linux-kernel - Re: [PATCH] ACPI/IORT: Fix iort_node_get

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Mon, 9 Jan 2017 14:48:51 +0800
From:   Hanjun Guo <hanjun.guo@...aro.org>
To:     Sinan Kaya <okaya@...eaurora.org>,
        Lorenzo Pieralisi <lorenzo.pieralisi@....com>,
        linux-acpi@...r.kernel.org
Cc:     linux-kernel@...r.kernel.org, Tomasz Nowicki <tn@...ihalf.com>,
        Nate Watterson <nwatters@...eaurora.org>,
        "Rafael J. Wysocki" <rjw@...ysocki.net>
Subject: Re: [PATCH] ACPI/IORT: Fix iort_node_get_id() mapping entries
 indexing

On 2017/1/9 14:34, Hanjun Guo wrote:
> Hi Sinan,
>
> On 2017/1/8 5:09, Sinan Kaya wrote:
>> On 1/5/2017 1:29 PM, Lorenzo Pieralisi wrote:
>>> Commit 618f535a6062 ("ACPI/IORT: Add single mapping function")
>>> introduced a function (iort_node_get_id()) to retrieve ids for IORT
>>> named components.
>>>
>>> iort_node_get_id() takes an index as input to refer to a specific
>>> mapping entry in the mapping array to retrieve the id at a specific
>>> index provided the index is below the total mapping count; currently the
>>> index is used to retrieve the mapping value from the correct entry but
>>> not to dereference the correct entry while retrieving the mapping
>>> output_reference (ie IORT parent pointer), which consequently always
>>> resolves to the output_reference of the first entry in the mapping
>>> array.
>>>
>>> Update the map array entry pointer computation in iort_node_get_id() to
>>> take into account the index value, fixing the issue.
>>>
>>> Fixes: 618f535a6062 ("ACPI/IORT: Add single mapping function")
>>> Reported-by: Hanjun Guo <hanjun.guo@...aro.org>
>>> Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@....com>
>>> Cc: Hanjun Guo <hanjun.guo@...aro.org>
>>> Cc: Sinan Kaya <okaya@...eaurora.org>
>>> Cc: Tomasz Nowicki <tn@...ihalf.com>
>>> Cc: Nate Watterson <nwatters@...eaurora.org>
>>> Cc: "Rafael J. Wysocki" <rjw@...ysocki.net>
>>> ---
>>>  drivers/acpi/arm64/iort.c | 6 +++---
>>>  1 file changed, 3 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/drivers/acpi/arm64/iort.c b/drivers/acpi/arm64/iort.c
>>> index e0d2e6e..ba156c5 100644
>>> --- a/drivers/acpi/arm64/iort.c
>>> +++ b/drivers/acpi/arm64/iort.c
>>> @@ -333,7 +333,7 @@ struct acpi_iort_node *iort_node_get_id(struct
>>> acpi_iort_node *node,
>>>          return NULL;
>>>
>>>      map = ACPI_ADD_PTR(struct acpi_iort_id_mapping, node,
>>> -               node->mapping_offset);
>>> +               node->mapping_offset + index * sizeof(*map));
>>
>> What does this give us that the previous code didn't do?
>
> Fir example, if you have multi mappings ids under platform device:
>
> |-------------|
> |  SMMU  2    |<-------
> |-------------|       |
>                       |
>                       |
> |-------------|       |
> |  SMMU 1     |<----  |
> |-------------|    |  |
>                    |  |
>                    |  |
> |-------------|    |  |
> |  platform   |    |  |
> |  device     |    |  |
> |-------------|    |  |
> | stream id   |    |  |
> | 1           |    |  |
> | parent------|----|  |
> |-------------|       |
> |  stream id  |       |
> |  2          |       |
> |  parent-----|-------|
> |-------------|
>
> For now, we just use the first entry in the mapping entry to get
> the parent, and always point to the same parent, as above, we will
> always map to SMMU 1 even if you connect to different SMMUs. (Although
> we may don't have such device topology yet)
>
>
>>
>> You are using map as a pointer and returning the offset of the first
>> map entry above
>> and then accessing the map at the indexed offset with map[index]
>>
>> The new code is using map as a plain pointer, calculating the pointer
>> location with ACPI_ADD_PTR
>> instead and then collecting the output parameter with map->output_base.
>>
>>>
>>>      /* Firmware bug! */
>>>      if (!map->output_reference) {
>>> @@ -348,10 +348,10 @@ struct acpi_iort_node *iort_node_get_id(struct
>>> acpi_iort_node *node,
>>>      if (!(IORT_TYPE_MASK(parent->type) & type_mask))
>>>          return NULL;
>>>
>>> -    if (map[index].flags & ACPI_IORT_ID_SINGLE_MAPPING) {
>>> +    if (map->flags & ACPI_IORT_ID_SINGLE_MAPPING) {
>>>          if (node->type == ACPI_IORT_NODE_NAMED_COMPONENT ||
>>>              node->type == ACPI_IORT_NODE_PCI_ROOT_COMPLEX) {
>>> -            *id_out = map[index].output_base;
>>> +            *id_out = map->output_base;
>>
>> You are claiming that the existing code is collecting the output
>> parameter from the first mapping.
>> I don't see this happening above.
>>
>> What am I missing?
>
> It's not about the output id but it's about the parent returned
> by this function, it always return the first entry's parent in the
> mapping entry.

And as Lorenzo said in the commit message:
"always resolves to the output_reference of the first entry in the
mapping array."

Then always judge the same entry for:

if (!map->output_reference)
	pr_err();

...

Thanks
Hanjun