Build errors caused by modalias generation patch

Build errors caused by modalias generation patch

[Ralf Baechle]

Andreas,

doing builds of all MIPS defconfigs I observed a fair number of them
were failing to build with errors like these:

cobalt_defconfig:

  CC [M]  drivers/hid/usbhid/hid-quirks.o
  LD [M]  drivers/hid/usbhid/usbhid.o
FATAL: drivers/hid/usbhid/usbhid: sizeof(struct usb_device_id)=32 is not a modulo of the size of section __mod_usb_device_table=48.
Fix definition of struct usb_device_id in mod_devicetable.h
make[5]: *** [drivers/hid/usbhid/usbhid.o] Error 1

gpr_defconfig:

  LD [M]  drivers/atm/fore_200e.o
FATAL: drivers/atm/fore_200e: sizeof(struct pci_device_id)=32 is not a modulo of the size of section __mod_pci_device_table=56.
Fix definition of struct pci_device_id in mod_devicetable.h
make[4]: *** [drivers/atm/fore_200e.o] Error 1

jazz_defconfig:

  Building modules, stage 2.
  MODPOST 355 modules
FATAL: drivers/block/floppy: sizeof(struct pnp_device_id)=16 is not a modulo of the size of section __mod_pnp_device_table=24.
Fix definition of struct pnp_device_id in mod_devicetable.h
make[3]: *** [__modpost] Error 1

malta_defconfig:

  LD [M]  drivers/net/ethernet/chelsio/cxgb3/cxgb3.o
FATAL: drivers/net/ethernet/chelsio/cxgb3/cxgb3: sizeof(struct pci_device_id)=32 is not a modulo of the size of section __mod_pci_device_table=392.
Fix definition of struct pci_device_id in mod_devicetable.h
make[7]: *** [drivers/net/ethernet/chelsio/cxgb3/cxgb3.o] Error 1
make[6]: *** [drivers/net/ethernet/chelsio/cxgb3] Error 2
make[5]: *** [drivers/net/ethernet/chelsio] Error 2

etc.

I can reproduce the issue can only building with a separate object directory.

Reverting 6543becf26fff612cdadeed7250ccc8d49f67f27 [mod/file2alias: make
modalias generation safe for cross compiling] fixes these for me.

  Ralf

Re: Build errors caused by modalias generation patch

[Andreas Schwab]

Please try the patch in
<http://marc.info/?l=linux-kbuild&m=136767800809256&w=2>.

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: Build errors caused by modalias generation patch

[Ralf Baechle]

On Mon, May 06, 2013 at 07:19:53PM +0200, Andreas Schwab wrote:

> Please try the patch in
> <http://marc.info/?l=linux-kbuild&m=136767800809256&w=2>.

No change in observed behaviour.  I did all my builds in empty object
directories so I don't see why this patch would make any difference.

  Ralf

Re: Build errors caused by modalias generation patch

[Geert Uytterhoeven]

On Tue, May 7, 2013 at 12:04 PM, Ralf Baechle <ralf@linux-mips.org> wrote:
> On Mon, May 06, 2013 at 07:19:53PM +0200, Andreas Schwab wrote:
>> Please try the patch in
>> <http://marc.info/?l=linux-kbuild&m=136767800809256&w=2>.
>
> No change in observed behaviour.  I did all my builds in empty object
> directories so I don't see why this patch would make any difference.

I'm getting this in an ia64 build (don't ask me why I build for itanic ;-):

FATAL: drivers/acpi/button: sizeof(struct acpi_device_id)=14 is not a
modulo of the size of section __mod_acpi_device_table=144.
Fix definition of struct acpi_device_id in mod_devicetable.h

This is next-20130501, with or without Andreas' fix.

It works in my development tree based on yesterday's mainline
(d7ab7302f970a254997687a1cdede421a5635c68).

I'll try today's linux-next, and will bisect if I find time...

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

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

Re: Build errors caused by modalias generation patch

[Geert Uytterhoeven]

On Tue, May 7, 2013 at 12:58 PM, Geert Uytterhoeven
<geert@linux-m68k.org> wrote:
> On Tue, May 7, 2013 at 12:04 PM, Ralf Baechle <ralf@linux-mips.org> wrote:
>> On Mon, May 06, 2013 at 07:19:53PM +0200, Andreas Schwab wrote:
>>> Please try the patch in
>>> <http://marc.info/?l=linux-kbuild&m=136767800809256&w=2>.
>>
>> No change in observed behaviour.  I did all my builds in empty object
>> directories so I don't see why this patch would make any difference.
>
> I'm getting this in an ia64 build (don't ask me why I build for itanic ;-):
>
> FATAL: drivers/acpi/button: sizeof(struct acpi_device_id)=14 is not a
> modulo of the size of section __mod_acpi_device_table=144.
> Fix definition of struct acpi_device_id in mod_devicetable.h
>
> This is next-20130501, with or without Andreas' fix.
>
> It works in my development tree based on yesterday's mainline
> (d7ab7302f970a254997687a1cdede421a5635c68).
>
> I'll try today's linux-next, and will bisect if I find time...

This was also caused by a leftover scripts/mod/devicetable-offsets.h in the
source tree.

Andreas' fix works (sort of, read on), but only if you run "make mrproper"
_after_ applying it.

Still, if you have other leftovers from a previous build in the source tree,
you get:

  Using /scratch/geert/linux/linux-next as source for kernel
  /scratch/geert/linux/linux-next is not clean, please run 'make mrproper'
  in the '/scratch/geert/linux/linux-next' directory.

However, you don't get that message if there's still a stale
scripts/mod/devicetable-offsets.h in the source tree.
Any idea how to fix that?

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

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

Re: Build errors caused by modalias generation patch

[Geert Uytterhoeven]

On Tue, May 7, 2013 at 12:58 PM, Geert Uytterhoeven
<geert@linux-m68k.org> wrote:
> On Tue, May 7, 2013 at 12:04 PM, Ralf Baechle <ralf@linux-mips.org> wrote:
>> On Mon, May 06, 2013 at 07:19:53PM +0200, Andreas Schwab wrote:
>>> Please try the patch in
>>> <http://marc.info/?l=linux-kbuild&m=136767800809256&w=2>.
>>
>> No change in observed behaviour.  I did all my builds in empty object
>> directories so I don't see why this patch would make any difference.
>
> I'm getting this in an ia64 build (don't ask me why I build for itanic ;-):
>
> FATAL: drivers/acpi/button: sizeof(struct acpi_device_id)=14 is not a
> modulo of the size of section __mod_acpi_device_table=144.
> Fix definition of struct acpi_device_id in mod_devicetable.h
>
> This is next-20130501, with or without Andreas' fix.
>
> It works in my development tree based on yesterday's mainline
> (d7ab7302f970a254997687a1cdede421a5635c68).
>
> I'll try today's linux-next, and will bisect if I find time...

This was also caused by a leftover scripts/mod/devicetable-offsets.h in the
source tree.

Andreas' fix works (sort of, read on), but only if you run "make mrproper"
_after_ applying it.

Still, if you have other leftovers from a previous build in the source tree,
you get:

  Using /scratch/geert/linux/linux-next as source for kernel
  /scratch/geert/linux/linux-next is not clean, please run 'make mrproper'
  in the '/scratch/geert/linux/linux-next' directory.

However, you don't get that message if there's still a stale
scripts/mod/devicetable-offsets.h in the source tree.
Any idea how to fix that?

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

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

Re: Build errors caused by modalias generation patch

[Andreas Schwab]

Ralf Baechle <ralf@linux-mips.org> writes:

> cobalt_defconfig:
>
>   CC [M]  drivers/hid/usbhid/hid-quirks.o
>   LD [M]  drivers/hid/usbhid/usbhid.o
> FATAL: drivers/hid/usbhid/usbhid: sizeof(struct usb_device_id)=32 is not a modulo of the size of section __mod_usb_device_table=48.
> Fix definition of struct usb_device_id in mod_devicetable.h

What are the contents of devicetable-offsets.[sh]?

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."