Re: [coreboot] [RFC] Error out on implicit declarations

[Myles Watson]

> -----Original Message-----
> From: Carl-Daniel Hailfinger [mailto:[e-mail]]
> Sent: Monday, December 22, 2008 6:48 AM
> To: Myles Watson
> Cc: Marc Jones; Coreboot; [e-mail]
> Subject: Re: [coreboot] [RFC] Error out on implicit declarations
>
> On 18.12.2008 19:33, Myles Watson wrote:
> > On Thu, Dec 18, 2008 at 11:08 AM, Marc Jones  wrote:
> > > On Thu, Dec 18, 2008 at 10:22 AM, Myles Watson
>
> wrote:
> > > > On Thu, Dec 18, 2008 at 7:52 AM, Jordan Crouse
>
> <[e-mail]
> > > > wrote:
> > > > > Myles Watson wrote:
> > > > > > On Thu, Dec 18, 2008 at 3:18 AM, Carl-Daniel Hailfinger <
> > > > > > [e-mail]> wrote:
> > > > > >
> > > > > > > Bao, Zheng found a bug which killed SATA booting on my board.
> > > > > > >
> > > > > > > This happened because we do not error out on implicit function
> > > > > > > declarations. The linker has no way of checking whether the
>
> implicitly
> > > > > > > assumed function signature is identical to the real signature, so
> > > > > > > mismatches can occur and these mismatches are practically
>
> impossible
> > > > > > > to debug because the code looks completely correct.
> > > > > > >
> > > > > > > Adding -Werror-implicit-function-declaration to our CFLAGS would
>
> solve
> > > > > > > this problem nicely, but a lot of files in the tree need to be
>
> fixed.
> > > > > > >
> > > > > > >
> > > > > > I think this is a great idea.  Isn't the correct order to fix all
>
> the
> > > > > > warnings, then make it an error?
> > > > >
> > > > > Yeah - the unfortunate thing about changes like this is that you end
>
> up
> > > > > being responsible for fixing the errors.. :)
> > > > > Carl-Daniel - if you post a list of offending files, we'll all help
> > > > > clear them up.  Dumping the log through grep "implicit
> > > > > declaration of function" should suffice.
> >
> > I'm attaching a patch which I thought was trivial, but breaks things all
> > over the place.  I'm not sure how this will need to be done.  It looks
>
> like
> > a problem with romcc vs. CAR.
>
> Try this patch. It passes abuild.
>
> Regards,
> Carl-Daniel
>
> Fix implicit declarations of pci_read_config8 and pci_write_config8 in
> the following files:
> src/mainboard/intel/jarrell/reset.c
> src/mainboard/supermicro/x6dai_g/reset.c
> src/mainboard/supermicro/x6dhe_g2/reset.c
> src/mainboard/supermicro/x6dhe_g/reset.c
> src/mainboard/supermicro/x6dhr_ig2/reset.c
> src/mainboard/supermicro/x6dhr_ig/reset.c
>
> Signed-off-by: Carl-Daniel Hailfinger

Acked-by: Myles Watson

Thanks,
Myles