Re: [PATCH 1/7] dynamic debug v2 - infrastructure

[jbaron]

On Wed, Aug 13, 2008 at 10:00:47PM +0200, Sam Ravnborg wrote:
> On Tue, Aug 12, 2008 at 08:38:59PM -0700, Greg KH wrote:
> > On Tue, Aug 12, 2008 at 06:16:53PM -0700, Andrew Morton wrote:
> > > On Tue, 12 Aug 2008 18:08:04 -0700
> > > Greg KH <greg@xxxxxxxxx> wrote:
> > > 
> > > > On Tue, Aug 12, 2008 at 04:46:19PM -0400, Jason Baron wrote:
> > > > > On Tue, Aug 12, 2008 at 01:09:08PM -0700, Greg KH wrote:
> > > > > > So close, can I have a good changelog comment with the patch so people
> > > > > > know what it is when they look in the logs?
> > > > > > 
> > > > > > Care to resend it with that?
> > > > > > 
> > > > > 
> > > > > Base infrastructure to enable per-module debug messages.
> > > > 
> > > > Ah, so close...
> > > > 
> > > > With this patch, I get the following build error:
> > > > 
> > > >   CC [M]  drivers/usb/gadget/u_ether.o
> > > > drivers/usb/gadget/u_ether.c: In function ___gether_setup___:
> > > > drivers/usb/gadget/u_ether.c:787: error: ___KBUILD_MODNAME___ undeclared (first use in this function)
> > > > drivers/usb/gadget/u_ether.c:787: error: (Each undeclared identifier is reported only once
> > > > drivers/usb/gadget/u_ether.c:787: error: for each function it appears in.)
> > > > drivers/usb/gadget/u_ether.c:787: error: unknown field ___Usage___ specified in initializer
> > > > drivers/usb/gadget/u_ether.c:787: error: expected expression before ___.___ token
> > > > drivers/usb/gadget/u_ether.c:787: error: initializer element is not constant
> > > > drivers/usb/gadget/u_ether.c:787: error: (near initialization for ___descriptor.hash___)
> > > > drivers/usb/gadget/u_ether.c:787: error: ___Usage___ undeclared (first use in this function)
> > > > drivers/usb/gadget/u_ether.c:787: error: invalid operands to binary << (have ___long long int___ and ___char *___)
> > > > drivers/usb/gadget/u_ether.c:787: error: expected ___)___ before ___:___ token
> > > > drivers/usb/gadget/u_ether.c:787: error: invalid operands to binary & (have ___long long int___ and ___char *___)
> > > > drivers/usb/gadget/u_ether.c:787: error: invalid operands to binary << (have ___long long int___ and ___char *___)
> > > > drivers/usb/gadget/u_ether.c:787: error: expected ___)___ before ___:___ token
> > > > drivers/usb/gadget/u_ether.c:787: error: invalid operands to binary & (have ___long long int___ and ___char *___)
> > > > drivers/usb/gadget/u_ether.c:787: error: expected ___)___ before ___:___ token
> > > > drivers/usb/gadget/u_ether.c:787: warning: passing argument 4 of _____dynamic_dbg_enabled_helper___ makes integer from pointer without a cast
> > > > drivers/usb/gadget/u_ether.c:787: error: expected ___)___ before ___KBUILD_MODNAME___
> > > 
> > > (lol at gcc crap)
> > > 
> > > > drivers/usb/gadget/u_ether.c:787: warning: too few arguments for format
> > > 
> > > argh, not that again.  We had this with a patch from Jesse a while
> > > back, which I fixed thusly:
> > 
> > <patch moving KBUILD_MODNAME from a inline function to a macro snipped>
> > 
> > Yeah, I remember that too, but in this case, KBUILD_MODNAME is being
> > used in a macro, not an inline function, the exact opposite.
> > 
> > Sam, any ideas here?  I'll bounce the original patch at you if you
> > missed it last time.
> 
> I have not looked into the particular error but what happens is
> that u_ether.o is used by more than one module so kbuild does
> not know what value to assign the KBUILD_MODNAME and therefore
> decide not to assign KBUILD_MODNAME at all.
> 
> And because any usage of KBUILD_MODNAME is anyway flawed in u_ether.c
> as we do not know for what module it is being used it makes sense for
> kbuild to do so.
> 
> The above error happens because u_ether.c uses something that require KBUILD_MODNAME
> to be define.
> So the correct fix is to avid usage of this macro in the source.
> 
> Note: The old error Andrew mentions was caused by alpha indirectly
> pulling in a inline function that used KBUILD_MODNAME - something
> that did not happen on the other architectures.
> 
> 	Sam


the comment above KBUILD_MODNAME from scripts/Makefile.lib says:

# Note: It's possible that one object gets potentially linked into more
#       than one module. In that case KBUILD_MODNAME will be set to
#       foo_bar,
#       where foo and bar are the name of the modules.

That seems like a reasonable behavior to me...but that is not what the
code is doing...so at the very least the comment is wrong.

thanks,

-Jason

--
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/