netdev - Re: [PATCH net-next 5/7] bnxt_en: Change MSIX/NQs allocation policy

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <23709427-23c9-46e9-ae47-ead1ba5ab22c@intel.com>
Date: Wed, 10 Apr 2024 12:28:26 -0700
From: Jacob Keller <jacob.e.keller@...el.com>
To: Michael Chan <michael.chan@...adcom.com>
CC: <davem@...emloft.net>, <netdev@...r.kernel.org>, <edumazet@...gle.com>,
	<kuba@...nel.org>, <pabeni@...hat.com>, <pavan.chebbi@...adcom.com>,
	<andrew.gospodarek@...adcom.com>, Vikas Gupta <vikas.gupta@...adcom.com>
Subject: Re: [PATCH net-next 5/7] bnxt_en: Change MSIX/NQs allocation policy



On 4/9/2024 4:48 PM, Michael Chan wrote:
> On Tue, Apr 9, 2024 at 4:40 PM Jacob Keller <jacob.e.keller@...el.com> wrote:
>>
>>
>>
>> On 4/9/2024 2:54 PM, Michael Chan wrote:
>>> From: Vikas Gupta <vikas.gupta@...adcom.com>
>>>
>>> The existing scheme sets aside a number of MSIX/NQs for the RoCE
>>> driver whether the RoCE driver is registered or not.  This scheme
>>> is not flexible and limits the resources available for the L2 rings
>>> if RoCE is never used.
>>>
>>> Modify the scheme so that the RoCE MSIX/NQs can be used by the L2
>>> driver if they are not used for RoCE.  The MSIX/NQs are now
>>> represented by 3 fields.  bp->ulp_num_msix_want contains the
>>> desired default value, edev->ulp_num_msix_vec contains the
>>> available value (but not necessarily in use), and
>>> ulp_tbl->msix_requested contains the actual value in use by RoCE.
>>>
>>> The L2 driver can dip into edev->ulp_num_msix_vec if necessary.
>>>
>>> We need to add rtnl_lock() back in bnxt_register_dev() and
>>> bnxt_unregister_dev() to synchronize the MSIX usage between L2 and
>>> RoCE.
>>>
>>> Reviewed-by: Andy Gospodarek <andrew.gospodarek@...adcom.com>
>>> Signed-off-by: Vikas Gupta <vikas.gupta@...adcom.com>
>>> Signed-off-by: Michael Chan <michael.chan@...adcom.com>
>>> ---
>>
>> Whats the behavior if the L2 driver dips into this pool and then RoCE is
>> enabled later?
> 
> Thanks for the review.  If the user increases the L2 rings or enables
> XDP which will cause the driver to allocate a new set of XDP TX rings,
> it can now use the RoCE MSIX if needed and if they are not in use.
> 
>>
>> I guess RoCE would fail to get the resources it needs, but then system
>> administrator could re-configure the L2 device to use fewer resources?
> 
> If the above simply reduces the RoCE MSIX, the RoCE driver can still
> operate with fewer MSIX.  If the above has reduced the MSIX below the
> minimum required for RoCE, then RoCE will fail to initialize.  At that
> point, the user can reduce the L2 rings and reload the RoCE driver.

Sensible behavior. Thanks for the detailed explanation

-Jake