netdev - Re: [PATCH] bnxt_en: Fix RSS logic in __bnxt_reserve

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: <20240724172536.318fb6f8@kernel.org>
Date: Wed, 24 Jul 2024 17:25:36 -0700
From: Jakub Kicinski <kuba@...nel.org>
To: Michael Chan <michael.chan@...adcom.com>
Cc: davem@...emloft.net, netdev@...r.kernel.org, edumazet@...gle.com,
 pabeni@...hat.com, pavan.chebbi@...adcom.com,
 andrew.gospodarek@...adcom.com
Subject: Re: [PATCH] bnxt_en: Fix RSS logic in __bnxt_reserve_rings()

On Wed, 24 Jul 2024 15:21:06 -0700 Michael Chan wrote:
> Now, with RSS contexts support, if the user has added or deleted RSS
> contexts, we may now enter this path to reserve the new number of VNICs.
> However, netif_is_rxfh_configured() will not return the correct state if
> we are still in the middle of set_rxfh().  So the existing code may
> set the indirection table of the default RSS context to default by
> mistake.

I feel like my explanation was more clear :S

The key point is that ethtool::set_rxfh() calls the "reload" functions
and expects the scope of the "reload" to be quite narrow, because only
the RSS table has changed. Unfortunately the add / delete of additional
contexts de-sync the resource counts, so ethtool::set_rxfh() now ends
up "reloading" more than it intended. The "more than intended" includes
going down the RSS indir reset path, which calls netif_is_rxfh_configured().
Return value from netif_is_rxfh_configured() during ethtool::set_rxfh()
is undefined.

Reported tag would have been nice too..