>> net.ipv4.conf.default.accept_source_route = 0
>> net.ipv4.conf.default.log_martians = 1
>>
>> net.ipv4.conf.all.arp_filter = 1
>> -net.ipv4.conf.all.rp_filter = 0
>> +net.ipv4.conf.all.rp_filter = 2
>> net.ipv4.conf.all.accept_redirects = 0
>> net.ipv4.conf.all.accept_source_route = 0
>> net.ipv4.conf.all.log_martians = 1
>> -- 
>> 2.31.1
>
  

Hello,

Okay. Let’s give it a try. I will refer anyone running into problems to you :)

-Michael

> On 18 Jan 2022, at 21:18, Peter Müller <peter.mueller@ipfire.org> wrote:
> 
> Hello Michael,
> 
> thanks for your reply.
> 
> Well, besides amending the patch for Core Update 164 (or beyond), mention it in the changelog,
> and encourage people running special setups to test this update, I have no idea.
> 
> Since the vast majority of IPFire installations will have a default route set, Loose Reverse
> Path Filtering cannot break anything for them. Therefore, I am willing to risk the procedure
> mentioned above.
> 
> Thanks, and best regards,
> Peter Müller
> 
> 
>> Hello,
>> 
>> How are we going to test this with a wider audience?
>> 
>> I do not expect this to break anything, but I would like to make sure that this assumption holds true.
>> 
>> -Michael
>> 
>>> On 16 Jan 2022, at 14:47, Peter Müller <peter.mueller@ipfire.org> wrote:
>>> 
>>> For historical reasons, we were always reluctant to reverse path
>>> filtering, since configuration changes were tricky to evaluate for a
>>> larger userbase, IPFire permits a number of complex scenarios, and due
>>> to limited resources.
>>> 
>>> As a compromise, this patch suggests to enable Loose Reverse Path
>>> Filtering, as specified in RFC 3704 (section 2.4), to gain at least some
>>> security achievement on this end.
>>> 
>>> To quote from that:
>>> 
>>>  Loose Reverse Path Forwarding (Loose RPF) is algorithmically similar
>>>  to strict RPF, but differs in that it checks only for the existence
>>>  of a route (even a default route, if applicable), not where the route
>>>  points to.  Practically, this could be considered as a "route
>>>  presence check" ("loose RPF is a misnomer in a sense because there is
>>>  no "reverse path" check in the first place).
>>> 
>>>  The questionable benefit of Loose RPF is found in asymmetric routing
>>>  situations: a packet is dropped if there is no route at all, such as
>>>  to "Martian addresses" or addresses that are not currently routed,
>>>  but is not dropped if a route exists.
>>> 
>>> There is no legitimate reason why we cannot enable this: If IPFire
>>> receives a packet on some interface it cannot route on _any_ interface
>>> at all, there is no sense in processing it.
>>> 
>>> While testing this change, I was unable to produce a situation where it
>>> actually causes any harm. In theory, it shouldn't do so anyways.
>>> 
>>> In the future, we will hopefully be able to set these sysctl's to "1",
>>> using Strict Reverse Path Filtering, as specified in RFC 3704 (section
>>> 2.2). Doing so was found to work fine in my testing environment as well,
>>> but there is no asymmetric routing in place there.
>>> 
>>> Signed-off-by: Peter Müller <peter.mueller@ipfire.org>
>>> ---
>>> config/etc/sysctl.conf | 4 ++--
>>> 1 file changed, 2 insertions(+), 2 deletions(-)
>>> 
>>> diff --git a/config/etc/sysctl.conf b/config/etc/sysctl.conf
>>> index bc2d21c93..c8c775d13 100644
>>> --- a/config/etc/sysctl.conf
>>> +++ b/config/etc/sysctl.conf
>>> @@ -12,13 +12,13 @@ net.ipv4.tcp_syn_retries = 3
>>> net.ipv4.tcp_synack_retries = 3
>>> 
>>> net.ipv4.conf.default.arp_filter = 1
>>> -net.ipv4.conf.default.rp_filter = 0
>>> +net.ipv4.conf.default.rp_filter = 2
>>> net.ipv4.conf.default.accept_redirects = 0
>>> net.ipv4.conf.default.accept_source_route = 0
>>> net.ipv4.conf.default.log_martians = 1
>>> 
>>> net.ipv4.conf.all.arp_filter = 1
>>> -net.ipv4.conf.all.rp_filter = 0
>>> +net.ipv4.conf.all.rp_filter = 2
>>> net.ipv4.conf.all.accept_redirects = 0
>>> net.ipv4.conf.all.accept_source_route = 0
>>> net.ipv4.conf.all.log_martians = 1
>>> -- 
>>> 2.31.1
>>
  

Hello Michael,

please do. It would be interesting to see setups where this breaks anything...

Thanks, and best regards,
Peter Müller


> Hello,
> 
> Okay. Let’s give it a try. I will refer anyone running into problems to you :)
> 
> -Michael
> 
>> On 18 Jan 2022, at 21:18, Peter Müller <peter.mueller@ipfire.org> wrote:
>>
>> Hello Michael,
>>
>> thanks for your reply.
>>
>> Well, besides amending the patch for Core Update 164 (or beyond), mention it in the changelog,
>> and encourage people running special setups to test this update, I have no idea.
>>
>> Since the vast majority of IPFire installations will have a default route set, Loose Reverse
>> Path Filtering cannot break anything for them. Therefore, I am willing to risk the procedure
>> mentioned above.
>>
>> Thanks, and best regards,
>> Peter Müller
>>
>>
>>> Hello,
>>>
>>> How are we going to test this with a wider audience?
>>>
>>> I do not expect this to break anything, but I would like to make sure that this assumption holds true.
>>>
>>> -Michael
>>>
>>>> On 16 Jan 2022, at 14:47, Peter Müller <peter.mueller@ipfire.org> wrote:
>>>>
>>>> For historical reasons, we were always reluctant to reverse path
>>>> filtering, since configuration changes were tricky to evaluate for a
>>>> larger userbase, IPFire permits a number of complex scenarios, and due
>>>> to limited resources.
>>>>
>>>> As a compromise, this patch suggests to enable Loose Reverse Path
>>>> Filtering, as specified in RFC 3704 (section 2.4), to gain at least some
>>>> security achievement on this end.
>>>>
>>>> To quote from that:
>>>>
>>>>  Loose Reverse Path Forwarding (Loose RPF) is algorithmically similar
>>>>  to strict RPF, but differs in that it checks only for the existence
>>>>  of a route (even a default route, if applicable), not where the route
>>>>  points to.  Practically, this could be considered as a "route
>>>>  presence check" ("loose RPF is a misnomer in a sense because there is
>>>>  no "reverse path" check in the first place).
>>>>
>>>>  The questionable benefit of Loose RPF is found in asymmetric routing
>>>>  situations: a packet is dropped if there is no route at all, such as
>>>>  to "Martian addresses" or addresses that are not currently routed,
>>>>  but is not dropped if a route exists.
>>>>
>>>> There is no legitimate reason why we cannot enable this: If IPFire
>>>> receives a packet on some interface it cannot route on _any_ interface
>>>> at all, there is no sense in processing it.
>>>>
>>>> While testing this change, I was unable to produce a situation where it
>>>> actually causes any harm. In theory, it shouldn't do so anyways.
>>>>
>>>> In the future, we will hopefully be able to set these sysctl's to "1",
>>>> using Strict Reverse Path Filtering, as specified in RFC 3704 (section
>>>> 2.2). Doing so was found to work fine in my testing environment as well,
>>>> but there is no asymmetric routing in place there.
>>>>
>>>> Signed-off-by: Peter Müller <peter.mueller@ipfire.org>
>>>> ---
>>>> config/etc/sysctl.conf | 4 ++--
>>>> 1 file changed, 2 insertions(+), 2 deletions(-)
>>>>
>>>> diff --git a/config/etc/sysctl.conf b/config/etc/sysctl.conf
>>>> index bc2d21c93..c8c775d13 100644
>>>> --- a/config/etc/sysctl.conf
>>>> +++ b/config/etc/sysctl.conf
>>>> @@ -12,13 +12,13 @@ net.ipv4.tcp_syn_retries = 3
>>>> net.ipv4.tcp_synack_retries = 3
>>>>
>>>> net.ipv4.conf.default.arp_filter = 1
>>>> -net.ipv4.conf.default.rp_filter = 0
>>>> +net.ipv4.conf.default.rp_filter = 2
>>>> net.ipv4.conf.default.accept_redirects = 0
>>>> net.ipv4.conf.default.accept_source_route = 0
>>>> net.ipv4.conf.default.log_martians = 1
>>>>
>>>> net.ipv4.conf.all.arp_filter = 1
>>>> -net.ipv4.conf.all.rp_filter = 0
>>>> +net.ipv4.conf.all.rp_filter = 2
>>>> net.ipv4.conf.all.accept_redirects = 0
>>>> net.ipv4.conf.all.accept_source_route = 0
>>>> net.ipv4.conf.all.log_martians = 1
>>>> -- 
>>>> 2.31.1
>>>
>