The receive buffers of the NETSEC driver are owned by the driver itself
(as opposed to the protocol client in the case of the transmit path),
and so the descriptors and the buffers (which are of a fixed size) are
allocated in one go. The idea is that the 'buffer' member of the
descriptor should point to a DMA aligned offset into the same allocation,
but the code in pfdep_alloc_pkt_buf() calculates the value incorrectly,
resulting in corruption of the descriptor metadata if the pool allocation
happens to be DMA aligned.

Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Ard Biesheuvel <ard.biesheu...@linaro.org>
---
 Silicon/Socionext/SynQuacer/Drivers/Net/NetsecDxe/netsec_for_uefi/pfdep_uefi.c 
| 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git 
a/Silicon/Socionext/SynQuacer/Drivers/Net/NetsecDxe/netsec_for_uefi/pfdep_uefi.c
 
b/Silicon/Socionext/SynQuacer/Drivers/Net/NetsecDxe/netsec_for_uefi/pfdep_uefi.c
index b43d1aaff037..0875558c08ea 100644
--- 
a/Silicon/Socionext/SynQuacer/Drivers/Net/NetsecDxe/netsec_for_uefi/pfdep_uefi.c
+++ 
b/Silicon/Socionext/SynQuacer/Drivers/Net/NetsecDxe/netsec_for_uefi/pfdep_uefi.c
@@ -134,7 +134,7 @@ pfdep_alloc_pkt_buf (
       return PFDEP_ERR_ALLOC;
     }
 
-    (*pkt_handle_p)->Buffer = ALIGN_POINTER (*pkt_handle_p,
+    (*pkt_handle_p)->Buffer = ALIGN_POINTER (*pkt_handle_p + 1,
                                              mCpu->DmaBufferAlignment);
   }
 
-- 
2.17.0

_______________________________________________
edk2-devel mailing list
edk2-devel@lists.01.org
https://lists.01.org/mailman/listinfo/edk2-devel

Reply via email to