xfs
[Top] [All Lists]

Re: CVS patches == b0rken SCSI drivers?

Subject: Re: CVS patches == b0rken SCSI drivers?
From: "D. Stimits" <stimits@xxxxxxxxxx>
Date: Sun, 16 Dec 2001 10:32:25 -0700
Cc: linux-xfs@xxxxxxxxxxx
References: <31889.1008503791@ocs3.intra.ocs.com.au>
Reply-to: stimits@xxxxxxxxxx
Sender: owner-linux-xfs@xxxxxxxxxxx
Keith Owens wrote:
> 
> On Sun, 16 Dec 2001 05:04:30 -0600,
> Kelledin <kelledin@xxxxxxxxxxxxxxxxxxxxx> wrote:
> >This is something that's annoyed me for a long time...the development
> >XFS patches apparently break a few SCSI drivers (like NCR and SymBios).
> 
> Because those drivers ship generated files which they then overwrite,
> which messes up source repositories.  The generated files have been
> deleted from the XFS source repository.  They should regenerate when
> required, but you need Perl to do so, and the existing makefiles do not
> correctly handle parallel running in the scsi directory.  Do not use
> make -j if you need to regenerate more than one set of scsi firmware.

I normally use aic7xxx and make -j3. Is it correct to say that the -j is
not a problem if I build only the aic7xxx and no other scsi?

D. Stimits, stimits@xxxxxxxxxx

PS: I haven't checked out kbuild 2.5, but I know it has been needed for
a long time. "make" is itself no better than a very sophisticated macro,
it tends to be annoying to get it to do what is needed on large
projected.

> 
> Plus there is a missing dependency in drivers/scsi/Makefile, although I
> finally got Linus to accept the patch for that in 2.4.15.
> 
> --- 14.1/drivers/scsi/Makefile
> +++ 15.1(w)/drivers/scsi/Makefile
> @@ -170,6 +174,8 @@ cpqfc.o: $(cpqfc-objs)
>         rm fake8.c
> 
>  53c8xx_u.h: 53c8xx_d.h
> +
> +53c7,8xx.o: 53c8xx_u.h
> 
>  53c7xx_d.h: 53c7xx.scr script_asm.pl
>         ln -sf 53c7xx.scr fake7.c
> 
> All the problems are caused by a bad scsi/Makefile.  Of course, this is
> all fixed in kbuild 2.5.


<Prev in Thread] Current Thread [Next in Thread>