netdev - Re: [PATCH net] r8169: fix NAPI handling under high load

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: <92db72c9-1f8c-d6a5-bcf4-241fa4c5a310@gmail.com>
Date:   Tue, 16 Oct 2018 16:08:12 -0700
From:   Florian Fainelli <f.fainelli@...il.com>
To:     Stephen Hemminger <stephen@...workplumber.org>,
        Holger Hoffstätte <holger@...lied-asynchrony.com>
Cc:     Heiner Kallweit <hkallweit1@...il.com>,
        David Miller <davem@...emloft.net>,
        Realtek linux nic maintainers <nic_swsd@...ltek.com>,
        "netdev@...r.kernel.org" <netdev@...r.kernel.org>
Subject: Re: [PATCH net] r8169: fix NAPI handling under high load

On 10/16/2018 04:03 PM, Stephen Hemminger wrote:
> On Tue, 16 Oct 2018 23:17:31 +0200
> Holger Hoffstätte <holger@...lied-asynchrony.com> wrote:
> 
>> On 10/16/18 22:37, Heiner Kallweit wrote:
>>> rtl_rx() and rtl_tx() are called only if the respective bits are set
>>> in the interrupt status register. Under high load NAPI may not be
>>> able to process all data (work_done == budget) and it will schedule
>>> subsequent calls to the poll callback.
>>> rtl_ack_events() however resets the bits in the interrupt status
>>> register, therefore subsequent calls to rtl8169_poll() won't call
>>> rtl_rx() and rtl_tx() - chip interrupts are still disabled.  
>>
>> Very interesting! Could this be the reason for the mysterious
>> hangs & resets we experienced when enabling BQL for r8169?
>> They happened more often with TSO/GSO enabled and several people
>> attempted to fix those hangs unsuccessfully; it was later reverted
>> and has been since then (#87cda7cb43).
>> If this bug has been there "forever" it might be tempting to
>> re-apply BQL and see what happens. Any chance you could give that
>> a try? I'll gladly test patches, just like I'll run this one.
>>
>> cheers
>> Holger
> 
> Many drivers have buggy usage of napi_complete_done.
> 
> Might even be worth forcing all network drivers to check the return
> value. But fixing 150 broken drivers will be a nuisance.

I had started doing that about a month ago in light of the ixbge
ndo_poll_controller vs. napi problem, but have not had time to submit
that series yet:

https://github.com/ffainelli/linux/commits/napi-check

feel free to piggy back on top of that series if you would like to
address this.

> 
> diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
> index dc1d9ed33b31..c38bc66ffe74 100644
> --- a/include/linux/netdevice.h
> +++ b/include/linux/netdevice.h
> @@ -466,7 +466,8 @@ static inline bool napi_reschedule(struct napi_struct *napi)
>         return false;
>  }
>  
> -bool napi_complete_done(struct napi_struct *n, int work_done);
> +bool __must_check napi_complete_done(struct napi_struct *n, int work_done);
> +
>  /**
>   *     napi_complete - NAPI processing complete
>   *     @n: NAPI context
> 


-- 
Florian