The MAX10 BMC Secure Update driver instantiates the new Firmware
Upload functionality of the Firmware Loader and provides the callback
functions required to support secure updates on Intel n3000 PAC
devices. This driver is implemented as a sub-driver of the Intel MAX10
BMC mfd driver.
This driver interacts with the HW secure update engine of the FPGA
card BMC in order to transfer new FPGA and BMC images to FLASH on
the FPGA card. Security is enforced by hardware and firmware. The
FPGA Card BMC Secure Update driver interacts with the firmware to
initiate an update, pass in the necessary data, and collect status
on the update.
This driver provides sysfs files for displaying the flash count, the
root entry hashes (REH), and the code-signing-key (CSK) cancellation
vectors.
Changelog v19 -> v20:
- Added text to commit messages to describe Root Entry Hashes (REH) and
Code Signing Key (CSK) cancellation.
- Use reverse christmas tree format for local variable declarations in
show_root_entry_hash().
- Remove WARN_ON() from show_root_entry_hash() and return -EINVAL if
sha_num_bytes is not a multiple of stride.
- Move MODULE_DEVICE_TABLE() macro to just beneath the definition of
intel_m10bmc_sec_ids[].
Changelog v18 -> v19:
- Change "card bmc" naming back to "m10 bmc" naming to be consistent
with the parent driver.
Changelog v17 -> v18:
- Changed the ABI documentation for the Root Entry Hashes to specify
string as the format for the output.
- Updated comments, strings and config options to more consistently
refer to the driver as the Intel FPGA Card BMC Secure Update driver.
- Removed an instance of dev_dbg().
- Deferred the call to firmware_upload_register() to a later patch
where the required ops are provided.
- Switched from MODULE_ALIAS() to MODULE_DEVICE_TABLE() in anticipation
of additional cards to be supported by the same driver.
Changelog v16 -> v17:
- Change m10bmc to cardbmc to reflect the fact that the future devices
will not necessarily use the MAX10. This affects filenames, configs,
symbol names, and the driver name.
- Update the Date and KernelVersion for the ABI documentation to Jul 2022
and 5.19 respectively.
- Updated the copyright end-date to 2022 for the secure update driver.
- Removed references to the FPGA Image Load class driver and replaced
them with the new firmware-upload service from the firmware loader.
- Use xarray_alloc to generate a unique number/name firmware-upload.
- Chang the license from GPL to GPLv2 per commit bf7fbeeae6db ("module:
Cure the MODULE_LICENSE "GPL" vs. "GPL v2" bogosity")
- fw_upload_ops functions will return "enum fw_upload_err" data types
instead of integer values.
Changelog v15 -> v16:
- Use 0 instead of FPGA_IMAGE_ERR_NONE to indicate success.
- The size alignment check was moved from the FPGA Image Load framework
to the prepare() op.
- Added cancel_request boolean flag to struct m10bmc_sec.
- Moved the RSU cancellation logic from m10bmc_sec_cancel() to a new
rsu_cancel() function.
- The m10bmc_sec_cancel() function ONLY sets the cancel_request flag.
The cancel_request flag is checked at the beginning of the
m10bmc_sec_write() and m10bmc_sec_poll_complete() functions.
- Adapt to changed prototypes for the prepare() and write() ops. The
m10bmc_sec_write_blk() function has been renamed to
m10bmc_sec_write().
- Created a cleanup op, m10bmc_sec_cleanup(), to attempt to cancel an
ongoing op during when exiting the update process.
Changelog v14 -> v15:
- Updated the Dates and KernelVersions in the ABI documentation
- Change driver name from "n3000bmc-secure" to "n3000bmc-sec-update".
- Change CONFIG_FPGA_M10_BMC_SECURE to CONFIG_FPGA_M10_BMC_SEC_UPDATE.
- Change instances of *bmc-secure to *bmc-sec-update in file name
and symbol names.
- Change instances of *m10bmc_secure* to *m10bmc-sec_update* in symbol
names.
- Adapted to changes in the FPGA Image Load framework:
(1) All enum types (progress and errors) are now type u32
(2) m10bmc_sec_write_blk() adds *blk_size and max_size parameters
and uses *blk_size as provided by the caller.
(3) m10bmc_sec_poll_complete() no long checks the driver_unload
flag.
Changelog v13 -> v14:
- Changed symbol and text references to reflect the renaming of the
Security Manager Class driver to FPGA Image Load.
Changelog v12 -> v13:
- Updated copyright to 2021
- Updated Date and KernelVersion fields in ABI documentation
- Call updated fpga_sec_mgr_register() and fpga_sec_mgr_unregister()
functions instead of devm_fpga_sec_mgr_create() and
devm_fpga_sec_mgr_register().
Changelog v11 -> v12:
- Updated Date and KernelVersion fields in ABI documentation
- Removed size parameter from the write_blk() op. m10bmc_sec_write_blk()
no longer has a size parameter, and the block size is determined
in this (the lower-level) driver.
Changelog v10 -> v11:
- Added Reviewed-by tag to patch #1
Changelog v9 -> v10:
- Changed the path expressions in the sysfs documentation to
replace the n3000 reference with something more generic to
accommodate other devices that use the same driver.
Changelog v8 -> v9:
- Rebased to 5.12-rc2 next
- Updated Date and KernelVersion in ABI documentation
Changelog v7 -> v8:
- Split out patch "mfd: intel-m10-bmc: support for MAX10 BMC Secure
Updates" and submitted it separately:
https://marc.info/?l=linux-kernel&m=161126987101096&w=2
Changelog v6 -> v7:
- Rebased patches for 5.11-rc2
- Updated Date and KernelVersion in ABI documentation
Changelog v5 -> v6:
- Added WARN_ON() prior to several calls to regmap_bulk_read()
to assert that the (SIZE / stride) calculations did not result
in remainders.
- Changed the (size / stride) calculation in regmap_bulk_write()
call to ensure that we don't write one less than intended.
- Changed flash_count_show() parameter list to achieve
reverse-christmas tree format.
- Removed unnecessary call to rsu_check_complete() in
m10bmc_sec_poll_complete() and changed while loop to
do/while loop.
- Initialized auth_result and doorbell to HW_ERRINFO_POISON
in m10bmc_sec_hw_errinfo() and removed unnecessary if statements.
Changelog v4 -> v5:
- Renamed sysfs node user_flash_count to flash_count and updated
the sysfs documentation accordingly to more accurately descirbe
the purpose of the count.
Changelog v3 -> v4:
- Moved sysfs files for displaying the flash count, the root
entry hashes (REH), and the code-signing-key (CSK) cancellation
vectors from the FPGA Security Manager class driver to this
driver (as they are not generic enough for the class driver).
- Added a new ABI documentation file with informtaion about the
new sysfs entries: sysfs-driver-intel-m10-bmc-secure
- Updated the MAINTAINERS file to add the new ABI documentation
file: sysfs-driver-intel-m10-bmc-secure
- Removed unnecessary ret variable from m10bmc_secure_probe()
- Incorporated new devm_fpga_sec_mgr_register() function into
m10bmc_secure_probe() and removed the m10bmc_secure_remove()
function.
Changelog v2 -> v3:
- Changed "MAX10 BMC Security Engine driver" to "MAX10 BMC Secure
Update driver"
- Changed from "Intel FPGA Security Manager" to FPGA Security Manager"
- Changed: iops -> sops, imgr -> smgr, IFPGA_ -> FPGA_, ifpga_ to fpga_
- Removed wrapper functions (m10bmc_raw_*, m10bmc_sys_*). The
underlying functions are now called directly.
- Changed "_root_entry_hash" to "_reh", with a comment explaining
what reh is.
- Renamed get_csk_vector() to m10bmc_csk_vector()
- Changed calling functions of functions that return "enum fpga_sec_err"
to check for (ret != FPGA_SEC_ERR_NONE) instead of (ret)
Changelog v1 -> v2:
- These patches were previously submitted as part of a larger V1
patch set under the title "Intel FPGA Security Manager Class Driver".
- Grouped all changes to include/linux/mfd/intel-m10-bmc.h into a
single patch: "mfd: intel-m10-bmc: support for MAX10 BMC Security
Engine".
- Removed ifpga_sec_mgr_init() and ifpga_sec_mgr_uinit() functions.
- Adapted to changes in the Intel FPGA Security Manager by splitting
the single call to ifpga_sec_mgr_register() into two function
calls: devm_ifpga_sec_mgr_create() and ifpga_sec_mgr_register().
- Replaced small function-creation macros for explicit function
declarations.
- Bug fix for the get_csk_vector() function to properly apply the
stride variable in calls to m10bmc_raw_bulk_read().
- Added m10bmc_ prefix to functions in m10bmc_iops structure
- Implemented HW_ERRINFO_POISON for m10bmc_sec_hw_errinfo() to
ensure that corresponding bits are set to 1 if we are unable
to read the doorbell or auth_result registers.
- Added comments and additional code cleanup per V1 review.
Russ Weight (5):
mfd: intel-m10-bmc: Rename n3000bmc-secure driver
fpga: m10bmc-sec: create max10 bmc secure update
fpga: m10bmc-sec: expose max10 flash update count
fpga: m10bmc-sec: expose max10 canceled keys in sysfs
fpga: m10bmc-sec: add max10 secure update functions
.../sysfs-driver-intel-m10-bmc-sec-update | 61 ++
MAINTAINERS | 7 +
drivers/fpga/Kconfig | 12 +
drivers/fpga/Makefile | 3 +
drivers/fpga/intel-m10-bmc-sec-update.c | 595 ++++++++++++++++++
drivers/mfd/intel-m10-bmc.c | 2 +-
6 files changed, 679 insertions(+), 1 deletion(-)
create mode 100644 Documentation/ABI/testing/sysfs-driver-intel-m10-bmc-sec-update
create mode 100644 drivers/fpga/intel-m10-bmc-sec-update.c
base-commit: 3f7bdc402fb06f897ff1f492a2d42e1f7c2efedb
--
2.25.1
Extend the MAX10 BMC Secure Update driver to provide sysfs files to
expose the 128 bit code signing key (CSK) cancellation vectors. These use
the standard bitmap list format (e.g. 1,2-6,9).
Each CSK is assigned an ID, a number between 0-127, during the signing
process. CSK ID cancellation information is stored in 128-bit fields in
write-once locations in flash. The cancellation of a CSK can be used
to prevent the card from being rolled back to older images that were
signed with a CSK that is now cancelled.
Signed-off-by: Russ Weight <[email protected]>
Reviewed-by: Tom Rix <[email protected]>
---
v20:
- Added text to the commit message to further describe the cancellation of
code signing keys.
v19:
- Change "card bmc" naming back to "m10 bmc" naming to be consistent
with the parent driver.
v18:
- No change
v17:
- Update the Date and KernelVersion for the ABI documentation to Jul 2022
and 5.19 respectively.
- Change "m10bmc" in symbol names to "cardbmc" to reflect the fact that the
future devices will not necessarily use the MAX10.
v16:
- No Change
v15:
- Updated the Dates and KernelVersions in the ABI documentation
v14:
- No changes
v13:
- Updated ABI documentation date and kernel version
v12:
- Updated Date and KernelVersion fields in ABI documentation
v11:
- No change
v10:
- Changed the path expressions in the sysfs documentation to
replace the n3000 reference with something more generic to
accomodate other devices that use the same driver.
v9:
- Rebased to 5.12-rc2 next
- Updated Date and KernelVersion in ABI documentation
v8:
- Previously patch 4/6, otherwise no change
v7:
- Updated Date and KernelVersion in ABI documentation
v6:
- Added WARN_ON() call for (size / stride) to ensure
that the proper count is passed to regmap_bulk_read().
v5:
- No change
v4:
- Moved sysfs files for displaying the code-signing-key (CSK)
cancellation vectors from the FPGA Security Manger class driver
to here. The m10bmc_csk_vector() and m10bmc_csk_cancel_nbits()
functions are removed and the functionality from these functions
is moved into a show_canceled_csk() function for for displaying
the CSK vectors.
- Added ABI documentation for new sysfs entries
v3:
- Changed: iops -> sops, imgr -> smgr, IFPGA_ -> FPGA_, ifpga_ to fpga_
- Changed "MAX10 BMC Secure Engine driver" to "MAX10 BMC Secure Update
driver"
- Removed wrapper functions (m10bmc_raw_*, m10bmc_sys_*). The
underlying functions are now called directly.
- Renamed get_csk_vector() to m10bmc_csk_vector()
v2:
- Replaced small function-creation macros for explicit function
declarations.
- Fixed get_csk_vector() function to properly apply the stride
variable in calls to m10bmc_raw_bulk_read()
- Added m10bmc_ prefix to functions in m10bmc_iops structure
---
.../sysfs-driver-intel-m10-bmc-sec-update | 24 ++++++++++
drivers/fpga/intel-m10-bmc-sec-update.c | 48 +++++++++++++++++++
2 files changed, 72 insertions(+)
diff --git a/Documentation/ABI/testing/sysfs-driver-intel-m10-bmc-sec-update b/Documentation/ABI/testing/sysfs-driver-intel-m10-bmc-sec-update
index 1132e39b2125..ca5a34c1c31f 100644
--- a/Documentation/ABI/testing/sysfs-driver-intel-m10-bmc-sec-update
+++ b/Documentation/ABI/testing/sysfs-driver-intel-m10-bmc-sec-update
@@ -28,6 +28,30 @@ Description: Read only. Returns the root entry hash for the BMC image
underlying device supports it.
Format: string.
+What: /sys/bus/platform/drivers/intel-m10bmc-sec-update/.../security/sr_canceled_csks
+Date: Jul 2022
+KernelVersion: 5.19
+Contact: Russ Weight <[email protected]>
+Description: Read only. Returns a list of indices for canceled code
+ signing keys for the static region. The standard bitmap
+ list format is used (e.g. "1,2-6,9").
+
+What: /sys/bus/platform/drivers/intel-m10bmc-sec-update/.../security/pr_canceled_csks
+Date: Jul 2022
+KernelVersion: 5.19
+Contact: Russ Weight <[email protected]>
+Description: Read only. Returns a list of indices for canceled code
+ signing keys for the partial reconfiguration region. The
+ standard bitmap list format is used (e.g. "1,2-6,9").
+
+What: /sys/bus/platform/drivers/intel-m10bmc-sec-update/.../security/bmc_canceled_csks
+Date: Jul 2022
+KernelVersion: 5.19
+Contact: Russ Weight <[email protected]>
+Description: Read only. Returns a list of indices for canceled code
+ signing keys for the BMC. The standard bitmap list format
+ is used (e.g. "1,2-6,9").
+
What: /sys/bus/platform/drivers/intel-m10bmc-sec-update/.../security/flash_count
Date: Jul 2022
KernelVersion: 5.19
diff --git a/drivers/fpga/intel-m10-bmc-sec-update.c b/drivers/fpga/intel-m10-bmc-sec-update.c
index 3f183202de3b..6c39adc6492d 100644
--- a/drivers/fpga/intel-m10-bmc-sec-update.c
+++ b/drivers/fpga/intel-m10-bmc-sec-update.c
@@ -78,6 +78,51 @@ DEVICE_ATTR_SEC_REH_RO(bmc, BMC_PROG_MAGIC, BMC_PROG_ADDR, BMC_REH_ADDR);
DEVICE_ATTR_SEC_REH_RO(sr, SR_PROG_MAGIC, SR_PROG_ADDR, SR_REH_ADDR);
DEVICE_ATTR_SEC_REH_RO(pr, PR_PROG_MAGIC, PR_PROG_ADDR, PR_REH_ADDR);
+#define CSK_BIT_LEN 128U
+#define CSK_32ARRAY_SIZE DIV_ROUND_UP(CSK_BIT_LEN, 32)
+
+static ssize_t
+show_canceled_csk(struct device *dev, u32 addr, char *buf)
+{
+ unsigned int i, stride, size = CSK_32ARRAY_SIZE * sizeof(u32);
+ struct m10bmc_sec *sec = dev_get_drvdata(dev);
+ DECLARE_BITMAP(csk_map, CSK_BIT_LEN);
+ __le32 csk_le32[CSK_32ARRAY_SIZE];
+ u32 csk32[CSK_32ARRAY_SIZE];
+ int ret;
+
+ stride = regmap_get_reg_stride(sec->m10bmc->regmap);
+
+ WARN_ON(size % stride);
+ ret = regmap_bulk_read(sec->m10bmc->regmap, addr, csk_le32,
+ size / stride);
+ if (ret) {
+ dev_err(sec->dev, "failed to read CSK vector: %x cnt %x: %d\n",
+ addr, size / stride, ret);
+ return ret;
+ }
+
+ for (i = 0; i < CSK_32ARRAY_SIZE; i++)
+ csk32[i] = le32_to_cpu(((csk_le32[i])));
+
+ bitmap_from_arr32(csk_map, csk32, CSK_BIT_LEN);
+ bitmap_complement(csk_map, csk_map, CSK_BIT_LEN);
+ return bitmap_print_to_pagebuf(1, buf, csk_map, CSK_BIT_LEN);
+}
+
+#define DEVICE_ATTR_SEC_CSK_RO(_name, _addr) \
+static ssize_t _name##_canceled_csks_show(struct device *dev, \
+ struct device_attribute *attr, \
+ char *buf) \
+{ return show_canceled_csk(dev, _addr, buf); } \
+static DEVICE_ATTR_RO(_name##_canceled_csks)
+
+#define CSK_VEC_OFFSET 0x34
+
+DEVICE_ATTR_SEC_CSK_RO(bmc, BMC_PROG_ADDR + CSK_VEC_OFFSET);
+DEVICE_ATTR_SEC_CSK_RO(sr, SR_PROG_ADDR + CSK_VEC_OFFSET);
+DEVICE_ATTR_SEC_CSK_RO(pr, PR_PROG_ADDR + CSK_VEC_OFFSET);
+
#define FLASH_COUNT_SIZE 4096 /* count stored as inverted bit vector */
static ssize_t flash_count_show(struct device *dev,
@@ -118,6 +163,9 @@ static struct attribute *m10bmc_security_attrs[] = {
&dev_attr_bmc_root_entry_hash.attr,
&dev_attr_sr_root_entry_hash.attr,
&dev_attr_pr_root_entry_hash.attr,
+ &dev_attr_sr_canceled_csks.attr,
+ &dev_attr_pr_canceled_csks.attr,
+ &dev_attr_bmc_canceled_csks.attr,
NULL,
};
--
2.25.1
Create firmware upload ops and call the Firmware Upload support of the
Firmware Loader subsystem to enable FPGA image uploads for secure
updates of BMC images, FPGA images, etc.
Signed-off-by: Russ Weight <[email protected]>
---
v20:
- No change.
v19:
- Change "card bmc" naming back to "m10 bmc" naming to be consistent
with the parent driver.
v18:
- Moved the firmware_upload_register() function here from an earlier
patch since this is where the required ops are provided.
- Moved the bmc_sec_remove() function here from an earlier patch to
unregister the firmware driver and do cleanup.
v17:
- Change "m10bmc" in symbol names to "cardbmc" to reflect the fact that the
future devices will not necessarily use the MAX10.
- Change from image_load class driver to the new firmware_upload
functionality of the firmware_loader.
- fw_upload_ops functions will return "enum fw_upload_err" data types
instead of integer values.
v16:
- Use 0 instead of FPGA_IMAGE_ERR_NONE to indicate success.
- The size alignment check was moved from the FPGA Image Load framework
to the prepare() op.
- Added cancel_request boolean flag to struct m10bmc_sec.
- Moved the RSU cancellation logic from m10bmc_sec_cancel() to a new
rsu_cancel() function.
- The m10bmc_sec_cancel() function ONLY sets the cancel_request flag.
The cancel_request flag is checked at the beginning of the
m10bmc_sec_write() and m10bmc_sec_poll_complete() functions.
- Adapt to changed prototypes for the prepare() and write() ops. The
m10bmc_sec_write_blk() function has been renamed to
m10bmc_sec_write().
- Created a cleanup() op, m10bmc_sec_cleanup(), to attempt to cancel an
ongoing op during when exiting the update process.
v15:
- Adapted to changes in the FPGA Image Load framework:
(1) All enum types (progress and errors) are now type u32
(2) m10bmc_sec_write_blk() adds *blk_size and max_size parameters
and uses *blk_size as provided by the caller.
(3) m10bmc_sec_poll_complete() no long checks the driver_unload
flag.
v14:
- Changed symbol names to reflect the renaming of the Security Manager
Class driver to FPGA Image Load.
v13:
- No change
v12:
- Updated Date and KernelVersion fields in ABI documentation
- Removed size parameter from the write_blk() op. m10bmc_sec_write_blk()
no longer has a size parameter, and the block size is determined
in this (the lower-level) driver.
v11:
- No change
v10:
- No change
v9:
- No change
v8:
- Previously patch 5/6, otherwise no change
v7:
- No change
v6:
- Changed (size / stride) calculation to ((size + stride - 1) / stride)
to ensure that the proper count is passed to regmap_bulk_write().
- Removed unnecessary call to rsu_check_complete() in
m10bmc_sec_poll_complete() and changed while loop to
do/while loop.
v5:
- No change
v4:
- No change
v3:
- Changed: iops -> sops, imgr -> smgr, IFPGA_ -> FPGA_, ifpga_ to fpga_
- Changed "MAX10 BMC Secure Engine driver" to "MAX10 BMC Secure Update
driver"
- Removed wrapper functions (m10bmc_raw_*, m10bmc_sys_*). The
underlying functions are now called directly.
- Changed calling functions of functions that return "enum fpga_sec_err"
to check for (ret != FPGA_SEC_ERR_NONE) instead of (ret)
v2:
- Reworked the rsu_start_done() function to make it more readable
- Reworked while-loop condition/content in rsu_prog_ready()
- Minor code cleanup per review comments
- Added a comment to the m10bmc_sec_poll_complete() function to
explain the context (could take 30+ minutes to complete).
- Added m10bmc_ prefix to functions in m10bmc_iops structure
- Moved MAX10 BMC address and function definitions to a separate
patch.
---
drivers/fpga/intel-m10-bmc-sec-update.c | 377 ++++++++++++++++++++++++
1 file changed, 377 insertions(+)
diff --git a/drivers/fpga/intel-m10-bmc-sec-update.c b/drivers/fpga/intel-m10-bmc-sec-update.c
index 6c39adc6492d..a4dc5f47e398 100644
--- a/drivers/fpga/intel-m10-bmc-sec-update.c
+++ b/drivers/fpga/intel-m10-bmc-sec-update.c
@@ -17,8 +17,14 @@
struct m10bmc_sec {
struct device *dev;
struct intel_m10bmc *m10bmc;
+ struct fw_upload *fwl;
+ char *fw_name;
+ u32 fw_name_id;
+ bool cancel_request;
};
+static DEFINE_XARRAY_ALLOC(fw_upload_xa);
+
/* Root Entry Hash (REH) support */
#define REH_SHA256_SIZE 32
#define REH_SHA384_SIZE 48
@@ -179,9 +185,349 @@ static const struct attribute_group *m10bmc_sec_attr_groups[] = {
NULL,
};
+static void log_error_regs(struct m10bmc_sec *sec, u32 doorbell)
+{
+ u32 auth_result;
+
+ dev_err(sec->dev, "RSU error status: 0x%08x\n", doorbell);
+
+ if (!m10bmc_sys_read(sec->m10bmc, M10BMC_AUTH_RESULT, &auth_result))
+ dev_err(sec->dev, "RSU auth result: 0x%08x\n", auth_result);
+}
+
+static enum fw_upload_err rsu_check_idle(struct m10bmc_sec *sec)
+{
+ u32 doorbell;
+ int ret;
+
+ ret = m10bmc_sys_read(sec->m10bmc, M10BMC_DOORBELL, &doorbell);
+ if (ret)
+ return FW_UPLOAD_ERR_RW_ERROR;
+
+ if (rsu_prog(doorbell) != RSU_PROG_IDLE &&
+ rsu_prog(doorbell) != RSU_PROG_RSU_DONE) {
+ log_error_regs(sec, doorbell);
+ return FW_UPLOAD_ERR_BUSY;
+ }
+
+ return FW_UPLOAD_ERR_NONE;
+}
+
+static inline bool rsu_start_done(u32 doorbell)
+{
+ u32 status, progress;
+
+ if (doorbell & DRBL_RSU_REQUEST)
+ return false;
+
+ status = rsu_stat(doorbell);
+ if (status == RSU_STAT_ERASE_FAIL || status == RSU_STAT_WEAROUT)
+ return true;
+
+ progress = rsu_prog(doorbell);
+ if (progress != RSU_PROG_IDLE && progress != RSU_PROG_RSU_DONE)
+ return true;
+
+ return false;
+}
+
+static enum fw_upload_err rsu_update_init(struct m10bmc_sec *sec)
+{
+ u32 doorbell, status;
+ int ret;
+
+ ret = regmap_update_bits(sec->m10bmc->regmap,
+ M10BMC_SYS_BASE + M10BMC_DOORBELL,
+ DRBL_RSU_REQUEST | DRBL_HOST_STATUS,
+ DRBL_RSU_REQUEST |
+ FIELD_PREP(DRBL_HOST_STATUS,
+ HOST_STATUS_IDLE));
+ if (ret)
+ return FW_UPLOAD_ERR_RW_ERROR;
+
+ ret = regmap_read_poll_timeout(sec->m10bmc->regmap,
+ M10BMC_SYS_BASE + M10BMC_DOORBELL,
+ doorbell,
+ rsu_start_done(doorbell),
+ NIOS_HANDSHAKE_INTERVAL_US,
+ NIOS_HANDSHAKE_TIMEOUT_US);
+
+ if (ret == -ETIMEDOUT) {
+ log_error_regs(sec, doorbell);
+ return FW_UPLOAD_ERR_TIMEOUT;
+ } else if (ret) {
+ return FW_UPLOAD_ERR_RW_ERROR;
+ }
+
+ status = rsu_stat(doorbell);
+ if (status == RSU_STAT_WEAROUT) {
+ dev_warn(sec->dev, "Excessive flash update count detected\n");
+ return FW_UPLOAD_ERR_WEAROUT;
+ } else if (status == RSU_STAT_ERASE_FAIL) {
+ log_error_regs(sec, doorbell);
+ return FW_UPLOAD_ERR_HW_ERROR;
+ }
+
+ return FW_UPLOAD_ERR_NONE;
+}
+
+static enum fw_upload_err rsu_prog_ready(struct m10bmc_sec *sec)
+{
+ unsigned long poll_timeout;
+ u32 doorbell, progress;
+ int ret;
+
+ ret = m10bmc_sys_read(sec->m10bmc, M10BMC_DOORBELL, &doorbell);
+ if (ret)
+ return FW_UPLOAD_ERR_RW_ERROR;
+
+ poll_timeout = jiffies + msecs_to_jiffies(RSU_PREP_TIMEOUT_MS);
+ while (rsu_prog(doorbell) == RSU_PROG_PREPARE) {
+ msleep(RSU_PREP_INTERVAL_MS);
+ if (time_after(jiffies, poll_timeout))
+ break;
+
+ ret = m10bmc_sys_read(sec->m10bmc, M10BMC_DOORBELL, &doorbell);
+ if (ret)
+ return FW_UPLOAD_ERR_RW_ERROR;
+ }
+
+ progress = rsu_prog(doorbell);
+ if (progress == RSU_PROG_PREPARE) {
+ log_error_regs(sec, doorbell);
+ return FW_UPLOAD_ERR_TIMEOUT;
+ } else if (progress != RSU_PROG_READY) {
+ log_error_regs(sec, doorbell);
+ return FW_UPLOAD_ERR_HW_ERROR;
+ }
+
+ return FW_UPLOAD_ERR_NONE;
+}
+
+static enum fw_upload_err rsu_send_data(struct m10bmc_sec *sec)
+{
+ u32 doorbell;
+ int ret;
+
+ ret = regmap_update_bits(sec->m10bmc->regmap,
+ M10BMC_SYS_BASE + M10BMC_DOORBELL,
+ DRBL_HOST_STATUS,
+ FIELD_PREP(DRBL_HOST_STATUS,
+ HOST_STATUS_WRITE_DONE));
+ if (ret)
+ return FW_UPLOAD_ERR_RW_ERROR;
+
+ ret = regmap_read_poll_timeout(sec->m10bmc->regmap,
+ M10BMC_SYS_BASE + M10BMC_DOORBELL,
+ doorbell,
+ rsu_prog(doorbell) != RSU_PROG_READY,
+ NIOS_HANDSHAKE_INTERVAL_US,
+ NIOS_HANDSHAKE_TIMEOUT_US);
+
+ if (ret == -ETIMEDOUT) {
+ log_error_regs(sec, doorbell);
+ return FW_UPLOAD_ERR_TIMEOUT;
+ } else if (ret) {
+ return FW_UPLOAD_ERR_RW_ERROR;
+ }
+
+ switch (rsu_stat(doorbell)) {
+ case RSU_STAT_NORMAL:
+ case RSU_STAT_NIOS_OK:
+ case RSU_STAT_USER_OK:
+ case RSU_STAT_FACTORY_OK:
+ break;
+ default:
+ log_error_regs(sec, doorbell);
+ return FW_UPLOAD_ERR_HW_ERROR;
+ }
+
+ return FW_UPLOAD_ERR_NONE;
+}
+
+static int rsu_check_complete(struct m10bmc_sec *sec, u32 *doorbell)
+{
+ if (m10bmc_sys_read(sec->m10bmc, M10BMC_DOORBELL, doorbell))
+ return -EIO;
+
+ switch (rsu_stat(*doorbell)) {
+ case RSU_STAT_NORMAL:
+ case RSU_STAT_NIOS_OK:
+ case RSU_STAT_USER_OK:
+ case RSU_STAT_FACTORY_OK:
+ break;
+ default:
+ return -EINVAL;
+ }
+
+ switch (rsu_prog(*doorbell)) {
+ case RSU_PROG_IDLE:
+ case RSU_PROG_RSU_DONE:
+ return 0;
+ case RSU_PROG_AUTHENTICATING:
+ case RSU_PROG_COPYING:
+ case RSU_PROG_UPDATE_CANCEL:
+ case RSU_PROG_PROGRAM_KEY_HASH:
+ return -EAGAIN;
+ default:
+ return -EINVAL;
+ }
+}
+
+static enum fw_upload_err rsu_cancel(struct m10bmc_sec *sec)
+{
+ u32 doorbell;
+ int ret;
+
+ ret = m10bmc_sys_read(sec->m10bmc, M10BMC_DOORBELL, &doorbell);
+ if (ret)
+ return FW_UPLOAD_ERR_RW_ERROR;
+
+ if (rsu_prog(doorbell) != RSU_PROG_READY)
+ return FW_UPLOAD_ERR_BUSY;
+
+ ret = regmap_update_bits(sec->m10bmc->regmap,
+ M10BMC_SYS_BASE + M10BMC_DOORBELL,
+ DRBL_HOST_STATUS,
+ FIELD_PREP(DRBL_HOST_STATUS,
+ HOST_STATUS_ABORT_RSU));
+ if (ret)
+ return FW_UPLOAD_ERR_RW_ERROR;
+
+ return FW_UPLOAD_ERR_CANCELED;
+}
+
+static enum fw_upload_err m10bmc_sec_prepare(struct fw_upload *fwl,
+ const u8 *data, u32 size)
+{
+ struct m10bmc_sec *sec = fwl->dd_handle;
+ u32 ret;
+
+ sec->cancel_request = false;
+
+ if (!size || size & 0x3 || size > M10BMC_STAGING_SIZE)
+ return FW_UPLOAD_ERR_INVALID_SIZE;
+
+ ret = rsu_check_idle(sec);
+ if (ret != FW_UPLOAD_ERR_NONE)
+ return ret;
+
+ ret = rsu_update_init(sec);
+ if (ret != FW_UPLOAD_ERR_NONE)
+ return ret;
+
+ ret = rsu_prog_ready(sec);
+ if (ret != FW_UPLOAD_ERR_NONE)
+ return ret;
+
+ if (sec->cancel_request)
+ return rsu_cancel(sec);
+
+ return FW_UPLOAD_ERR_NONE;
+}
+
+#define WRITE_BLOCK_SIZE 0x4000 /* Default write-block size is 0x4000 bytes */
+
+static enum fw_upload_err m10bmc_sec_write(struct fw_upload *fwl, const u8 *data,
+ u32 offset, u32 size, u32 *written)
+{
+ struct m10bmc_sec *sec = fwl->dd_handle;
+ unsigned int stride = regmap_get_reg_stride(sec->m10bmc->regmap);
+ u32 blk_size, doorbell;
+ int ret;
+
+ if (sec->cancel_request)
+ return rsu_cancel(sec);
+
+ ret = m10bmc_sys_read(sec->m10bmc, M10BMC_DOORBELL, &doorbell);
+ if (ret) {
+ return FW_UPLOAD_ERR_RW_ERROR;
+ } else if (rsu_prog(doorbell) != RSU_PROG_READY) {
+ log_error_regs(sec, doorbell);
+ return FW_UPLOAD_ERR_HW_ERROR;
+ }
+
+ blk_size = min_t(u32, WRITE_BLOCK_SIZE, size);
+ ret = regmap_bulk_write(sec->m10bmc->regmap,
+ M10BMC_STAGING_BASE + offset,
+ (void *)data + offset,
+ (blk_size + stride - 1) / stride);
+
+ if (ret)
+ return FW_UPLOAD_ERR_RW_ERROR;
+
+ *written = blk_size;
+ return FW_UPLOAD_ERR_NONE;
+}
+
+static enum fw_upload_err m10bmc_sec_poll_complete(struct fw_upload *fwl)
+{
+ struct m10bmc_sec *sec = fwl->dd_handle;
+ unsigned long poll_timeout;
+ u32 doorbell, result;
+ int ret;
+
+ if (sec->cancel_request)
+ return rsu_cancel(sec);
+
+ result = rsu_send_data(sec);
+ if (result != FW_UPLOAD_ERR_NONE)
+ return result;
+
+ poll_timeout = jiffies + msecs_to_jiffies(RSU_COMPLETE_TIMEOUT_MS);
+ do {
+ msleep(RSU_COMPLETE_INTERVAL_MS);
+ ret = rsu_check_complete(sec, &doorbell);
+ } while (ret == -EAGAIN && !time_after(jiffies, poll_timeout));
+
+ if (ret == -EAGAIN) {
+ log_error_regs(sec, doorbell);
+ return FW_UPLOAD_ERR_TIMEOUT;
+ } else if (ret == -EIO) {
+ return FW_UPLOAD_ERR_RW_ERROR;
+ } else if (ret) {
+ log_error_regs(sec, doorbell);
+ return FW_UPLOAD_ERR_HW_ERROR;
+ }
+
+ return FW_UPLOAD_ERR_NONE;
+}
+
+/*
+ * m10bmc_sec_cancel() may be called asynchronously with an on-going update.
+ * All other functions are called sequentially in a single thread. To avoid
+ * contention on register accesses, m10bmc_sec_cancel() must only update
+ * the cancel_request flag. Other functions will check this flag and handle
+ * the cancel request synchronously.
+ */
+static void m10bmc_sec_cancel(struct fw_upload *fwl)
+{
+ struct m10bmc_sec *sec = fwl->dd_handle;
+
+ sec->cancel_request = true;
+}
+
+static void m10bmc_sec_cleanup(struct fw_upload *fwl)
+{
+ struct m10bmc_sec *sec = fwl->dd_handle;
+
+ (void)rsu_cancel(sec);
+}
+
+static const struct fw_upload_ops m10bmc_ops = {
+ .prepare = m10bmc_sec_prepare,
+ .write = m10bmc_sec_write,
+ .poll_complete = m10bmc_sec_poll_complete,
+ .cancel = m10bmc_sec_cancel,
+ .cleanup = m10bmc_sec_cleanup,
+};
+
#define SEC_UPDATE_LEN_MAX 32
static int m10bmc_sec_probe(struct platform_device *pdev)
{
+ char buf[SEC_UPDATE_LEN_MAX];
+ struct fw_upload *fwl;
+ unsigned int len, ret;
struct m10bmc_sec *sec;
sec = devm_kzalloc(&pdev->dev, sizeof(*sec), GFP_KERNEL);
@@ -192,6 +538,36 @@ static int m10bmc_sec_probe(struct platform_device *pdev)
sec->m10bmc = dev_get_drvdata(pdev->dev.parent);
dev_set_drvdata(&pdev->dev, sec);
+ ret = xa_alloc(&fw_upload_xa, &sec->fw_name_id, sec,
+ xa_limit_32b, GFP_KERNEL);
+ if (ret)
+ return ret;
+
+ len = scnprintf(buf, SEC_UPDATE_LEN_MAX, "secure-update%d",
+ sec->fw_name_id);
+ sec->fw_name = kmemdup_nul(buf, len, GFP_KERNEL);
+
+ fwl = firmware_upload_register(THIS_MODULE, sec->dev, sec->fw_name,
+ &m10bmc_ops, sec);
+ if (IS_ERR(fwl)) {
+ dev_err(sec->dev, "Firmware Upload driver failed to start\n");
+ kfree(sec->fw_name);
+ xa_erase(&fw_upload_xa, sec->fw_name_id);
+ return PTR_ERR(fwl);
+ }
+
+ sec->fwl = fwl;
+ return 0;
+}
+
+static int m10bmc_sec_remove(struct platform_device *pdev)
+{
+ struct m10bmc_sec *sec = dev_get_drvdata(&pdev->dev);
+
+ firmware_upload_unregister(sec->fwl);
+ kfree(sec->fw_name);
+ xa_erase(&fw_upload_xa, sec->fw_name_id);
+
return 0;
}
@@ -205,6 +581,7 @@ MODULE_DEVICE_TABLE(platform, intel_m10bmc_sec_ids);
static struct platform_driver intel_m10bmc_sec_driver = {
.probe = m10bmc_sec_probe,
+ .remove = m10bmc_sec_remove,
.driver = {
.name = "intel-m10bmc-sec-update",
.dev_groups = m10bmc_sec_attr_groups,
--
2.25.1
On Mon, May 16, 2022 at 04:49:40PM -0700, Russ Weight wrote:
> Extend the MAX10 BMC Secure Update driver to provide sysfs files to
> expose the 128 bit code signing key (CSK) cancellation vectors. These use
> the standard bitmap list format (e.g. 1,2-6,9).
>
> Each CSK is assigned an ID, a number between 0-127, during the signing
> process. CSK ID cancellation information is stored in 128-bit fields in
> write-once locations in flash. The cancellation of a CSK can be used
> to prevent the card from being rolled back to older images that were
> signed with a CSK that is now cancelled.
>
> Signed-off-by: Russ Weight <[email protected]>
> Reviewed-by: Tom Rix <[email protected]>
> ---
> v20:
> - Added text to the commit message to further describe the cancellation of
> code signing keys.
> v19:
> - Change "card bmc" naming back to "m10 bmc" naming to be consistent
> with the parent driver.
> v18:
> - No change
> v17:
> - Update the Date and KernelVersion for the ABI documentation to Jul 2022
> and 5.19 respectively.
> - Change "m10bmc" in symbol names to "cardbmc" to reflect the fact that the
> future devices will not necessarily use the MAX10.
> v16:
> - No Change
> v15:
> - Updated the Dates and KernelVersions in the ABI documentation
> v14:
> - No changes
> v13:
> - Updated ABI documentation date and kernel version
> v12:
> - Updated Date and KernelVersion fields in ABI documentation
> v11:
> - No change
> v10:
> - Changed the path expressions in the sysfs documentation to
> replace the n3000 reference with something more generic to
> accomodate other devices that use the same driver.
> v9:
> - Rebased to 5.12-rc2 next
> - Updated Date and KernelVersion in ABI documentation
> v8:
> - Previously patch 4/6, otherwise no change
> v7:
> - Updated Date and KernelVersion in ABI documentation
> v6:
> - Added WARN_ON() call for (size / stride) to ensure
> that the proper count is passed to regmap_bulk_read().
> v5:
> - No change
> v4:
> - Moved sysfs files for displaying the code-signing-key (CSK)
> cancellation vectors from the FPGA Security Manger class driver
> to here. The m10bmc_csk_vector() and m10bmc_csk_cancel_nbits()
> functions are removed and the functionality from these functions
> is moved into a show_canceled_csk() function for for displaying
> the CSK vectors.
> - Added ABI documentation for new sysfs entries
> v3:
> - Changed: iops -> sops, imgr -> smgr, IFPGA_ -> FPGA_, ifpga_ to fpga_
> - Changed "MAX10 BMC Secure Engine driver" to "MAX10 BMC Secure Update
> driver"
> - Removed wrapper functions (m10bmc_raw_*, m10bmc_sys_*). The
> underlying functions are now called directly.
> - Renamed get_csk_vector() to m10bmc_csk_vector()
> v2:
> - Replaced small function-creation macros for explicit function
> declarations.
> - Fixed get_csk_vector() function to properly apply the stride
> variable in calls to m10bmc_raw_bulk_read()
> - Added m10bmc_ prefix to functions in m10bmc_iops structure
> ---
> .../sysfs-driver-intel-m10-bmc-sec-update | 24 ++++++++++
> drivers/fpga/intel-m10-bmc-sec-update.c | 48 +++++++++++++++++++
> 2 files changed, 72 insertions(+)
>
> diff --git a/Documentation/ABI/testing/sysfs-driver-intel-m10-bmc-sec-update b/Documentation/ABI/testing/sysfs-driver-intel-m10-bmc-sec-update
> index 1132e39b2125..ca5a34c1c31f 100644
> --- a/Documentation/ABI/testing/sysfs-driver-intel-m10-bmc-sec-update
> +++ b/Documentation/ABI/testing/sysfs-driver-intel-m10-bmc-sec-update
> @@ -28,6 +28,30 @@ Description: Read only. Returns the root entry hash for the BMC image
> underlying device supports it.
> Format: string.
>
> +What: /sys/bus/platform/drivers/intel-m10bmc-sec-update/.../security/sr_canceled_csks
> +Date: Jul 2022
> +KernelVersion: 5.19
> +Contact: Russ Weight <[email protected]>
> +Description: Read only. Returns a list of indices for canceled code
> + signing keys for the static region. The standard bitmap
> + list format is used (e.g. "1,2-6,9").
> +
> +What: /sys/bus/platform/drivers/intel-m10bmc-sec-update/.../security/pr_canceled_csks
> +Date: Jul 2022
> +KernelVersion: 5.19
> +Contact: Russ Weight <[email protected]>
> +Description: Read only. Returns a list of indices for canceled code
> + signing keys for the partial reconfiguration region. The
> + standard bitmap list format is used (e.g. "1,2-6,9").
> +
> +What: /sys/bus/platform/drivers/intel-m10bmc-sec-update/.../security/bmc_canceled_csks
> +Date: Jul 2022
> +KernelVersion: 5.19
> +Contact: Russ Weight <[email protected]>
> +Description: Read only. Returns a list of indices for canceled code
> + signing keys for the BMC. The standard bitmap list format
> + is used (e.g. "1,2-6,9").
> +
> What: /sys/bus/platform/drivers/intel-m10bmc-sec-update/.../security/flash_count
> Date: Jul 2022
> KernelVersion: 5.19
> diff --git a/drivers/fpga/intel-m10-bmc-sec-update.c b/drivers/fpga/intel-m10-bmc-sec-update.c
> index 3f183202de3b..6c39adc6492d 100644
> --- a/drivers/fpga/intel-m10-bmc-sec-update.c
> +++ b/drivers/fpga/intel-m10-bmc-sec-update.c
> @@ -78,6 +78,51 @@ DEVICE_ATTR_SEC_REH_RO(bmc, BMC_PROG_MAGIC, BMC_PROG_ADDR, BMC_REH_ADDR);
> DEVICE_ATTR_SEC_REH_RO(sr, SR_PROG_MAGIC, SR_PROG_ADDR, SR_REH_ADDR);
> DEVICE_ATTR_SEC_REH_RO(pr, PR_PROG_MAGIC, PR_PROG_ADDR, PR_REH_ADDR);
>
> +#define CSK_BIT_LEN 128U
> +#define CSK_32ARRAY_SIZE DIV_ROUND_UP(CSK_BIT_LEN, 32)
> +
> +static ssize_t
> +show_canceled_csk(struct device *dev, u32 addr, char *buf)
> +{
> + unsigned int i, stride, size = CSK_32ARRAY_SIZE * sizeof(u32);
> + struct m10bmc_sec *sec = dev_get_drvdata(dev);
> + DECLARE_BITMAP(csk_map, CSK_BIT_LEN);
> + __le32 csk_le32[CSK_32ARRAY_SIZE];
> + u32 csk32[CSK_32ARRAY_SIZE];
> + int ret;
> +
> + stride = regmap_get_reg_stride(sec->m10bmc->regmap);
> +
> + WARN_ON(size % stride);
Same concern.
> + ret = regmap_bulk_read(sec->m10bmc->regmap, addr, csk_le32,
> + size / stride);
> + if (ret) {
> + dev_err(sec->dev, "failed to read CSK vector: %x cnt %x: %d\n",
> + addr, size / stride, ret);
> + return ret;
> + }
> +
> + for (i = 0; i < CSK_32ARRAY_SIZE; i++)
> + csk32[i] = le32_to_cpu(((csk_le32[i])));
> +
> + bitmap_from_arr32(csk_map, csk32, CSK_BIT_LEN);
> + bitmap_complement(csk_map, csk_map, CSK_BIT_LEN);
> + return bitmap_print_to_pagebuf(1, buf, csk_map, CSK_BIT_LEN);
> +}
> +
> +#define DEVICE_ATTR_SEC_CSK_RO(_name, _addr) \
> +static ssize_t _name##_canceled_csks_show(struct device *dev, \
> + struct device_attribute *attr, \
> + char *buf) \
> +{ return show_canceled_csk(dev, _addr, buf); } \
> +static DEVICE_ATTR_RO(_name##_canceled_csks)
> +
> +#define CSK_VEC_OFFSET 0x34
> +
> +DEVICE_ATTR_SEC_CSK_RO(bmc, BMC_PROG_ADDR + CSK_VEC_OFFSET);
> +DEVICE_ATTR_SEC_CSK_RO(sr, SR_PROG_ADDR + CSK_VEC_OFFSET);
> +DEVICE_ATTR_SEC_CSK_RO(pr, PR_PROG_ADDR + CSK_VEC_OFFSET);
> +
> #define FLASH_COUNT_SIZE 4096 /* count stored as inverted bit vector */
>
> static ssize_t flash_count_show(struct device *dev,
> @@ -118,6 +163,9 @@ static struct attribute *m10bmc_security_attrs[] = {
> &dev_attr_bmc_root_entry_hash.attr,
> &dev_attr_sr_root_entry_hash.attr,
> &dev_attr_pr_root_entry_hash.attr,
> + &dev_attr_sr_canceled_csks.attr,
> + &dev_attr_pr_canceled_csks.attr,
> + &dev_attr_bmc_canceled_csks.attr,
> NULL,
> };
>
> --
> 2.25.1
On Mon, May 16, 2022 at 04:49:41PM -0700, Russ Weight wrote:
> Create firmware upload ops and call the Firmware Upload support of the
> Firmware Loader subsystem to enable FPGA image uploads for secure
> updates of BMC images, FPGA images, etc.
>
> Signed-off-by: Russ Weight <[email protected]>
> ---
> v20:
> - No change.
> v19:
> - Change "card bmc" naming back to "m10 bmc" naming to be consistent
> with the parent driver.
> v18:
> - Moved the firmware_upload_register() function here from an earlier
> patch since this is where the required ops are provided.
> - Moved the bmc_sec_remove() function here from an earlier patch to
> unregister the firmware driver and do cleanup.
> v17:
> - Change "m10bmc" in symbol names to "cardbmc" to reflect the fact that the
> future devices will not necessarily use the MAX10.
> - Change from image_load class driver to the new firmware_upload
> functionality of the firmware_loader.
> - fw_upload_ops functions will return "enum fw_upload_err" data types
> instead of integer values.
> v16:
> - Use 0 instead of FPGA_IMAGE_ERR_NONE to indicate success.
> - The size alignment check was moved from the FPGA Image Load framework
> to the prepare() op.
> - Added cancel_request boolean flag to struct m10bmc_sec.
> - Moved the RSU cancellation logic from m10bmc_sec_cancel() to a new
> rsu_cancel() function.
> - The m10bmc_sec_cancel() function ONLY sets the cancel_request flag.
> The cancel_request flag is checked at the beginning of the
> m10bmc_sec_write() and m10bmc_sec_poll_complete() functions.
> - Adapt to changed prototypes for the prepare() and write() ops. The
> m10bmc_sec_write_blk() function has been renamed to
> m10bmc_sec_write().
> - Created a cleanup() op, m10bmc_sec_cleanup(), to attempt to cancel an
> ongoing op during when exiting the update process.
> v15:
> - Adapted to changes in the FPGA Image Load framework:
> (1) All enum types (progress and errors) are now type u32
> (2) m10bmc_sec_write_blk() adds *blk_size and max_size parameters
> and uses *blk_size as provided by the caller.
> (3) m10bmc_sec_poll_complete() no long checks the driver_unload
> flag.
> v14:
> - Changed symbol names to reflect the renaming of the Security Manager
> Class driver to FPGA Image Load.
> v13:
> - No change
> v12:
> - Updated Date and KernelVersion fields in ABI documentation
> - Removed size parameter from the write_blk() op. m10bmc_sec_write_blk()
> no longer has a size parameter, and the block size is determined
> in this (the lower-level) driver.
> v11:
> - No change
> v10:
> - No change
> v9:
> - No change
> v8:
> - Previously patch 5/6, otherwise no change
> v7:
> - No change
> v6:
> - Changed (size / stride) calculation to ((size + stride - 1) / stride)
> to ensure that the proper count is passed to regmap_bulk_write().
> - Removed unnecessary call to rsu_check_complete() in
> m10bmc_sec_poll_complete() and changed while loop to
> do/while loop.
> v5:
> - No change
> v4:
> - No change
> v3:
> - Changed: iops -> sops, imgr -> smgr, IFPGA_ -> FPGA_, ifpga_ to fpga_
> - Changed "MAX10 BMC Secure Engine driver" to "MAX10 BMC Secure Update
> driver"
> - Removed wrapper functions (m10bmc_raw_*, m10bmc_sys_*). The
> underlying functions are now called directly.
> - Changed calling functions of functions that return "enum fpga_sec_err"
> to check for (ret != FPGA_SEC_ERR_NONE) instead of (ret)
> v2:
> - Reworked the rsu_start_done() function to make it more readable
> - Reworked while-loop condition/content in rsu_prog_ready()
> - Minor code cleanup per review comments
> - Added a comment to the m10bmc_sec_poll_complete() function to
> explain the context (could take 30+ minutes to complete).
> - Added m10bmc_ prefix to functions in m10bmc_iops structure
> - Moved MAX10 BMC address and function definitions to a separate
> patch.
> ---
> drivers/fpga/intel-m10-bmc-sec-update.c | 377 ++++++++++++++++++++++++
> 1 file changed, 377 insertions(+)
>
> diff --git a/drivers/fpga/intel-m10-bmc-sec-update.c b/drivers/fpga/intel-m10-bmc-sec-update.c
> index 6c39adc6492d..a4dc5f47e398 100644
> --- a/drivers/fpga/intel-m10-bmc-sec-update.c
> +++ b/drivers/fpga/intel-m10-bmc-sec-update.c
> @@ -17,8 +17,14 @@
> struct m10bmc_sec {
> struct device *dev;
> struct intel_m10bmc *m10bmc;
> + struct fw_upload *fwl;
> + char *fw_name;
> + u32 fw_name_id;
> + bool cancel_request;
> };
>
> +static DEFINE_XARRAY_ALLOC(fw_upload_xa);
> +
> /* Root Entry Hash (REH) support */
> #define REH_SHA256_SIZE 32
> #define REH_SHA384_SIZE 48
> @@ -179,9 +185,349 @@ static const struct attribute_group *m10bmc_sec_attr_groups[] = {
> NULL,
> };
>
> +static void log_error_regs(struct m10bmc_sec *sec, u32 doorbell)
> +{
> + u32 auth_result;
> +
> + dev_err(sec->dev, "RSU error status: 0x%08x\n", doorbell);
> +
> + if (!m10bmc_sys_read(sec->m10bmc, M10BMC_AUTH_RESULT, &auth_result))
> + dev_err(sec->dev, "RSU auth result: 0x%08x\n", auth_result);
> +}
> +
> +static enum fw_upload_err rsu_check_idle(struct m10bmc_sec *sec)
> +{
> + u32 doorbell;
> + int ret;
> +
> + ret = m10bmc_sys_read(sec->m10bmc, M10BMC_DOORBELL, &doorbell);
> + if (ret)
> + return FW_UPLOAD_ERR_RW_ERROR;
> +
> + if (rsu_prog(doorbell) != RSU_PROG_IDLE &&
> + rsu_prog(doorbell) != RSU_PROG_RSU_DONE) {
> + log_error_regs(sec, doorbell);
> + return FW_UPLOAD_ERR_BUSY;
> + }
> +
> + return FW_UPLOAD_ERR_NONE;
> +}
> +
> +static inline bool rsu_start_done(u32 doorbell)
> +{
> + u32 status, progress;
> +
> + if (doorbell & DRBL_RSU_REQUEST)
> + return false;
> +
> + status = rsu_stat(doorbell);
> + if (status == RSU_STAT_ERASE_FAIL || status == RSU_STAT_WEAROUT)
> + return true;
> +
> + progress = rsu_prog(doorbell);
> + if (progress != RSU_PROG_IDLE && progress != RSU_PROG_RSU_DONE)
> + return true;
> +
> + return false;
> +}
> +
> +static enum fw_upload_err rsu_update_init(struct m10bmc_sec *sec)
> +{
> + u32 doorbell, status;
> + int ret;
> +
> + ret = regmap_update_bits(sec->m10bmc->regmap,
> + M10BMC_SYS_BASE + M10BMC_DOORBELL,
> + DRBL_RSU_REQUEST | DRBL_HOST_STATUS,
> + DRBL_RSU_REQUEST |
> + FIELD_PREP(DRBL_HOST_STATUS,
> + HOST_STATUS_IDLE));
> + if (ret)
> + return FW_UPLOAD_ERR_RW_ERROR;
> +
> + ret = regmap_read_poll_timeout(sec->m10bmc->regmap,
> + M10BMC_SYS_BASE + M10BMC_DOORBELL,
> + doorbell,
> + rsu_start_done(doorbell),
> + NIOS_HANDSHAKE_INTERVAL_US,
> + NIOS_HANDSHAKE_TIMEOUT_US);
> +
> + if (ret == -ETIMEDOUT) {
> + log_error_regs(sec, doorbell);
> + return FW_UPLOAD_ERR_TIMEOUT;
> + } else if (ret) {
> + return FW_UPLOAD_ERR_RW_ERROR;
> + }
> +
> + status = rsu_stat(doorbell);
> + if (status == RSU_STAT_WEAROUT) {
> + dev_warn(sec->dev, "Excessive flash update count detected\n");
> + return FW_UPLOAD_ERR_WEAROUT;
> + } else if (status == RSU_STAT_ERASE_FAIL) {
> + log_error_regs(sec, doorbell);
> + return FW_UPLOAD_ERR_HW_ERROR;
> + }
> +
> + return FW_UPLOAD_ERR_NONE;
> +}
> +
> +static enum fw_upload_err rsu_prog_ready(struct m10bmc_sec *sec)
> +{
> + unsigned long poll_timeout;
> + u32 doorbell, progress;
> + int ret;
> +
> + ret = m10bmc_sys_read(sec->m10bmc, M10BMC_DOORBELL, &doorbell);
> + if (ret)
> + return FW_UPLOAD_ERR_RW_ERROR;
> +
> + poll_timeout = jiffies + msecs_to_jiffies(RSU_PREP_TIMEOUT_MS);
> + while (rsu_prog(doorbell) == RSU_PROG_PREPARE) {
> + msleep(RSU_PREP_INTERVAL_MS);
> + if (time_after(jiffies, poll_timeout))
> + break;
> +
> + ret = m10bmc_sys_read(sec->m10bmc, M10BMC_DOORBELL, &doorbell);
> + if (ret)
> + return FW_UPLOAD_ERR_RW_ERROR;
> + }
> +
> + progress = rsu_prog(doorbell);
> + if (progress == RSU_PROG_PREPARE) {
> + log_error_regs(sec, doorbell);
> + return FW_UPLOAD_ERR_TIMEOUT;
> + } else if (progress != RSU_PROG_READY) {
> + log_error_regs(sec, doorbell);
> + return FW_UPLOAD_ERR_HW_ERROR;
> + }
> +
> + return FW_UPLOAD_ERR_NONE;
> +}
> +
> +static enum fw_upload_err rsu_send_data(struct m10bmc_sec *sec)
> +{
> + u32 doorbell;
> + int ret;
> +
> + ret = regmap_update_bits(sec->m10bmc->regmap,
> + M10BMC_SYS_BASE + M10BMC_DOORBELL,
> + DRBL_HOST_STATUS,
> + FIELD_PREP(DRBL_HOST_STATUS,
> + HOST_STATUS_WRITE_DONE));
> + if (ret)
> + return FW_UPLOAD_ERR_RW_ERROR;
> +
> + ret = regmap_read_poll_timeout(sec->m10bmc->regmap,
> + M10BMC_SYS_BASE + M10BMC_DOORBELL,
> + doorbell,
> + rsu_prog(doorbell) != RSU_PROG_READY,
> + NIOS_HANDSHAKE_INTERVAL_US,
> + NIOS_HANDSHAKE_TIMEOUT_US);
> +
> + if (ret == -ETIMEDOUT) {
> + log_error_regs(sec, doorbell);
> + return FW_UPLOAD_ERR_TIMEOUT;
> + } else if (ret) {
> + return FW_UPLOAD_ERR_RW_ERROR;
> + }
> +
> + switch (rsu_stat(doorbell)) {
> + case RSU_STAT_NORMAL:
> + case RSU_STAT_NIOS_OK:
> + case RSU_STAT_USER_OK:
> + case RSU_STAT_FACTORY_OK:
> + break;
> + default:
> + log_error_regs(sec, doorbell);
> + return FW_UPLOAD_ERR_HW_ERROR;
> + }
> +
> + return FW_UPLOAD_ERR_NONE;
> +}
> +
> +static int rsu_check_complete(struct m10bmc_sec *sec, u32 *doorbell)
> +{
> + if (m10bmc_sys_read(sec->m10bmc, M10BMC_DOORBELL, doorbell))
> + return -EIO;
> +
> + switch (rsu_stat(*doorbell)) {
> + case RSU_STAT_NORMAL:
> + case RSU_STAT_NIOS_OK:
> + case RSU_STAT_USER_OK:
> + case RSU_STAT_FACTORY_OK:
> + break;
> + default:
> + return -EINVAL;
> + }
> +
> + switch (rsu_prog(*doorbell)) {
> + case RSU_PROG_IDLE:
> + case RSU_PROG_RSU_DONE:
> + return 0;
> + case RSU_PROG_AUTHENTICATING:
> + case RSU_PROG_COPYING:
> + case RSU_PROG_UPDATE_CANCEL:
> + case RSU_PROG_PROGRAM_KEY_HASH:
> + return -EAGAIN;
> + default:
> + return -EINVAL;
> + }
> +}
> +
> +static enum fw_upload_err rsu_cancel(struct m10bmc_sec *sec)
> +{
> + u32 doorbell;
> + int ret;
> +
> + ret = m10bmc_sys_read(sec->m10bmc, M10BMC_DOORBELL, &doorbell);
> + if (ret)
> + return FW_UPLOAD_ERR_RW_ERROR;
> +
> + if (rsu_prog(doorbell) != RSU_PROG_READY)
> + return FW_UPLOAD_ERR_BUSY;
> +
> + ret = regmap_update_bits(sec->m10bmc->regmap,
> + M10BMC_SYS_BASE + M10BMC_DOORBELL,
> + DRBL_HOST_STATUS,
> + FIELD_PREP(DRBL_HOST_STATUS,
> + HOST_STATUS_ABORT_RSU));
> + if (ret)
> + return FW_UPLOAD_ERR_RW_ERROR;
> +
> + return FW_UPLOAD_ERR_CANCELED;
> +}
> +
> +static enum fw_upload_err m10bmc_sec_prepare(struct fw_upload *fwl,
> + const u8 *data, u32 size)
> +{
> + struct m10bmc_sec *sec = fwl->dd_handle;
> + u32 ret;
> +
> + sec->cancel_request = false;
> +
> + if (!size || size & 0x3 || size > M10BMC_STAGING_SIZE)
Why the size should be 4 bytes aligned? I assume this relates to max10
register stride. If so, I suggest we don't use instant number here.
> + return FW_UPLOAD_ERR_INVALID_SIZE;
> +
> + ret = rsu_check_idle(sec);
> + if (ret != FW_UPLOAD_ERR_NONE)
> + return ret;
> +
> + ret = rsu_update_init(sec);
> + if (ret != FW_UPLOAD_ERR_NONE)
> + return ret;
> +
> + ret = rsu_prog_ready(sec);
> + if (ret != FW_UPLOAD_ERR_NONE)
> + return ret;
> +
> + if (sec->cancel_request)
> + return rsu_cancel(sec);
> +
> + return FW_UPLOAD_ERR_NONE;
> +}
> +
> +#define WRITE_BLOCK_SIZE 0x4000 /* Default write-block size is 0x4000 bytes */
> +
> +static enum fw_upload_err m10bmc_sec_write(struct fw_upload *fwl, const u8 *data,
> + u32 offset, u32 size, u32 *written)
> +{
> + struct m10bmc_sec *sec = fwl->dd_handle;
> + unsigned int stride = regmap_get_reg_stride(sec->m10bmc->regmap);
reverse xmax tree please.
> + u32 blk_size, doorbell;
> + int ret;
> +
> + if (sec->cancel_request)
> + return rsu_cancel(sec);
> +
> + ret = m10bmc_sys_read(sec->m10bmc, M10BMC_DOORBELL, &doorbell);
> + if (ret) {
> + return FW_UPLOAD_ERR_RW_ERROR;
> + } else if (rsu_prog(doorbell) != RSU_PROG_READY) {
> + log_error_regs(sec, doorbell);
> + return FW_UPLOAD_ERR_HW_ERROR;
> + }
> +
> + blk_size = min_t(u32, WRITE_BLOCK_SIZE, size);
> + ret = regmap_bulk_write(sec->m10bmc->regmap,
> + M10BMC_STAGING_BASE + offset,
> + (void *)data + offset,
> + (blk_size + stride - 1) / stride);
You mean to DIV_ROUND_UP the write size here? So you may write more
bytes than required?
Also possible to access out of bounds for 'data', is it?
> +
> + if (ret)
> + return FW_UPLOAD_ERR_RW_ERROR;
> +
> + *written = blk_size;
> + return FW_UPLOAD_ERR_NONE;
> +}
> +
> +static enum fw_upload_err m10bmc_sec_poll_complete(struct fw_upload *fwl)
> +{
> + struct m10bmc_sec *sec = fwl->dd_handle;
> + unsigned long poll_timeout;
> + u32 doorbell, result;
> + int ret;
> +
> + if (sec->cancel_request)
> + return rsu_cancel(sec);
> +
> + result = rsu_send_data(sec);
> + if (result != FW_UPLOAD_ERR_NONE)
> + return result;
> +
> + poll_timeout = jiffies + msecs_to_jiffies(RSU_COMPLETE_TIMEOUT_MS);
> + do {
> + msleep(RSU_COMPLETE_INTERVAL_MS);
> + ret = rsu_check_complete(sec, &doorbell);
> + } while (ret == -EAGAIN && !time_after(jiffies, poll_timeout));
> +
> + if (ret == -EAGAIN) {
> + log_error_regs(sec, doorbell);
> + return FW_UPLOAD_ERR_TIMEOUT;
> + } else if (ret == -EIO) {
> + return FW_UPLOAD_ERR_RW_ERROR;
> + } else if (ret) {
> + log_error_regs(sec, doorbell);
> + return FW_UPLOAD_ERR_HW_ERROR;
> + }
> +
> + return FW_UPLOAD_ERR_NONE;
> +}
> +
> +/*
> + * m10bmc_sec_cancel() may be called asynchronously with an on-going update.
> + * All other functions are called sequentially in a single thread. To avoid
> + * contention on register accesses, m10bmc_sec_cancel() must only update
> + * the cancel_request flag. Other functions will check this flag and handle
> + * the cancel request synchronously.
> + */
> +static void m10bmc_sec_cancel(struct fw_upload *fwl)
> +{
> + struct m10bmc_sec *sec = fwl->dd_handle;
> +
> + sec->cancel_request = true;
Do we need some atomic_xx functions to ensure concurrent correctness?
> +}
> +
> +static void m10bmc_sec_cleanup(struct fw_upload *fwl)
> +{
> + struct m10bmc_sec *sec = fwl->dd_handle;
> +
> + (void)rsu_cancel(sec);
> +}
> +
> +static const struct fw_upload_ops m10bmc_ops = {
> + .prepare = m10bmc_sec_prepare,
> + .write = m10bmc_sec_write,
> + .poll_complete = m10bmc_sec_poll_complete,
> + .cancel = m10bmc_sec_cancel,
> + .cleanup = m10bmc_sec_cleanup,
> +};
> +
> #define SEC_UPDATE_LEN_MAX 32
> static int m10bmc_sec_probe(struct platform_device *pdev)
> {
> + char buf[SEC_UPDATE_LEN_MAX];
> + struct fw_upload *fwl;
> + unsigned int len, ret;
Why the ret is unsigned int?
> struct m10bmc_sec *sec;
>
> sec = devm_kzalloc(&pdev->dev, sizeof(*sec), GFP_KERNEL);
> @@ -192,6 +538,36 @@ static int m10bmc_sec_probe(struct platform_device *pdev)
> sec->m10bmc = dev_get_drvdata(pdev->dev.parent);
> dev_set_drvdata(&pdev->dev, sec);
>
> + ret = xa_alloc(&fw_upload_xa, &sec->fw_name_id, sec,
> + xa_limit_32b, GFP_KERNEL);
> + if (ret)
> + return ret;
> +
> + len = scnprintf(buf, SEC_UPDATE_LEN_MAX, "secure-update%d",
> + sec->fw_name_id);
> + sec->fw_name = kmemdup_nul(buf, len, GFP_KERNEL);
Should we check this memory allocation?
Thanks,
Yilun
> +
> + fwl = firmware_upload_register(THIS_MODULE, sec->dev, sec->fw_name,
> + &m10bmc_ops, sec);
> + if (IS_ERR(fwl)) {
> + dev_err(sec->dev, "Firmware Upload driver failed to start\n");
> + kfree(sec->fw_name);
> + xa_erase(&fw_upload_xa, sec->fw_name_id);
> + return PTR_ERR(fwl);
> + }
> +
> + sec->fwl = fwl;
> + return 0;
> +}
> +
> +static int m10bmc_sec_remove(struct platform_device *pdev)
> +{
> + struct m10bmc_sec *sec = dev_get_drvdata(&pdev->dev);
> +
> + firmware_upload_unregister(sec->fwl);
> + kfree(sec->fw_name);
> + xa_erase(&fw_upload_xa, sec->fw_name_id);
> +
> return 0;
> }
>
> @@ -205,6 +581,7 @@ MODULE_DEVICE_TABLE(platform, intel_m10bmc_sec_ids);
>
> static struct platform_driver intel_m10bmc_sec_driver = {
> .probe = m10bmc_sec_probe,
> + .remove = m10bmc_sec_remove,
> .driver = {
> .name = "intel-m10bmc-sec-update",
> .dev_groups = m10bmc_sec_attr_groups,
> --
> 2.25.1
On 5/16/22 21:24, Xu Yilun wrote:
> On Mon, May 16, 2022 at 04:49:40PM -0700, Russ Weight wrote:
>> Extend the MAX10 BMC Secure Update driver to provide sysfs files to
>> expose the 128 bit code signing key (CSK) cancellation vectors. These use
>> the standard bitmap list format (e.g. 1,2-6,9).
>>
>> Each CSK is assigned an ID, a number between 0-127, during the signing
>> process. CSK ID cancellation information is stored in 128-bit fields in
>> write-once locations in flash. The cancellation of a CSK can be used
>> to prevent the card from being rolled back to older images that were
>> signed with a CSK that is now cancelled.
>>
>> Signed-off-by: Russ Weight <[email protected]>
>> Reviewed-by: Tom Rix <[email protected]>
>> ---
>> v20:
>> - Added text to the commit message to further describe the cancellation of
>> code signing keys.
>> v19:
>> - Change "card bmc" naming back to "m10 bmc" naming to be consistent
>> with the parent driver.
>> v18:
>> - No change
>> v17:
>> - Update the Date and KernelVersion for the ABI documentation to Jul 2022
>> and 5.19 respectively.
>> - Change "m10bmc" in symbol names to "cardbmc" to reflect the fact that the
>> future devices will not necessarily use the MAX10.
>> v16:
>> - No Change
>> v15:
>> - Updated the Dates and KernelVersions in the ABI documentation
>> v14:
>> - No changes
>> v13:
>> - Updated ABI documentation date and kernel version
>> v12:
>> - Updated Date and KernelVersion fields in ABI documentation
>> v11:
>> - No change
>> v10:
>> - Changed the path expressions in the sysfs documentation to
>> replace the n3000 reference with something more generic to
>> accomodate other devices that use the same driver.
>> v9:
>> - Rebased to 5.12-rc2 next
>> - Updated Date and KernelVersion in ABI documentation
>> v8:
>> - Previously patch 4/6, otherwise no change
>> v7:
>> - Updated Date and KernelVersion in ABI documentation
>> v6:
>> - Added WARN_ON() call for (size / stride) to ensure
>> that the proper count is passed to regmap_bulk_read().
>> v5:
>> - No change
>> v4:
>> - Moved sysfs files for displaying the code-signing-key (CSK)
>> cancellation vectors from the FPGA Security Manger class driver
>> to here. The m10bmc_csk_vector() and m10bmc_csk_cancel_nbits()
>> functions are removed and the functionality from these functions
>> is moved into a show_canceled_csk() function for for displaying
>> the CSK vectors.
>> - Added ABI documentation for new sysfs entries
>> v3:
>> - Changed: iops -> sops, imgr -> smgr, IFPGA_ -> FPGA_, ifpga_ to fpga_
>> - Changed "MAX10 BMC Secure Engine driver" to "MAX10 BMC Secure Update
>> driver"
>> - Removed wrapper functions (m10bmc_raw_*, m10bmc_sys_*). The
>> underlying functions are now called directly.
>> - Renamed get_csk_vector() to m10bmc_csk_vector()
>> v2:
>> - Replaced small function-creation macros for explicit function
>> declarations.
>> - Fixed get_csk_vector() function to properly apply the stride
>> variable in calls to m10bmc_raw_bulk_read()
>> - Added m10bmc_ prefix to functions in m10bmc_iops structure
>> ---
>> .../sysfs-driver-intel-m10-bmc-sec-update | 24 ++++++++++
>> drivers/fpga/intel-m10-bmc-sec-update.c | 48 +++++++++++++++++++
>> 2 files changed, 72 insertions(+)
>>
>> diff --git a/Documentation/ABI/testing/sysfs-driver-intel-m10-bmc-sec-update b/Documentation/ABI/testing/sysfs-driver-intel-m10-bmc-sec-update
>> index 1132e39b2125..ca5a34c1c31f 100644
>> --- a/Documentation/ABI/testing/sysfs-driver-intel-m10-bmc-sec-update
>> +++ b/Documentation/ABI/testing/sysfs-driver-intel-m10-bmc-sec-update
>> @@ -28,6 +28,30 @@ Description: Read only. Returns the root entry hash for the BMC image
>> underlying device supports it.
>> Format: string.
>>
>> +What: /sys/bus/platform/drivers/intel-m10bmc-sec-update/.../security/sr_canceled_csks
>> +Date: Jul 2022
>> +KernelVersion: 5.19
>> +Contact: Russ Weight <[email protected]>
>> +Description: Read only. Returns a list of indices for canceled code
>> + signing keys for the static region. The standard bitmap
>> + list format is used (e.g. "1,2-6,9").
>> +
>> +What: /sys/bus/platform/drivers/intel-m10bmc-sec-update/.../security/pr_canceled_csks
>> +Date: Jul 2022
>> +KernelVersion: 5.19
>> +Contact: Russ Weight <[email protected]>
>> +Description: Read only. Returns a list of indices for canceled code
>> + signing keys for the partial reconfiguration region. The
>> + standard bitmap list format is used (e.g. "1,2-6,9").
>> +
>> +What: /sys/bus/platform/drivers/intel-m10bmc-sec-update/.../security/bmc_canceled_csks
>> +Date: Jul 2022
>> +KernelVersion: 5.19
>> +Contact: Russ Weight <[email protected]>
>> +Description: Read only. Returns a list of indices for canceled code
>> + signing keys for the BMC. The standard bitmap list format
>> + is used (e.g. "1,2-6,9").
>> +
>> What: /sys/bus/platform/drivers/intel-m10bmc-sec-update/.../security/flash_count
>> Date: Jul 2022
>> KernelVersion: 5.19
>> diff --git a/drivers/fpga/intel-m10-bmc-sec-update.c b/drivers/fpga/intel-m10-bmc-sec-update.c
>> index 3f183202de3b..6c39adc6492d 100644
>> --- a/drivers/fpga/intel-m10-bmc-sec-update.c
>> +++ b/drivers/fpga/intel-m10-bmc-sec-update.c
>> @@ -78,6 +78,51 @@ DEVICE_ATTR_SEC_REH_RO(bmc, BMC_PROG_MAGIC, BMC_PROG_ADDR, BMC_REH_ADDR);
>> DEVICE_ATTR_SEC_REH_RO(sr, SR_PROG_MAGIC, SR_PROG_ADDR, SR_REH_ADDR);
>> DEVICE_ATTR_SEC_REH_RO(pr, PR_PROG_MAGIC, PR_PROG_ADDR, PR_REH_ADDR);
>>
>> +#define CSK_BIT_LEN 128U
>> +#define CSK_32ARRAY_SIZE DIV_ROUND_UP(CSK_BIT_LEN, 32)
>> +
>> +static ssize_t
>> +show_canceled_csk(struct device *dev, u32 addr, char *buf)
>> +{
>> + unsigned int i, stride, size = CSK_32ARRAY_SIZE * sizeof(u32);
>> + struct m10bmc_sec *sec = dev_get_drvdata(dev);
>> + DECLARE_BITMAP(csk_map, CSK_BIT_LEN);
>> + __le32 csk_le32[CSK_32ARRAY_SIZE];
>> + u32 csk32[CSK_32ARRAY_SIZE];
>> + int ret;
>> +
>> + stride = regmap_get_reg_stride(sec->m10bmc->regmap);
>> +
>> + WARN_ON(size % stride);
> Same concern.
OK - I'll handle it as an error case.
Thanks,
- Russ
>
>> + ret = regmap_bulk_read(sec->m10bmc->regmap, addr, csk_le32,
>> + size / stride);
>> + if (ret) {
>> + dev_err(sec->dev, "failed to read CSK vector: %x cnt %x: %d\n",
>> + addr, size / stride, ret);
>> + return ret;
>> + }
>> +
>> + for (i = 0; i < CSK_32ARRAY_SIZE; i++)
>> + csk32[i] = le32_to_cpu(((csk_le32[i])));
>> +
>> + bitmap_from_arr32(csk_map, csk32, CSK_BIT_LEN);
>> + bitmap_complement(csk_map, csk_map, CSK_BIT_LEN);
>> + return bitmap_print_to_pagebuf(1, buf, csk_map, CSK_BIT_LEN);
>> +}
>> +
>> +#define DEVICE_ATTR_SEC_CSK_RO(_name, _addr) \
>> +static ssize_t _name##_canceled_csks_show(struct device *dev, \
>> + struct device_attribute *attr, \
>> + char *buf) \
>> +{ return show_canceled_csk(dev, _addr, buf); } \
>> +static DEVICE_ATTR_RO(_name##_canceled_csks)
>> +
>> +#define CSK_VEC_OFFSET 0x34
>> +
>> +DEVICE_ATTR_SEC_CSK_RO(bmc, BMC_PROG_ADDR + CSK_VEC_OFFSET);
>> +DEVICE_ATTR_SEC_CSK_RO(sr, SR_PROG_ADDR + CSK_VEC_OFFSET);
>> +DEVICE_ATTR_SEC_CSK_RO(pr, PR_PROG_ADDR + CSK_VEC_OFFSET);
>> +
>> #define FLASH_COUNT_SIZE 4096 /* count stored as inverted bit vector */
>>
>> static ssize_t flash_count_show(struct device *dev,
>> @@ -118,6 +163,9 @@ static struct attribute *m10bmc_security_attrs[] = {
>> &dev_attr_bmc_root_entry_hash.attr,
>> &dev_attr_sr_root_entry_hash.attr,
>> &dev_attr_pr_root_entry_hash.attr,
>> + &dev_attr_sr_canceled_csks.attr,
>> + &dev_attr_pr_canceled_csks.attr,
>> + &dev_attr_bmc_canceled_csks.attr,
>> NULL,
>> };
>>
>> --
>> 2.25.1
On 5/17/22 02:48, Xu Yilun wrote:
> On Mon, May 16, 2022 at 04:49:41PM -0700, Russ Weight wrote:
>> Create firmware upload ops and call the Firmware Upload support of the
>> Firmware Loader subsystem to enable FPGA image uploads for secure
>> updates of BMC images, FPGA images, etc.
>>
>> Signed-off-by: Russ Weight <[email protected]>
>> ---
>> v20:
>> - No change.
>> v19:
>> - Change "card bmc" naming back to "m10 bmc" naming to be consistent
>> with the parent driver.
>> v18:
>> - Moved the firmware_upload_register() function here from an earlier
>> patch since this is where the required ops are provided.
>> - Moved the bmc_sec_remove() function here from an earlier patch to
>> unregister the firmware driver and do cleanup.
>> v17:
>> - Change "m10bmc" in symbol names to "cardbmc" to reflect the fact that the
>> future devices will not necessarily use the MAX10.
>> - Change from image_load class driver to the new firmware_upload
>> functionality of the firmware_loader.
>> - fw_upload_ops functions will return "enum fw_upload_err" data types
>> instead of integer values.
>> v16:
>> - Use 0 instead of FPGA_IMAGE_ERR_NONE to indicate success.
>> - The size alignment check was moved from the FPGA Image Load framework
>> to the prepare() op.
>> - Added cancel_request boolean flag to struct m10bmc_sec.
>> - Moved the RSU cancellation logic from m10bmc_sec_cancel() to a new
>> rsu_cancel() function.
>> - The m10bmc_sec_cancel() function ONLY sets the cancel_request flag.
>> The cancel_request flag is checked at the beginning of the
>> m10bmc_sec_write() and m10bmc_sec_poll_complete() functions.
>> - Adapt to changed prototypes for the prepare() and write() ops. The
>> m10bmc_sec_write_blk() function has been renamed to
>> m10bmc_sec_write().
>> - Created a cleanup() op, m10bmc_sec_cleanup(), to attempt to cancel an
>> ongoing op during when exiting the update process.
>> v15:
>> - Adapted to changes in the FPGA Image Load framework:
>> (1) All enum types (progress and errors) are now type u32
>> (2) m10bmc_sec_write_blk() adds *blk_size and max_size parameters
>> and uses *blk_size as provided by the caller.
>> (3) m10bmc_sec_poll_complete() no long checks the driver_unload
>> flag.
>> v14:
>> - Changed symbol names to reflect the renaming of the Security Manager
>> Class driver to FPGA Image Load.
>> v13:
>> - No change
>> v12:
>> - Updated Date and KernelVersion fields in ABI documentation
>> - Removed size parameter from the write_blk() op. m10bmc_sec_write_blk()
>> no longer has a size parameter, and the block size is determined
>> in this (the lower-level) driver.
>> v11:
>> - No change
>> v10:
>> - No change
>> v9:
>> - No change
>> v8:
>> - Previously patch 5/6, otherwise no change
>> v7:
>> - No change
>> v6:
>> - Changed (size / stride) calculation to ((size + stride - 1) / stride)
>> to ensure that the proper count is passed to regmap_bulk_write().
>> - Removed unnecessary call to rsu_check_complete() in
>> m10bmc_sec_poll_complete() and changed while loop to
>> do/while loop.
>> v5:
>> - No change
>> v4:
>> - No change
>> v3:
>> - Changed: iops -> sops, imgr -> smgr, IFPGA_ -> FPGA_, ifpga_ to fpga_
>> - Changed "MAX10 BMC Secure Engine driver" to "MAX10 BMC Secure Update
>> driver"
>> - Removed wrapper functions (m10bmc_raw_*, m10bmc_sys_*). The
>> underlying functions are now called directly.
>> - Changed calling functions of functions that return "enum fpga_sec_err"
>> to check for (ret != FPGA_SEC_ERR_NONE) instead of (ret)
>> v2:
>> - Reworked the rsu_start_done() function to make it more readable
>> - Reworked while-loop condition/content in rsu_prog_ready()
>> - Minor code cleanup per review comments
>> - Added a comment to the m10bmc_sec_poll_complete() function to
>> explain the context (could take 30+ minutes to complete).
>> - Added m10bmc_ prefix to functions in m10bmc_iops structure
>> - Moved MAX10 BMC address and function definitions to a separate
>> patch.
>> ---
>> drivers/fpga/intel-m10-bmc-sec-update.c | 377 ++++++++++++++++++++++++
>> 1 file changed, 377 insertions(+)
>>
>> diff --git a/drivers/fpga/intel-m10-bmc-sec-update.c b/drivers/fpga/intel-m10-bmc-sec-update.c
>> index 6c39adc6492d..a4dc5f47e398 100644
>> --- a/drivers/fpga/intel-m10-bmc-sec-update.c
>> +++ b/drivers/fpga/intel-m10-bmc-sec-update.c
>> @@ -17,8 +17,14 @@
>> struct m10bmc_sec {
>> struct device *dev;
>> struct intel_m10bmc *m10bmc;
>> + struct fw_upload *fwl;
>> + char *fw_name;
>> + u32 fw_name_id;
>> + bool cancel_request;
>> };
>>
>> +static DEFINE_XARRAY_ALLOC(fw_upload_xa);
>> +
>> /* Root Entry Hash (REH) support */
>> #define REH_SHA256_SIZE 32
>> #define REH_SHA384_SIZE 48
>> @@ -179,9 +185,349 @@ static const struct attribute_group *m10bmc_sec_attr_groups[] = {
>> NULL,
>> };
>>
>> +static void log_error_regs(struct m10bmc_sec *sec, u32 doorbell)
>> +{
>> + u32 auth_result;
>> +
>> + dev_err(sec->dev, "RSU error status: 0x%08x\n", doorbell);
>> +
>> + if (!m10bmc_sys_read(sec->m10bmc, M10BMC_AUTH_RESULT, &auth_result))
>> + dev_err(sec->dev, "RSU auth result: 0x%08x\n", auth_result);
>> +}
>> +
>> +static enum fw_upload_err rsu_check_idle(struct m10bmc_sec *sec)
>> +{
>> + u32 doorbell;
>> + int ret;
>> +
>> + ret = m10bmc_sys_read(sec->m10bmc, M10BMC_DOORBELL, &doorbell);
>> + if (ret)
>> + return FW_UPLOAD_ERR_RW_ERROR;
>> +
>> + if (rsu_prog(doorbell) != RSU_PROG_IDLE &&
>> + rsu_prog(doorbell) != RSU_PROG_RSU_DONE) {
>> + log_error_regs(sec, doorbell);
>> + return FW_UPLOAD_ERR_BUSY;
>> + }
>> +
>> + return FW_UPLOAD_ERR_NONE;
>> +}
>> +
>> +static inline bool rsu_start_done(u32 doorbell)
>> +{
>> + u32 status, progress;
>> +
>> + if (doorbell & DRBL_RSU_REQUEST)
>> + return false;
>> +
>> + status = rsu_stat(doorbell);
>> + if (status == RSU_STAT_ERASE_FAIL || status == RSU_STAT_WEAROUT)
>> + return true;
>> +
>> + progress = rsu_prog(doorbell);
>> + if (progress != RSU_PROG_IDLE && progress != RSU_PROG_RSU_DONE)
>> + return true;
>> +
>> + return false;
>> +}
>> +
>> +static enum fw_upload_err rsu_update_init(struct m10bmc_sec *sec)
>> +{
>> + u32 doorbell, status;
>> + int ret;
>> +
>> + ret = regmap_update_bits(sec->m10bmc->regmap,
>> + M10BMC_SYS_BASE + M10BMC_DOORBELL,
>> + DRBL_RSU_REQUEST | DRBL_HOST_STATUS,
>> + DRBL_RSU_REQUEST |
>> + FIELD_PREP(DRBL_HOST_STATUS,
>> + HOST_STATUS_IDLE));
>> + if (ret)
>> + return FW_UPLOAD_ERR_RW_ERROR;
>> +
>> + ret = regmap_read_poll_timeout(sec->m10bmc->regmap,
>> + M10BMC_SYS_BASE + M10BMC_DOORBELL,
>> + doorbell,
>> + rsu_start_done(doorbell),
>> + NIOS_HANDSHAKE_INTERVAL_US,
>> + NIOS_HANDSHAKE_TIMEOUT_US);
>> +
>> + if (ret == -ETIMEDOUT) {
>> + log_error_regs(sec, doorbell);
>> + return FW_UPLOAD_ERR_TIMEOUT;
>> + } else if (ret) {
>> + return FW_UPLOAD_ERR_RW_ERROR;
>> + }
>> +
>> + status = rsu_stat(doorbell);
>> + if (status == RSU_STAT_WEAROUT) {
>> + dev_warn(sec->dev, "Excessive flash update count detected\n");
>> + return FW_UPLOAD_ERR_WEAROUT;
>> + } else if (status == RSU_STAT_ERASE_FAIL) {
>> + log_error_regs(sec, doorbell);
>> + return FW_UPLOAD_ERR_HW_ERROR;
>> + }
>> +
>> + return FW_UPLOAD_ERR_NONE;
>> +}
>> +
>> +static enum fw_upload_err rsu_prog_ready(struct m10bmc_sec *sec)
>> +{
>> + unsigned long poll_timeout;
>> + u32 doorbell, progress;
>> + int ret;
>> +
>> + ret = m10bmc_sys_read(sec->m10bmc, M10BMC_DOORBELL, &doorbell);
>> + if (ret)
>> + return FW_UPLOAD_ERR_RW_ERROR;
>> +
>> + poll_timeout = jiffies + msecs_to_jiffies(RSU_PREP_TIMEOUT_MS);
>> + while (rsu_prog(doorbell) == RSU_PROG_PREPARE) {
>> + msleep(RSU_PREP_INTERVAL_MS);
>> + if (time_after(jiffies, poll_timeout))
>> + break;
>> +
>> + ret = m10bmc_sys_read(sec->m10bmc, M10BMC_DOORBELL, &doorbell);
>> + if (ret)
>> + return FW_UPLOAD_ERR_RW_ERROR;
>> + }
>> +
>> + progress = rsu_prog(doorbell);
>> + if (progress == RSU_PROG_PREPARE) {
>> + log_error_regs(sec, doorbell);
>> + return FW_UPLOAD_ERR_TIMEOUT;
>> + } else if (progress != RSU_PROG_READY) {
>> + log_error_regs(sec, doorbell);
>> + return FW_UPLOAD_ERR_HW_ERROR;
>> + }
>> +
>> + return FW_UPLOAD_ERR_NONE;
>> +}
>> +
>> +static enum fw_upload_err rsu_send_data(struct m10bmc_sec *sec)
>> +{
>> + u32 doorbell;
>> + int ret;
>> +
>> + ret = regmap_update_bits(sec->m10bmc->regmap,
>> + M10BMC_SYS_BASE + M10BMC_DOORBELL,
>> + DRBL_HOST_STATUS,
>> + FIELD_PREP(DRBL_HOST_STATUS,
>> + HOST_STATUS_WRITE_DONE));
>> + if (ret)
>> + return FW_UPLOAD_ERR_RW_ERROR;
>> +
>> + ret = regmap_read_poll_timeout(sec->m10bmc->regmap,
>> + M10BMC_SYS_BASE + M10BMC_DOORBELL,
>> + doorbell,
>> + rsu_prog(doorbell) != RSU_PROG_READY,
>> + NIOS_HANDSHAKE_INTERVAL_US,
>> + NIOS_HANDSHAKE_TIMEOUT_US);
>> +
>> + if (ret == -ETIMEDOUT) {
>> + log_error_regs(sec, doorbell);
>> + return FW_UPLOAD_ERR_TIMEOUT;
>> + } else if (ret) {
>> + return FW_UPLOAD_ERR_RW_ERROR;
>> + }
>> +
>> + switch (rsu_stat(doorbell)) {
>> + case RSU_STAT_NORMAL:
>> + case RSU_STAT_NIOS_OK:
>> + case RSU_STAT_USER_OK:
>> + case RSU_STAT_FACTORY_OK:
>> + break;
>> + default:
>> + log_error_regs(sec, doorbell);
>> + return FW_UPLOAD_ERR_HW_ERROR;
>> + }
>> +
>> + return FW_UPLOAD_ERR_NONE;
>> +}
>> +
>> +static int rsu_check_complete(struct m10bmc_sec *sec, u32 *doorbell)
>> +{
>> + if (m10bmc_sys_read(sec->m10bmc, M10BMC_DOORBELL, doorbell))
>> + return -EIO;
>> +
>> + switch (rsu_stat(*doorbell)) {
>> + case RSU_STAT_NORMAL:
>> + case RSU_STAT_NIOS_OK:
>> + case RSU_STAT_USER_OK:
>> + case RSU_STAT_FACTORY_OK:
>> + break;
>> + default:
>> + return -EINVAL;
>> + }
>> +
>> + switch (rsu_prog(*doorbell)) {
>> + case RSU_PROG_IDLE:
>> + case RSU_PROG_RSU_DONE:
>> + return 0;
>> + case RSU_PROG_AUTHENTICATING:
>> + case RSU_PROG_COPYING:
>> + case RSU_PROG_UPDATE_CANCEL:
>> + case RSU_PROG_PROGRAM_KEY_HASH:
>> + return -EAGAIN;
>> + default:
>> + return -EINVAL;
>> + }
>> +}
>> +
>> +static enum fw_upload_err rsu_cancel(struct m10bmc_sec *sec)
>> +{
>> + u32 doorbell;
>> + int ret;
>> +
>> + ret = m10bmc_sys_read(sec->m10bmc, M10BMC_DOORBELL, &doorbell);
>> + if (ret)
>> + return FW_UPLOAD_ERR_RW_ERROR;
>> +
>> + if (rsu_prog(doorbell) != RSU_PROG_READY)
>> + return FW_UPLOAD_ERR_BUSY;
>> +
>> + ret = regmap_update_bits(sec->m10bmc->regmap,
>> + M10BMC_SYS_BASE + M10BMC_DOORBELL,
>> + DRBL_HOST_STATUS,
>> + FIELD_PREP(DRBL_HOST_STATUS,
>> + HOST_STATUS_ABORT_RSU));
>> + if (ret)
>> + return FW_UPLOAD_ERR_RW_ERROR;
>> +
>> + return FW_UPLOAD_ERR_CANCELED;
>> +}
>> +
>> +static enum fw_upload_err m10bmc_sec_prepare(struct fw_upload *fwl,
>> + const u8 *data, u32 size)
>> +{
>> + struct m10bmc_sec *sec = fwl->dd_handle;
>> + u32 ret;
>> +
>> + sec->cancel_request = false;
>> +
>> + if (!size || size & 0x3 || size > M10BMC_STAGING_SIZE)
> Why the size should be 4 bytes aligned? I assume this relates to max10
> register stride. If so, I suggest we don't use instant number here.
This check was not related to stride. All of the image files for these
devices are 4-byte aligned. I suppose there is no reason to enforce this
as the BMC will detect if a secure image is invalid. I can remove this
check.
>> + return FW_UPLOAD_ERR_INVALID_SIZE;
>> +
>> + ret = rsu_check_idle(sec);
>> + if (ret != FW_UPLOAD_ERR_NONE)
>> + return ret;
>> +
>> + ret = rsu_update_init(sec);
>> + if (ret != FW_UPLOAD_ERR_NONE)
>> + return ret;
>> +
>> + ret = rsu_prog_ready(sec);
>> + if (ret != FW_UPLOAD_ERR_NONE)
>> + return ret;
>> +
>> + if (sec->cancel_request)
>> + return rsu_cancel(sec);
>> +
>> + return FW_UPLOAD_ERR_NONE;
>> +}
>> +
>> +#define WRITE_BLOCK_SIZE 0x4000 /* Default write-block size is 0x4000 bytes */
>> +
>> +static enum fw_upload_err m10bmc_sec_write(struct fw_upload *fwl, const u8 *data,
>> + u32 offset, u32 size, u32 *written)
>> +{
>> + struct m10bmc_sec *sec = fwl->dd_handle;
>> + unsigned int stride = regmap_get_reg_stride(sec->m10bmc->regmap);
> reverse xmax tree please.
OK - I'll fix it.
>
>> + u32 blk_size, doorbell;
>> + int ret;
>> +
>> + if (sec->cancel_request)
>> + return rsu_cancel(sec);
>> +
>> + ret = m10bmc_sys_read(sec->m10bmc, M10BMC_DOORBELL, &doorbell);
>> + if (ret) {
>> + return FW_UPLOAD_ERR_RW_ERROR;
>> + } else if (rsu_prog(doorbell) != RSU_PROG_READY) {
>> + log_error_regs(sec, doorbell);
>> + return FW_UPLOAD_ERR_HW_ERROR;
>> + }
>> +
>> + blk_size = min_t(u32, WRITE_BLOCK_SIZE, size);
>> + ret = regmap_bulk_write(sec->m10bmc->regmap,
>> + M10BMC_STAGING_BASE + offset,
>> + (void *)data + offset,
>> + (blk_size + stride - 1) / stride);
> You mean to DIV_ROUND_UP the write size here? So you may write more
> bytes than required?
>
> Also possible to access out of bounds for 'data', is it?
All of the writes, except possibly the last write, will be of size
WRITE_BLOCK_SIZE, which is page aligned. Only the last write could
potentially round up on the stride. We have already verified that
the data written will fit in the staging area (in the prepare function),
so additional bytes to complete the stride will not exceed the staging
area which is the target for these writes.
>> +
>> + if (ret)
>> + return FW_UPLOAD_ERR_RW_ERROR;
>> +
>> + *written = blk_size;
>> + return FW_UPLOAD_ERR_NONE;
>> +}
>> +
>> +static enum fw_upload_err m10bmc_sec_poll_complete(struct fw_upload *fwl)
>> +{
>> + struct m10bmc_sec *sec = fwl->dd_handle;
>> + unsigned long poll_timeout;
>> + u32 doorbell, result;
>> + int ret;
>> +
>> + if (sec->cancel_request)
>> + return rsu_cancel(sec);
>> +
>> + result = rsu_send_data(sec);
>> + if (result != FW_UPLOAD_ERR_NONE)
>> + return result;
>> +
>> + poll_timeout = jiffies + msecs_to_jiffies(RSU_COMPLETE_TIMEOUT_MS);
>> + do {
>> + msleep(RSU_COMPLETE_INTERVAL_MS);
>> + ret = rsu_check_complete(sec, &doorbell);
>> + } while (ret == -EAGAIN && !time_after(jiffies, poll_timeout));
>> +
>> + if (ret == -EAGAIN) {
>> + log_error_regs(sec, doorbell);
>> + return FW_UPLOAD_ERR_TIMEOUT;
>> + } else if (ret == -EIO) {
>> + return FW_UPLOAD_ERR_RW_ERROR;
>> + } else if (ret) {
>> + log_error_regs(sec, doorbell);
>> + return FW_UPLOAD_ERR_HW_ERROR;
>> + }
>> +
>> + return FW_UPLOAD_ERR_NONE;
>> +}
>> +
>> +/*
>> + * m10bmc_sec_cancel() may be called asynchronously with an on-going update.
>> + * All other functions are called sequentially in a single thread. To avoid
>> + * contention on register accesses, m10bmc_sec_cancel() must only update
>> + * the cancel_request flag. Other functions will check this flag and handle
>> + * the cancel request synchronously.
>> + */
>> +static void m10bmc_sec_cancel(struct fw_upload *fwl)
>> +{
>> + struct m10bmc_sec *sec = fwl->dd_handle;
>> +
>> + sec->cancel_request = true;
> Do we need some atomic_xx functions to ensure concurrent correctness?
In the context of an update, this is once-written and only if the secure update
is cancelled. We are only checking true vs false so I don't think atomicity
is an issue. Do you recommend that I use atomic_read() and atomic_write() for
the cancel_request flag?
>
>> +}
>> +
>> +static void m10bmc_sec_cleanup(struct fw_upload *fwl)
>> +{
>> + struct m10bmc_sec *sec = fwl->dd_handle;
>> +
>> + (void)rsu_cancel(sec);
>> +}
>> +
>> +static const struct fw_upload_ops m10bmc_ops = {
>> + .prepare = m10bmc_sec_prepare,
>> + .write = m10bmc_sec_write,
>> + .poll_complete = m10bmc_sec_poll_complete,
>> + .cancel = m10bmc_sec_cancel,
>> + .cleanup = m10bmc_sec_cleanup,
>> +};
>> +
>> #define SEC_UPDATE_LEN_MAX 32
>> static int m10bmc_sec_probe(struct platform_device *pdev)
>> {
>> + char buf[SEC_UPDATE_LEN_MAX];
>> + struct fw_upload *fwl;
>> + unsigned int len, ret;
> Why the ret is unsigned int?
I will change it to int.
>
>> struct m10bmc_sec *sec;
>>
>> sec = devm_kzalloc(&pdev->dev, sizeof(*sec), GFP_KERNEL);
>> @@ -192,6 +538,36 @@ static int m10bmc_sec_probe(struct platform_device *pdev)
>> sec->m10bmc = dev_get_drvdata(pdev->dev.parent);
>> dev_set_drvdata(&pdev->dev, sec);
>>
>> + ret = xa_alloc(&fw_upload_xa, &sec->fw_name_id, sec,
>> + xa_limit_32b, GFP_KERNEL);
>> + if (ret)
>> + return ret;
>> +
>> + len = scnprintf(buf, SEC_UPDATE_LEN_MAX, "secure-update%d",
>> + sec->fw_name_id);
>> + sec->fw_name = kmemdup_nul(buf, len, GFP_KERNEL);
> Should we check this memory allocation?
Yes! - I'll add a check.
Thanks for the review.
- Russ
>
> Thanks,
> Yilun
>
>> +
>> + fwl = firmware_upload_register(THIS_MODULE, sec->dev, sec->fw_name,
>> + &m10bmc_ops, sec);
>> + if (IS_ERR(fwl)) {
>> + dev_err(sec->dev, "Firmware Upload driver failed to start\n");
>> + kfree(sec->fw_name);
>> + xa_erase(&fw_upload_xa, sec->fw_name_id);
>> + return PTR_ERR(fwl);
>> + }
>> +
>> + sec->fwl = fwl;
>> + return 0;
>> +}
>> +
>> +static int m10bmc_sec_remove(struct platform_device *pdev)
>> +{
>> + struct m10bmc_sec *sec = dev_get_drvdata(&pdev->dev);
>> +
>> + firmware_upload_unregister(sec->fwl);
>> + kfree(sec->fw_name);
>> + xa_erase(&fw_upload_xa, sec->fw_name_id);
>> +
>> return 0;
>> }
>>
>> @@ -205,6 +581,7 @@ MODULE_DEVICE_TABLE(platform, intel_m10bmc_sec_ids);
>>
>> static struct platform_driver intel_m10bmc_sec_driver = {
>> .probe = m10bmc_sec_probe,
>> + .remove = m10bmc_sec_remove,
>> .driver = {
>> .name = "intel-m10bmc-sec-update",
>> .dev_groups = m10bmc_sec_attr_groups,
>> --
>> 2.25.1
On Tue, May 17, 2022 at 01:16:42PM -0700, Russ Weight wrote:
>
>
> On 5/17/22 02:48, Xu Yilun wrote:
> > On Mon, May 16, 2022 at 04:49:41PM -0700, Russ Weight wrote:
> >> Create firmware upload ops and call the Firmware Upload support of the
> >> Firmware Loader subsystem to enable FPGA image uploads for secure
> >> updates of BMC images, FPGA images, etc.
> >>
> >> Signed-off-by: Russ Weight <[email protected]>
> >> ---
> >> v20:
> >> - No change.
> >> v19:
> >> - Change "card bmc" naming back to "m10 bmc" naming to be consistent
> >> with the parent driver.
> >> v18:
> >> - Moved the firmware_upload_register() function here from an earlier
> >> patch since this is where the required ops are provided.
> >> - Moved the bmc_sec_remove() function here from an earlier patch to
> >> unregister the firmware driver and do cleanup.
> >> v17:
> >> - Change "m10bmc" in symbol names to "cardbmc" to reflect the fact that the
> >> future devices will not necessarily use the MAX10.
> >> - Change from image_load class driver to the new firmware_upload
> >> functionality of the firmware_loader.
> >> - fw_upload_ops functions will return "enum fw_upload_err" data types
> >> instead of integer values.
> >> v16:
> >> - Use 0 instead of FPGA_IMAGE_ERR_NONE to indicate success.
> >> - The size alignment check was moved from the FPGA Image Load framework
> >> to the prepare() op.
> >> - Added cancel_request boolean flag to struct m10bmc_sec.
> >> - Moved the RSU cancellation logic from m10bmc_sec_cancel() to a new
> >> rsu_cancel() function.
> >> - The m10bmc_sec_cancel() function ONLY sets the cancel_request flag.
> >> The cancel_request flag is checked at the beginning of the
> >> m10bmc_sec_write() and m10bmc_sec_poll_complete() functions.
> >> - Adapt to changed prototypes for the prepare() and write() ops. The
> >> m10bmc_sec_write_blk() function has been renamed to
> >> m10bmc_sec_write().
> >> - Created a cleanup() op, m10bmc_sec_cleanup(), to attempt to cancel an
> >> ongoing op during when exiting the update process.
> >> v15:
> >> - Adapted to changes in the FPGA Image Load framework:
> >> (1) All enum types (progress and errors) are now type u32
> >> (2) m10bmc_sec_write_blk() adds *blk_size and max_size parameters
> >> and uses *blk_size as provided by the caller.
> >> (3) m10bmc_sec_poll_complete() no long checks the driver_unload
> >> flag.
> >> v14:
> >> - Changed symbol names to reflect the renaming of the Security Manager
> >> Class driver to FPGA Image Load.
> >> v13:
> >> - No change
> >> v12:
> >> - Updated Date and KernelVersion fields in ABI documentation
> >> - Removed size parameter from the write_blk() op. m10bmc_sec_write_blk()
> >> no longer has a size parameter, and the block size is determined
> >> in this (the lower-level) driver.
> >> v11:
> >> - No change
> >> v10:
> >> - No change
> >> v9:
> >> - No change
> >> v8:
> >> - Previously patch 5/6, otherwise no change
> >> v7:
> >> - No change
> >> v6:
> >> - Changed (size / stride) calculation to ((size + stride - 1) / stride)
> >> to ensure that the proper count is passed to regmap_bulk_write().
> >> - Removed unnecessary call to rsu_check_complete() in
> >> m10bmc_sec_poll_complete() and changed while loop to
> >> do/while loop.
> >> v5:
> >> - No change
> >> v4:
> >> - No change
> >> v3:
> >> - Changed: iops -> sops, imgr -> smgr, IFPGA_ -> FPGA_, ifpga_ to fpga_
> >> - Changed "MAX10 BMC Secure Engine driver" to "MAX10 BMC Secure Update
> >> driver"
> >> - Removed wrapper functions (m10bmc_raw_*, m10bmc_sys_*). The
> >> underlying functions are now called directly.
> >> - Changed calling functions of functions that return "enum fpga_sec_err"
> >> to check for (ret != FPGA_SEC_ERR_NONE) instead of (ret)
> >> v2:
> >> - Reworked the rsu_start_done() function to make it more readable
> >> - Reworked while-loop condition/content in rsu_prog_ready()
> >> - Minor code cleanup per review comments
> >> - Added a comment to the m10bmc_sec_poll_complete() function to
> >> explain the context (could take 30+ minutes to complete).
> >> - Added m10bmc_ prefix to functions in m10bmc_iops structure
> >> - Moved MAX10 BMC address and function definitions to a separate
> >> patch.
> >> ---
> >> drivers/fpga/intel-m10-bmc-sec-update.c | 377 ++++++++++++++++++++++++
> >> 1 file changed, 377 insertions(+)
> >>
> >> diff --git a/drivers/fpga/intel-m10-bmc-sec-update.c b/drivers/fpga/intel-m10-bmc-sec-update.c
> >> index 6c39adc6492d..a4dc5f47e398 100644
> >> --- a/drivers/fpga/intel-m10-bmc-sec-update.c
> >> +++ b/drivers/fpga/intel-m10-bmc-sec-update.c
> >> @@ -17,8 +17,14 @@
> >> struct m10bmc_sec {
> >> struct device *dev;
> >> struct intel_m10bmc *m10bmc;
> >> + struct fw_upload *fwl;
> >> + char *fw_name;
> >> + u32 fw_name_id;
> >> + bool cancel_request;
> >> };
> >>
> >> +static DEFINE_XARRAY_ALLOC(fw_upload_xa);
> >> +
> >> /* Root Entry Hash (REH) support */
> >> #define REH_SHA256_SIZE 32
> >> #define REH_SHA384_SIZE 48
> >> @@ -179,9 +185,349 @@ static const struct attribute_group *m10bmc_sec_attr_groups[] = {
> >> NULL,
> >> };
> >>
> >> +static void log_error_regs(struct m10bmc_sec *sec, u32 doorbell)
> >> +{
> >> + u32 auth_result;
> >> +
> >> + dev_err(sec->dev, "RSU error status: 0x%08x\n", doorbell);
> >> +
> >> + if (!m10bmc_sys_read(sec->m10bmc, M10BMC_AUTH_RESULT, &auth_result))
> >> + dev_err(sec->dev, "RSU auth result: 0x%08x\n", auth_result);
> >> +}
> >> +
> >> +static enum fw_upload_err rsu_check_idle(struct m10bmc_sec *sec)
> >> +{
> >> + u32 doorbell;
> >> + int ret;
> >> +
> >> + ret = m10bmc_sys_read(sec->m10bmc, M10BMC_DOORBELL, &doorbell);
> >> + if (ret)
> >> + return FW_UPLOAD_ERR_RW_ERROR;
> >> +
> >> + if (rsu_prog(doorbell) != RSU_PROG_IDLE &&
> >> + rsu_prog(doorbell) != RSU_PROG_RSU_DONE) {
> >> + log_error_regs(sec, doorbell);
> >> + return FW_UPLOAD_ERR_BUSY;
> >> + }
> >> +
> >> + return FW_UPLOAD_ERR_NONE;
> >> +}
> >> +
> >> +static inline bool rsu_start_done(u32 doorbell)
> >> +{
> >> + u32 status, progress;
> >> +
> >> + if (doorbell & DRBL_RSU_REQUEST)
> >> + return false;
> >> +
> >> + status = rsu_stat(doorbell);
> >> + if (status == RSU_STAT_ERASE_FAIL || status == RSU_STAT_WEAROUT)
> >> + return true;
> >> +
> >> + progress = rsu_prog(doorbell);
> >> + if (progress != RSU_PROG_IDLE && progress != RSU_PROG_RSU_DONE)
> >> + return true;
> >> +
> >> + return false;
> >> +}
> >> +
> >> +static enum fw_upload_err rsu_update_init(struct m10bmc_sec *sec)
> >> +{
> >> + u32 doorbell, status;
> >> + int ret;
> >> +
> >> + ret = regmap_update_bits(sec->m10bmc->regmap,
> >> + M10BMC_SYS_BASE + M10BMC_DOORBELL,
> >> + DRBL_RSU_REQUEST | DRBL_HOST_STATUS,
> >> + DRBL_RSU_REQUEST |
> >> + FIELD_PREP(DRBL_HOST_STATUS,
> >> + HOST_STATUS_IDLE));
> >> + if (ret)
> >> + return FW_UPLOAD_ERR_RW_ERROR;
> >> +
> >> + ret = regmap_read_poll_timeout(sec->m10bmc->regmap,
> >> + M10BMC_SYS_BASE + M10BMC_DOORBELL,
> >> + doorbell,
> >> + rsu_start_done(doorbell),
> >> + NIOS_HANDSHAKE_INTERVAL_US,
> >> + NIOS_HANDSHAKE_TIMEOUT_US);
> >> +
> >> + if (ret == -ETIMEDOUT) {
> >> + log_error_regs(sec, doorbell);
> >> + return FW_UPLOAD_ERR_TIMEOUT;
> >> + } else if (ret) {
> >> + return FW_UPLOAD_ERR_RW_ERROR;
> >> + }
> >> +
> >> + status = rsu_stat(doorbell);
> >> + if (status == RSU_STAT_WEAROUT) {
> >> + dev_warn(sec->dev, "Excessive flash update count detected\n");
> >> + return FW_UPLOAD_ERR_WEAROUT;
> >> + } else if (status == RSU_STAT_ERASE_FAIL) {
> >> + log_error_regs(sec, doorbell);
> >> + return FW_UPLOAD_ERR_HW_ERROR;
> >> + }
> >> +
> >> + return FW_UPLOAD_ERR_NONE;
> >> +}
> >> +
> >> +static enum fw_upload_err rsu_prog_ready(struct m10bmc_sec *sec)
> >> +{
> >> + unsigned long poll_timeout;
> >> + u32 doorbell, progress;
> >> + int ret;
> >> +
> >> + ret = m10bmc_sys_read(sec->m10bmc, M10BMC_DOORBELL, &doorbell);
> >> + if (ret)
> >> + return FW_UPLOAD_ERR_RW_ERROR;
> >> +
> >> + poll_timeout = jiffies + msecs_to_jiffies(RSU_PREP_TIMEOUT_MS);
> >> + while (rsu_prog(doorbell) == RSU_PROG_PREPARE) {
> >> + msleep(RSU_PREP_INTERVAL_MS);
> >> + if (time_after(jiffies, poll_timeout))
> >> + break;
> >> +
> >> + ret = m10bmc_sys_read(sec->m10bmc, M10BMC_DOORBELL, &doorbell);
> >> + if (ret)
> >> + return FW_UPLOAD_ERR_RW_ERROR;
> >> + }
> >> +
> >> + progress = rsu_prog(doorbell);
> >> + if (progress == RSU_PROG_PREPARE) {
> >> + log_error_regs(sec, doorbell);
> >> + return FW_UPLOAD_ERR_TIMEOUT;
> >> + } else if (progress != RSU_PROG_READY) {
> >> + log_error_regs(sec, doorbell);
> >> + return FW_UPLOAD_ERR_HW_ERROR;
> >> + }
> >> +
> >> + return FW_UPLOAD_ERR_NONE;
> >> +}
> >> +
> >> +static enum fw_upload_err rsu_send_data(struct m10bmc_sec *sec)
> >> +{
> >> + u32 doorbell;
> >> + int ret;
> >> +
> >> + ret = regmap_update_bits(sec->m10bmc->regmap,
> >> + M10BMC_SYS_BASE + M10BMC_DOORBELL,
> >> + DRBL_HOST_STATUS,
> >> + FIELD_PREP(DRBL_HOST_STATUS,
> >> + HOST_STATUS_WRITE_DONE));
> >> + if (ret)
> >> + return FW_UPLOAD_ERR_RW_ERROR;
> >> +
> >> + ret = regmap_read_poll_timeout(sec->m10bmc->regmap,
> >> + M10BMC_SYS_BASE + M10BMC_DOORBELL,
> >> + doorbell,
> >> + rsu_prog(doorbell) != RSU_PROG_READY,
> >> + NIOS_HANDSHAKE_INTERVAL_US,
> >> + NIOS_HANDSHAKE_TIMEOUT_US);
> >> +
> >> + if (ret == -ETIMEDOUT) {
> >> + log_error_regs(sec, doorbell);
> >> + return FW_UPLOAD_ERR_TIMEOUT;
> >> + } else if (ret) {
> >> + return FW_UPLOAD_ERR_RW_ERROR;
> >> + }
> >> +
> >> + switch (rsu_stat(doorbell)) {
> >> + case RSU_STAT_NORMAL:
> >> + case RSU_STAT_NIOS_OK:
> >> + case RSU_STAT_USER_OK:
> >> + case RSU_STAT_FACTORY_OK:
> >> + break;
> >> + default:
> >> + log_error_regs(sec, doorbell);
> >> + return FW_UPLOAD_ERR_HW_ERROR;
> >> + }
> >> +
> >> + return FW_UPLOAD_ERR_NONE;
> >> +}
> >> +
> >> +static int rsu_check_complete(struct m10bmc_sec *sec, u32 *doorbell)
> >> +{
> >> + if (m10bmc_sys_read(sec->m10bmc, M10BMC_DOORBELL, doorbell))
> >> + return -EIO;
> >> +
> >> + switch (rsu_stat(*doorbell)) {
> >> + case RSU_STAT_NORMAL:
> >> + case RSU_STAT_NIOS_OK:
> >> + case RSU_STAT_USER_OK:
> >> + case RSU_STAT_FACTORY_OK:
> >> + break;
> >> + default:
> >> + return -EINVAL;
> >> + }
> >> +
> >> + switch (rsu_prog(*doorbell)) {
> >> + case RSU_PROG_IDLE:
> >> + case RSU_PROG_RSU_DONE:
> >> + return 0;
> >> + case RSU_PROG_AUTHENTICATING:
> >> + case RSU_PROG_COPYING:
> >> + case RSU_PROG_UPDATE_CANCEL:
> >> + case RSU_PROG_PROGRAM_KEY_HASH:
> >> + return -EAGAIN;
> >> + default:
> >> + return -EINVAL;
> >> + }
> >> +}
> >> +
> >> +static enum fw_upload_err rsu_cancel(struct m10bmc_sec *sec)
> >> +{
> >> + u32 doorbell;
> >> + int ret;
> >> +
> >> + ret = m10bmc_sys_read(sec->m10bmc, M10BMC_DOORBELL, &doorbell);
> >> + if (ret)
> >> + return FW_UPLOAD_ERR_RW_ERROR;
> >> +
> >> + if (rsu_prog(doorbell) != RSU_PROG_READY)
> >> + return FW_UPLOAD_ERR_BUSY;
> >> +
> >> + ret = regmap_update_bits(sec->m10bmc->regmap,
> >> + M10BMC_SYS_BASE + M10BMC_DOORBELL,
> >> + DRBL_HOST_STATUS,
> >> + FIELD_PREP(DRBL_HOST_STATUS,
> >> + HOST_STATUS_ABORT_RSU));
> >> + if (ret)
> >> + return FW_UPLOAD_ERR_RW_ERROR;
> >> +
> >> + return FW_UPLOAD_ERR_CANCELED;
> >> +}
> >> +
> >> +static enum fw_upload_err m10bmc_sec_prepare(struct fw_upload *fwl,
> >> + const u8 *data, u32 size)
> >> +{
> >> + struct m10bmc_sec *sec = fwl->dd_handle;
> >> + u32 ret;
> >> +
> >> + sec->cancel_request = false;
> >> +
> >> + if (!size || size & 0x3 || size > M10BMC_STAGING_SIZE)
> > Why the size should be 4 bytes aligned? I assume this relates to max10
> > register stride. If so, I suggest we don't use instant number here.
>
> This check was not related to stride. All of the image files for these
> devices are 4-byte aligned. I suppose there is no reason to enforce this
> as the BMC will detect if a secure image is invalid. I can remove this
> check.
>
> >> + return FW_UPLOAD_ERR_INVALID_SIZE;
> >> +
> >> + ret = rsu_check_idle(sec);
> >> + if (ret != FW_UPLOAD_ERR_NONE)
> >> + return ret;
> >> +
> >> + ret = rsu_update_init(sec);
> >> + if (ret != FW_UPLOAD_ERR_NONE)
> >> + return ret;
> >> +
> >> + ret = rsu_prog_ready(sec);
> >> + if (ret != FW_UPLOAD_ERR_NONE)
> >> + return ret;
> >> +
> >> + if (sec->cancel_request)
> >> + return rsu_cancel(sec);
> >> +
> >> + return FW_UPLOAD_ERR_NONE;
> >> +}
> >> +
> >> +#define WRITE_BLOCK_SIZE 0x4000 /* Default write-block size is 0x4000 bytes */
> >> +
> >> +static enum fw_upload_err m10bmc_sec_write(struct fw_upload *fwl, const u8 *data,
> >> + u32 offset, u32 size, u32 *written)
> >> +{
> >> + struct m10bmc_sec *sec = fwl->dd_handle;
> >> + unsigned int stride = regmap_get_reg_stride(sec->m10bmc->regmap);
> > reverse xmax tree please.
> OK - I'll fix it.
> >
> >> + u32 blk_size, doorbell;
> >> + int ret;
> >> +
> >> + if (sec->cancel_request)
> >> + return rsu_cancel(sec);
> >> +
> >> + ret = m10bmc_sys_read(sec->m10bmc, M10BMC_DOORBELL, &doorbell);
> >> + if (ret) {
> >> + return FW_UPLOAD_ERR_RW_ERROR;
> >> + } else if (rsu_prog(doorbell) != RSU_PROG_READY) {
> >> + log_error_regs(sec, doorbell);
> >> + return FW_UPLOAD_ERR_HW_ERROR;
> >> + }
> >> +
> >> + blk_size = min_t(u32, WRITE_BLOCK_SIZE, size);
> >> + ret = regmap_bulk_write(sec->m10bmc->regmap,
> >> + M10BMC_STAGING_BASE + offset,
> >> + (void *)data + offset,
> >> + (blk_size + stride - 1) / stride);
> > You mean to DIV_ROUND_UP the write size here? So you may write more
> > bytes than required?
> >
> > Also possible to access out of bounds for 'data', is it?
>
> All of the writes, except possibly the last write, will be of size
> WRITE_BLOCK_SIZE, which is page aligned. Only the last write could
> potentially round up on the stride. We have already verified that
> the data written will fit in the staging area (in the prepare function),
> so additional bytes to complete the stride will not exceed the staging
> area which is the target for these writes.
Is it possible the source memory - 'data' be accessed out of bounds?
>
> >> +
> >> + if (ret)
> >> + return FW_UPLOAD_ERR_RW_ERROR;
> >> +
> >> + *written = blk_size;
> >> + return FW_UPLOAD_ERR_NONE;
> >> +}
> >> +
> >> +static enum fw_upload_err m10bmc_sec_poll_complete(struct fw_upload *fwl)
> >> +{
> >> + struct m10bmc_sec *sec = fwl->dd_handle;
> >> + unsigned long poll_timeout;
> >> + u32 doorbell, result;
> >> + int ret;
> >> +
> >> + if (sec->cancel_request)
> >> + return rsu_cancel(sec);
> >> +
> >> + result = rsu_send_data(sec);
> >> + if (result != FW_UPLOAD_ERR_NONE)
> >> + return result;
> >> +
> >> + poll_timeout = jiffies + msecs_to_jiffies(RSU_COMPLETE_TIMEOUT_MS);
> >> + do {
> >> + msleep(RSU_COMPLETE_INTERVAL_MS);
> >> + ret = rsu_check_complete(sec, &doorbell);
> >> + } while (ret == -EAGAIN && !time_after(jiffies, poll_timeout));
> >> +
> >> + if (ret == -EAGAIN) {
> >> + log_error_regs(sec, doorbell);
> >> + return FW_UPLOAD_ERR_TIMEOUT;
> >> + } else if (ret == -EIO) {
> >> + return FW_UPLOAD_ERR_RW_ERROR;
> >> + } else if (ret) {
> >> + log_error_regs(sec, doorbell);
> >> + return FW_UPLOAD_ERR_HW_ERROR;
> >> + }
> >> +
> >> + return FW_UPLOAD_ERR_NONE;
> >> +}
> >> +
> >> +/*
> >> + * m10bmc_sec_cancel() may be called asynchronously with an on-going update.
> >> + * All other functions are called sequentially in a single thread. To avoid
> >> + * contention on register accesses, m10bmc_sec_cancel() must only update
> >> + * the cancel_request flag. Other functions will check this flag and handle
> >> + * the cancel request synchronously.
> >> + */
> >> +static void m10bmc_sec_cancel(struct fw_upload *fwl)
> >> +{
> >> + struct m10bmc_sec *sec = fwl->dd_handle;
> >> +
> >> + sec->cancel_request = true;
> > Do we need some atomic_xx functions to ensure concurrent correctness?
>
> In the context of an update, this is once-written and only if the secure update
> is cancelled. We are only checking true vs false so I don't think atomicity
> is an issue. Do you recommend that I use atomic_read() and atomic_write() for
I see the flag is set false on prepare. So if cancel is not allowed in
prepare phase, there will be no concurrent writing problem.
Now there seems a problem that cancel_request is allowed to be set during
prepare phase, but may be overriden by its initialization to false in
prepare(). Could we just prevent cancel in prepare phase in framework?
This is not all about the atomicity for cancel_request but also helps
solve the concern.
Thanks,
Yilun
> the cancel_request flag?
>
> >
> >> +}
> >> +
> >> +static void m10bmc_sec_cleanup(struct fw_upload *fwl)
> >> +{
> >> + struct m10bmc_sec *sec = fwl->dd_handle;
> >> +
> >> + (void)rsu_cancel(sec);
> >> +}
> >> +
> >> +static const struct fw_upload_ops m10bmc_ops = {
> >> + .prepare = m10bmc_sec_prepare,
> >> + .write = m10bmc_sec_write,
> >> + .poll_complete = m10bmc_sec_poll_complete,
> >> + .cancel = m10bmc_sec_cancel,
> >> + .cleanup = m10bmc_sec_cleanup,
> >> +};
> >> +
> >> #define SEC_UPDATE_LEN_MAX 32
> >> static int m10bmc_sec_probe(struct platform_device *pdev)
> >> {
> >> + char buf[SEC_UPDATE_LEN_MAX];
> >> + struct fw_upload *fwl;
> >> + unsigned int len, ret;
> > Why the ret is unsigned int?
> I will change it to int.
>
> >
> >> struct m10bmc_sec *sec;
> >>
> >> sec = devm_kzalloc(&pdev->dev, sizeof(*sec), GFP_KERNEL);
> >> @@ -192,6 +538,36 @@ static int m10bmc_sec_probe(struct platform_device *pdev)
> >> sec->m10bmc = dev_get_drvdata(pdev->dev.parent);
> >> dev_set_drvdata(&pdev->dev, sec);
> >>
> >> + ret = xa_alloc(&fw_upload_xa, &sec->fw_name_id, sec,
> >> + xa_limit_32b, GFP_KERNEL);
> >> + if (ret)
> >> + return ret;
> >> +
> >> + len = scnprintf(buf, SEC_UPDATE_LEN_MAX, "secure-update%d",
> >> + sec->fw_name_id);
> >> + sec->fw_name = kmemdup_nul(buf, len, GFP_KERNEL);
> > Should we check this memory allocation?
> Yes! - I'll add a check.
>
> Thanks for the review.
> - Russ
> >
> > Thanks,
> > Yilun
> >
> >> +
> >> + fwl = firmware_upload_register(THIS_MODULE, sec->dev, sec->fw_name,
> >> + &m10bmc_ops, sec);
> >> + if (IS_ERR(fwl)) {
> >> + dev_err(sec->dev, "Firmware Upload driver failed to start\n");
> >> + kfree(sec->fw_name);
> >> + xa_erase(&fw_upload_xa, sec->fw_name_id);
> >> + return PTR_ERR(fwl);
> >> + }
> >> +
> >> + sec->fwl = fwl;
> >> + return 0;
> >> +}
> >> +
> >> +static int m10bmc_sec_remove(struct platform_device *pdev)
> >> +{
> >> + struct m10bmc_sec *sec = dev_get_drvdata(&pdev->dev);
> >> +
> >> + firmware_upload_unregister(sec->fwl);
> >> + kfree(sec->fw_name);
> >> + xa_erase(&fw_upload_xa, sec->fw_name_id);
> >> +
> >> return 0;
> >> }
> >>
> >> @@ -205,6 +581,7 @@ MODULE_DEVICE_TABLE(platform, intel_m10bmc_sec_ids);
> >>
> >> static struct platform_driver intel_m10bmc_sec_driver = {
> >> .probe = m10bmc_sec_probe,
> >> + .remove = m10bmc_sec_remove,
> >> .driver = {
> >> .name = "intel-m10bmc-sec-update",
> >> .dev_groups = m10bmc_sec_attr_groups,
> >> --
> >> 2.25.1
On 5/18/22 01:40, Xu Yilun wrote:
> On Tue, May 17, 2022 at 01:16:42PM -0700, Russ Weight wrote:
>>
>> On 5/17/22 02:48, Xu Yilun wrote:
>>> On Mon, May 16, 2022 at 04:49:41PM -0700, Russ Weight wrote:
>>>> Create firmware upload ops and call the Firmware Upload support of the
>>>> Firmware Loader subsystem to enable FPGA image uploads for secure
>>>> updates of BMC images, FPGA images, etc.
>>>>
>>>> Signed-off-by: Russ Weight <[email protected]>
>>>> ---
>>>> v20:
>>>> - No change.
>>>> v19:
>>>> - Change "card bmc" naming back to "m10 bmc" naming to be consistent
>>>> with the parent driver.
>>>> v18:
>>>> - Moved the firmware_upload_register() function here from an earlier
>>>> patch since this is where the required ops are provided.
>>>> - Moved the bmc_sec_remove() function here from an earlier patch to
>>>> unregister the firmware driver and do cleanup.
>>>> v17:
>>>> - Change "m10bmc" in symbol names to "cardbmc" to reflect the fact that the
>>>> future devices will not necessarily use the MAX10.
>>>> - Change from image_load class driver to the new firmware_upload
>>>> functionality of the firmware_loader.
>>>> - fw_upload_ops functions will return "enum fw_upload_err" data types
>>>> instead of integer values.
>>>> v16:
>>>> - Use 0 instead of FPGA_IMAGE_ERR_NONE to indicate success.
>>>> - The size alignment check was moved from the FPGA Image Load framework
>>>> to the prepare() op.
>>>> - Added cancel_request boolean flag to struct m10bmc_sec.
>>>> - Moved the RSU cancellation logic from m10bmc_sec_cancel() to a new
>>>> rsu_cancel() function.
>>>> - The m10bmc_sec_cancel() function ONLY sets the cancel_request flag.
>>>> The cancel_request flag is checked at the beginning of the
>>>> m10bmc_sec_write() and m10bmc_sec_poll_complete() functions.
>>>> - Adapt to changed prototypes for the prepare() and write() ops. The
>>>> m10bmc_sec_write_blk() function has been renamed to
>>>> m10bmc_sec_write().
>>>> - Created a cleanup() op, m10bmc_sec_cleanup(), to attempt to cancel an
>>>> ongoing op during when exiting the update process.
>>>> v15:
>>>> - Adapted to changes in the FPGA Image Load framework:
>>>> (1) All enum types (progress and errors) are now type u32
>>>> (2) m10bmc_sec_write_blk() adds *blk_size and max_size parameters
>>>> and uses *blk_size as provided by the caller.
>>>> (3) m10bmc_sec_poll_complete() no long checks the driver_unload
>>>> flag.
>>>> v14:
>>>> - Changed symbol names to reflect the renaming of the Security Manager
>>>> Class driver to FPGA Image Load.
>>>> v13:
>>>> - No change
>>>> v12:
>>>> - Updated Date and KernelVersion fields in ABI documentation
>>>> - Removed size parameter from the write_blk() op. m10bmc_sec_write_blk()
>>>> no longer has a size parameter, and the block size is determined
>>>> in this (the lower-level) driver.
>>>> v11:
>>>> - No change
>>>> v10:
>>>> - No change
>>>> v9:
>>>> - No change
>>>> v8:
>>>> - Previously patch 5/6, otherwise no change
>>>> v7:
>>>> - No change
>>>> v6:
>>>> - Changed (size / stride) calculation to ((size + stride - 1) / stride)
>>>> to ensure that the proper count is passed to regmap_bulk_write().
>>>> - Removed unnecessary call to rsu_check_complete() in
>>>> m10bmc_sec_poll_complete() and changed while loop to
>>>> do/while loop.
>>>> v5:
>>>> - No change
>>>> v4:
>>>> - No change
>>>> v3:
>>>> - Changed: iops -> sops, imgr -> smgr, IFPGA_ -> FPGA_, ifpga_ to fpga_
>>>> - Changed "MAX10 BMC Secure Engine driver" to "MAX10 BMC Secure Update
>>>> driver"
>>>> - Removed wrapper functions (m10bmc_raw_*, m10bmc_sys_*). The
>>>> underlying functions are now called directly.
>>>> - Changed calling functions of functions that return "enum fpga_sec_err"
>>>> to check for (ret != FPGA_SEC_ERR_NONE) instead of (ret)
>>>> v2:
>>>> - Reworked the rsu_start_done() function to make it more readable
>>>> - Reworked while-loop condition/content in rsu_prog_ready()
>>>> - Minor code cleanup per review comments
>>>> - Added a comment to the m10bmc_sec_poll_complete() function to
>>>> explain the context (could take 30+ minutes to complete).
>>>> - Added m10bmc_ prefix to functions in m10bmc_iops structure
>>>> - Moved MAX10 BMC address and function definitions to a separate
>>>> patch.
>>>> ---
>>>> drivers/fpga/intel-m10-bmc-sec-update.c | 377 ++++++++++++++++++++++++
>>>> 1 file changed, 377 insertions(+)
>>>>
>>>> diff --git a/drivers/fpga/intel-m10-bmc-sec-update.c b/drivers/fpga/intel-m10-bmc-sec-update.c
>>>> index 6c39adc6492d..a4dc5f47e398 100644
>>>> --- a/drivers/fpga/intel-m10-bmc-sec-update.c
>>>> +++ b/drivers/fpga/intel-m10-bmc-sec-update.c
>>>> @@ -17,8 +17,14 @@
>>>> struct m10bmc_sec {
>>>> struct device *dev;
>>>> struct intel_m10bmc *m10bmc;
>>>> + struct fw_upload *fwl;
>>>> + char *fw_name;
>>>> + u32 fw_name_id;
>>>> + bool cancel_request;
>>>> };
>>>>
>>>> +static DEFINE_XARRAY_ALLOC(fw_upload_xa);
>>>> +
>>>> /* Root Entry Hash (REH) support */
>>>> #define REH_SHA256_SIZE 32
>>>> #define REH_SHA384_SIZE 48
>>>> @@ -179,9 +185,349 @@ static const struct attribute_group *m10bmc_sec_attr_groups[] = {
>>>> NULL,
>>>> };
>>>>
>>>> +static void log_error_regs(struct m10bmc_sec *sec, u32 doorbell)
>>>> +{
>>>> + u32 auth_result;
>>>> +
>>>> + dev_err(sec->dev, "RSU error status: 0x%08x\n", doorbell);
>>>> +
>>>> + if (!m10bmc_sys_read(sec->m10bmc, M10BMC_AUTH_RESULT, &auth_result))
>>>> + dev_err(sec->dev, "RSU auth result: 0x%08x\n", auth_result);
>>>> +}
>>>> +
>>>> +static enum fw_upload_err rsu_check_idle(struct m10bmc_sec *sec)
>>>> +{
>>>> + u32 doorbell;
>>>> + int ret;
>>>> +
>>>> + ret = m10bmc_sys_read(sec->m10bmc, M10BMC_DOORBELL, &doorbell);
>>>> + if (ret)
>>>> + return FW_UPLOAD_ERR_RW_ERROR;
>>>> +
>>>> + if (rsu_prog(doorbell) != RSU_PROG_IDLE &&
>>>> + rsu_prog(doorbell) != RSU_PROG_RSU_DONE) {
>>>> + log_error_regs(sec, doorbell);
>>>> + return FW_UPLOAD_ERR_BUSY;
>>>> + }
>>>> +
>>>> + return FW_UPLOAD_ERR_NONE;
>>>> +}
>>>> +
>>>> +static inline bool rsu_start_done(u32 doorbell)
>>>> +{
>>>> + u32 status, progress;
>>>> +
>>>> + if (doorbell & DRBL_RSU_REQUEST)
>>>> + return false;
>>>> +
>>>> + status = rsu_stat(doorbell);
>>>> + if (status == RSU_STAT_ERASE_FAIL || status == RSU_STAT_WEAROUT)
>>>> + return true;
>>>> +
>>>> + progress = rsu_prog(doorbell);
>>>> + if (progress != RSU_PROG_IDLE && progress != RSU_PROG_RSU_DONE)
>>>> + return true;
>>>> +
>>>> + return false;
>>>> +}
>>>> +
>>>> +static enum fw_upload_err rsu_update_init(struct m10bmc_sec *sec)
>>>> +{
>>>> + u32 doorbell, status;
>>>> + int ret;
>>>> +
>>>> + ret = regmap_update_bits(sec->m10bmc->regmap,
>>>> + M10BMC_SYS_BASE + M10BMC_DOORBELL,
>>>> + DRBL_RSU_REQUEST | DRBL_HOST_STATUS,
>>>> + DRBL_RSU_REQUEST |
>>>> + FIELD_PREP(DRBL_HOST_STATUS,
>>>> + HOST_STATUS_IDLE));
>>>> + if (ret)
>>>> + return FW_UPLOAD_ERR_RW_ERROR;
>>>> +
>>>> + ret = regmap_read_poll_timeout(sec->m10bmc->regmap,
>>>> + M10BMC_SYS_BASE + M10BMC_DOORBELL,
>>>> + doorbell,
>>>> + rsu_start_done(doorbell),
>>>> + NIOS_HANDSHAKE_INTERVAL_US,
>>>> + NIOS_HANDSHAKE_TIMEOUT_US);
>>>> +
>>>> + if (ret == -ETIMEDOUT) {
>>>> + log_error_regs(sec, doorbell);
>>>> + return FW_UPLOAD_ERR_TIMEOUT;
>>>> + } else if (ret) {
>>>> + return FW_UPLOAD_ERR_RW_ERROR;
>>>> + }
>>>> +
>>>> + status = rsu_stat(doorbell);
>>>> + if (status == RSU_STAT_WEAROUT) {
>>>> + dev_warn(sec->dev, "Excessive flash update count detected\n");
>>>> + return FW_UPLOAD_ERR_WEAROUT;
>>>> + } else if (status == RSU_STAT_ERASE_FAIL) {
>>>> + log_error_regs(sec, doorbell);
>>>> + return FW_UPLOAD_ERR_HW_ERROR;
>>>> + }
>>>> +
>>>> + return FW_UPLOAD_ERR_NONE;
>>>> +}
>>>> +
>>>> +static enum fw_upload_err rsu_prog_ready(struct m10bmc_sec *sec)
>>>> +{
>>>> + unsigned long poll_timeout;
>>>> + u32 doorbell, progress;
>>>> + int ret;
>>>> +
>>>> + ret = m10bmc_sys_read(sec->m10bmc, M10BMC_DOORBELL, &doorbell);
>>>> + if (ret)
>>>> + return FW_UPLOAD_ERR_RW_ERROR;
>>>> +
>>>> + poll_timeout = jiffies + msecs_to_jiffies(RSU_PREP_TIMEOUT_MS);
>>>> + while (rsu_prog(doorbell) == RSU_PROG_PREPARE) {
>>>> + msleep(RSU_PREP_INTERVAL_MS);
>>>> + if (time_after(jiffies, poll_timeout))
>>>> + break;
>>>> +
>>>> + ret = m10bmc_sys_read(sec->m10bmc, M10BMC_DOORBELL, &doorbell);
>>>> + if (ret)
>>>> + return FW_UPLOAD_ERR_RW_ERROR;
>>>> + }
>>>> +
>>>> + progress = rsu_prog(doorbell);
>>>> + if (progress == RSU_PROG_PREPARE) {
>>>> + log_error_regs(sec, doorbell);
>>>> + return FW_UPLOAD_ERR_TIMEOUT;
>>>> + } else if (progress != RSU_PROG_READY) {
>>>> + log_error_regs(sec, doorbell);
>>>> + return FW_UPLOAD_ERR_HW_ERROR;
>>>> + }
>>>> +
>>>> + return FW_UPLOAD_ERR_NONE;
>>>> +}
>>>> +
>>>> +static enum fw_upload_err rsu_send_data(struct m10bmc_sec *sec)
>>>> +{
>>>> + u32 doorbell;
>>>> + int ret;
>>>> +
>>>> + ret = regmap_update_bits(sec->m10bmc->regmap,
>>>> + M10BMC_SYS_BASE + M10BMC_DOORBELL,
>>>> + DRBL_HOST_STATUS,
>>>> + FIELD_PREP(DRBL_HOST_STATUS,
>>>> + HOST_STATUS_WRITE_DONE));
>>>> + if (ret)
>>>> + return FW_UPLOAD_ERR_RW_ERROR;
>>>> +
>>>> + ret = regmap_read_poll_timeout(sec->m10bmc->regmap,
>>>> + M10BMC_SYS_BASE + M10BMC_DOORBELL,
>>>> + doorbell,
>>>> + rsu_prog(doorbell) != RSU_PROG_READY,
>>>> + NIOS_HANDSHAKE_INTERVAL_US,
>>>> + NIOS_HANDSHAKE_TIMEOUT_US);
>>>> +
>>>> + if (ret == -ETIMEDOUT) {
>>>> + log_error_regs(sec, doorbell);
>>>> + return FW_UPLOAD_ERR_TIMEOUT;
>>>> + } else if (ret) {
>>>> + return FW_UPLOAD_ERR_RW_ERROR;
>>>> + }
>>>> +
>>>> + switch (rsu_stat(doorbell)) {
>>>> + case RSU_STAT_NORMAL:
>>>> + case RSU_STAT_NIOS_OK:
>>>> + case RSU_STAT_USER_OK:
>>>> + case RSU_STAT_FACTORY_OK:
>>>> + break;
>>>> + default:
>>>> + log_error_regs(sec, doorbell);
>>>> + return FW_UPLOAD_ERR_HW_ERROR;
>>>> + }
>>>> +
>>>> + return FW_UPLOAD_ERR_NONE;
>>>> +}
>>>> +
>>>> +static int rsu_check_complete(struct m10bmc_sec *sec, u32 *doorbell)
>>>> +{
>>>> + if (m10bmc_sys_read(sec->m10bmc, M10BMC_DOORBELL, doorbell))
>>>> + return -EIO;
>>>> +
>>>> + switch (rsu_stat(*doorbell)) {
>>>> + case RSU_STAT_NORMAL:
>>>> + case RSU_STAT_NIOS_OK:
>>>> + case RSU_STAT_USER_OK:
>>>> + case RSU_STAT_FACTORY_OK:
>>>> + break;
>>>> + default:
>>>> + return -EINVAL;
>>>> + }
>>>> +
>>>> + switch (rsu_prog(*doorbell)) {
>>>> + case RSU_PROG_IDLE:
>>>> + case RSU_PROG_RSU_DONE:
>>>> + return 0;
>>>> + case RSU_PROG_AUTHENTICATING:
>>>> + case RSU_PROG_COPYING:
>>>> + case RSU_PROG_UPDATE_CANCEL:
>>>> + case RSU_PROG_PROGRAM_KEY_HASH:
>>>> + return -EAGAIN;
>>>> + default:
>>>> + return -EINVAL;
>>>> + }
>>>> +}
>>>> +
>>>> +static enum fw_upload_err rsu_cancel(struct m10bmc_sec *sec)
>>>> +{
>>>> + u32 doorbell;
>>>> + int ret;
>>>> +
>>>> + ret = m10bmc_sys_read(sec->m10bmc, M10BMC_DOORBELL, &doorbell);
>>>> + if (ret)
>>>> + return FW_UPLOAD_ERR_RW_ERROR;
>>>> +
>>>> + if (rsu_prog(doorbell) != RSU_PROG_READY)
>>>> + return FW_UPLOAD_ERR_BUSY;
>>>> +
>>>> + ret = regmap_update_bits(sec->m10bmc->regmap,
>>>> + M10BMC_SYS_BASE + M10BMC_DOORBELL,
>>>> + DRBL_HOST_STATUS,
>>>> + FIELD_PREP(DRBL_HOST_STATUS,
>>>> + HOST_STATUS_ABORT_RSU));
>>>> + if (ret)
>>>> + return FW_UPLOAD_ERR_RW_ERROR;
>>>> +
>>>> + return FW_UPLOAD_ERR_CANCELED;
>>>> +}
>>>> +
>>>> +static enum fw_upload_err m10bmc_sec_prepare(struct fw_upload *fwl,
>>>> + const u8 *data, u32 size)
>>>> +{
>>>> + struct m10bmc_sec *sec = fwl->dd_handle;
>>>> + u32 ret;
>>>> +
>>>> + sec->cancel_request = false;
>>>> +
>>>> + if (!size || size & 0x3 || size > M10BMC_STAGING_SIZE)
>>> Why the size should be 4 bytes aligned? I assume this relates to max10
>>> register stride. If so, I suggest we don't use instant number here.
>> This check was not related to stride. All of the image files for these
>> devices are 4-byte aligned. I suppose there is no reason to enforce this
>> as the BMC will detect if a secure image is invalid. I can remove this
>> check.
I'm going to add a check here to make sure that the base address is
aligned.
>>
>>>> + return FW_UPLOAD_ERR_INVALID_SIZE;
>>>> +
>>>> + ret = rsu_check_idle(sec);
>>>> + if (ret != FW_UPLOAD_ERR_NONE)
>>>> + return ret;
>>>> +
>>>> + ret = rsu_update_init(sec);
>>>> + if (ret != FW_UPLOAD_ERR_NONE)
>>>> + return ret;
>>>> +
>>>> + ret = rsu_prog_ready(sec);
>>>> + if (ret != FW_UPLOAD_ERR_NONE)
>>>> + return ret;
>>>> +
>>>> + if (sec->cancel_request)
>>>> + return rsu_cancel(sec);
>>>> +
>>>> + return FW_UPLOAD_ERR_NONE;
>>>> +}
>>>> +
>>>> +#define WRITE_BLOCK_SIZE 0x4000 /* Default write-block size is 0x4000 bytes */
>>>> +
>>>> +static enum fw_upload_err m10bmc_sec_write(struct fw_upload *fwl, const u8 *data,
>>>> + u32 offset, u32 size, u32 *written)
>>>> +{
>>>> + struct m10bmc_sec *sec = fwl->dd_handle;
>>>> + unsigned int stride = regmap_get_reg_stride(sec->m10bmc->regmap);
>>> reverse xmax tree please.
>> OK - I'll fix it.
>>>> + u32 blk_size, doorbell;
>>>> + int ret;
>>>> +
>>>> + if (sec->cancel_request)
>>>> + return rsu_cancel(sec);
>>>> +
>>>> + ret = m10bmc_sys_read(sec->m10bmc, M10BMC_DOORBELL, &doorbell);
>>>> + if (ret) {
>>>> + return FW_UPLOAD_ERR_RW_ERROR;
>>>> + } else if (rsu_prog(doorbell) != RSU_PROG_READY) {
>>>> + log_error_regs(sec, doorbell);
>>>> + return FW_UPLOAD_ERR_HW_ERROR;
>>>> + }
>>>> +
>>>> + blk_size = min_t(u32, WRITE_BLOCK_SIZE, size);
>>>> + ret = regmap_bulk_write(sec->m10bmc->regmap,
>>>> + M10BMC_STAGING_BASE + offset,
>>>> + (void *)data + offset,
>>>> + (blk_size + stride - 1) / stride);
>>> You mean to DIV_ROUND_UP the write size here? So you may write more
>>> bytes than required?
>>>
>>> Also possible to access out of bounds for 'data', is it?
>> All of the writes, except possibly the last write, will be of size
>> WRITE_BLOCK_SIZE, which is page aligned. Only the last write could
>> potentially round up on the stride. We have already verified that
>> the data written will fit in the staging area (in the prepare function),
>> so additional bytes to complete the stride will not exceed the staging
>> area which is the target for these writes.
> Is it possible the source memory - 'data' be accessed out of bounds?
Yes, I think it is. I have modified the code to allocate a zero-filled
buffer with size alignment and copy the data in. This should only ever
happen on the last block that is written, since the other writes should
be of size: WRITE_BLOCK_SIZE.
>
>>>> +
>>>> + if (ret)
>>>> + return FW_UPLOAD_ERR_RW_ERROR;
>>>> +
>>>> + *written = blk_size;
>>>> + return FW_UPLOAD_ERR_NONE;
>>>> +}
>>>> +
>>>> +static enum fw_upload_err m10bmc_sec_poll_complete(struct fw_upload *fwl)
>>>> +{
>>>> + struct m10bmc_sec *sec = fwl->dd_handle;
>>>> + unsigned long poll_timeout;
>>>> + u32 doorbell, result;
>>>> + int ret;
>>>> +
>>>> + if (sec->cancel_request)
>>>> + return rsu_cancel(sec);
>>>> +
>>>> + result = rsu_send_data(sec);
>>>> + if (result != FW_UPLOAD_ERR_NONE)
>>>> + return result;
>>>> +
>>>> + poll_timeout = jiffies + msecs_to_jiffies(RSU_COMPLETE_TIMEOUT_MS);
>>>> + do {
>>>> + msleep(RSU_COMPLETE_INTERVAL_MS);
>>>> + ret = rsu_check_complete(sec, &doorbell);
>>>> + } while (ret == -EAGAIN && !time_after(jiffies, poll_timeout));
>>>> +
>>>> + if (ret == -EAGAIN) {
>>>> + log_error_regs(sec, doorbell);
>>>> + return FW_UPLOAD_ERR_TIMEOUT;
>>>> + } else if (ret == -EIO) {
>>>> + return FW_UPLOAD_ERR_RW_ERROR;
>>>> + } else if (ret) {
>>>> + log_error_regs(sec, doorbell);
>>>> + return FW_UPLOAD_ERR_HW_ERROR;
>>>> + }
>>>> +
>>>> + return FW_UPLOAD_ERR_NONE;
>>>> +}
>>>> +
>>>> +/*
>>>> + * m10bmc_sec_cancel() may be called asynchronously with an on-going update.
>>>> + * All other functions are called sequentially in a single thread. To avoid
>>>> + * contention on register accesses, m10bmc_sec_cancel() must only update
>>>> + * the cancel_request flag. Other functions will check this flag and handle
>>>> + * the cancel request synchronously.
>>>> + */
>>>> +static void m10bmc_sec_cancel(struct fw_upload *fwl)
>>>> +{
>>>> + struct m10bmc_sec *sec = fwl->dd_handle;
>>>> +
>>>> + sec->cancel_request = true;
>>> Do we need some atomic_xx functions to ensure concurrent correctness?
>> In the context of an update, this is once-written and only if the secure update
>> is cancelled. We are only checking true vs false so I don't think atomicity
>> is an issue. Do you recommend that I use atomic_read() and atomic_write() for
> I see the flag is set false on prepare. So if cancel is not allowed in
> prepare phase, there will be no concurrent writing problem.
Cancel is allowed in the prepare phase, but it won't be handled until
the hardware has completed the "prepare" phase. It is being set to
false in prepare, because that is the earliest place in the context
of a new secure update where the flag can be initialized.
> Now there seems a problem that cancel_request is allowed to be set during
> prepare phase, but may be overriden by its initialization to false in
> prepare(). Could we just prevent cancel in prepare phase in framework?
> This is not all about the atomicity for cancel_request but also helps
> solve the concern.
The initialization of the flag is the first thing that happens in
the lower-level driver in the context of a new secure update. Considering
that a user would have to initiate the cancel (CTRL-C or echo 1 > cancel),
I think the window for a race condition is so small that it can be ignored.
If you think this needs to be changed, where would you recommend that
request_cancel be initialized for a new secure-update? I suppose
it could be cleared at the time it is handled, but it seems safer to
explicitly initialize it at the beginning of a new update.
Thanks,
- Russ
>
> Thanks,
> Yilun
>
>> the cancel_request flag?
>>
>>>> +}
>>>> +
>>>> +static void m10bmc_sec_cleanup(struct fw_upload *fwl)
>>>> +{
>>>> + struct m10bmc_sec *sec = fwl->dd_handle;
>>>> +
>>>> + (void)rsu_cancel(sec);
>>>> +}
>>>> +
>>>> +static const struct fw_upload_ops m10bmc_ops = {
>>>> + .prepare = m10bmc_sec_prepare,
>>>> + .write = m10bmc_sec_write,
>>>> + .poll_complete = m10bmc_sec_poll_complete,
>>>> + .cancel = m10bmc_sec_cancel,
>>>> + .cleanup = m10bmc_sec_cleanup,
>>>> +};
>>>> +
>>>> #define SEC_UPDATE_LEN_MAX 32
>>>> static int m10bmc_sec_probe(struct platform_device *pdev)
>>>> {
>>>> + char buf[SEC_UPDATE_LEN_MAX];
>>>> + struct fw_upload *fwl;
>>>> + unsigned int len, ret;
>>> Why the ret is unsigned int?
>> I will change it to int.
>>
>>>> struct m10bmc_sec *sec;
>>>>
>>>> sec = devm_kzalloc(&pdev->dev, sizeof(*sec), GFP_KERNEL);
>>>> @@ -192,6 +538,36 @@ static int m10bmc_sec_probe(struct platform_device *pdev)
>>>> sec->m10bmc = dev_get_drvdata(pdev->dev.parent);
>>>> dev_set_drvdata(&pdev->dev, sec);
>>>>
>>>> + ret = xa_alloc(&fw_upload_xa, &sec->fw_name_id, sec,
>>>> + xa_limit_32b, GFP_KERNEL);
>>>> + if (ret)
>>>> + return ret;
>>>> +
>>>> + len = scnprintf(buf, SEC_UPDATE_LEN_MAX, "secure-update%d",
>>>> + sec->fw_name_id);
>>>> + sec->fw_name = kmemdup_nul(buf, len, GFP_KERNEL);
>>> Should we check this memory allocation?
>> Yes! - I'll add a check.
>>
>> Thanks for the review.
>> - Russ
>>> Thanks,
>>> Yilun
>>>
>>>> +
>>>> + fwl = firmware_upload_register(THIS_MODULE, sec->dev, sec->fw_name,
>>>> + &m10bmc_ops, sec);
>>>> + if (IS_ERR(fwl)) {
>>>> + dev_err(sec->dev, "Firmware Upload driver failed to start\n");
>>>> + kfree(sec->fw_name);
>>>> + xa_erase(&fw_upload_xa, sec->fw_name_id);
>>>> + return PTR_ERR(fwl);
>>>> + }
>>>> +
>>>> + sec->fwl = fwl;
>>>> + return 0;
>>>> +}
>>>> +
>>>> +static int m10bmc_sec_remove(struct platform_device *pdev)
>>>> +{
>>>> + struct m10bmc_sec *sec = dev_get_drvdata(&pdev->dev);
>>>> +
>>>> + firmware_upload_unregister(sec->fwl);
>>>> + kfree(sec->fw_name);
>>>> + xa_erase(&fw_upload_xa, sec->fw_name_id);
>>>> +
>>>> return 0;
>>>> }
>>>>
>>>> @@ -205,6 +581,7 @@ MODULE_DEVICE_TABLE(platform, intel_m10bmc_sec_ids);
>>>>
>>>> static struct platform_driver intel_m10bmc_sec_driver = {
>>>> .probe = m10bmc_sec_probe,
>>>> + .remove = m10bmc_sec_remove,
>>>> .driver = {
>>>> .name = "intel-m10bmc-sec-update",
>>>> .dev_groups = m10bmc_sec_attr_groups,
>>>> --
>>>> 2.25.1