From: Yan Burman <burman.yan@xxxxxxxxx>
Date: Fri, 24 Nov 2006 21:02:53 +0200
> Stefan Richter wrote:
> > Yan Burman wrote:
> > ...
> >
> >> --- linux-2.6.19-rc5_orig/arch/sparc/kernel/sun4d_irq.c 2006-11-09 12:16:21.000000000 +0200
> >> +++ linux-2.6.19-rc5_kzalloc/arch/sparc/kernel/sun4d_irq.c 2006-11-11 22:44:04.000000000 +0200
> >> @@ -545,8 +545,7 @@ void __init sun4d_init_sbi_irq(void)
> >> nsbi = 0;
> >> for_each_sbus(sbus)
> >> nsbi++;
> >> - sbus_actions = (struct sbus_action *)kmalloc (nsbi * 8 * 4 * sizeof(struct sbus_action), GFP_ATOMIC);
> >> - memset (sbus_actions, 0, (nsbi * 8 * 4 * sizeof(struct sbus_action)));
> >> + sbus_actions = kzalloc (nsbi * 8 * 4 * sizeof(struct sbus_action), GFP_ATOMIC);
> >> for_each_sbus(sbus) {
> >> #ifdef CONFIG_SMP
> >> extern unsigned char boot_cpu_id;
> >>
> >
> > I'm not sure about this ^ hunk, but...
> >
> >
> >> diff -rubp linux-2.6.19-rc5_orig/arch/sparc/mm/io-unit.c linux-2.6.19-rc5_kzalloc/arch/sparc/mm/io-unit.c
> >> --- linux-2.6.19-rc5_orig/arch/sparc/mm/io-unit.c 2006-11-09 12:16:21.000000000 +0200
> >> +++ linux-2.6.19-rc5_kzalloc/arch/sparc/mm/io-unit.c 2006-11-11 22:44:04.000000000 +0200
> >> @@ -41,9 +41,8 @@ iounit_init(int sbi_node, int io_node, s
> >> struct linux_prom_registers iommu_promregs[PROMREG_MAX];
> >> struct resource r;
> >>
> >> - iounit = kmalloc(sizeof(struct iounit_struct), GFP_ATOMIC);
> >> + iounit = kzalloc(sizeof(struct iounit_struct), GFP_ATOMIC);
> >>
> >> - memset(iounit, 0, sizeof(*iounit));
> >> iounit->limit[0] = IOUNIT_BMAP1_START;
> >> iounit->limit[1] = IOUNIT_BMAP2_START;
> >> iounit->limit[2] = IOUNIT_BMAPM_START;
> >>
> >
> > ...in this ^, the old code and your update don't check for NULL return.
> >
>
> Both of this parts are done at early stages, so it is probably:
> a) Impossible to recover from failure
> b) If you run out of memory at this stage, you are probably in very big
> trouble anyway
>
> I could modify it to check and panic if the check fails.
> Would that be better?
Don't panic, call prom_printf() with a suitable message and then
prom_halt() just like all other early-boot failures do on sparc.
-
To unsubscribe from this list: send the line "unsubscribe sparclinux" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
