[patch] Paul's rsync kernel compile fix

[patch] Paul's rsync kernel compile fix

[Shaw Terwilliger]

[-- Attachment #1: Type: text/plain, Size: 621 bytes --]

I've been compiling Paul's rsync kernel on my PowerBook G3 and I've
prepared a patch that blocks out some undeclared (and reference)
variable initializations in linux/arch/ppc/kernel/pmac_setup.c.  A quick
look at the other ????_setup.c (prep, chrp, apus) shows this might also
be useful there, but I didn't run through the compiles to check.

Has everyone else had the same problem compiling these kernels?  

This is the first time I've dug around in "kernel" kernel stuff,
and I think I'm clear on why these variables really don't mean
anything, but I could always be wrong.  :)

-- 
Shaw Terwilliger (sterwill@io.nu)

[-- Attachment #2: pmac_setup.c.diff --]
[-- Type: text/plain, Size: 591 bytes --]

--- pmac_setup.c.orig	Fri Sep 24 18:27:25 1999
+++ pmac_setup.c	Fri Sep 24 18:29:27 1999
@@ -594,9 +594,15 @@
 	/* isa_io_base gets set in pmac_find_bridges */
 	isa_mem_base = PMAC_ISA_MEM_BASE;
 	pci_dram_offset = PMAC_PCI_DRAM_OFFSET;
+
+	/* These three variables don't seem to be used in pmac_setup.c and 
+	   aren't declared anywhere.  Do we need them at all?  They're 
+	   declared "unsigned long" in arch/ppc/kernel/setup.c */
+/*
 	ISA_DMA_THRESHOLD = ~0L;
 	DMA_MODE_READ = 1;
 	DMA_MODE_WRITE = 2;
+*/
 
 	ppc_md.setup_arch     = pmac_setup_arch;
 	ppc_md.setup_residual = NULL;

Re: [patch] Paul's rsync kernel compile fix

[Martin Costabel]

Shaw Terwilliger wrote:
> 
> I've been compiling Paul's rsync kernel on my PowerBook G3 and I've
> prepared a patch that blocks out some undeclared (and reference)
> variable initializations in linux/arch/ppc/kernel/pmac_setup.c.  A quick
> look at the other ????_setup.c (prep, chrp, apus) shows this might also
> be useful there, but I didn't run through the compiles to check.
> 
> Has everyone else had the same problem compiling these kernels?
> 
> This is the first time I've dug around in "kernel" kernel stuff,
> and I think I'm clear on why these variables really don't mean
> anything, but I could always be wrong.  :)

I think you are: 
- these variables are used by any number of drivers
- they are declared in <asm/dma.h>

Maybe you should rsync again?

--
Martin

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: [patch] Paul's rsync kernel compile fix

[Shaw Terwilliger]

Michael Fenske wrote:
> I rsync'd today. Nevertheless I get warnings various variables are
> declared but not defined or defined but not used. I think (hope) this is
> because I selected only the necessary drivers for my Wallstreet PB to
> compile into the kernel.

The patch I keep around actually fixes errors in my compile.  Without
it, pmac_setup.c is referencing variables that aren't defined
anywhere in its scope.  The variables are declared in asm/dma.h, 
which isn't included.  I just can't compile a kernel without a fix.  :)


-- 
Shaw Terwilliger (sterwill@io.nu)

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: [patch] Paul's rsync kernel compile fix

[Shaw Terwilliger]

Martin Costabel wrote:
> Oh, I see. We are indeed not talking about the same thing. I have the
> development tree (2.3.18), and there arch/ppc/kernel/pmac_setup.c does
> include <asm/dma.h>. 
> What if you just plug a "#include <asm/dma.h>" in there? But then I
> don't know whether the declarations in question are there in this
> version.

Yep, dma.h holds the declarations, and sticking it in pmac_setup.c
corrects my problems.  I thought there must have been a much more
complicated solution because I couldn't imagine I was the only one
seeing this error.  :)

> BTW, Paul's 2.3.18 compiles cleanly (and boots, in contrast to the
> 2.3.18 version from the cvs tree at openprojects), at least if HFS
> support is switched off.

I've got 2.2.12 working very nicely with my Lombard, and generally
I'm aware of the differences between stable/dev kernels, but is
there anything really cool in 2.3.18 for the PPC (from Paul's
archive) that might lead me to switch to it?

-- 
Shaw Terwilliger (sterwill@io.nu)

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: [patch] Paul's rsync kernel compile fix

[Michael Fenske]

Martin Costabel wrote:
> 
> I think you are:
> - these variables are used by any number of drivers
> - they are declared in <asm/dma.h>
> 
> Maybe you should rsync again?

I rsync'd today. Nevertheless I get warnings various variables are
declared but not defined or defined but not used. I think (hope) this is
because I selected only the necessary drivers for my Wallstreet PB to
compile into the kernel.

Michael


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/