[PATCH 04/13] powerpc/476: add machine check handler for 47x core

[Josh Boyer]

On Wed, May 05, 2010 at 08:27:01AM -0400, Josh Boyer wrote:
>>+int machine_check_47x(struct pt_regs *regs)
>>+{
>>+	unsigned long reason = get_mc_reason(regs);
>>+	u32 mcsr;
>>+
>>+	printk(KERN_ERR "Machine check in kernel mode.\n");
>>+	if (reason & ESR_IMCP) {
>>+		printk(KERN_ERR
>>+		       "Instruction Synchronous Machine Check exception\n");
>>+		mtspr(SPRN_ESR, reason & ~ESR_IMCP);
>>+		return 0;
>>+	}
>>+	mcsr = mfspr(SPRN_MCSR);
>>+	if (mcsr & MCSR_IB)
>>+		printk(KERN_ERR "Instruction Read PLB Error\n");
>>+	if (mcsr & MCSR_DRB)
>>+		printk(KERN_ERR "Data Read PLB Error\n");
>>+	if (mcsr & MCSR_DWB)
>>+		printk(KERN_ERR "Data Write PLB Error\n");
>>+	if (mcsr & MCSR_TLBP)
>>+		printk(KERN_ERR "TLB Parity Error\n");
>>+	if (mcsr & MCSR_ICP) {
>>+		flush_instruction_cache();
>>+		printk(KERN_ERR "I-Cache Parity Error\n");
>>+	}
>>+	if (mcsr & MCSR_DCSP)
>>+		printk(KERN_ERR "D-Cache Search Parity Error\n");
>>+	if (mcsr & PPC47x_MCSR_GPR)
>>+		printk(KERN_ERR "GPR Parity Error\n");
>>+	if (mcsr & PPC47x_MCSR_FPR)
>>+		printk(KERN_ERR "FPR Parity Error\n");
>>+	if (mcsr & PPC47x_MCSR_IPR)
>>+		printk(KERN_ERR "Machine Check exception is imprecise\n");
>
>This function isn't guarded by CONFIG_PPC_47x, however the defines for 
>PPC47x_MCSR_* are in include/asm/reg_booke.h.  They were added that way in patch
>4.  As it stands, we get this when trying to build for plain 44x:
>
> CC      arch/powerpc/kernel/traps.o
>arch/powerpc/kernel/traps.c: In function 'machine_check_47x':
>arch/powerpc/kernel/traps.c:411: error: 'PPC47x_MCSR_GPR' undeclared (first use in this function)
>arch/powerpc/kernel/traps.c:411: error: (Each undeclared identifier is reported only once
>arch/powerpc/kernel/traps.c:411: error: for each function it appears in.)
>arch/powerpc/kernel/traps.c:413: error: 'PPC47x_MCSR_FPR' undeclared (first use in this function)
>arch/powerpc/kernel/traps.c:415: error: 'PPC47x_MCSR_IPR' undeclared (first use in this function)
>make[1]: *** [arch/powerpc/kernel/traps.o] Error 1
>make: *** [arch/powerpc/kernel] Error 2
>make: *** Waiting for unfinished jobs....
>
>
>We should probably guard this function behind CONFIG_PPC_47x, right?  If so, I
>can fix that up locally quickly enough.

Actually, that won't work unless we wrap the .machine_check addition in cputable
as well.  That looks ugly.  I'm guessing the easiest and cleanest fix is to
not guard the #defines for those regs in reg_booke.h.  Anyone have problems with
that?

josh