mbox series

[net,v2,0/3] net: dsa: bcm_sf2: CFP fixes

Message ID 20180515230125.25886-1-f.fainelli@gmail.com
Headers show
Series net: dsa: bcm_sf2: CFP fixes | expand

Message

Florian Fainelli May 15, 2018, 11:01 p.m. UTC 
Hi David,

This patch series fixes a number of usability issues with the SF2 Compact Field
Processor code:

- we would not be properly bound checking the location when we let the kernel
  automatically place rules with RX_CLS_LOC_ANY

- when using IPv6 rules and user space specifies a location identifier we
  would be off by one in what the chain ID (within the Broadcom tag) indicates

- it would be possible to delete one of the two slices of an IPv6 while leaving
  the other one programming leading to various problems

Florian Fainelli (3):
  net: dsa: bcm_sf2: Fix RX_CLS_LOC_ANY overwrite for last rule
  net: dsa: bcm_sf2: Fix IPv6 rules and chain ID
  net: dsa: bcm_sf2: Fix IPv6 rule half deletion

 drivers/net/dsa/bcm_sf2_cfp.c | 36 ++++++++++++++++++++++--------------
 1 file changed, 22 insertions(+), 14 deletions(-)

Comments

David Miller May 16, 2018, 6:12 p.m. UTC | #1 
From: Florian Fainelli <f.fainelli@gmail.com>
Date: Tue, 15 May 2018 16:01:22 -0700

> This patch series fixes a number of usability issues with the SF2 Compact Field
> Processor code:
> 
> - we would not be properly bound checking the location when we let the kernel
>   automatically place rules with RX_CLS_LOC_ANY
> 
> - when using IPv6 rules and user space specifies a location identifier we
>   would be off by one in what the chain ID (within the Broadcom tag) indicates
> 
> - it would be possible to delete one of the two slices of an IPv6 while leaving
>   the other one programming leading to various problems

Series applied, thanks.

Queue up for -stable?
Florian Fainelli May 16, 2018, 6:15 p.m. UTC | #2 
On 05/16/2018 11:12 AM, David Miller wrote:
> From: Florian Fainelli <f.fainelli@gmail.com>
> Date: Tue, 15 May 2018 16:01:22 -0700
> 
>> This patch series fixes a number of usability issues with the SF2 Compact Field
>> Processor code:
>>
>> - we would not be properly bound checking the location when we let the kernel
>>   automatically place rules with RX_CLS_LOC_ANY
>>
>> - when using IPv6 rules and user space specifies a location identifier we
>>   would be off by one in what the chain ID (within the Broadcom tag) indicates
>>
>> - it would be possible to delete one of the two slices of an IPv6 while leaving
>>   the other one programming leading to various problems
> 
> Series applied, thanks.
> 
> Queue up for -stable?

Yes please!
David Miller May 16, 2018, 6:18 p.m. UTC | #3 
From: Florian Fainelli <f.fainelli@gmail.com>
Date: Wed, 16 May 2018 11:15:37 -0700

> On 05/16/2018 11:12 AM, David Miller wrote:
>> Queue up for -stable?
> 
> Yes please!

Done.