[RFC PATCH] PCI: imx6: Handle the abort from user-space

[Lucas Stach]

Am Freitag, dem 28.01.2022 um 09:29 +0100 schrieb Francesco Dolcini:
> From: Jason Liu <jason.hui.liu at nxp.com>
> 
> The driver install one hook to handle the external abort, but issue
> is that if the abort introduced from user space code, the following
> code unsigned long instr = *(unsigned long *)pc; which will created
> another data-abort(page domain fault) if CONFIG_CPU_SW_DOMAIN_PAN.
> 
> The patch does not intent to use copy_from_user and then do the hack
> due to the security consideration. In fact, we can just return and
> report the external abort to user-space.
> 
I'm not sure how userspace would be able to trigger this abort. Maybe
invalid access to a device cfg space via sysfs?

However the patch seems to do the right thing in that case.

Acked-by: Lucas Stach <l.stach at pengutronix.de>
 
> Signed-off-by: Jason Liu <jason.hui.liu at nxp.com>
> Reviewed-by: Richard Zhu <hongxing.zhu at nxp.com>
> Signed-off-by: Francesco Dolcini <francesco.dolcini at toradex.com>
> ---
> We have this patch from NXP downstream kernel [1] in our kernel branch [2]
> since a long time, to me it would make sense to upstream it. Any concern?
> 
> [1] https://source.codeaurora.org/external/imxsupport/linux-imx/commit/?id=62dfb2fb953463dd1b6710567c9e174672a98f24
> [2] https://git.toradex.com/cgit/linux-toradex.git/commit/?id=2b42547cf659f979be2defdff6a99f921b33d0f1
> ---
>  drivers/pci/controller/dwc/pci-imx6.c | 12 ++++++++++--
>  1 file changed, 10 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/pci/controller/dwc/pci-imx6.c b/drivers/pci/controller/dwc/pci-imx6.c
> index 6974bd5aa116..6b178a29e253 100644
> --- a/drivers/pci/controller/dwc/pci-imx6.c
> +++ b/drivers/pci/controller/dwc/pci-imx6.c
> @@ -27,6 +27,7 @@
>  #include <linux/resource.h>
>  #include <linux/signal.h>
>  #include <linux/types.h>
> +#include <linux/uaccess.h>
>  #include <linux/interrupt.h>
>  #include <linux/reset.h>
>  #include <linux/phy/phy.h>
> @@ -297,8 +298,15 @@ static int imx6q_pcie_abort_handler(unsigned long addr,
>  		unsigned int fsr, struct pt_regs *regs)
>  {
>  	unsigned long pc = instruction_pointer(regs);
> -	unsigned long instr = *(unsigned long *)pc;
> -	int reg = (instr >> 12) & 15;
> +	unsigned long instr;
> +	int reg;
> +
> +	/* if the abort from user-space, just return and report it */
> +	if (user_mode(regs))
> +		return 1;
> +
> +	instr = *(unsigned long *)pc;
> +	reg = (instr >> 12) & 15;
>  
>  	/*
>  	 * If the instruction being executed was a read,