Re: Kernel Panic in skb_release_data using genet

[Maxime Ripard <maxime@cerno.tech>]

[-- Attachment #1: Type: text/plain, Size: 9480 bytes --]

Hi Florian,

Sorry for reviving this old thread...

On Thu, Jun 10, 2021 at 02:33:17PM -0700, Florian Fainelli wrote:
> 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);

It looks like current upstream still has this issue, which also upsets KASAN:

[   16.798433] ==================================================================
[   16.809347] BUG: KASAN: wild-memory-access in skb_release_data+0x124/0x270
[   16.816379] Read of size 8 at addr 80800000807f2e0c by task swapper/0/0
[   16.823122]
[   16.824655] CPU: 0 PID: 0 Comm: swapper/0 Not tainted 5.18.0-rc5-v8+ #210
[   16.831581] Hardware name: Raspberry Pi 4 Model B Rev 1.1 (DT)
[   16.837525] Call trace:
[   16.840025]  dump_backtrace.part.0+0x1dc/0x1f0
[   16.844576]  show_stack+0x24/0x80
[   16.847974]  dump_stack_lvl+0x8c/0xb8
[   16.851735]  print_report+0x1cc/0x240
[   16.855494]  kasan_report+0xb4/0x120
[   16.859161]  __asan_load8+0xa0/0xc4
[   16.862743]  skb_release_data+0x124/0x270
[   16.866849]  consume_skb+0x74/0xe0
[   16.870337]  __dev_kfree_skb_any+0x74/0x90
[   16.874538]  bcmgenet_desc_rx+0x4b4/0x620
[   16.878642]  bcmgenet_rx_poll+0x78/0x150
[   16.882657]  __napi_poll.constprop.0+0x64/0x240
[   16.887290]  net_rx_action+0x4d4/0x590
[   16.891127]  __do_softirq+0x228/0x4d8
[   16.894875]  __irq_exit_rcu+0x1e4/0x24c
[   16.898806]  irq_exit_rcu+0x20/0x54
[   16.902382]  el1_interrupt+0x38/0x50
[   16.906051]  el1h_64_irq_handler+0x18/0x2c
[   16.910250]  el1h_64_irq+0x64/0x68
[   16.913733]  arch_local_irq_enable+0xc/0x20
[   16.918010]  default_idle_call+0x80/0x114
[   16.922118]  cpuidle_idle_call+0x1e0/0x224
[   16.926310]  do_idle+0x104/0x14c
[   16.929616]  cpu_startup_entry+0x34/0x3c
[   16.933630]  rest_init+0x180/0x200
[   16.937113]  arch_post_acpi_subsys_init+0x0/0x30
[   16.941840]  start_kernel+0x3c8/0x400
[   16.945592]  __primary_switched+0xa8/0xb0
[   16.949699] ==================================================================
[   16.957052] Disabling lock debugging due to kernel taint
[   16.962507] Unable to handle kernel paging request at virtual address 80800000807f2e0c
[   16.970590] Mem abort info:
[   16.973461]   ESR = 0x96000004
[   16.976602]   EC = 0x25: DABT (current EL), IL = 32 bits
[   16.982038]   SET = 0, FnV = 0
[   16.985176]   EA = 0, S1PTW = 0
[   16.988403]   FSC = 0x04: level 0 translation fault
[   16.993440] Data abort info:
[   16.996401]   ISV = 0, ISS = 0x00000004
[   17.000333]   CM = 0, WnR = 0
[   17.003384] [80800000807f2e0c] address between user and kernel address ranges
[   17.010674] Internal error: Oops: 96000004 [#1] SMP
[   17.015651] Modules linked in:
[   17.018784] CPU: 0 PID: 0 Comm: swapper/0 Tainted: G    B             5.18.0-rc5-v8+ #210
[   17.027115] Hardware name: Raspberry Pi 4 Model B Rev 1.1 (DT)
[   17.033052] pstate: 40000005 (nZcv daif -PAN -UAO -TCO -DIT -SSBS BTYPE=--)
[   17.040148] pc : skb_release_data+0x124/0x270
[   17.044603] lr : skb_release_data+0x124/0x270
[   17.049055] sp : ffffffc00a477690
[   17.052434] x29: ffffffc00a477690 x28: 0000000000000000 x27: ffffff8043cfeb42
[   17.059744] x26: 0000000000000001 x25: ffffff8040a5d5be x24: 00000000ffffffff
[   17.067049] x23: ffffff8043cfeb40 x22: 0000000000000000 x21: ffffff8040a5d540
[   17.074355] x20: ffffff8043cfeb70 x19: 80800000807f2e04 x18: 00000000ee6397dc
[   17.081661] x17: 0000000000000000 x16: 0000000000000000 x15: 0000000000000000
[   17.088961] x14: 0000000000000000 x13: 746e696174206c65 x12: ffffffb801567cf9
[   17.096265] x11: 1ffffff801567cf8 x10: ffffffb801567cf8 x9 : dfffffc000000000
[   17.103572] x8 : ffffffc00ab3e7c7 x7 : 0000000000000001 x6 : ffffffb801567cf8
[   17.110875] x5 : ffffffc00ab3e7c0 x4 : ffffffb801567cf9 x3 : 0000000000000000
[   17.118178] x2 : 0000000000000020 x1 : ffffffc00a48e7c0 x0 : 0000000000000001
[   17.125481] Call trace:
[   17.127978]  skb_release_data+0x124/0x270
[   17.132083]  consume_skb+0x74/0xe0
[   17.135567]  __dev_kfree_skb_any+0x74/0x90
[   17.139764]  bcmgenet_desc_rx+0x4b4/0x620
[   17.143863]  bcmgenet_rx_poll+0x78/0x150
[   17.147873]  __napi_poll.constprop.0+0x64/0x240
[   17.152503]  net_rx_action+0x4d4/0x590
[   17.156338]  __do_softirq+0x228/0x4d8
[   17.160083]  __irq_exit_rcu+0x1e4/0x24c
[   17.164008]  irq_exit_rcu+0x20/0x54
[   17.167580]  el1_interrupt+0x38/0x50
[   17.171247]  el1h_64_irq_handler+0x18/0x2c
[   17.175444]  el1h_64_irq+0x64/0x68
[   17.178923]  arch_local_irq_enable+0xc/0x20
[   17.183197]  default_idle_call+0x80/0x114
[   17.187301]  cpuidle_idle_call+0x1e0/0x224
[   17.191490]  do_idle+0x104/0x14c
[   17.194793]  cpu_startup_entry+0x34/0x3c
[   17.198803]  rest_init+0x180/0x200
[   17.202283]  arch_post_acpi_subsys_init+0x0/0x30
[   17.207006]  start_kernel+0x3c8/0x400
[   17.210755]  __primary_switched+0xa8/0xb0
[   17.214872] Code: 72001c1f 540001e1 91002260 97d3086f (f9400660)
[   17.221083] ---[ end trace 0000000000000000 ]---
[   17.225791] Kernel panic - not syncing: Oops: Fatal exception in interrupt
[   17.232785] SMP: stopping secondary CPUs
[   17.236795] Kernel Offset: disabled
[   17.240348] CPU features: 0x100,00000d08,00001086
[   17.245143] Memory Limit: none
[   17.248273] ---[ end Kernel panic - not syncing: Oops: Fatal exception in interrupt ]---

This is at boot, over TFTP and NFS on a RaspberryPi4

Maxime

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 228 bytes --]