netdev - Re: [PATCH net-next 0/9] Add support of VF Reset to HNS3 VF driver

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] [day] [month] [year] [list]

Date:   Thu, 22 Mar 2018 15:34:42 -0400 (EDT)
From:   David Miller <davem@...emloft.net>
To:     salil.mehta@...wei.com
Cc:     yisen.zhuang@...wei.com, mehta.salil.lnk@...il.com,
        netdev@...r.kernel.org, linux-kernel@...r.kernel.org,
        linuxarm@...wei.com
Subject: Re: [PATCH net-next 0/9] Add support of VF Reset to HNS3 VF driver

From: Salil Mehta <salil.mehta@...wei.com>
Date: Thu, 22 Mar 2018 14:28:51 +0000

> This patch-set adds the support of VF reset to the existing VF driver.
> VF Reset can be triggered due to TX watchdog firing  as a result of TX
> data-path not working. VF reset could also be a result of some internal
> configuration changes if that requires reset, or as a result of the
> PF/Core/Global/IMP(Integrated Management Processor) reset happened in
> the PF. 
> 
> Summary of Patches:
> * Watchdog timer trigger chnages are present in Patch 1.
> * Reset Service Task and related Event handling is present in Patches {2,3}
> * Changes to send reset request to PF, reset stack and re-initialization
>   of the hclge device is present in Patches {4,5,6}
> * Changes related to ARQ (Asynchronous Receive Queue) and its event handling
>   are present in Patches {7,8}
> * Changes required in PF to handle the VF Reset request and actually perform
>   hardware VF reset is there in Patch 9.
> 
> 
> NOTE: This patch depends upon "[PATCH net-next 00/11] fix some bugs for HNS3 driver"
> 	Link: https://lkml.org/lkml/2018/3/21/72

Series applied, thank you.

Please audit your usage of the various reset state bits.

In most places you do the correct atomic sequence by making use of
test_and_set_bit() and test_and_clear_bit().

But in some cases you do things like:

	if (!test_bit(X) && !test_bit(Y)) {
	...
		set_bit(Y);
	}

which is racy.