Re: [kvm-unit-tests PATCH v3 04/17] devicetree: fix dt_get_memory_params

[Thomas Huth <thuth@redhat.com>]

[-- Attachment #1: Type: text/plain, Size: 1811 bytes --]

On 15.02.2016 00:00, David Gibson wrote:
> On Sat, Feb 13, 2016 at 05:28:17PM +0100, Andrew Jones wrote:
>> Index nr was overreaching with its indexing. It should only
>> index the caller's regs array. We need a different index for
>> the memory nodes' reg properties.
>>
>> Signed-off-by: Andrew Jones <drjones@redhat.com>
> 
> I think I'm missing something, I don't see how reg_idx will ever take
> a different value from nr.

There is one difference ...

>> ---
>> The next patch needs this fix.
>> ---
>>  lib/devicetree.c | 8 +++++---
>>  1 file changed, 5 insertions(+), 3 deletions(-)
>>
>> diff --git a/lib/devicetree.c b/lib/devicetree.c
>> index 36cb28610ff41..a5c7f7c69ddfd 100644
>> --- a/lib/devicetree.c
>> +++ b/lib/devicetree.c
>> @@ -165,22 +165,24 @@ int dt_pbus_get_base_compatible(const char *compatible,
>>  int dt_get_memory_params(struct dt_pbus_reg *regs, int nr_regs)
>>  {
>>  	const char *pn = "device_type", *pv = "memory";
>> -	int node, ret, pl = strlen(pv) + 1, nr = 0;
>> +	int node, ret, reg_idx, pl = strlen(pv) + 1, nr = 0;
>>  	struct dt_pbus_reg reg;
>>  
>>  	node = fdt_node_offset_by_prop_value(fdt, -1, pn, pv, pl);
>>  
>>  	while (node >= 0) {
>>  
>> +		reg_idx = 0;

... here: reg_idx is reset to zero, but nr is not.

>>  		while (nr < nr_regs) {
>> -			ret = dt_pbus_translate_node(node, nr, &reg);
>> +			ret = dt_pbus_translate_node(node, reg_idx, &reg);
>>  			if (ret == -FDT_ERR_NOTFOUND)
>>  				break;
>>  			if (ret < 0)
>>  				return ret;
>>  			regs[nr].addr = reg.addr;
>>  			regs[nr].size = reg.size;
>> -			++nr;
>> +			++nr, ++reg_idx;
>>  		}
>>  
>>  		node = fdt_node_offset_by_prop_value(fdt, node, pn, pv, pl);

Sounds reasonable to me.

Reviewed-by: Thomas Huth <thuth@redhat.com>



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]