Re: [PATCH v2 2/3] platform/x86: wmi: add WMI support to MODULE_DEVICE_TABLE()

From: Mattias Jacobsson
Date: Mon Jan 28 2019 - 10:10:06 EST


On 2019-01-28, Pali Rohár wrote:
> On Monday 28 January 2019 15:09:11 Mattias Jacobsson wrote:
> > Hi,
> >
> > On 2019-01-27, Andy Shevchenko wrote:
> > > On Sun, Jan 27, 2019 at 9:04 PM Mattias Jacobsson <2pi@xxxxxx> wrote:
> > > >
> > > > The kernel provides the macro MODULE_DEVICE_TABLE() where driver authors
> > > > can specify their device type and their array of device_ids and thereby
> > > > trigger the generation of the appropriate MODULE_ALIAS() output. This is
> > > > opposed to having to specify one MODULE_ALIAS() for each device. The WMI
> > > > device type is currently not supported.
> > > >
> > > > While using MODULE_DEVICE_TABLE() does increase the complexity as well
> > > > as spreading out the implementation across the kernel, it does come with
> > > > some benefits too;
> > > > * It makes different drivers look more similar; if you can specify the
> > > > array of device_ids any device type specific input to MODULE_ALIAS()
> > > > will automatically be generated for you.
> > > > * It helps each driver avoid keeping multiple versions of the same
> > > > information in sync. That is, both the array of device_ids and the
> > > > potential multitude of MODULE_ALIAS()'s.
> > > >
> > > > Add WMI support to MODULE_DEVICE_TABLE() by adding info about struct
> > > > wmi_device_id in devicetable-offsets.c and add a WMI entry point in
> > > > file2alias.c.
> > > >
> > > > The type argument for MODULE_DEVICE_TABLE(type, name) is wmi.
> > >
> > > > +/* Looks like: wmi:guid */
> > > > +static int do_wmi_entry(const char *filename, void *symval, char *alias)
> > > > +{
> > > > + DEF_FIELD_ADDR(symval, wmi_device_id, guid_string);
> > > > + if (strlen(*guid_string) != WMI_GUID_STRING_LEN) {
> > > > + warn("Invalid WMI device id 'wmi:%s' in '%s'\n",
> > > > + *guid_string, filename);
> > > > + return 0;
> > > > + }
> > >
> > > > + if (snprintf(alias, 500, WMI_MODULE_PREFIX "%s", *guid_string) < 0) {
> > >
> > > What the point to use snprintf here with arbitrary buffer size if we
> > > exactly know 2 facts:
> > > 1. UUID string is 36 characters
> > > 2. buffer is long enough
> > >
> > > ?
> >
> > As long as no one changes the code, not much.
>
> At least instead of hardcoded number 500, you should use pass size of alias:

Just a note; 500 comes from a few lines below in the do_table()
function. It is the actual size of alias we get in do_wmi_entry().

>
> static int do_wmi_entry(const char *filename, void *symval, char *alias, size_t alias_size)
>
> if (snprintf(alias, alias_size, WMI_MODULE_PREFIX "%s", *guid_string) < 0) {

That is a good idea, but requires changing all other entry points.
I was thinking of defining a DO_ENTRY_ALIAS_SIZE macro to replace
all/any 500 in file2alias.c. However that is a separate patch.

>
> Or pass buffer of constant size and then you do not need to use snprintf:
>
> #define ALIAS_SIZE (sizeof(WMI_MODULE_PREFIX)+WMI_GUID_STRING_LEN)

I could use ALIAS_SIZE and add that to snprintf() instead of 500.
While I guess it is unlikely that alias ever will be changed to be too
short for us, using ALIAS_SIZE doesn't "guarantee" that we are within
the bounds of alias if anything changes.

>
> static int do_wmi_entry(const char *filename, void *symval, char alias[ALIAS_SIZE])
>
> This should not break even when code around changes.
>
> > >
> > > > + warn("Could not generate all MODULE_ALIAS's in '%s'\n",
> > > > + filename);
> > > > + return 0;
> > > > + }
> > > > + return 1;
> > > > +}
> > >
> > > --
> > > With Best Regards,
> > > Andy Shevchenko
> >
> > Thanks,
> > Mattias
>
> --
> Pali Rohár
> pali.rohar@xxxxxxxxx

Thanks,
Mattias