[PATCH] SecurityPkg/Tpm2Help.c: Add boundary check for array

[PATCH] SecurityPkg/Tpm2Help.c: Add boundary check for array

[Zhang, Shenglei]

Add 'Index < HASH_COUNT' to ensure things out of boundary
of digests[] can not be visited.

Cc: Jiewen Yao <jiewen.yao@intel.com>
Cc: Jian J Wang <jian.j.wang@intel.com>
Cc: Chao Zhang <chao.b.zhang@intel.com>
Signed-off-by: Shenglei Zhang <shenglei.zhang@intel.com>
---
 SecurityPkg/Library/Tpm2CommandLib/Tpm2Help.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/SecurityPkg/Library/Tpm2CommandLib/Tpm2Help.c b/SecurityPkg/Library/Tpm2CommandLib/Tpm2Help.c
index 36c240d1221c..a7d4e3ab5373 100644
--- a/SecurityPkg/Library/Tpm2CommandLib/Tpm2Help.c
+++ b/SecurityPkg/Library/Tpm2CommandLib/Tpm2Help.c
@@ -299,7 +299,7 @@ GetDigestListSize (
   UINT32 TotalSize;
 
   TotalSize = sizeof(DigestList->count);
-  for (Index = 0; Index < DigestList->count; Index++) {
+  for (Index = 0; Index < DigestList->count, Index < HASH_COUNT; Index++) {
     DigestSize = GetHashSizeFromAlgo (DigestList->digests[Index].hashAlg);
     TotalSize += sizeof(DigestList->digests[Index].hashAlg) + DigestSize;
   }
-- 
2.18.0.windows.1

Re: [PATCH] SecurityPkg/Tpm2Help.c: Add boundary check for array

[Yao, Jiewen]

Hi
May I know where is the data from? Trusted region or non-trusted region?

I am thinking if we need use ASSERT to avoid user mistake.
But want to check the API input assumption at first...



> -----Original Message-----
> From: Zhang, Shenglei <shenglei.zhang@intel.com>
> Sent: Friday, December 6, 2019 9:50 AM
> To: devel@edk2.groups.io
> Cc: Yao, Jiewen <jiewen.yao@intel.com>; Wang, Jian J <jian.j.wang@intel.com>;
> Zhang, Chao B <chao.b.zhang@intel.com>
> Subject: [PATCH] SecurityPkg/Tpm2Help.c: Add boundary check for array
> 
> Add 'Index < HASH_COUNT' to ensure things out of boundary
> of digests[] can not be visited.
> 
> Cc: Jiewen Yao <jiewen.yao@intel.com>
> Cc: Jian J Wang <jian.j.wang@intel.com>
> Cc: Chao Zhang <chao.b.zhang@intel.com>
> Signed-off-by: Shenglei Zhang <shenglei.zhang@intel.com>
> ---
>  SecurityPkg/Library/Tpm2CommandLib/Tpm2Help.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/SecurityPkg/Library/Tpm2CommandLib/Tpm2Help.c
> b/SecurityPkg/Library/Tpm2CommandLib/Tpm2Help.c
> index 36c240d1221c..a7d4e3ab5373 100644
> --- a/SecurityPkg/Library/Tpm2CommandLib/Tpm2Help.c
> +++ b/SecurityPkg/Library/Tpm2CommandLib/Tpm2Help.c
> @@ -299,7 +299,7 @@ GetDigestListSize (
>    UINT32 TotalSize;
> 
>    TotalSize = sizeof(DigestList->count);
> -  for (Index = 0; Index < DigestList->count; Index++) {
> +  for (Index = 0; Index < DigestList->count, Index < HASH_COUNT; Index++) {
>      DigestSize = GetHashSizeFromAlgo (DigestList->digests[Index].hashAlg);
>      TotalSize += sizeof(DigestList->digests[Index].hashAlg) + DigestSize;
>    }
> --
> 2.18.0.windows.1

Re: [PATCH] SecurityPkg/Tpm2Help.c: Add boundary check for array

[Zhang, Shenglei]

> -----Original Message-----
> From: Yao, Jiewen
> Sent: Friday, December 6, 2019 10:04 AM
> To: Zhang, Shenglei <shenglei.zhang@intel.com>; devel@edk2.groups.io
> Cc: Wang, Jian J <jian.j.wang@intel.com>; Zhang, Chao B
> <chao.b.zhang@intel.com>
> Subject: RE: [PATCH] SecurityPkg/Tpm2Help.c: Add boundary check for array
> 
> Hi
> May I know where is the data from? Trusted region or non-trusted region?
> 
> I am thinking if we need use ASSERT to avoid user mistake.
> But want to check the API input assumption at first...

Hi Jiewen,

I don't think DigestList->count can be trusted. We can add Index < HASH_COUNT into the
for(...) statement.

Thanks,
Shenglei

> 
> 
> 
> > -----Original Message-----
> > From: Zhang, Shenglei <shenglei.zhang@intel.com>
> > Sent: Friday, December 6, 2019 9:50 AM
> > To: devel@edk2.groups.io
> > Cc: Yao, Jiewen <jiewen.yao@intel.com>; Wang, Jian J
> <jian.j.wang@intel.com>;
> > Zhang, Chao B <chao.b.zhang@intel.com>
> > Subject: [PATCH] SecurityPkg/Tpm2Help.c: Add boundary check for array
> >
> > Add 'Index < HASH_COUNT' to ensure things out of boundary
> > of digests[] can not be visited.
> >
> > Cc: Jiewen Yao <jiewen.yao@intel.com>
> > Cc: Jian J Wang <jian.j.wang@intel.com>
> > Cc: Chao Zhang <chao.b.zhang@intel.com>
> > Signed-off-by: Shenglei Zhang <shenglei.zhang@intel.com>
> > ---
> >  SecurityPkg/Library/Tpm2CommandLib/Tpm2Help.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/SecurityPkg/Library/Tpm2CommandLib/Tpm2Help.c
> > b/SecurityPkg/Library/Tpm2CommandLib/Tpm2Help.c
> > index 36c240d1221c..a7d4e3ab5373 100644
> > --- a/SecurityPkg/Library/Tpm2CommandLib/Tpm2Help.c
> > +++ b/SecurityPkg/Library/Tpm2CommandLib/Tpm2Help.c
> > @@ -299,7 +299,7 @@ GetDigestListSize (
> >    UINT32 TotalSize;
> >
> >    TotalSize = sizeof(DigestList->count);
> > -  for (Index = 0; Index < DigestList->count; Index++) {
> > +  for (Index = 0; Index < DigestList->count, Index < HASH_COUNT; Index++)
> {
> >      DigestSize = GetHashSizeFromAlgo (DigestList->digests[Index].hashAlg);
> >      TotalSize += sizeof(DigestList->digests[Index].hashAlg) + DigestSize;
> >    }
> > --
> > 2.18.0.windows.1

Re: [edk2-devel] [PATCH] SecurityPkg/Tpm2Help.c: Add boundary check for array

[Laszlo Ersek]

On 12/06/19 02:49, Zhang, Shenglei wrote:
> Add 'Index < HASH_COUNT' to ensure things out of boundary
> of digests[] can not be visited.
> 
> Cc: Jiewen Yao <jiewen.yao@intel.com>
> Cc: Jian J Wang <jian.j.wang@intel.com>
> Cc: Chao Zhang <chao.b.zhang@intel.com>
> Signed-off-by: Shenglei Zhang <shenglei.zhang@intel.com>
> ---
>  SecurityPkg/Library/Tpm2CommandLib/Tpm2Help.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/SecurityPkg/Library/Tpm2CommandLib/Tpm2Help.c b/SecurityPkg/Library/Tpm2CommandLib/Tpm2Help.c
> index 36c240d1221c..a7d4e3ab5373 100644
> --- a/SecurityPkg/Library/Tpm2CommandLib/Tpm2Help.c
> +++ b/SecurityPkg/Library/Tpm2CommandLib/Tpm2Help.c
> @@ -299,7 +299,7 @@ GetDigestListSize (
>    UINT32 TotalSize;
>  
>    TotalSize = sizeof(DigestList->count);
> -  for (Index = 0; Index < DigestList->count; Index++) {
> +  for (Index = 0; Index < DigestList->count, Index < HASH_COUNT; Index++) {
>      DigestSize = GetHashSizeFromAlgo (DigestList->digests[Index].hashAlg);
>      TotalSize += sizeof(DigestList->digests[Index].hashAlg) + DigestSize;
>    }
> 

Nacked-by: Laszlo Ersek <lersek@redhat.com>

The comma operator is either functionally wrong in this context, or
stylistically wrong. From the C standard:

    The left operand of a comma operator is evaluated as a void
    expression; there is a sequence point after its evaluation. Then the
    right operand is evaluated; the result has its type and value. [...]

In case we *only* need to check (Index < HASH_COUNT), then the patch is
stylistically incorrect: the (Index < DigestList->count) condition
should simply be removed.

In case we need to check *both* conditions, then the patch is
functionally wrong: we should either use the logical AND (&&) operator,
instead of the comma:

  Index < DigestList->count && Index < HASH_COUNT

or invoke the MIN() function-like macro:

  Index < MIN ((UINTN)DigestList->count, (UINTN)HASH_COUNT)

Thanks
Laszlo

Re: [edk2-devel] [PATCH] SecurityPkg/Tpm2Help.c: Add boundary check for array

[Zhang, Shenglei]

> -----Original Message-----
> From: devel@edk2.groups.io [mailto:devel@edk2.groups.io] On Behalf Of
> Laszlo Ersek
> Sent: Friday, December 6, 2019 5:27 PM
> To: devel@edk2.groups.io; Zhang, Shenglei <shenglei.zhang@intel.com>
> Cc: Yao, Jiewen <jiewen.yao@intel.com>; Wang, Jian J
> <jian.j.wang@intel.com>; Zhang, Chao B <chao.b.zhang@intel.com>
> Subject: Re: [edk2-devel] [PATCH] SecurityPkg/Tpm2Help.c: Add boundary
> check for array
> 
> On 12/06/19 02:49, Zhang, Shenglei wrote:
> > Add 'Index < HASH_COUNT' to ensure things out of boundary
> > of digests[] can not be visited.
> >
> > Cc: Jiewen Yao <jiewen.yao@intel.com>
> > Cc: Jian J Wang <jian.j.wang@intel.com>
> > Cc: Chao Zhang <chao.b.zhang@intel.com>
> > Signed-off-by: Shenglei Zhang <shenglei.zhang@intel.com>
> > ---
> >  SecurityPkg/Library/Tpm2CommandLib/Tpm2Help.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/SecurityPkg/Library/Tpm2CommandLib/Tpm2Help.c
> b/SecurityPkg/Library/Tpm2CommandLib/Tpm2Help.c
> > index 36c240d1221c..a7d4e3ab5373 100644
> > --- a/SecurityPkg/Library/Tpm2CommandLib/Tpm2Help.c
> > +++ b/SecurityPkg/Library/Tpm2CommandLib/Tpm2Help.c
> > @@ -299,7 +299,7 @@ GetDigestListSize (
> >    UINT32 TotalSize;
> >
> >    TotalSize = sizeof(DigestList->count);
> > -  for (Index = 0; Index < DigestList->count; Index++) {
> > +  for (Index = 0; Index < DigestList->count, Index < HASH_COUNT; Index++)
> {
> >      DigestSize = GetHashSizeFromAlgo (DigestList->digests[Index].hashAlg);
> >      TotalSize += sizeof(DigestList->digests[Index].hashAlg) + DigestSize;
> >    }
> >
> 
> Nacked-by: Laszlo Ersek <lersek@redhat.com>
> 
> The comma operator is either functionally wrong in this context, or
> stylistically wrong. From the C standard:
> 
>     The left operand of a comma operator is evaluated as a void
>     expression; there is a sequence point after its evaluation. Then the
>     right operand is evaluated; the result has its type and value. [...]
> 
> In case we *only* need to check (Index < HASH_COUNT), then the patch is
> stylistically incorrect: the (Index < DigestList->count) condition
> should simply be removed.
> 
> In case we need to check *both* conditions, then the patch is
> functionally wrong: we should either use the logical AND (&&) operator,
> instead of the comma:
> 
>   Index < DigestList->count && Index < HASH_COUNT
> 

Hi Laszlo,

You are right. I'll change the statement to include both conditions.

Thanks,
Shenglei

> or invoke the MIN() function-like macro:
> 
>   Index < MIN ((UINTN)DigestList->count, (UINTN)HASH_COUNT)
> 
> Thanks
> Laszlo
> 
> 
>