Re: [PATCH v2 0/2] acpi/nfit: Fix command-supported detection

[Jeff Moyer]

Dan Williams <dan.j.williams@xxxxxxxxx> writes:

> On Tue, Jan 15, 2019 at 6:16 AM Jeff Moyer <jmoyer@xxxxxxxxxx> wrote:
>>
>> Dan Williams <dan.j.williams@xxxxxxxxx> writes:
>>
>> > Changes since v1 [1]:
>> > * Include another patch make sure that function-number zero can be
>> >   safely used as an invalid function number (Jeff)
>> > * Add a comment clarifying why zero is an invalid function number (Jeff)
>> > * Pass nfit_mem to cmd_to_func() (Jeff)
>> > * Collect a Tested-by from Sujith
>> > [1]: https://lists.01.org/pipermail/linux-nvdimm/2019-January/019435.html
>>
>> For the series:
>>
>> Reviewed-by: Jeff Moyer <jmoyer@xxxxxxxxxx>
>>
>> Thanks, Dan!
>
> Thanks, although I just realized one more change. The ND_CMD_CALL case
> should zero out command after the function translation, otherwise
> userspace can call functions that the kernel is blocking in the
> dsm_mask.
>
> Holler if this invalidates your "Reviewed-by".

AAAAAAAHHHHHHHHHH!!!!

:)

> diff --git a/drivers/acpi/nfit/core.c b/drivers/acpi/nfit/core.c
> index 87e02f281e51..d7747aceb7ab 100644
> --- a/drivers/acpi/nfit/core.c
> +++ b/drivers/acpi/nfit/core.c
> @@ -463,6 +463,12 @@ int acpi_nfit_ctl(struct nvdimm_bus_descriptor
> *nd_desc, struct nvdimm *nvdimm,
>                 func = cmd_to_func(nfit_mem, cmd, buf);
>                 if (func < 0)
>                         return func;
> +               /*
> +                * In the ND_CMD_CALL case we're now dependent on 'func'
> +                * being validated by the dimm's dsm_mask
> +                */
> +               if (cmd == ND_CMD_CALL)
> +                       cmd = 0;
>                 dimm_name = nvdimm_name(nvdimm);
>                 cmd_name = nvdimm_cmd_name(cmd);
>                 cmd_mask = nvdimm_cmd_mask(nvdimm);
                dsm_mask = nfit_mem->dsm_mask;
                desc = nd_cmd_dimm_desc(cmd);

That sure doesn't look right.  Now cmd_name and desc will be wrong.

> @@ -477,8 +483,10 @@ int acpi_nfit_ctl(struct nvdimm_bus_descriptor
> *nd_desc, struct nvdimm *nvdimm,
>                 cmd_name = nvdimm_bus_cmd_name(cmd);
>                 cmd_mask = nd_desc->cmd_mask;
>                 dsm_mask = cmd_mask;
> -               if (cmd == ND_CMD_CALL)
> +               if (cmd == ND_CMD_CALL) {
>                         dsm_mask = nd_desc->bus_dsm_mask;
> +                       cmd = 0;
> +               }
>                 desc = nd_cmd_bus_desc(cmd);

And again here.

We could reorder the zeroing, or you could modify the check for a valid
comand/function.  Something like this?

	/*
         * Check for a valid command.  For ND_CMD_CALL, we also
         * have to make sure that the DSM function is supported.
         */
	if (cmd == ND_CMD_CALL && !test_bit(func, &dsm_mask))
        	return -ENOTTY;
	else if (!test_bit(cmd, &cmd_mask))
		return -ENOTTY;

Which way do you think is cleaner?

Cheers,
Jeff