linux-kernel - Re: [PATCH] net: stmmac: Fix RX Coalesce IOC always true issue

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: <1591752616.29387.26.camel@mhfsdcap03>
Date:   Wed, 10 Jun 2020 09:30:16 +0800
From:   biao huang <biao.huang@...iatek.com>
To:     Jose Abreu <Jose.Abreu@...opsys.com>
CC:     Giuseppe Cavallaro <peppe.cavallaro@...com>,
        Alexandre Torgue <alexandre.torgue@...com>,
        "David S . Miller" <davem@...emloft.net>,
        "linux-mediatek@...ts.infradead.org" 
        <linux-mediatek@...ts.infradead.org>,
        "yt.shen@...iatek.com" <yt.shen@...iatek.com>,
        Matthias Brugger <matthias.bgg@...il.com>,
        "netdev@...r.kernel.org" <netdev@...r.kernel.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] net: stmmac: Fix RX Coalesce IOC always true issue

On Tue, 2020-06-09 at 14:20 +0000, Jose Abreu wrote:
> From: Biao Huang <biao.huang@...iatek.com>
> Date: Jun/09/2020, 10:41:33 (UTC+00:00)
> 
> > -		rx_q->rx_count_frames += priv->rx_coal_frames;
> > -		if (rx_q->rx_count_frames > priv->rx_coal_frames)
> > +		if (rx_q->rx_count_frames >= priv->rx_coal_frames)
> 
> This is no right. If you want to RX IC bit to not always be set you need 
> to change coalesce parameters using ethtool.

let's take look at these lines:

 	1. rx_q->rx_count_frames += priv->rx_coal_frames;
	2. 	if (rx_q->rx_count_frames > priv->rx_coal_frames)
	3. 		rx_q->rx_count_frames = 0;

before the if condition(line 2 above), there is "rx_q->rx_count_frames
+= priv->rx_coal_frames"(line 1 above) sentence; so the if condition
always true, and the assignment rx_q->rx_count_frames = 0 happens.

Here is the result on our platform, all desc3[30] = 1; you can also
check it on your platform. no matter what rx-frames you set with
"ethtool -C eth0 rx-frames", desc3[30] always true.

# ethtool -c eth0
Coalesce parameters for eth0:
Adaptive RX: off  TX: off
stats-block-usecs: 0
sample-interval: 0
pkt-rate-low: 0
pkt-rate-high: 0
 
rx-usecs: 301
rx-frames: 25
rx-usecs-irq: 0
rx-frames-irq: 0

#cd /sys/kernel/debug/stmmaceth/eth0
# cat descriptors_status
RX Queue 0:
Descriptor ring:
0 [0x5038b000]: 0xa8541a82 0x0 0x0 0xc1000000
 
1 [0x5038b010]: 0xa8541202 0x0 0x0 0xc1000000
 
2 [0x5038b020]: 0xa8540982 0x0 0x0 0xc1000000
 
3 [0x5038b030]: 0xa8540102 0x0 0x0 0xc1000000
 
4 [0x5038b040]: 0xa658f802 0x0 0x0 0xc1000000
 
5 [0x5038b050]: 0xa658ef82 0x0 0x0 0xc1000000
 
6 [0x5038b060]: 0xa658e702 0x0 0x0 0xc1000000
 
7 [0x5038b070]: 0xa658de82 0x0 0x0 0xc1000000
 
8 [0x5038b080]: 0xa658d602 0x0 0x0 0xc1000000
 
9 [0x5038b090]: 0xa658cd82 0x0 0x0 0xc1000000
 
10 [0x5038b0a0]: 0xa658c502 0x0 0x0 0xc1000000
 
11 [0x5038b0b0]: 0xa658bc82 0x0 0x0 0xc1000000
 
12 [0x5038b0c0]: 0xa658b402 0x0 0x0 0xc1000000
 
13 [0x5038b0d0]: 0xa658ab82 0x0 0x0 0xc1000000
 
14 [0x5038b0e0]: 0xa658a302 0x0 0x0 0xc1000000
 
15 [0x5038b0f0]: 0xa6589a82 0x0 0x0 0xc1000000
 
16 [0x5038b100]: 0xa6589202 0x0 0x0 0xc1000000
 
17 [0x5038b110]: 0xa6588982 0x0 0x0 0xc1000000
 
18 [0x5038b120]: 0xa6588102 0x0 0x0 0xc1000000
 
19 [0x5038b130]: 0xa46df802 0x0 0x0 0xc1000000
 
20 [0x5038b140]: 0xa46def82 0x0 0x0 0xc1000000
 
21 [0x5038b150]: 0xa46de702 0x0 0x0 0xc1000000
 
22 [0x5038b160]: 0xa46dde82 0x0 0x0 0xc1000000
 
23 [0x5038b170]: 0xa46dd602 0x0 0x0 0xc1000000
 
24 [0x5038b180]: 0xa46dcd82 0x0 0x0 0xc1000000
 
25 [0x5038b190]: 0xa46dc502 0x0 0x0 0xc1000000
 
26 [0x5038b1a0]: 0xa46dbc82 0x0 0x0 0xc1000000
 
27 [0x5038b1b0]: 0xa46db402 0x0 0x0 0xc1000000
 
28 [0x5038b1c0]: 0xa46dab82 0x0 0x0 0xc1000000
 
29 [0x5038b1d0]: 0xa46da302 0x0 0x0 0xc1000000
 
30 [0x5038b1e0]: 0xa46d9a82 0x0 0x0 0xc1000000
...

> ---
> Thanks,
> Jose Miguel Abreu