Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Subject: Re: Kernel Panic in skb_release_data using genet
To:     Maxime Ripard <maxime@cerno.tech>,
        nicolas saenz julienne <nsaenz@kernel.org>
Cc:     Florian Fainelli <f.fainelli@gmail.com>,
        Doug Berger <opendmb@gmail.com>,
        bcm-kernel-feedback-list@broadcom.com,
        linux-kernel@vger.kernel.org, netdev@vger.kernel.org
References: <20210524130147.7xv6ih2e3apu2zvu@gilmour>
 <a53f6192-3520-d5f8-df4b-786b3e4e8707@gmail.com>
 <20210524151329.5ummh4dfui6syme3@gilmour>
 <1482eff4-c5f4-66d9-237c-55a096ae2eb4@gmail.com>
 <6caa98e7-28ba-520c-f0cc-ee1219305c17@gmail.com>
 <20210528163219.x6yn44aimvdxlp6j@gilmour>
 <77d412b4-cdd6-ea86-d7fd-adb3af8970d9@gmail.com>
 <9e99ade5-ebfc-133e-ac61-1aba07ca80a2@gmail.com>
 <483c73edf02fa0139aae2b81e797534817655ea0.camel@kernel.org>
 <20210602132822.5hw4yynjgoomcfbg@gilmour>
From:   Florian Fainelli <f.fainelli@gmail.com>
Message-ID: <681f7369-90c7-0d2a-18a3-9a10917ce5f3@gmail.com>
Date:   Thu, 10 Jun 2021 14:33:17 -0700
User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:78.0) Gecko/20100101
 Firefox/78.0 Thunderbird/78.10.2
MIME-Version: 1.0
In-Reply-To: <20210602132822.5hw4yynjgoomcfbg@gilmour>
Content-Type: text/plain; charset=utf-8
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Precedence: bulk


On 6/2/2021 6:28 AM, Maxime Ripard wrote:
> On Tue, Jun 01, 2021 at 11:33:18AM +0200, nicolas saenz julienne wrote:
>> On Mon, 2021-05-31 at 19:36 -0700, Florian Fainelli wrote:
>>>> That is also how I boot my Pi4 at home, and I suspect you are right, if
>>>> the VPU does not shut down GENET's DMA, and leaves buffer addresses in
>>>> the on-chip descriptors that point to an address space that is managed
>>>> totally differently by Linux, then we can have a serious problem and
>>>> create some memory corruption when the ring is being reclaimed. I will
>>>> run a few experiments to test that theory and there may be a solution
>>>> using the SW_INIT reset controller to have a big reset of the controller
>>>> before handing it over to the Linux driver.
>>>
>>> Adding a WARN_ON(reg & DMA_EN) in bcmgenet_dma_disable() has not shown
>>> that the TX or RX DMA have been left running during the hand over from
>>> the VPU to the kernel. I checked out drm-misc-next-2021-05-17 to reduce
>>> as much as possible the differences between your set-up and my set-up
>>> but so far have not been able to reproduce the crash in booting from NFS
>>> repeatedly, I will try again.
>>
>> FWIW I can reproduce the error too. That said it's rather hard to reproduce,
>> something in the order of 1 failure every 20 tries.
> 
> Yeah, it looks like it's only from a cold boot and comes in "bursts",
> where you would get like 5 in a row and be done with it for a while.

Here are two patches that you could try exclusive from one another

1) Limit GENET to a single queue

diff --git a/drivers/net/ethernet/broadcom/genet/bcmgenet.c
b/drivers/net/ethernet/broadcom/genet/bcmgenet.c
index fcca023f22e5..e400c12e6868 100644
--- a/drivers/net/ethernet/broadcom/genet/bcmgenet.c
+++ b/drivers/net/ethernet/broadcom/genet/bcmgenet.c
@@ -3652,6 +3652,12 @@ static int bcmgenet_change_carrier(struct
net_device *dev, bool new_carrier)
        return 0;
 }

+static u16 bcmgenet_select_queue(struct net_device *dev, struct sk_buff
*skb,
+                                struct net_device *sb_dev)
+{
+       return 0;
+}
+
 static const struct net_device_ops bcmgenet_netdev_ops = {
        .ndo_open               = bcmgenet_open,
        .ndo_stop               = bcmgenet_close,
@@ -3666,6 +3672,7 @@ static const struct net_device_ops
bcmgenet_netdev_ops = {
 #endif
        .ndo_get_stats          = bcmgenet_get_stats,
        .ndo_change_carrier     = bcmgenet_change_carrier,
+       .ndo_select_queue       = bcmgenet_select_queue,
 };

 /* Array of GENET hardware parameters/characteristics */

2) Ensure that all TX/RX queues are disabled upon DMA initialization

diff --git a/drivers/net/ethernet/broadcom/genet/bcmgenet.c
b/drivers/net/ethernet/broadcom/genet/bcmgenet.c
index fcca023f22e5..7f8a5996fbbb 100644
--- a/drivers/net/ethernet/broadcom/genet/bcmgenet.c
+++ b/drivers/net/ethernet/broadcom/genet/bcmgenet.c
@@ -3237,15 +3237,21 @@ static void bcmgenet_get_hw_addr(struct
bcmgenet_priv *priv,
 /* Returns a reusable dma control register value */
 static u32 bcmgenet_dma_disable(struct bcmgenet_priv *priv)
 {
+       unsigned int i;
        u32 reg;
        u32 dma_ctrl;

        /* disable DMA */
        dma_ctrl = 1 << (DESC_INDEX + DMA_RING_BUF_EN_SHIFT) | DMA_EN;
+       for (i = 0; i < priv->hw_params->tx_queues; i++)
+               dma_ctrl |= (1 << (i + DMA_RING_BUF_EN_SHIFT));
        reg = bcmgenet_tdma_readl(priv, DMA_CTRL);
        reg &= ~dma_ctrl;
        bcmgenet_tdma_writel(priv, reg, DMA_CTRL);

+       dma_ctrl = 1 << (DESC_INDEX + DMA_RING_BUF_EN_SHIFT) | DMA_EN;
+       for (i = 0; i < priv->hw_params->rx_queues; i++)
+               dma_ctrl |= (1 << (i + DMA_RING_BUF_EN_SHIFT));
        reg = bcmgenet_rdma_readl(priv, DMA_CTRL);
        reg &= ~dma_ctrl;
        bcmgenet_rdma_writel(priv, reg, DMA_CTRL);
-- 
Florian