Re: [PATCH] mod/file2alias: make modalias generation safe for crosscompiling

[Geert Uytterhoeven]

	Hi Andreas,

On Mon, 25 Jun 2012, Andreas Schwab wrote:
> Use the target compiler to compute the offsets of the fields of the
> device_id structures, so that it won't be broken by different alignments
> between the host and target ABIs.

Thanks, very nice!

> --- a/include/linux/mod_devicetable.h
> +++ b/include/linux/mod_devicetable.h
> @@ -33,8 +33,7 @@ struct ieee1394_device_id {
>  	__u32 model_id;
>  	__u32 specifier_id;
>  	__u32 version;
> -	kernel_ulong_t driver_data
> -		__attribute__((aligned(sizeof(kernel_ulong_t))));
> +	kernel_ulong_t driver_data;

After this, we can also get rid of kernel_ulong_t, but that's a more intruisive
patch all over the place.

> --- /dev/null
> +++ b/scripts/mod/devicetable-offsets.c
> @@ -0,0 +1,171 @@
> +#include <linux/kbuild.h>
> +#include <linux/mod_devicetable.h>
> +
> +#define DEVID(devid) DEFINE(SIZE_##devid, sizeof(struct devid))
> +#define DEVID_FIELD(devid, field) \
> +	DEFINE(OFF_##devid##_##field, offsetof(struct devid, field))
> +
> +int main(void)
> +{
> +	DEVID(usb_device_id);
> +	DEVID_FIELD(usb_device_id, match_flags);
> +	DEVID_FIELD(usb_device_id, idVendor);
> +	DEVID_FIELD(usb_device_id, idProduct);
> +	DEVID_FIELD(usb_device_id, bcdDevice_lo);
> +	DEVID_FIELD(usb_device_id, bcdDevice_hi);
> +	DEVID_FIELD(usb_device_id, bDeviceClass);
> +	DEVID_FIELD(usb_device_id, bDeviceSubClass);
> +	DEVID_FIELD(usb_device_id, bDeviceProtocol);
> +	DEVID_FIELD(usb_device_id, bInterfaceClass);
> +	DEVID_FIELD(usb_device_id, bInterfaceSubClass);
> +	DEVID_FIELD(usb_device_id, bInterfaceProtocol);

These are all duplicates of the structures in <linux/mod_devicetable.h>.
Can't you get rid of the duplication by putting e.g. the following
in <linux/mod_devicetable.h>:

    DEVID_START(usb_device_id);
    DEVID_FIELD(usb_device_id, __u16, match_flags);
    DEVID_FIELD(usb_device_id, __u16, idVendor);
    DEVID_FIELD(usb_device_id, __u16, idProduct);
    DEVID_FIELD(usb_device_id, __u16, bcdDevice_lo);
    DEVID_FIELD(usb_device_id, __u16, bcdDevice_hi);
    DEVID_FIELD(usb_device_id, __u8, bDeviceClass);
    DEVID_FIELD(usb_device_id, __u8, bDeviceSubClass);
    DEVID_FIELD(usb_device_id, __u8, bDeviceProtocol);
    DEVID_FIELD(usb_device_id, __u8, bInterfaceClass);
    DEVID_FIELD(usb_device_id, __u8, bInterfaceSubClass);
    DEVID_FIELD(usb_device_id, __u8, bInterfaceProtocol);
    DEVID_END();

and redefining the DEVID_*() macros depending on
  1. __KERNEL__, to generate the C struct definitions for kernel builds,
  2. first inclusion by devicetable-offsets.c, also to generate C struct
     definitions,
  3. second inclusion by devicetable-offsets.c, to generate the sizes and
     offset definitions for modalias?

> --- a/scripts/mod/file2alias.c
> +++ b/scripts/mod/file2alias.c
> @@ -116,7 +129,6 @@ static inline void add_wildcard(char *str)
>  		strcat(str + len, "*");
>  }
>  
> -unsigned int cross_build = 0;
>  /**
>   * Check that sizeof(device_id type) are consistent with size of section
>   * in .o file. If in-consistent then userspace and kernel does not agree
> @@ -131,8 +143,6 @@ static void device_id_check(const char *modname, const char *device_id,
>  	int i;
>  
>  	if (size % id_size || size < id_size) {
> -		if (cross_build != 0)
> -			return;
>  		fatal("%s: sizeof(struct %s_device_id)=%lu is not a modulo "
>  		      "of the size of section __mod_%s_device_table=%lu.\n"
>  		      "Fix definition of struct %s_device_id "

Funny, I didn't know there was a check for cross-compiling, supposedly to
ignore the warnings.
And it didn't work, as the problem was noticed originally on the linux-next
build service (which uses cross-compiling), and also showed up for my own
cross-builds.

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/