| Message ID | 20230214131959.40298-1-neal.frager@amd.com |
|---|---|
| State | Changes Requested |
| Delegated to: | Michal Simek |
| Headers | show |
| Series | [v1,1/1] fpga: zynqmppl: fix fpga loads command for unencrypted use case | expand |
Acked-by: Ashok Reddy Soma <ashok.reddy.soma@amd.com> Thanks, Ashok > -----Original Message----- > From: Neal Frager <neal.frager@amd.com> > Sent: Tuesday, February 14, 2023 6:50 PM > To: u-boot@lists.denx.de; git-dev (AMD-Xilinx) <git-dev@amd.com> > Cc: Simek, Michal <michal.simek@amd.com>; Soma, Ashok Reddy > <ashok.reddy.soma@amd.com>; Manne, Nava kishore > <nava.kishore.manne@amd.com>; Erkiaga Elorza, Ibai <ibai.erkiaga- > elorza@amd.com>; Frager, Neal <neal.frager@amd.com> > Subject: [PATCH v1 1/1] fpga: zynqmppl: fix fpga loads command for > unencrypted use case > > When using the fpga loads command, the driver is passing the AES > encryption key address is all cases. However, for the authenticated, but not > encrypted use case, there is no AES encryption key, and this value is 0. > > When AES encryption is not used on the fpga bitstream, the pmufw assumes > that the AES key address is a bitstream size value like what is used by the > unsecure fpga load command. > > To fix the problem, this patch checks to see if the AES key address is zero. > If the AES key address is zero, it means that AES is not being used on the > bitstream and the bitstream size should be passed instead. Thus, matching > the fpga load functionality. > > Signed-off-by: Neal Frager <neal.frager@amd.com> > --- > drivers/fpga/zynqmppl.c | 12 +++++++++--- > 1 file changed, 9 insertions(+), 3 deletions(-) > > diff --git a/drivers/fpga/zynqmppl.c b/drivers/fpga/zynqmppl.c index > 6b394869db..718c04b22c 100644 > --- a/drivers/fpga/zynqmppl.c > +++ b/drivers/fpga/zynqmppl.c > @@ -278,10 +278,16 @@ static int zynqmp_loads(xilinx_desc *desc, const > void *buf, size_t bsize, > buf_lo = lower_32_bits((ulong)buf); > buf_hi = upper_32_bits((ulong)buf); > > - ret = xilinx_pm_request(PM_FPGA_LOAD, buf_lo, > + if ((u32)(uintptr_t)fpga_sec_info->userkey_addr) > + ret = xilinx_pm_request(PM_FPGA_LOAD, buf_lo, > buf_hi, > - (u32)(uintptr_t)fpga_sec_info->userkey_addr, > - flag, ret_payload); > + (u32)(uintptr_t)fpga_sec_info- > >userkey_addr, > + flag, ret_payload); > + else > + ret = xilinx_pm_request(PM_FPGA_LOAD, buf_lo, > + buf_hi, (u32)bsize, > + flag, ret_payload); > + > if (ret) > puts("PL FPGA LOAD fail\n"); > else > -- > 2.17.1
On 2/14/23 14:19, Neal Frager wrote: > When using the fpga loads command, the driver is passing the AES encryption > key address is all cases. However, for the authenticated, but not encrypted > use case, there is no AES encryption key, and this value is 0. > > When AES encryption is not used on the fpga bitstream, the pmufw assumes that > the AES key address is a bitstream size value like what is used by the > unsecure fpga load command. > > To fix the problem, this patch checks to see if the AES key address is zero. > If the AES key address is zero, it means that AES is not being used on the > bitstream and the bitstream size should be passed instead. Thus, matching > the fpga load functionality. > > Signed-off-by: Neal Frager <neal.frager@amd.com> > --- > drivers/fpga/zynqmppl.c | 12 +++++++++--- > 1 file changed, 9 insertions(+), 3 deletions(-) > > diff --git a/drivers/fpga/zynqmppl.c b/drivers/fpga/zynqmppl.c > index 6b394869db..718c04b22c 100644 > --- a/drivers/fpga/zynqmppl.c > +++ b/drivers/fpga/zynqmppl.c > @@ -278,10 +278,16 @@ static int zynqmp_loads(xilinx_desc *desc, const void *buf, size_t bsize, > buf_lo = lower_32_bits((ulong)buf); > buf_hi = upper_32_bits((ulong)buf); > > - ret = xilinx_pm_request(PM_FPGA_LOAD, buf_lo, > + if ((u32)(uintptr_t)fpga_sec_info->userkey_addr) > + ret = xilinx_pm_request(PM_FPGA_LOAD, buf_lo, > buf_hi, > - (u32)(uintptr_t)fpga_sec_info->userkey_addr, > - flag, ret_payload); > + (u32)(uintptr_t)fpga_sec_info->userkey_addr, > + flag, ret_payload); > + else > + ret = xilinx_pm_request(PM_FPGA_LOAD, buf_lo, > + buf_hi, (u32)bsize, > + flag, ret_payload); > + > if (ret) > puts("PL FPGA LOAD fail\n"); > else applied. M
diff --git a/drivers/fpga/zynqmppl.c b/drivers/fpga/zynqmppl.c index 6b394869db..718c04b22c 100644 --- a/drivers/fpga/zynqmppl.c +++ b/drivers/fpga/zynqmppl.c @@ -278,10 +278,16 @@ static int zynqmp_loads(xilinx_desc *desc, const void *buf, size_t bsize, buf_lo = lower_32_bits((ulong)buf); buf_hi = upper_32_bits((ulong)buf); - ret = xilinx_pm_request(PM_FPGA_LOAD, buf_lo, + if ((u32)(uintptr_t)fpga_sec_info->userkey_addr) + ret = xilinx_pm_request(PM_FPGA_LOAD, buf_lo, buf_hi, - (u32)(uintptr_t)fpga_sec_info->userkey_addr, - flag, ret_payload); + (u32)(uintptr_t)fpga_sec_info->userkey_addr, + flag, ret_payload); + else + ret = xilinx_pm_request(PM_FPGA_LOAD, buf_lo, + buf_hi, (u32)bsize, + flag, ret_payload); + if (ret) puts("PL FPGA LOAD fail\n"); else
When using the fpga loads command, the driver is passing the AES encryption key address is all cases. However, for the authenticated, but not encrypted use case, there is no AES encryption key, and this value is 0. When AES encryption is not used on the fpga bitstream, the pmufw assumes that the AES key address is a bitstream size value like what is used by the unsecure fpga load command. To fix the problem, this patch checks to see if the AES key address is zero. If the AES key address is zero, it means that AES is not being used on the bitstream and the bitstream size should be passed instead. Thus, matching the fpga load functionality. Signed-off-by: Neal Frager <neal.frager@amd.com> --- drivers/fpga/zynqmppl.c | 12 +++++++++--- 1 file changed, 9 insertions(+), 3 deletions(-)