2024-02-13 08:00:42

by Auditya Bhattaram

[permalink] [raw]
Subject: [PATCH v4] soc: qcom: mdt_loader: Add Upperbounds check for program header access

hash_index is evaluated by looping phdrs till QCOM_MDT_TYPE_HASH
is found. Add an upperbound check to phdrs to access within elf size.

Fixes: 64fb5eb87d58 ("soc: qcom: mdt_loader: Allow hash to reside in any segment")
Cc: <[email protected]>
Signed-off-by: Auditya Bhattaram <[email protected]>
Acked-by: Mukesh Ojha <[email protected]>
---
Changes in v4:
- Added additional prints incase of Invalid access.
Link to v3 https://lore.kernel.org/stable/[email protected]
Link to v2 https://lore.kernel.org/linux-arm-msm/[email protected]/T/#t
Link to v1 https://lore.kernel.org/linux-arm-msm/[email protected]/T/#t
---
drivers/soc/qcom/mdt_loader.c | 7 +++++++
1 file changed, 7 insertions(+)

diff --git a/drivers/soc/qcom/mdt_loader.c b/drivers/soc/qcom/mdt_loader.c
index 6f177e46fa0f..1a79a7bba468 100644
--- a/drivers/soc/qcom/mdt_loader.c
+++ b/drivers/soc/qcom/mdt_loader.c
@@ -145,6 +143,13 @@ void *qcom_mdt_read_metadata(const struct firmware *fw, size_t *data_len,
if (phdrs[0].p_type == PT_LOAD)
return ERR_PTR(-EINVAL);

+ if (((size_t)(phdrs + ehdr->e_phnum)) > ((size_t)ehdr + fw->size)) {
+ dev_err(dev,
+ "Invalid phdrs access for fw: %s, e_phnum: %u, fw->size: %zu\n",
+ fw_name, ehdr->e_phnum, fw->size);
+ return ERR_PTR(-EINVAL);
+ }
+
for (i = 1; i < ehdr->e_phnum; i++) {
if ((phdrs[i].p_flags & QCOM_MDT_TYPE_MASK) == QCOM_MDT_TYPE_HASH) {
hash_segment = i;
--
2.17.1



2024-02-14 05:42:29

by Bjorn Andersson

[permalink] [raw]
Subject: Re: [PATCH v4] soc: qcom: mdt_loader: Add Upperbounds check for program header access

On Tue, Feb 13, 2024 at 01:30:10PM +0530, Auditya Bhattaram wrote:
> hash_index is evaluated by looping phdrs till QCOM_MDT_TYPE_HASH
> is found. Add an upperbound check to phdrs to access within elf size.
>

How is this compatible with what is being observed on SM8450 and
implemented in commit 8bd42e2341a7 ("soc: qcom: mdt_loader: Allow hash
segment to be split out"?

Regards,
Bjorn

> Fixes: 64fb5eb87d58 ("soc: qcom: mdt_loader: Allow hash to reside in any segment")
> Cc: <[email protected]>
> Signed-off-by: Auditya Bhattaram <[email protected]>
> Acked-by: Mukesh Ojha <[email protected]>
> ---
> Changes in v4:
> - Added additional prints incase of Invalid access.
> Link to v3 https://lore.kernel.org/stable/[email protected]
> Link to v2 https://lore.kernel.org/linux-arm-msm/[email protected]/T/#t
> Link to v1 https://lore.kernel.org/linux-arm-msm/[email protected]/T/#t
> ---
> drivers/soc/qcom/mdt_loader.c | 7 +++++++
> 1 file changed, 7 insertions(+)
>
> diff --git a/drivers/soc/qcom/mdt_loader.c b/drivers/soc/qcom/mdt_loader.c
> index 6f177e46fa0f..1a79a7bba468 100644
> --- a/drivers/soc/qcom/mdt_loader.c
> +++ b/drivers/soc/qcom/mdt_loader.c
> @@ -145,6 +143,13 @@ void *qcom_mdt_read_metadata(const struct firmware *fw, size_t *data_len,
> if (phdrs[0].p_type == PT_LOAD)
> return ERR_PTR(-EINVAL);
>
> + if (((size_t)(phdrs + ehdr->e_phnum)) > ((size_t)ehdr + fw->size)) {
> + dev_err(dev,
> + "Invalid phdrs access for fw: %s, e_phnum: %u, fw->size: %zu\n",
> + fw_name, ehdr->e_phnum, fw->size);
> + return ERR_PTR(-EINVAL);
> + }
> +
> for (i = 1; i < ehdr->e_phnum; i++) {
> if ((phdrs[i].p_flags & QCOM_MDT_TYPE_MASK) == QCOM_MDT_TYPE_HASH) {
> hash_segment = i;
> --
> 2.17.1
>

2024-02-14 05:57:22

by Auditya Bhattaram

[permalink] [raw]
Subject: Re: [PATCH v4] soc: qcom: mdt_loader: Add Upperbounds check for program header access



On 2/14/2024 11:12 AM, Bjorn Andersson wrote:
> On Tue, Feb 13, 2024 at 01:30:10PM +0530, Auditya Bhattaram wrote:
>> hash_index is evaluated by looping phdrs till QCOM_MDT_TYPE_HASH
>> is found. Add an upperbound check to phdrs to access within elf size.
>>
>
> How is this compatible with what is being observed on SM8450 and
> implemented in commit 8bd42e2341a7 ("soc: qcom: mdt_loader: Allow hash
> segment to be split out"?
>
> Regards,
> Bjorn
>

Calculating hash_index is introduced with this commit 8bd42e2341a7
("soc: qcom: mdt_loader: Allow hash segment to be split out"

for (i = 1; i < ehdr->e_phnum; i++) {
if ((phdrs[i].p_flags & QCOM_MDT_TYPE_MASK) ...

I'm trying to add an upper bound for this access "phdrs[i]"

>> Fixes: 64fb5eb87d58 ("soc: qcom: mdt_loader: Allow hash to reside in any segment")
>> Cc: <[email protected]>
>> Signed-off-by: Auditya Bhattaram <[email protected]>
>> Acked-by: Mukesh Ojha <[email protected]>
>> ---
>> Changes in v4:
>> - Added additional prints incase of Invalid access.
>> Link to v3 https://lore.kernel.org/stable/[email protected]
>> Link to v2 https://lore.kernel.org/linux-arm-msm/[email protected]/T/#t
>> Link to v1 https://lore.kernel.org/linux-arm-msm/[email protected]/T/#t
>> ---
>> drivers/soc/qcom/mdt_loader.c | 7 +++++++
>> 1 file changed, 7 insertions(+)
>>
>> diff --git a/drivers/soc/qcom/mdt_loader.c b/drivers/soc/qcom/mdt_loader.c
>> index 6f177e46fa0f..1a79a7bba468 100644
>> --- a/drivers/soc/qcom/mdt_loader.c
>> +++ b/drivers/soc/qcom/mdt_loader.c
>> @@ -145,6 +143,13 @@ void *qcom_mdt_read_metadata(const struct firmware *fw, size_t *data_len,
>> if (phdrs[0].p_type == PT_LOAD)
>> return ERR_PTR(-EINVAL);
>>
>> + if (((size_t)(phdrs + ehdr->e_phnum)) > ((size_t)ehdr + fw->size)) {
>> + dev_err(dev,
>> + "Invalid phdrs access for fw: %s, e_phnum: %u, fw->size: %zu\n",
>> + fw_name, ehdr->e_phnum, fw->size);
>> + return ERR_PTR(-EINVAL);
>> + }
>> +
>> for (i = 1; i < ehdr->e_phnum; i++) {
>> if ((phdrs[i].p_flags & QCOM_MDT_TYPE_MASK) == QCOM_MDT_TYPE_HASH) {
>> hash_segment = i;
>> --
>> 2.17.1
>>

2024-03-12 07:06:15

by Auditya Bhattaram

[permalink] [raw]
Subject: Re: [PATCH v4] soc: qcom: mdt_loader: Add Upperbounds check for program header access



On 2/14/2024 11:27 AM, Auditya Bhattaram wrote:
>
>
> On 2/14/2024 11:12 AM, Bjorn Andersson wrote:
>> On Tue, Feb 13, 2024 at 01:30:10PM +0530, Auditya Bhattaram wrote:
>>> hash_index is evaluated by looping phdrs till QCOM_MDT_TYPE_HASH
>>> is found. Add an upperbound check to phdrs to access within elf size.
>>>
>>
>> How is this compatible with what is being observed on SM8450 and
>> implemented in commit 8bd42e2341a7 ("soc: qcom: mdt_loader: Allow hash
>> segment to be split out"?
>>
>> Regards,
>> Bjorn
>>
>
> Calculating hash_index is introduced with this commit 8bd42e2341a7
> ("soc: qcom: mdt_loader: Allow hash segment to be split out"
>
>     for (i = 1; i < ehdr->e_phnum; i++) {
>      if ((phdrs[i].p_flags & QCOM_MDT_TYPE_MASK) ...
>
> I'm trying to add an upper bound for this access "phdrs[i]"
>

Any further questions on this Bjorn.

>>> Fixes: 64fb5eb87d58 ("soc: qcom: mdt_loader: Allow hash to reside in
>>> any segment")
>>> Cc: <[email protected]>
>>> Signed-off-by: Auditya Bhattaram <[email protected]>
>>> Acked-by: Mukesh Ojha <[email protected]>
>>> ---
>>> Changes in v4:
>>>   - Added additional prints incase of Invalid access.
>>> Link to v3
>>> https://lore.kernel.org/stable/[email protected]
>>> Link to v2
>>> https://lore.kernel.org/linux-arm-msm/[email protected]/T/#t
>>> Link to v1
>>> https://lore.kernel.org/linux-arm-msm/[email protected]/T/#t
>>> ---
>>>   drivers/soc/qcom/mdt_loader.c | 7 +++++++
>>>   1 file changed, 7 insertions(+)
>>>
>>> diff --git a/drivers/soc/qcom/mdt_loader.c
>>> b/drivers/soc/qcom/mdt_loader.c
>>> index 6f177e46fa0f..1a79a7bba468 100644
>>> --- a/drivers/soc/qcom/mdt_loader.c
>>> +++ b/drivers/soc/qcom/mdt_loader.c
>>> @@ -145,6 +143,13 @@ void *qcom_mdt_read_metadata(const struct
>>> firmware *fw, size_t *data_len,
>>>       if (phdrs[0].p_type == PT_LOAD)
>>>           return ERR_PTR(-EINVAL);
>>>
>>> +    if (((size_t)(phdrs + ehdr->e_phnum)) > ((size_t)ehdr +
>>> fw->size)) {
>>> +        dev_err(dev,
>>> +            "Invalid phdrs access for fw: %s, e_phnum: %u, fw->size:
>>> %zu\n",
>>> +            fw_name, ehdr->e_phnum, fw->size);
>>> +        return ERR_PTR(-EINVAL);
>>> +    }
>>> +
>>>       for (i = 1; i < ehdr->e_phnum; i++) {
>>>           if ((phdrs[i].p_flags & QCOM_MDT_TYPE_MASK) ==
>>> QCOM_MDT_TYPE_HASH) {
>>>               hash_segment = i;
>>> --
>>> 2.17.1
>>>

2024-03-16 17:17:34

by Bjorn Andersson

[permalink] [raw]
Subject: Re: [PATCH v4] soc: qcom: mdt_loader: Add Upperbounds check for program header access

On Wed, Feb 14, 2024 at 11:27:01AM +0530, Auditya Bhattaram wrote:
>
>
> On 2/14/2024 11:12 AM, Bjorn Andersson wrote:
> > On Tue, Feb 13, 2024 at 01:30:10PM +0530, Auditya Bhattaram wrote:
> > > hash_index is evaluated by looping phdrs till QCOM_MDT_TYPE_HASH
> > > is found. Add an upperbound check to phdrs to access within elf size.
> > >
> >
> > How is this compatible with what is being observed on SM8450 and
> > implemented in commit 8bd42e2341a7 ("soc: qcom: mdt_loader: Allow hash
> > segment to be split out"?
> >
> > Regards,
> > Bjorn
> >
>
> Calculating hash_index is introduced with this commit 8bd42e2341a7 ("soc:
> qcom: mdt_loader: Allow hash segment to be split out"
>
> for (i = 1; i < ehdr->e_phnum; i++) {
> if ((phdrs[i].p_flags & QCOM_MDT_TYPE_MASK) ...
>
> I'm trying to add an upper bound for this access "phdrs[i]"
>

Ahh, sorry, you're of course correct. I think I would have preferred a
more obvious check of offset >= fw->size though...

That said, this code used to sit behind rproc_elf_sanity_check(), but
that seem to have been lost in the refactorings. As such, your patch is
incomplete and we should reintroduce a whole bunch of these sanity
checks!

Regards,
Bjorn

> > > Fixes: 64fb5eb87d58 ("soc: qcom: mdt_loader: Allow hash to reside in any segment")
> > > Cc: <[email protected]>
> > > Signed-off-by: Auditya Bhattaram <[email protected]>
> > > Acked-by: Mukesh Ojha <[email protected]>
> > > ---
> > > Changes in v4:
> > > - Added additional prints incase of Invalid access.
> > > Link to v3 https://lore.kernel.org/stable/[email protected]
> > > Link to v2 https://lore.kernel.org/linux-arm-msm/[email protected]/T/#t
> > > Link to v1 https://lore.kernel.org/linux-arm-msm/[email protected]/T/#t
> > > ---
> > > drivers/soc/qcom/mdt_loader.c | 7 +++++++
> > > 1 file changed, 7 insertions(+)
> > >
> > > diff --git a/drivers/soc/qcom/mdt_loader.c b/drivers/soc/qcom/mdt_loader.c
> > > index 6f177e46fa0f..1a79a7bba468 100644
> > > --- a/drivers/soc/qcom/mdt_loader.c
> > > +++ b/drivers/soc/qcom/mdt_loader.c
> > > @@ -145,6 +143,13 @@ void *qcom_mdt_read_metadata(const struct firmware *fw, size_t *data_len,
> > > if (phdrs[0].p_type == PT_LOAD)
> > > return ERR_PTR(-EINVAL);
> > >
> > > + if (((size_t)(phdrs + ehdr->e_phnum)) > ((size_t)ehdr + fw->size)) {
> > > + dev_err(dev,
> > > + "Invalid phdrs access for fw: %s, e_phnum: %u, fw->size: %zu\n",
> > > + fw_name, ehdr->e_phnum, fw->size);
> > > + return ERR_PTR(-EINVAL);
> > > + }
> > > +
> > > for (i = 1; i < ehdr->e_phnum; i++) {
> > > if ((phdrs[i].p_flags & QCOM_MDT_TYPE_MASK) == QCOM_MDT_TYPE_HASH) {
> > > hash_segment = i;
> > > --
> > > 2.17.1
> > >