[PATCH v2] virt/coco/sev-guest: Remove err in handle_guest_request

[PATCH v2] virt/coco/sev-guest: Remove err in handle_guest_request

[Dionna Glaze]

The err variable may not be set in the call to snp_issue_guest_request,
yet it is unconditionally written back to fw_err if fw_err is non-null.
This is undefined behavior, and currently returns uninitialized kernel
stack memory to user space.

The fw_err argument is better to just pass through to
snp_issue_guest_request, so we do that.

Cc: Tom Lendacky <Thomas.Lendacky@amd.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Joerg Roedel <jroedel@suse.de>
Cc: Peter Gonda <pgonda@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Dave Hansen <dave.hansen@linux.intel.com>

Signed-off-by: Dionna Glaze <dionnaglaze@google.com>
---
 drivers/virt/coco/sev-guest/sev-guest.c | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)

diff --git a/drivers/virt/coco/sev-guest/sev-guest.c b/drivers/virt/coco/sev-guest/sev-guest.c
index f422f9c58ba7..00708b4491e0 100644
--- a/drivers/virt/coco/sev-guest/sev-guest.c
+++ b/drivers/virt/coco/sev-guest/sev-guest.c
@@ -305,7 +305,6 @@ static int handle_guest_request(struct snp_guest_dev *snp_dev, u64 exit_code, in
 				u8 type, void *req_buf, size_t req_sz, void *resp_buf,
 				u32 resp_sz, __u64 *fw_err)
 {
-	unsigned long err;
 	u64 seqno;
 	int rc;
 
@@ -322,9 +321,7 @@ static int handle_guest_request(struct snp_guest_dev *snp_dev, u64 exit_code, in
 		return rc;
 
 	/* Call firmware to process the request */
-	rc = snp_issue_guest_request(exit_code, &snp_dev->input, &err);
-	if (fw_err)
-		*fw_err = err;
+	rc = snp_issue_guest_request(exit_code, &snp_dev->input, fw_err);
 
 	if (rc)
 		return rc;
-- 
2.38.0.413.g74048e4d9e-goog

Re: [PATCH v2] virt/coco/sev-guest: Remove err in handle_guest_request

[Tom Lendacky]

On 10/19/22 14:58, Dionna Glaze wrote:
> The err variable may not be set in the call to snp_issue_guest_request,
> yet it is unconditionally written back to fw_err if fw_err is non-null.
> This is undefined behavior, and currently returns uninitialized kernel
> stack memory to user space.
> 
> The fw_err argument is better to just pass through to
> snp_issue_guest_request, so we do that.
> 
> Cc: Tom Lendacky <Thomas.Lendacky@amd.com>
> Cc: Paolo Bonzini <pbonzini@redhat.com>
> Cc: Joerg Roedel <jroedel@suse.de>
> Cc: Peter Gonda <pgonda@google.com>
> Cc: Thomas Gleixner <tglx@linutronix.de>
> Cc: Dave Hansen <dave.hansen@linux.intel.com>
> 
> Signed-off-by: Dionna Glaze <dionnaglaze@google.com>

You should include a Fixes: tag.

With that,
Reviewed-by: Tom Lendacky <thomas.lendacky@amd.com>

Also, you may want to base your other series [1] on this one, since you'll
likely get conflicts if this is applied first.

[1] https://lore.kernel.org/lkml/20221019173537.1238121-1-dionnaglaze@google.com/

Thanks,
Tom

> ---
>   drivers/virt/coco/sev-guest/sev-guest.c | 5 +----
>   1 file changed, 1 insertion(+), 4 deletions(-)
> 
> diff --git a/drivers/virt/coco/sev-guest/sev-guest.c b/drivers/virt/coco/sev-guest/sev-guest.c
> index f422f9c58ba7..00708b4491e0 100644
> --- a/drivers/virt/coco/sev-guest/sev-guest.c
> +++ b/drivers/virt/coco/sev-guest/sev-guest.c
> @@ -305,7 +305,6 @@ static int handle_guest_request(struct snp_guest_dev *snp_dev, u64 exit_code, in
>   				u8 type, void *req_buf, size_t req_sz, void *resp_buf,
>   				u32 resp_sz, __u64 *fw_err)
>   {
> -	unsigned long err;
>   	u64 seqno;
>   	int rc;
>   
> @@ -322,9 +321,7 @@ static int handle_guest_request(struct snp_guest_dev *snp_dev, u64 exit_code, in
>   		return rc;
>   
>   	/* Call firmware to process the request */
> -	rc = snp_issue_guest_request(exit_code, &snp_dev->input, &err);
> -	if (fw_err)
> -		*fw_err = err;
> +	rc = snp_issue_guest_request(exit_code, &snp_dev->input, fw_err);
>   
>   	if (rc)
>   		return rc;

Re: [PATCH v2] virt/coco/sev-guest: Remove err in handle_guest_request

[Dionna Amalie Glaze]

>
> You should include a Fixes: tag.
>
> With that,
> Reviewed-by: Tom Lendacky <thomas.lendacky@amd.com>
>
> Also, you may want to base your other series [1] on this one, since you'll
> likely get conflicts if this is applied first.
>
> [1] https://lore.kernel.org/lkml/20221019173537.1238121-1-dionnaglaze@google.com/
>

Let's call this a wash and I'll roll this into that other series then.
By passing through fw_err directly, I get an unsigned long vs unsigned
long long type error (in this v2, mea culpa). So I'll put the rename
and re-typing in the x86 patch before this one.

-- 
-Dionna Glaze, PhD (she/her)