diff mbox series

[RFC,v2,12/26] x86/sgx: Add helper to update SGX_LEPUBKEYHASHn MSRs

Message ID 5116fdc732e8e14b3378c44e3b461a43f330ed0c.1610935432.git.kai.huang@intel.com (mailing list archive)
State New, archived
Headers show
Series KVM SGX virtualization support | expand

Commit Message

Kai Huang Jan. 18, 2021, 3:28 a.m. UTC
Add a helper to update SGX_LEPUBKEYHASHn MSRs.  SGX virtualization also
needs to update those MSRs based on guest's "virtual" SGX_LEPUBKEYHASHn
before EINIT from guest.

Acked-by: Dave Hansen <dave.hansen@intel.com>
Signed-off-by: Kai Huang <kai.huang@intel.com>
---
 arch/x86/kernel/cpu/sgx/ioctl.c | 5 ++---
 arch/x86/kernel/cpu/sgx/main.c  | 8 ++++++++
 arch/x86/kernel/cpu/sgx/sgx.h   | 2 ++
 3 files changed, 12 insertions(+), 3 deletions(-)

Comments

Jarkko Sakkinen Jan. 20, 2021, 12:03 p.m. UTC | #1
On Mon, Jan 18, 2021 at 04:28:05PM +1300, Kai Huang wrote:
> Add a helper to update SGX_LEPUBKEYHASHn MSRs.  SGX virtualization also
> needs to update those MSRs based on guest's "virtual" SGX_LEPUBKEYHASHn
> before EINIT from guest.
> 
> Acked-by: Dave Hansen <dave.hansen@intel.com>
> Signed-off-by: Kai Huang <kai.huang@intel.com>
> ---
>  arch/x86/kernel/cpu/sgx/ioctl.c | 5 ++---
>  arch/x86/kernel/cpu/sgx/main.c  | 8 ++++++++
>  arch/x86/kernel/cpu/sgx/sgx.h   | 2 ++
>  3 files changed, 12 insertions(+), 3 deletions(-)
> 
> diff --git a/arch/x86/kernel/cpu/sgx/ioctl.c b/arch/x86/kernel/cpu/sgx/ioctl.c
> index e5977752c7be..1bae754268d1 100644
> --- a/arch/x86/kernel/cpu/sgx/ioctl.c
> +++ b/arch/x86/kernel/cpu/sgx/ioctl.c
> @@ -495,7 +495,7 @@ static int sgx_encl_init(struct sgx_encl *encl, struct sgx_sigstruct *sigstruct,
>  			 void *token)
>  {
>  	u64 mrsigner[4];
> -	int i, j, k;
> +	int i, j;
>  	void *addr;
>  	int ret;
>  
> @@ -544,8 +544,7 @@ static int sgx_encl_init(struct sgx_encl *encl, struct sgx_sigstruct *sigstruct,
>  
>  			preempt_disable();
>  
> -			for (k = 0; k < 4; k++)
> -				wrmsrl(MSR_IA32_SGXLEPUBKEYHASH0 + k, mrsigner[k]);
> +			sgx_update_lepubkeyhash(mrsigner);
>  
>  			ret = __einit(sigstruct, token, addr);
>  
> diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
> index bdda631c975b..1cf1f0f058b8 100644
> --- a/arch/x86/kernel/cpu/sgx/main.c
> +++ b/arch/x86/kernel/cpu/sgx/main.c
> @@ -697,6 +697,14 @@ static bool __init sgx_page_cache_init(void)
>  	return true;
>  }
>  
> +void sgx_update_lepubkeyhash(u64 *lepubkeyhash)
> +{
> +	int i;
> +
> +	for (i = 0; i < 4; i++)
> +		wrmsrl(MSR_IA32_SGXLEPUBKEYHASH0 + i, lepubkeyhash[i]);
> +}

Missing kdoc.

> +
>  static void __init sgx_init(void)
>  {
>  	int ret;
> diff --git a/arch/x86/kernel/cpu/sgx/sgx.h b/arch/x86/kernel/cpu/sgx/sgx.h
> index 509f2af33e1d..ccd4f145c464 100644
> --- a/arch/x86/kernel/cpu/sgx/sgx.h
> +++ b/arch/x86/kernel/cpu/sgx/sgx.h
> @@ -83,4 +83,6 @@ void sgx_mark_page_reclaimable(struct sgx_epc_page *page);
>  int sgx_unmark_page_reclaimable(struct sgx_epc_page *page);
>  struct sgx_epc_page *sgx_alloc_epc_page(void *owner, bool reclaim);
>  
> +void sgx_update_lepubkeyhash(u64 *lepubkeyhash);
> +
>  #endif /* _X86_SGX_H */
> -- 
> 2.29.2
> 
> 

/Jarkko
Dave Hansen Jan. 20, 2021, 6:36 p.m. UTC | #2
On 1/20/21 4:03 AM, Jarkko Sakkinen wrote:
>> +void sgx_update_lepubkeyhash(u64 *lepubkeyhash)
>> +{
>> +	int i;
>> +
>> +	for (i = 0; i < 4; i++)
>> +		wrmsrl(MSR_IA32_SGXLEPUBKEYHASH0 + i, lepubkeyhash[i]);
>> +}
> Missing kdoc.

I dunno... kdoc is nice, but I'm not sure its verbosity is useful here,
even if this function is called from more than one .c file.

I'd be happy with a single-line comment, personally.
Kai Huang Jan. 20, 2021, 11:36 p.m. UTC | #3
On Wed, 20 Jan 2021 10:36:09 -0800 Dave Hansen wrote:
> On 1/20/21 4:03 AM, Jarkko Sakkinen wrote:
> >> +void sgx_update_lepubkeyhash(u64 *lepubkeyhash)
> >> +{
> >> +	int i;
> >> +
> >> +	for (i = 0; i < 4; i++)
> >> +		wrmsrl(MSR_IA32_SGXLEPUBKEYHASH0 + i, lepubkeyhash[i]);
> >> +}
> > Missing kdoc.
> 
> I dunno... kdoc is nice, but I'm not sure its verbosity is useful here,
> even if this function is called from more than one .c file.
> 
> I'd be happy with a single-line comment, personally.
> 

I actually feel the function name already explains what the function does
clearly, therefore I don't think even comment is needed. To be honest I
don't know how to rephrase here. Perhaps:

/* Update SGX LEPUBKEYHASH MSRs of the platform. */

?
Dave Hansen Jan. 20, 2021, 11:50 p.m. UTC | #4
On 1/20/21 3:36 PM, Kai Huang wrote:
> I actually feel the function name already explains what the function does
> clearly, therefore I don't think even comment is needed. To be honest I
> don't know how to rephrase here. Perhaps:
> 
> /* Update SGX LEPUBKEYHASH MSRs of the platform. */

Whee!  I'm gonna write me a function comment!

/*
 * A Launch Enclave (LE) must be signed with a public key
 * that matches this SHA256 hash.  Usually overwrites Intel's
 * default signing key.
 */

So, this isn't a one-liner.  *But*, it tells us what "le" means, what
"pubkey" means and implies that there need to be 4x64-bits worth of MSR
writes to get to a SHA256 hash.  It also tells what it's usually doing
here: overwriting Intel's blasted hash.

It sure beats the entirely uncommented for loop that we've got today.
Kai Huang Jan. 21, 2021, 1:06 a.m. UTC | #5
On Wed, 20 Jan 2021 15:50:31 -0800 Dave Hansen wrote:
> On 1/20/21 3:36 PM, Kai Huang wrote:
> > I actually feel the function name already explains what the function does
> > clearly, therefore I don't think even comment is needed. To be honest I
> > don't know how to rephrase here. Perhaps:
> > 
> > /* Update SGX LEPUBKEYHASH MSRs of the platform. */
> 
> Whee!  I'm gonna write me a function comment!
> 
> /*
>  * A Launch Enclave (LE) must be signed with a public key
>  * that matches this SHA256 hash.  Usually overwrites Intel's
>  * default signing key.
>  */
> 
> So, this isn't a one-liner.  *But*, it tells us what "le" means, what
> "pubkey" means and implies that there need to be 4x64-bits worth of MSR
> writes to get to a SHA256 hash.  

In current linux driver implementation, LE is effectively abandoned, because
the initialization of any enclave doesn't take a valid TOKEN, making
initializing enclave requires hash of enclave's signer equal to the hash in
SGX_LEPUBKEYHASH MSRs. 

I written the function name based on SDM's description, to reflect the fact
that we are updating the SGX_LEPUBKEYHASH MSRs, but nothing more.

So perhaps below?

/*
 * Update the SGX_LEPUBKEYHASH MSRs to the values specified by caller.
 *
 * EINITTOKEN is not used in enclave initialization, which requires
 * hash of enclave's signer must match values in SGX_LEPUBKEYHASH MSRs
 * to make EINIT be successful.
 */


It also tells what it's usually doing
> here: overwriting Intel's blasted hash.

Technically, only initial value is intel's pubkey hash. This function
overwrites whatever pubkey hash that used to sign previous enclave.

> 
> It sure beats the entirely uncommented for loop that we've got today.

Agreed, although to me it seems the comment is a little bit out of the scope
of this function itself, but is more about the logic of the caller.
Jarkko Sakkinen Jan. 21, 2021, 1:08 a.m. UTC | #6
On Wed, Jan 20, 2021 at 10:36:09AM -0800, Dave Hansen wrote:
> On 1/20/21 4:03 AM, Jarkko Sakkinen wrote:
> >> +void sgx_update_lepubkeyhash(u64 *lepubkeyhash)
> >> +{
> >> +	int i;
> >> +
> >> +	for (i = 0; i < 4; i++)
> >> +		wrmsrl(MSR_IA32_SGXLEPUBKEYHASH0 + i, lepubkeyhash[i]);
> >> +}
> > Missing kdoc.
> 
> I dunno... kdoc is nice, but I'm not sure its verbosity is useful here,
> even if this function is called from more than one .c file.
> 
> I'd be happy with a single-line comment, personally.

WFM.

/Jarkko
Jarkko Sakkinen Jan. 21, 2021, 1:09 a.m. UTC | #7
On Thu, Jan 21, 2021 at 12:36:25PM +1300, Kai Huang wrote:
> On Wed, 20 Jan 2021 10:36:09 -0800 Dave Hansen wrote:
> > On 1/20/21 4:03 AM, Jarkko Sakkinen wrote:
> > >> +void sgx_update_lepubkeyhash(u64 *lepubkeyhash)
> > >> +{
> > >> +	int i;
> > >> +
> > >> +	for (i = 0; i < 4; i++)
> > >> +		wrmsrl(MSR_IA32_SGXLEPUBKEYHASH0 + i, lepubkeyhash[i]);
> > >> +}
> > > Missing kdoc.
> > 
> > I dunno... kdoc is nice, but I'm not sure its verbosity is useful here,
> > even if this function is called from more than one .c file.
> > 
> > I'd be happy with a single-line comment, personally.
> > 
> 
> I actually feel the function name already explains what the function does
> clearly, therefore I don't think even comment is needed. To be honest I
> don't know how to rephrase here. Perhaps:
> 
> /* Update SGX LEPUBKEYHASH MSRs of the platform. */
> 
> ? 

WFM, thanks.

/Jarkko
Dave Hansen Jan. 21, 2021, 1:15 a.m. UTC | #8
On 1/20/21 5:06 PM, Kai Huang wrote:
> 
> /*
>  * Update the SGX_LEPUBKEYHASH MSRs to the values specified by caller.
>  *
>  * EINITTOKEN is not used in enclave initialization, which requires
>  * hash of enclave's signer must match values in SGX_LEPUBKEYHASH MSRs
>  * to make EINIT be successful.
>  */

I'm grumpy, but I hate it.

I'll stop the bike shedding for now, though.
Kai Huang Jan. 21, 2021, 1:18 a.m. UTC | #9
On Thu, 21 Jan 2021 14:06:38 +1300 Kai Huang wrote:
> On Wed, 20 Jan 2021 15:50:31 -0800 Dave Hansen wrote:
> > On 1/20/21 3:36 PM, Kai Huang wrote:
> > > I actually feel the function name already explains what the function does
> > > clearly, therefore I don't think even comment is needed. To be honest I
> > > don't know how to rephrase here. Perhaps:
> > > 
> > > /* Update SGX LEPUBKEYHASH MSRs of the platform. */
> > 
> > Whee!  I'm gonna write me a function comment!
> > 
> > /*
> >  * A Launch Enclave (LE) must be signed with a public key
> >  * that matches this SHA256 hash.  Usually overwrites Intel's
> >  * default signing key.
> >  */
> > 
> > So, this isn't a one-liner.  *But*, it tells us what "le" means, what
> > "pubkey" means and implies that there need to be 4x64-bits worth of MSR
> > writes to get to a SHA256 hash.  
> 
> In current linux driver implementation, LE is effectively abandoned, because
> the initialization of any enclave doesn't take a valid TOKEN, making
> initializing enclave requires hash of enclave's signer equal to the hash in
> SGX_LEPUBKEYHASH MSRs. 
> 
> I written the function name based on SDM's description, to reflect the fact
> that we are updating the SGX_LEPUBKEYHASH MSRs, but nothing more.
> 
> So perhaps below?
> 
> /*
>  * Update the SGX_LEPUBKEYHASH MSRs to the values specified by caller.
>  *
>  * EINITTOKEN is not used in enclave initialization, which requires
>  * hash of enclave's signer must match values in SGX_LEPUBKEYHASH MSRs
>  * to make EINIT be successful.
>  */
> 

Actually I take it back. This is only valid for bare-metal driver. For KVM
guest, it should be: 

  /*
   * Update the SGX_LEPUBKEYHASH MSRs according to guest's *virtual*
   * SGX_LEPUBKEYHASH MSRs values, to make EINIT from guest consistent
   * with hardware behavior.
   */

So like I said below, the comment is actually more reasonable for the logic of
caller of this function.

Makes sense?

> 
> It also tells what it's usually doing
> > here: overwriting Intel's blasted hash.
> 
> Technically, only initial value is intel's pubkey hash. This function
> overwrites whatever pubkey hash that used to sign previous enclave.
> 
> > 
> > It sure beats the entirely uncommented for loop that we've got today.
> 
> Agreed, although to me it seems the comment is a little bit out of the scope
> of this function itself, but is more about the logic of the caller.
Kai Huang Jan. 21, 2021, 1:44 a.m. UTC | #10
On Wed, 20 Jan 2021 17:15:35 -0800 Dave Hansen wrote:
> On 1/20/21 5:06 PM, Kai Huang wrote:
> > 
> > /*
> >  * Update the SGX_LEPUBKEYHASH MSRs to the values specified by caller.
> >  *
> >  * EINITTOKEN is not used in enclave initialization, which requires
> >  * hash of enclave's signer must match values in SGX_LEPUBKEYHASH MSRs
> >  * to make EINIT be successful.
> >  */
> 
> I'm grumpy, but I hate it.
> 
> I'll stop the bike shedding for now, though.

Jarkko and Dave,

I'll change to use below:

 /*
  * Update the SGX_LEPUBKEYHASH MSRs to the values specified by caller.
  * Bare-metal driver requires to update them to hash of enclave's signer
  * before EINIT. KVM needs to update them to guest's virtual MSR values
  * before doing EINIT from guest.
  */

Please let me know if are not OK with this.
Jarkko Sakkinen Jan. 21, 2021, 2:36 p.m. UTC | #11
On Thu, Jan 21, 2021 at 02:44:26PM +1300, Kai Huang wrote:
> On Wed, 20 Jan 2021 17:15:35 -0800 Dave Hansen wrote:
> > On 1/20/21 5:06 PM, Kai Huang wrote:
> > > 
> > > /*
> > >  * Update the SGX_LEPUBKEYHASH MSRs to the values specified by caller.
> > >  *
> > >  * EINITTOKEN is not used in enclave initialization, which requires
> > >  * hash of enclave's signer must match values in SGX_LEPUBKEYHASH MSRs
> > >  * to make EINIT be successful.
> > >  */
> > 
> > I'm grumpy, but I hate it.
> > 
> > I'll stop the bike shedding for now, though.
> 
> Jarkko and Dave,
> 
> I'll change to use below:
> 
>  /*
>   * Update the SGX_LEPUBKEYHASH MSRs to the values specified by caller.
>   * Bare-metal driver requires to update them to hash of enclave's signer
>   * before EINIT. KVM needs to update them to guest's virtual MSR values
>   * before doing EINIT from guest.
>   */
>
> Please let me know if are not OK with this.

I am.

/Jarkko
diff mbox series

Patch

diff --git a/arch/x86/kernel/cpu/sgx/ioctl.c b/arch/x86/kernel/cpu/sgx/ioctl.c
index e5977752c7be..1bae754268d1 100644
--- a/arch/x86/kernel/cpu/sgx/ioctl.c
+++ b/arch/x86/kernel/cpu/sgx/ioctl.c
@@ -495,7 +495,7 @@  static int sgx_encl_init(struct sgx_encl *encl, struct sgx_sigstruct *sigstruct,
 			 void *token)
 {
 	u64 mrsigner[4];
-	int i, j, k;
+	int i, j;
 	void *addr;
 	int ret;
 
@@ -544,8 +544,7 @@  static int sgx_encl_init(struct sgx_encl *encl, struct sgx_sigstruct *sigstruct,
 
 			preempt_disable();
 
-			for (k = 0; k < 4; k++)
-				wrmsrl(MSR_IA32_SGXLEPUBKEYHASH0 + k, mrsigner[k]);
+			sgx_update_lepubkeyhash(mrsigner);
 
 			ret = __einit(sigstruct, token, addr);
 
diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
index bdda631c975b..1cf1f0f058b8 100644
--- a/arch/x86/kernel/cpu/sgx/main.c
+++ b/arch/x86/kernel/cpu/sgx/main.c
@@ -697,6 +697,14 @@  static bool __init sgx_page_cache_init(void)
 	return true;
 }
 
+void sgx_update_lepubkeyhash(u64 *lepubkeyhash)
+{
+	int i;
+
+	for (i = 0; i < 4; i++)
+		wrmsrl(MSR_IA32_SGXLEPUBKEYHASH0 + i, lepubkeyhash[i]);
+}
+
 static void __init sgx_init(void)
 {
 	int ret;
diff --git a/arch/x86/kernel/cpu/sgx/sgx.h b/arch/x86/kernel/cpu/sgx/sgx.h
index 509f2af33e1d..ccd4f145c464 100644
--- a/arch/x86/kernel/cpu/sgx/sgx.h
+++ b/arch/x86/kernel/cpu/sgx/sgx.h
@@ -83,4 +83,6 @@  void sgx_mark_page_reclaimable(struct sgx_epc_page *page);
 int sgx_unmark_page_reclaimable(struct sgx_epc_page *page);
 struct sgx_epc_page *sgx_alloc_epc_page(void *owner, bool reclaim);
 
+void sgx_update_lepubkeyhash(u64 *lepubkeyhash);
+
 #endif /* _X86_SGX_H */