RE: [PATCH net v2] can: flexcan: fix synchronous external abort in flexcan_get_berr_counter()

-----Original Message-----
From: ChunYuZhiQiang <redacted>
Sent: Tuesday, September 29, 2026 3:13 PM
To: netdev@vger.kernel.org
Cc: linux-can@vger.kernel.org; mkl@pengutronix.de; kernel@pengutronix.de;
Bough Chen (OSS) [off-list ref]; shawnguo@kernel.org;
imx@lists.linux.dev; ChunYuZhiQiang [off-list ref]
Subject: [PATCH net v2] can: flexcan: fix synchronous external abort in
flexcan_get_berr_counter()

Bad title.

When the interface is down, can_fill_info() still calls
flexcan_get_berr_counter() to fill the error counters for a netlink dump. On
some SoCs (e.g. i.MX8QXP) this can trigger a synchronous external abort.

The root cause is platform-specific clocking. On i.MX8QXP the CAN clocks are
gated through the SCU LPCG and the whole subsystem shares the
IMX_SC_R_CAN_0 power domain. flexcan_runtime_resume() only calls
flexcan_clks_enable(); combined with the LPCG e10858 synchronization
erratum, there is a window where pm_runtime_resume_and_get() has
returned but the clock has not actually reached the module yet. The
immediate readl(&regs->ecr) in that window hits an un-clocked target and
faults with the external abort:

    Internal error: synchronous external abort: 0000000096000210 [#1]
PREEMPT SMP
    pc : flexcan_read_le+0x0/0x18
    lr : flexcan_get_berr_counter+0x4c/0x8c
    Call trace:
     flexcan_read_le+0x0/0x18
     can_fill_info+0x1f8/0x434
     rtnl_fill_ifinfo+0x8fc/0xbb4
     rtnl_dump_ifinfo+0x364/0x448
     ...

This is the same class of problem that was already fixed for m_can:

  commit 91a55c72a821d ("can: m_can: m_can_get_berr_counter(): don't
  wake up controller if interface is down")

Remove this in commit log.

Fix this by returning early if the interface is not running.

Tested on i.MX8QXP: after the patch, `ip link show` no longer triggers the
abort.

Fixes: ec56acfef2af1 ("can: flexcan: flexcan_get_berr_counter(): switch on
clocks before accessing ecr register")
Link: https://lore.kernel.org/netdev/20260915063511.3010642-1-
chunyuzhiqiang@yinhe.ht/

Please drop this link. It will be added by the maintainer when the patch is merged.

For the commit, I suggest you change like this:

can: flexcan: flexcan_get_berr_counter: don't wakeup controller if interface is down

If the interface is down, the CAN controller might be powered down, the
 clock disabled, and/or it's external reset asserted.

Don't wake up the controller to read the CAN bus error counters, if the
interface is down.


I don't think we need to mention the 8QXP dump in the commit log. 
The 8QXP-specific erratum only served as a trigger for uncovering this issue.
The patch itself fixes a generic logic problem that exists independently of the
 erratum. For example, even on i.MX95, which is not affected by the 8QXP 
erratum, the same logical fix is still needed.

Regards
Haibo Chen