2020-05-12 23:57:55

by Tuan Phan

Subject: [PATCH v2] ACPI/IORT: Fix PMCG node always look for a single ID mapping.

PMCG node can have zero ID mapping if its overflow interrupt
is wire based. The code to parse PMCG node can not assume it will
have a single ID mapping.

Signed-off-by: Tuan Phan <[email protected]>
---
Changes in v2:
- Used pmcg node to detect wired base overflow interrupt.

drivers/acpi/arm64/iort.c | 5 +++++
1 file changed, 5 insertions(+)

diff --git a/drivers/acpi/arm64/iort.c b/drivers/acpi/arm64/iort.c
index ed3d2d1..11a4e8e 100644
--- a/drivers/acpi/arm64/iort.c
+++ b/drivers/acpi/arm64/iort.c
@@ -414,6 +414,7 @@ static struct acpi_iort_node *iort_node_get_id(struct acpi_iort_node *node,
static int iort_get_id_mapping_index(struct acpi_iort_node *node)
{
struct acpi_iort_smmu_v3 *smmu;
+ struct acpi_iort_pmcg *pmcg;

switch (node->type) {
case ACPI_IORT_NODE_SMMU_V3:
@@ -441,6 +442,10 @@ static int iort_get_id_mapping_index(struct acpi_iort_node *node)

return smmu->id_mapping_index;
case ACPI_IORT_NODE_PMCG:
+ pmcg = (struct acpi_iort_pmcg *)node->node_data;
+ if (pmcg->overflow_gsiv)
+ return -EINVAL;
+
return 0;
default:
return -EINVAL;
--
2.7.4

2020-05-13 03:19:16

by Hanjun Guo

Subject: Re: [PATCH v2] ACPI/IORT: Fix PMCG node always look for a single ID mapping.

On 2020/5/13 7:56, Tuan Phan wrote:
> PMCG node can have zero ID mapping if its overflow interrupt
> is wire based. The code to parse PMCG node can not assume it will
> have a single ID mapping.
>
> Signed-off-by: Tuan Phan <[email protected]>

It's better to add

Fixes: 24e516049360 ("ACPI/IORT: Add support for PMCG")

> ---
> Changes in v2:
> - Used pmcg node to detect wired base overflow interrupt.
>
> drivers/acpi/arm64/iort.c | 5 +++++
> 1 file changed, 5 insertions(+)
>
> diff --git a/drivers/acpi/arm64/iort.c b/drivers/acpi/arm64/iort.c
> index ed3d2d1..11a4e8e 100644
> --- a/drivers/acpi/arm64/iort.c
> +++ b/drivers/acpi/arm64/iort.c
> @@ -414,6 +414,7 @@ static struct acpi_iort_node *iort_node_get_id(struct acpi_iort_node *node,
> static int iort_get_id_mapping_index(struct acpi_iort_node *node)
> {
> struct acpi_iort_smmu_v3 *smmu;
> + struct acpi_iort_pmcg *pmcg;
>
> switch (node->type) {
> case ACPI_IORT_NODE_SMMU_V3:
> @@ -441,6 +442,10 @@ static int iort_get_id_mapping_index(struct acpi_iort_node *node)
>
> return smmu->id_mapping_index;
> case ACPI_IORT_NODE_PMCG:
> + pmcg = (struct acpi_iort_pmcg *)node->node_data;
> + if (pmcg->overflow_gsiv)
> + return -EINVAL;
> +
> return 0;
> default:
> return -EINVAL;

With my comments addressed,

Reviewed-by: Hanjun Guo <[email protected]>

2020-05-13 10:33:00

by Robin Murphy

Subject: Re: [PATCH v2] ACPI/IORT: Fix PMCG node always look for a single ID mapping.

On 2020-05-13 4:16 am, Hanjun Guo wrote:
> On 2020/5/13 7:56, Tuan Phan wrote:
>> PMCG node can have zero ID mapping if its overflow interrupt
>> is wire based. The code to parse PMCG node can not assume it will
>> have a single ID mapping.
>>
>> Signed-off-by: Tuan Phan <[email protected]>
>
> It's better to add
>
> Fixes: 24e516049360 ("ACPI/IORT: Add support for PMCG")
>
>> ---
>> Changes in v2:
>> - Used pmcg node to detect wired base overflow interrupt.
>>   drivers/acpi/arm64/iort.c | 5 +++++
>>   1 file changed, 5 insertions(+)
>>
>> diff --git a/drivers/acpi/arm64/iort.c b/drivers/acpi/arm64/iort.c
>> index ed3d2d1..11a4e8e 100644
>> --- a/drivers/acpi/arm64/iort.c
>> +++ b/drivers/acpi/arm64/iort.c
>> @@ -414,6 +414,7 @@ static struct acpi_iort_node
>> *iort_node_get_id(struct acpi_iort_node *node,
>>   static int iort_get_id_mapping_index(struct acpi_iort_node *node)
>>   {
>>       struct acpi_iort_smmu_v3 *smmu;
>> +    struct acpi_iort_pmcg *pmcg;
>>       switch (node->type) {
>>       case ACPI_IORT_NODE_SMMU_V3:
>> @@ -441,6 +442,10 @@ static int iort_get_id_mapping_index(struct
>> acpi_iort_node *node)
>>           return smmu->id_mapping_index;
>>       case ACPI_IORT_NODE_PMCG:
>> +        pmcg = (struct acpi_iort_pmcg *)node->node_data;
>> +        if (pmcg->overflow_gsiv)

if (pmcg->overflow_gsiv || node->mapping_count == 0)

...if there is no GSIV, we should still *also* check that index 0 is
valid before we return it.

Robin.

>> +            return -EINVAL;
>> +
>>           return 0;
>>       default:
>>           return -EINVAL;
>
> With my comments addressed,
>
> Reviewed-by: Hanjun Guo <[email protected]>
>
>
> _______________________________________________
> linux-arm-kernel mailing list
> [email protected]
> http://lists.infradead.org/mailman/listinfo/linux-arm-kernel