Re: kobject_register failed for Promise_Old_IDE (-17)

[Greg KH <greg@kroah.com>]

On Fri, Jan 27, 2006 at 12:01:52AM +0100, Adrian Bunk wrote:
> On Wed, Jan 25, 2006 at 09:54:40AM +0200, Vasil Kolev wrote:
> > ?? ????, 2006-01-24 ?? 23:58 +0100, Bartlomiej Zolnierkiewicz ????????????:
> > > On 1/24/06, Greg KH <greg@kroah.com> wrote:
> > > > On Tue, Jan 24, 2006 at 11:08:47AM +0200, Vasil Kolev wrote:
> > > > > Hello,
> > > > > I have a machine that's currently running 2.4.28 with the promise_old
> > > > > driver, which runs ok. I tried upgrading it last night to 2.6.15, and
> > > > > the following error occured, and no drives were detected/made available:
> > > > >
> > > > >  [17179598.940000] kobject_register failed for Promise_Old_IDE (-17)
> > > > >  [17179598.940000]  [dump_stack+21/23] dump_stack+0x15/0x17
> > > > >  [17179598.940000]  [kobject_register+52/64] kobject_register+0x34/0x40
> > > > >  [17179598.940000]  [bus_add_driver+69/163] bus_add_driver+0x45/0xa3
> > > > >  [17179598.940000]  [driver_register+57/60] driver_register+0x39/0x3c
> > > > >  [17179598.940000]  [__pci_register_driver+125/132] __pci_register_driver+0x7d/0x84
> > > > >  [17179598.940000]  [__ide_pci_register_driver+19/53] __ide_pci_register_driver+0x13/0x35
> > > > >  [17179598.940000]  [pg0+945449588/1069855744] pdc202xx_ide_init+0x12/0x16 [pdc202xx_old]
> > > > >  [17179598.940000]  [sys_init_module+193/430] sys_init_module+0xc1/0x1ae
> > > > >  [17179598.940000]  [syscall_call+7/11] syscall_call+0x7/0xb
> > > >
> > > > This means that some other driver tried to register with the same exact
> > > > name for the same bus.  As it looks like this is the ide bus, I suggest
> > > > asking on the linux ide mailing list.
> > > 
> > 
> > Well, now I remember that in /sys in the proper place there were two
> > directories called Promise_Old_IDE, maybe the driver tried to register
> > twice?
> >...
> 
> Greg, IIRC, weren't there plans to turn this case into a BUG()?

No, we dump the stack trace so that people can see what is happening,
but if the caller has done their error handling correctly, the kernel
will not crash.

I don't like adding BUG() calls for stuff that it should not be needed
for (like this.)

thanks,

greg k-h