diff options
author | Andrew Morton <akpm@linux-foundation.org> | 2008-07-30 12:07:04 -0700 |
---|---|---|
committer | Jesse Barnes <jbarnes@virtuousgeek.org> | 2008-08-07 06:52:01 -0700 |
commit | bba81165867313766534dd31603de51bdd36ef9b (patch) | |
tree | f6233ae2f717c49b8642009f1762599d30415a9c /kernel/dma-coherent.c | |
parent | 9d82d8eaec032bf935144e0d5789a5cd4b95e958 (diff) |
PCI: make pci_register_driver() a macro
alpha:
CC [M] drivers/usb/gadget/u_ether.o
In file included from include/asm/dma-mapping.h:7,
from include/linux/dma-mapping.h:52,
from include/linux/dmaengine.h:29,
from include/linux/skbuff.h:29,
from include/linux/if_ether.h:114,
from include/linux/etherdevice.h:27,
from drivers/usb/gadget/u_ether.c:29:
include/linux/pci.h: In function 'pci_register_driver':
include/linux/pci.h:673: error: 'KBUILD_MODNAME' undeclared (first use in this function)
include/linux/pci.h:673: error: (Each undeclared identifier is reported only once
include/linux/pci.h:673: error: for each function it appears in.)
Sam says:
The problem is that u_ether.o is used by two modules so when we build it
KBUILD_MODNAME is not defined because kbuild does not know what value to
use.
And in pci.h we have the following inline:
static inline int __must_check pci_register_driver(struct pci_driver *driver)
{
return __pci_register_driver(driver, THIS_MODULE, KBUILD_MODNAME);
}
And alpha uses dma-mapping.h to nullify a number of functions that seem to
require something from pci.h.
Making it a macro fixes this particular problem. However, the underlying issue
of a file using KBUILD_MODNAME and being shared between multiple modules is
*not* addressed. I guess the answer there is "don't do that".
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Diffstat (limited to 'kernel/dma-coherent.c')
0 files changed, 0 insertions, 0 deletions