netdev - Re: [PATCH] r8169: use RxFIFO overflow workaround for 8168c chipset

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]

Message-ID: <1296556238.19128.27.camel@cera.brq.redhat.com>
Date:	Tue, 01 Feb 2011 11:30:38 +0100
From:	Ivan Vecera <ivecera@...hat.com>
To:	Francois Romieu <romieu@...zoreil.com>
Cc:	netdev@...r.kernel.org, Hayes <hayeswang@...ltek.com>
Subject: Re: [PATCH] r8169: use RxFIFO overflow workaround for 8168c chipset

On Fri, 2011-01-28 at 18:28 +0100, Francois Romieu wrote:
> Ivan Vecera <ivecera@...hat.com> :
> [...]
> > The test case was: Migration of the several kvm guests at the same time
> > between two hosts.
> 
> :o(
> 
> Could you check if simply leaving the irq handler when status == RxFIFOOver
> works by any chance ?
It does not help, I tried two versions:
1) Simply leave the loop (only "break;") when status == RxFIFOOver
...
if (unlikely(status & RxFIFOOver)) {
	static int rxfifo_count = 0; 
        printk("r8169: Rx FIFO overflows detected: %d\n", 
        	++rxfifo_count);
	break;
}
...

2) the same but with interrupt confirmation before leave
...
if (unlikely(status & RxFIFOOver)) {
	static int rxfifo_count = 0; 
        printk("r8169: Rx FIFO overflows detected: %d\n", 
        	++rxfifo_count);
	RTL_W16(IntrStatus,
		(status & RxFIFOOver) ?
		 (status | RxOverflow) : status);
	break;
}
...

In both cases the NIC continues in generating Rx FIFO overflows.

Ivan

> 
> Faced with pktgen, RTL_GIGA_MAC_VER_26 spits RxFIFO overflow errors quite
> fast (10000 packets kill it, reliably). The same fix as your avoids the crash.
> 
> As is, RTL_GIGA_MAC_VER_12 seems to survive at 900kpps. It signals RXFIFO
> overflow, loses half the packets and sends pause control frames but it does
> not crash. I have made it leave the irq handler when status == RxFIFOOver.
> I'll see if it can stand a few hours like that.
> 


--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html