On Thu, Apr 30, 2020 at 11:54:15AM +0530, Allen wrote:
> Hi,
>
> $ rmmod dsa-loop
> [ 50.688935] Unable to handle kernel read from unreadable memory at
> virtual address 0000000000000040
> [ 50.718185] pstate: 60400005 (nZCv daif +PAN -UAO)
> [ 50.719274] pc : __dev_set_rx_mode+0x48/0xa0
ffff8000113adb60 <__dev_set_rx_mode>:
Function starts at ffff8000113adb60, the instruction with problems is
ffff8000113adb60+48.
{
ffff8000113adb60: a9bd7bfd stp x29, x30, [sp,#-48]!
ffff8000113adb64: 910003fd mov x29, sp
ffff8000113adb68: a90153f3 stp x19, x20, [sp,#16]
ffff8000113adb6c: f90013f5 str x21, [sp,#32]
ffff8000113adb70: aa0003f3 mov x19, x0
ffff8000113adb74: aa1e03e0 mov x0, x30
ffff8000113adb78: 94000000 bl 0 <_mcount>
ffff8000113adb78: R_AARCH64_CALL26 _mcount
if (!(dev->flags&IFF_UP))
ffff8000113adb7c: b9422a60 ldr w0, [x19,#552]
const struct net_device_ops *ops = dev->netdev_ops;
ffff8000113adb80: f940fa74 ldr x20, [x19,#496]
if (!(dev->flags&IFF_UP))
ffff8000113adb84: 360001a0 tbz w0, #0, ffff8000113adbb8
<__dev_set_rx_mode+0x58>
ffff8000113adb88: f9402260 ldr x0, [x19,#64]
if (!netif_device_present(dev))
ffff8000113adb8c: 36080160 tbz w0, #1, ffff8000113adbb8
<__dev_set_rx_mode+0x58>
if (!(dev->priv_flags & IFF_UNICAST_FLT)) {
ffff8000113adb90: b9422e60 ldr w0, [x19,#556]
ffff8000113adb94: 376000a0 tbnz w0, #12, ffff8000113adba8
<__dev_set_rx_mode+0x48>
if (!netdev_uc_empty(dev) && !dev->uc_promisc) {
ffff8000113adb98: b9429275 ldr w21, [x19,#656]
ffff8000113adb9c: 35000175 cbnz w21, ffff8000113adbc8
<__dev_set_rx_mode+0x68>
} else if (netdev_uc_empty(dev) && dev->uc_promisc) {
ffff8000113adba0: 3949f660 ldrb w0, [x19,#637]
ffff8000113adba4: 35000220 cbnz w0, ffff8000113adbe8
<__dev_set_rx_mode+0x88>
if (ops->ndo_set_rx_mode)
ffff8000113adba8: f9402281 ldr x1, [x20,#64]
Which suggests ops is NULL. The #64 also matches with
> [ 50.688935] Unable to handle kernel read from unreadable memory at
> virtual address 0000000000000040
How did dev->netdev_ops become NULL?
Odd
Andrew
