Manage handshake register access on Max 10 FPGA cards that have a major
slowdown on reading handshake registers during secure update prepare and
write phases. The problem does not occur with PMCI-based cards.
Ilpo Järvinen (4):
mfd: intel-m10-bmc: Move core symbols to own namespace
mfd: intel-m10-bmc: Create m10bmc_sys_update_bits()
mfd: intel-m10-bmc: Move m10bmc_sys_read() away from header
mfd: intel-m10-bmc: Manage access to MAX 10 fw handshake registers
drivers/fpga/intel-m10-bmc-sec-update.c | 47 ++++++++------
drivers/hwmon/intel-m10-bmc-hwmon.c | 1 +
drivers/mfd/intel-m10-bmc-core.c | 84 ++++++++++++++++++++++++-
drivers/mfd/intel-m10-bmc-pmci.c | 5 ++
drivers/mfd/intel-m10-bmc-spi.c | 15 +++++
include/linux/mfd/intel-m10-bmc.h | 42 +++++++++----
6 files changed, 161 insertions(+), 33 deletions(-)
--
2.30.2
Wrap regmap_update_bits() with m10bmc_sys_update_bits() in order to be
able to add additional checks into it.
Co-developed-by: Russ Weight <[email protected]>
Signed-off-by: Russ Weight <[email protected]>
Signed-off-by: Ilpo Järvinen <[email protected]>
---
drivers/fpga/intel-m10-bmc-sec-update.c | 30 ++++++++++++-------------
drivers/mfd/intel-m10-bmc-core.c | 9 ++++++++
include/linux/mfd/intel-m10-bmc.h | 4 ++++
3 files changed, 27 insertions(+), 16 deletions(-)
diff --git a/drivers/fpga/intel-m10-bmc-sec-update.c b/drivers/fpga/intel-m10-bmc-sec-update.c
index d7e2f9f461bc..fe0127a58eff 100644
--- a/drivers/fpga/intel-m10-bmc-sec-update.c
+++ b/drivers/fpga/intel-m10-bmc-sec-update.c
@@ -376,12 +376,11 @@ static enum fw_upload_err rsu_update_init(struct m10bmc_sec *sec)
u32 doorbell_reg, progress, status;
int ret, err;
- ret = regmap_update_bits(sec->m10bmc->regmap,
- csr_map->base + csr_map->doorbell,
- DRBL_RSU_REQUEST | DRBL_HOST_STATUS,
- DRBL_RSU_REQUEST |
- FIELD_PREP(DRBL_HOST_STATUS,
- HOST_STATUS_IDLE));
+ ret = m10bmc_sys_update_bits(sec->m10bmc, csr_map->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;
@@ -450,11 +449,10 @@ static enum fw_upload_err rsu_send_data(struct m10bmc_sec *sec)
u32 doorbell_reg, status;
int ret;
- ret = regmap_update_bits(sec->m10bmc->regmap,
- csr_map->base + csr_map->doorbell,
- DRBL_HOST_STATUS,
- FIELD_PREP(DRBL_HOST_STATUS,
- HOST_STATUS_WRITE_DONE));
+ ret = m10bmc_sys_update_bits(sec->m10bmc, csr_map->doorbell,
+ DRBL_HOST_STATUS,
+ FIELD_PREP(DRBL_HOST_STATUS,
+ HOST_STATUS_WRITE_DONE));
if (ret)
return FW_UPLOAD_ERR_RW_ERROR;
@@ -517,11 +515,10 @@ static enum fw_upload_err rsu_cancel(struct m10bmc_sec *sec)
if (rsu_prog(doorbell) != RSU_PROG_READY)
return FW_UPLOAD_ERR_BUSY;
- ret = regmap_update_bits(sec->m10bmc->regmap,
- csr_map->base + csr_map->doorbell,
- DRBL_HOST_STATUS,
- FIELD_PREP(DRBL_HOST_STATUS,
- HOST_STATUS_ABORT_RSU));
+ ret = m10bmc_sys_update_bits(sec->m10bmc, csr_map->doorbell,
+ DRBL_HOST_STATUS,
+ FIELD_PREP(DRBL_HOST_STATUS,
+ HOST_STATUS_ABORT_RSU));
if (ret)
return FW_UPLOAD_ERR_RW_ERROR;
@@ -764,3 +761,4 @@ module_platform_driver(intel_m10bmc_sec_driver);
MODULE_AUTHOR("Intel Corporation");
MODULE_DESCRIPTION("Intel MAX10 BMC Secure Update");
MODULE_LICENSE("GPL");
+MODULE_IMPORT_NS(INTEL_M10_BMC_CORE);
diff --git a/drivers/mfd/intel-m10-bmc-core.c b/drivers/mfd/intel-m10-bmc-core.c
index b94412813887..879d98b9b14d 100644
--- a/drivers/mfd/intel-m10-bmc-core.c
+++ b/drivers/mfd/intel-m10-bmc-core.c
@@ -12,6 +12,15 @@
#include <linux/mfd/intel-m10-bmc.h>
#include <linux/module.h>
+int m10bmc_sys_update_bits(struct intel_m10bmc *m10bmc, unsigned int offset,
+ unsigned int msk, unsigned int val)
+{
+ const struct m10bmc_csr_map *csr_map = m10bmc->info->csr_map;
+
+ return regmap_update_bits(m10bmc->regmap, csr_map->base + offset, msk, val);
+}
+EXPORT_SYMBOL_NS_GPL(m10bmc_sys_update_bits, INTEL_M10_BMC_CORE);
+
static ssize_t bmc_version_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
diff --git a/include/linux/mfd/intel-m10-bmc.h b/include/linux/mfd/intel-m10-bmc.h
index 1812ebfa11a8..5418f7279ed0 100644
--- a/include/linux/mfd/intel-m10-bmc.h
+++ b/include/linux/mfd/intel-m10-bmc.h
@@ -251,6 +251,7 @@ struct intel_m10bmc {
*
* m10bmc_raw_read - read m10bmc register per addr
* m10bmc_sys_read - read m10bmc system register per offset
+ * m10bmc_sys_update_bits - update m10bmc system register per offset
*/
static inline int
m10bmc_raw_read(struct intel_m10bmc *m10bmc, unsigned int addr,
@@ -282,6 +283,9 @@ static inline int m10bmc_sys_read(struct intel_m10bmc *m10bmc, unsigned int offs
return m10bmc_raw_read(m10bmc, csr_map->base + offset, val);
}
+int m10bmc_sys_update_bits(struct intel_m10bmc *m10bmc, unsigned int offset,
+ unsigned int msk, unsigned int val);
+
/*
* MAX10 BMC Core support
*/
--
2.30.2
Move m10bmc_sys_read() out from the header to prepare it for adding
more code into the function which would make it too large to be a
static inline any more.
While at it, replace the vague wording in function comment with more
precise statements.
Signed-off-by: Ilpo Järvinen <[email protected]>
---
drivers/hwmon/intel-m10-bmc-hwmon.c | 1 +
drivers/mfd/intel-m10-bmc-core.c | 14 ++++++++++++++
include/linux/mfd/intel-m10-bmc.h | 17 +----------------
3 files changed, 16 insertions(+), 16 deletions(-)
diff --git a/drivers/hwmon/intel-m10-bmc-hwmon.c b/drivers/hwmon/intel-m10-bmc-hwmon.c
index 2f0323c14bab..92900ce7986b 100644
--- a/drivers/hwmon/intel-m10-bmc-hwmon.c
+++ b/drivers/hwmon/intel-m10-bmc-hwmon.c
@@ -794,3 +794,4 @@ MODULE_DEVICE_TABLE(platform, intel_m10bmc_hwmon_ids);
MODULE_AUTHOR("Intel Corporation");
MODULE_DESCRIPTION("Intel MAX 10 BMC hardware monitor");
MODULE_LICENSE("GPL");
+MODULE_IMPORT_NS(INTEL_M10_BMC_CORE);
diff --git a/drivers/mfd/intel-m10-bmc-core.c b/drivers/mfd/intel-m10-bmc-core.c
index 879d98b9b14d..4a1bfe135293 100644
--- a/drivers/mfd/intel-m10-bmc-core.c
+++ b/drivers/mfd/intel-m10-bmc-core.c
@@ -12,6 +12,20 @@
#include <linux/mfd/intel-m10-bmc.h>
#include <linux/module.h>
+/*
+ * This function helps to simplify the accessing of the system registers.
+ *
+ * The base of the system registers is configured through the struct
+ * csr_map.
+ */
+int m10bmc_sys_read(struct intel_m10bmc *m10bmc, unsigned int offset, unsigned int *val)
+{
+ const struct m10bmc_csr_map *csr_map = m10bmc->info->csr_map;
+
+ return m10bmc_raw_read(m10bmc, csr_map->base + offset, val);
+}
+EXPORT_SYMBOL_NS_GPL(m10bmc_sys_read, INTEL_M10_BMC_CORE);
+
int m10bmc_sys_update_bits(struct intel_m10bmc *m10bmc, unsigned int offset,
unsigned int msk, unsigned int val)
{
diff --git a/include/linux/mfd/intel-m10-bmc.h b/include/linux/mfd/intel-m10-bmc.h
index 5418f7279ed0..252644fa61be 100644
--- a/include/linux/mfd/intel-m10-bmc.h
+++ b/include/linux/mfd/intel-m10-bmc.h
@@ -267,22 +267,7 @@ m10bmc_raw_read(struct intel_m10bmc *m10bmc, unsigned int addr,
return ret;
}
-/*
- * The base of the system registers could be configured by HW developers, and
- * in HW SPEC, the base is not added to the addresses of the system registers.
- *
- * This function helps to simplify the accessing of the system registers. And if
- * the base is reconfigured in HW, SW developers could simply change the
- * csr_map's base accordingly.
- */
-static inline int m10bmc_sys_read(struct intel_m10bmc *m10bmc, unsigned int offset,
- unsigned int *val)
-{
- const struct m10bmc_csr_map *csr_map = m10bmc->info->csr_map;
-
- return m10bmc_raw_read(m10bmc, csr_map->base + offset, val);
-}
-
+int m10bmc_sys_read(struct intel_m10bmc *m10bmc, unsigned int offset, unsigned int *val);
int m10bmc_sys_update_bits(struct intel_m10bmc *m10bmc, unsigned int offset,
unsigned int msk, unsigned int val);
--
2.30.2
On some MAX 10 cards, the BMC firmware is not available to service
handshake registers during secure update erase and write phases at
normal speeds. This problem affects at least hwmon driver. When the MAX
10 hwmon driver tries to read the sensor values during a secure update,
the reads are slowed down (e.g., reading all D5005 sensors takes ~24s
which is magnitudes worse than the normal <0.02s).
Manage access to the handshake registers using a rw semaphore and a FW
state variable to prevent accesses during those secure update phases
and return -EBUSY instead.
Co-developed-by: Russ Weight <[email protected]>
Signed-off-by: Russ Weight <[email protected]>
Co-developed-by: Xu Yilun <[email protected]>
Signed-off-by: Xu Yilun <[email protected]>
Signed-off-by: Ilpo Järvinen <[email protected]>
---
drivers/fpga/intel-m10-bmc-sec-update.c | 17 +++++--
drivers/mfd/intel-m10-bmc-core.c | 63 ++++++++++++++++++++++++-
drivers/mfd/intel-m10-bmc-pmci.c | 4 ++
drivers/mfd/intel-m10-bmc-spi.c | 14 ++++++
include/linux/mfd/intel-m10-bmc.h | 27 +++++++++++
5 files changed, 120 insertions(+), 5 deletions(-)
diff --git a/drivers/fpga/intel-m10-bmc-sec-update.c b/drivers/fpga/intel-m10-bmc-sec-update.c
index fe0127a58eff..31af2e08c825 100644
--- a/drivers/fpga/intel-m10-bmc-sec-update.c
+++ b/drivers/fpga/intel-m10-bmc-sec-update.c
@@ -544,21 +544,28 @@ static enum fw_upload_err m10bmc_sec_prepare(struct fw_upload *fwl,
if (ret != FW_UPLOAD_ERR_NONE)
goto unlock_flash;
+ m10bmc_fw_state_set(sec->m10bmc, M10BMC_FW_STATE_SEC_UPDATE_PREPARE);
+
ret = rsu_update_init(sec);
if (ret != FW_UPLOAD_ERR_NONE)
- goto unlock_flash;
+ goto fw_state_exit;
ret = rsu_prog_ready(sec);
if (ret != FW_UPLOAD_ERR_NONE)
- goto unlock_flash;
+ goto fw_state_exit;
if (sec->cancel_request) {
ret = rsu_cancel(sec);
- goto unlock_flash;
+ goto fw_state_exit;
}
+ m10bmc_fw_state_set(sec->m10bmc, M10BMC_FW_STATE_SEC_UPDATE_WRITE);
+
return FW_UPLOAD_ERR_NONE;
+fw_state_exit:
+ m10bmc_fw_state_set(sec->m10bmc, M10BMC_FW_STATE_NORMAL);
+
unlock_flash:
if (sec->m10bmc->flash_bulk_ops)
sec->m10bmc->flash_bulk_ops->unlock_write(sec->m10bmc);
@@ -607,6 +614,8 @@ static enum fw_upload_err m10bmc_sec_poll_complete(struct fw_upload *fwl)
if (sec->cancel_request)
return rsu_cancel(sec);
+ m10bmc_fw_state_set(sec->m10bmc, M10BMC_FW_STATE_SEC_UPDATE_PROGRAM);
+
result = rsu_send_data(sec);
if (result != FW_UPLOAD_ERR_NONE)
return result;
@@ -650,6 +659,8 @@ static void m10bmc_sec_cleanup(struct fw_upload *fwl)
(void)rsu_cancel(sec);
+ m10bmc_fw_state_set(sec->m10bmc, M10BMC_FW_STATE_NORMAL);
+
if (sec->m10bmc->flash_bulk_ops)
sec->m10bmc->flash_bulk_ops->unlock_write(sec->m10bmc);
}
diff --git a/drivers/mfd/intel-m10-bmc-core.c b/drivers/mfd/intel-m10-bmc-core.c
index 4a1bfe135293..ae976fe0baed 100644
--- a/drivers/mfd/intel-m10-bmc-core.c
+++ b/drivers/mfd/intel-m10-bmc-core.c
@@ -12,6 +12,42 @@
#include <linux/mfd/intel-m10-bmc.h>
#include <linux/module.h>
+void m10bmc_fw_state_set(struct intel_m10bmc *m10bmc, enum m10bmc_fw_state new_state)
+{
+ down_write(&m10bmc->bmcfw_lock);
+ m10bmc->bmcfw_state = new_state;
+ up_write(&m10bmc->bmcfw_lock);
+}
+EXPORT_SYMBOL_NS_GPL(m10bmc_fw_state_set, INTEL_M10_BMC_CORE);
+
+/*
+ * For some Intel FPGA devices, the BMC firmware is not available to service
+ * handshake registers during a secure update.
+ */
+static bool m10bmc_reg_always_available(struct intel_m10bmc *m10bmc, unsigned int offset)
+{
+ if (!m10bmc->info->handshake_sys_reg_nranges)
+ return true;
+
+ return !regmap_reg_in_ranges(offset, m10bmc->info->handshake_sys_reg_ranges,
+ m10bmc->info->handshake_sys_reg_nranges);
+}
+
+/*
+ * m10bmc_handshake_reg_unavailable - Checks if reg access collides with secure update state
+ * @m10bmc: M10 BMC structure
+ *
+ * For some Intel FPGA devices, the BMC firmware is not available to service
+ * handshake registers during a secure update erase and write phases.
+ *
+ * Context: @m10bmc->bmcfw_lock must be held.
+ */
+static bool m10bmc_handshake_reg_unavailable(struct intel_m10bmc *m10bmc)
+{
+ return m10bmc->bmcfw_state == M10BMC_FW_STATE_SEC_UPDATE_PREPARE ||
+ m10bmc->bmcfw_state == M10BMC_FW_STATE_SEC_UPDATE_WRITE;
+}
+
/*
* This function helps to simplify the accessing of the system registers.
*
@@ -21,8 +57,19 @@
int m10bmc_sys_read(struct intel_m10bmc *m10bmc, unsigned int offset, unsigned int *val)
{
const struct m10bmc_csr_map *csr_map = m10bmc->info->csr_map;
+ int ret;
+
+ if (m10bmc_reg_always_available(m10bmc, offset))
+ return m10bmc_raw_read(m10bmc, csr_map->base + offset, val);
- return m10bmc_raw_read(m10bmc, csr_map->base + offset, val);
+ down_read(&m10bmc->bmcfw_lock);
+ if (m10bmc_handshake_reg_unavailable(m10bmc))
+ ret = -EBUSY; /* Reg not available during secure update */
+ else
+ ret = m10bmc_raw_read(m10bmc, csr_map->base + offset, val);
+ up_read(&m10bmc->bmcfw_lock);
+
+ return ret;
}
EXPORT_SYMBOL_NS_GPL(m10bmc_sys_read, INTEL_M10_BMC_CORE);
@@ -30,8 +77,19 @@ int m10bmc_sys_update_bits(struct intel_m10bmc *m10bmc, unsigned int offset,
unsigned int msk, unsigned int val)
{
const struct m10bmc_csr_map *csr_map = m10bmc->info->csr_map;
+ int ret;
- return regmap_update_bits(m10bmc->regmap, csr_map->base + offset, msk, val);
+ if (m10bmc_reg_always_available(m10bmc, offset))
+ return regmap_update_bits(m10bmc->regmap, csr_map->base + offset, msk, val);
+
+ down_read(&m10bmc->bmcfw_lock);
+ if (m10bmc_handshake_reg_unavailable(m10bmc))
+ ret = -EBUSY; /* Reg not available during secure update */
+ else
+ ret = regmap_update_bits(m10bmc->regmap, csr_map->base + offset, msk, val);
+ up_read(&m10bmc->bmcfw_lock);
+
+ return ret;
}
EXPORT_SYMBOL_NS_GPL(m10bmc_sys_update_bits, INTEL_M10_BMC_CORE);
@@ -129,6 +187,7 @@ int m10bmc_dev_init(struct intel_m10bmc *m10bmc, const struct intel_m10bmc_platf
m10bmc->info = info;
dev_set_drvdata(m10bmc->dev, m10bmc);
+ init_rwsem(&m10bmc->bmcfw_lock);
ret = devm_mfd_add_devices(m10bmc->dev, PLATFORM_DEVID_AUTO,
info->cells, info->n_cells,
diff --git a/drivers/mfd/intel-m10-bmc-pmci.c b/drivers/mfd/intel-m10-bmc-pmci.c
index 0392ef8b57d8..8cdbabe37708 100644
--- a/drivers/mfd/intel-m10-bmc-pmci.c
+++ b/drivers/mfd/intel-m10-bmc-pmci.c
@@ -352,6 +352,8 @@ static struct mfd_cell m10bmc_pmci_n6000_bmc_subdevs[] = {
{ .name = "n6000bmc-sec-update" },
};
+static const struct regmap_range null_fw_handshake_regs[0];
+
static const struct m10bmc_csr_map m10bmc_n6000_csr_map = {
.base = M10BMC_N6000_SYS_BASE,
.build_version = M10BMC_N6000_BUILD_VER,
@@ -375,6 +377,8 @@ static const struct m10bmc_csr_map m10bmc_n6000_csr_map = {
static const struct intel_m10bmc_platform_info m10bmc_pmci_n6000 = {
.cells = m10bmc_pmci_n6000_bmc_subdevs,
.n_cells = ARRAY_SIZE(m10bmc_pmci_n6000_bmc_subdevs),
+ .handshake_sys_reg_ranges = null_fw_handshake_regs,
+ .handshake_sys_reg_nranges = 0,
.csr_map = &m10bmc_n6000_csr_map,
};
diff --git a/drivers/mfd/intel-m10-bmc-spi.c b/drivers/mfd/intel-m10-bmc-spi.c
index edd266557ab9..cbeb7de9e041 100644
--- a/drivers/mfd/intel-m10-bmc-spi.c
+++ b/drivers/mfd/intel-m10-bmc-spi.c
@@ -116,12 +116,20 @@ static struct mfd_cell m10bmc_d5005_subdevs[] = {
{ .name = "d5005bmc-sec-update" },
};
+static const struct regmap_range m10bmc_d5005_fw_handshake_regs[] = {
+ regmap_reg_range(M10BMC_N3000_TELEM_START, M10BMC_D5005_TELEM_END),
+};
+
static struct mfd_cell m10bmc_pacn3000_subdevs[] = {
{ .name = "n3000bmc-hwmon" },
{ .name = "n3000bmc-retimer" },
{ .name = "n3000bmc-sec-update" },
};
+static const struct regmap_range m10bmc_n3000_fw_handshake_regs[] = {
+ regmap_reg_range(M10BMC_N3000_TELEM_START, M10BMC_N3000_TELEM_END),
+};
+
static struct mfd_cell m10bmc_n5010_subdevs[] = {
{ .name = "n5010bmc-hwmon" },
};
@@ -129,18 +137,24 @@ static struct mfd_cell m10bmc_n5010_subdevs[] = {
static const struct intel_m10bmc_platform_info m10bmc_spi_n3000 = {
.cells = m10bmc_pacn3000_subdevs,
.n_cells = ARRAY_SIZE(m10bmc_pacn3000_subdevs),
+ .handshake_sys_reg_ranges = m10bmc_n3000_fw_handshake_regs,
+ .handshake_sys_reg_nranges = ARRAY_SIZE(m10bmc_n3000_fw_handshake_regs),
.csr_map = &m10bmc_n3000_csr_map,
};
static const struct intel_m10bmc_platform_info m10bmc_spi_d5005 = {
.cells = m10bmc_d5005_subdevs,
.n_cells = ARRAY_SIZE(m10bmc_d5005_subdevs),
+ .handshake_sys_reg_ranges = m10bmc_d5005_fw_handshake_regs,
+ .handshake_sys_reg_nranges = ARRAY_SIZE(m10bmc_d5005_fw_handshake_regs),
.csr_map = &m10bmc_n3000_csr_map,
};
static const struct intel_m10bmc_platform_info m10bmc_spi_n5010 = {
.cells = m10bmc_n5010_subdevs,
.n_cells = ARRAY_SIZE(m10bmc_n5010_subdevs),
+ .handshake_sys_reg_ranges = m10bmc_n3000_fw_handshake_regs,
+ .handshake_sys_reg_nranges = ARRAY_SIZE(m10bmc_n3000_fw_handshake_regs),
.csr_map = &m10bmc_n3000_csr_map,
};
diff --git a/include/linux/mfd/intel-m10-bmc.h b/include/linux/mfd/intel-m10-bmc.h
index 252644fa61be..afa634c9981a 100644
--- a/include/linux/mfd/intel-m10-bmc.h
+++ b/include/linux/mfd/intel-m10-bmc.h
@@ -11,6 +11,7 @@
#include <linux/bits.h>
#include <linux/dev_printk.h>
#include <linux/regmap.h>
+#include <linux/rwsem.h>
#define M10BMC_N3000_LEGACY_BUILD_VER 0x300468
#define M10BMC_N3000_SYS_BASE 0x300800
@@ -39,6 +40,11 @@
#define M10BMC_N3000_VER_PCB_INFO_MSK GENMASK(31, 24)
#define M10BMC_N3000_VER_LEGACY_INVALID 0xffffffff
+/* Telemetry registers */
+#define M10BMC_N3000_TELEM_START 0x100
+#define M10BMC_N3000_TELEM_END 0x250
+#define M10BMC_D5005_TELEM_END 0x300
+
/* Secure update doorbell register, in system register region */
#define M10BMC_N3000_DOORBELL 0x400
@@ -205,11 +211,15 @@ struct m10bmc_csr_map {
* struct intel_m10bmc_platform_info - Intel MAX 10 BMC platform specific information
* @cells: MFD cells
* @n_cells: MFD cells ARRAY_SIZE()
+ * @handshake_sys_reg_ranges: array of register ranges for fw handshake regs
+ * @handshake_sys_reg_nranges: number of register ranges for fw handshake regs
* @csr_map: the mappings for register definition of MAX10 BMC
*/
struct intel_m10bmc_platform_info {
struct mfd_cell *cells;
int n_cells;
+ const struct regmap_range *handshake_sys_reg_ranges;
+ unsigned int handshake_sys_reg_nranges;
const struct m10bmc_csr_map *csr_map;
};
@@ -232,18 +242,29 @@ struct intel_m10bmc_flash_bulk_ops {
void (*unlock_write)(struct intel_m10bmc *m10bmc);
};
+enum m10bmc_fw_state {
+ M10BMC_FW_STATE_NORMAL,
+ M10BMC_FW_STATE_SEC_UPDATE_PREPARE,
+ M10BMC_FW_STATE_SEC_UPDATE_WRITE,
+ M10BMC_FW_STATE_SEC_UPDATE_PROGRAM,
+};
+
/**
* struct intel_m10bmc - Intel MAX 10 BMC parent driver data structure
* @dev: this device
* @regmap: the regmap used to access registers by m10bmc itself
* @info: the platform information for MAX10 BMC
* @flash_bulk_ops: optional device specific operations for flash R/W
+ * @bmcfw_lock: read/write semaphore to BMC firmware running state
+ * @bmcfw_state: BMC firmware running state
*/
struct intel_m10bmc {
struct device *dev;
struct regmap *regmap;
const struct intel_m10bmc_platform_info *info;
const struct intel_m10bmc_flash_bulk_ops *flash_bulk_ops;
+ struct rw_semaphore bmcfw_lock; /* Protects bmcfw_state */
+ enum m10bmc_fw_state bmcfw_state;
};
/*
@@ -271,6 +292,12 @@ int m10bmc_sys_read(struct intel_m10bmc *m10bmc, unsigned int offset, unsigned i
int m10bmc_sys_update_bits(struct intel_m10bmc *m10bmc, unsigned int offset,
unsigned int msk, unsigned int val);
+/*
+ * Track the state of the firmware, as it is not available for register
+ * handshakes during secure updates on some MAX 10 cards.
+ */
+void m10bmc_fw_state_set(struct intel_m10bmc *m10bmc, enum m10bmc_fw_state new_state);
+
/*
* MAX10 BMC Core support
*/
--
2.30.2
On 4/5/23 01:01, Ilpo Järvinen wrote:
> Move m10bmc_sys_read() out from the header to prepare it for adding
> more code into the function which would make it too large to be a
> static inline any more.
>
> While at it, replace the vague wording in function comment with more
> precise statements.
>
> Signed-off-by: Ilpo Järvinen <[email protected]>
> ---
> drivers/hwmon/intel-m10-bmc-hwmon.c | 1 +
For hwmon:
Acked-by: Guenter Roeck <[email protected]>
> drivers/mfd/intel-m10-bmc-core.c | 14 ++++++++++++++
> include/linux/mfd/intel-m10-bmc.h | 17 +----------------
> 3 files changed, 16 insertions(+), 16 deletions(-)
>
> diff --git a/drivers/hwmon/intel-m10-bmc-hwmon.c b/drivers/hwmon/intel-m10-bmc-hwmon.c
> index 2f0323c14bab..92900ce7986b 100644
> --- a/drivers/hwmon/intel-m10-bmc-hwmon.c
> +++ b/drivers/hwmon/intel-m10-bmc-hwmon.c
> @@ -794,3 +794,4 @@ MODULE_DEVICE_TABLE(platform, intel_m10bmc_hwmon_ids);
> MODULE_AUTHOR("Intel Corporation");
> MODULE_DESCRIPTION("Intel MAX 10 BMC hardware monitor");
> MODULE_LICENSE("GPL");
> +MODULE_IMPORT_NS(INTEL_M10_BMC_CORE);
> diff --git a/drivers/mfd/intel-m10-bmc-core.c b/drivers/mfd/intel-m10-bmc-core.c
> index 879d98b9b14d..4a1bfe135293 100644
> --- a/drivers/mfd/intel-m10-bmc-core.c
> +++ b/drivers/mfd/intel-m10-bmc-core.c
> @@ -12,6 +12,20 @@
> #include <linux/mfd/intel-m10-bmc.h>
> #include <linux/module.h>
>
> +/*
> + * This function helps to simplify the accessing of the system registers.
> + *
> + * The base of the system registers is configured through the struct
> + * csr_map.
> + */
> +int m10bmc_sys_read(struct intel_m10bmc *m10bmc, unsigned int offset, unsigned int *val)
> +{
> + const struct m10bmc_csr_map *csr_map = m10bmc->info->csr_map;
> +
> + return m10bmc_raw_read(m10bmc, csr_map->base + offset, val);
> +}
> +EXPORT_SYMBOL_NS_GPL(m10bmc_sys_read, INTEL_M10_BMC_CORE);
> +
> int m10bmc_sys_update_bits(struct intel_m10bmc *m10bmc, unsigned int offset,
> unsigned int msk, unsigned int val)
> {
> diff --git a/include/linux/mfd/intel-m10-bmc.h b/include/linux/mfd/intel-m10-bmc.h
> index 5418f7279ed0..252644fa61be 100644
> --- a/include/linux/mfd/intel-m10-bmc.h
> +++ b/include/linux/mfd/intel-m10-bmc.h
> @@ -267,22 +267,7 @@ m10bmc_raw_read(struct intel_m10bmc *m10bmc, unsigned int addr,
> return ret;
> }
>
> -/*
> - * The base of the system registers could be configured by HW developers, and
> - * in HW SPEC, the base is not added to the addresses of the system registers.
> - *
> - * This function helps to simplify the accessing of the system registers. And if
> - * the base is reconfigured in HW, SW developers could simply change the
> - * csr_map's base accordingly.
> - */
> -static inline int m10bmc_sys_read(struct intel_m10bmc *m10bmc, unsigned int offset,
> - unsigned int *val)
> -{
> - const struct m10bmc_csr_map *csr_map = m10bmc->info->csr_map;
> -
> - return m10bmc_raw_read(m10bmc, csr_map->base + offset, val);
> -}
> -
> +int m10bmc_sys_read(struct intel_m10bmc *m10bmc, unsigned int offset, unsigned int *val);
> int m10bmc_sys_update_bits(struct intel_m10bmc *m10bmc, unsigned int offset,
> unsigned int msk, unsigned int val);
>
On 4/5/23 01:01, Ilpo Järvinen wrote:
> Move m10bmc_sys_read() out from the header to prepare it for adding
> more code into the function which would make it too large to be a
> static inline any more.
>
> While at it, replace the vague wording in function comment with more
> precise statements.
Reviewed-by: Russ Weight <[email protected]>
>
> Signed-off-by: Ilpo Järvinen <[email protected]>
> ---
> drivers/hwmon/intel-m10-bmc-hwmon.c | 1 +
> drivers/mfd/intel-m10-bmc-core.c | 14 ++++++++++++++
> include/linux/mfd/intel-m10-bmc.h | 17 +----------------
> 3 files changed, 16 insertions(+), 16 deletions(-)
>
> diff --git a/drivers/hwmon/intel-m10-bmc-hwmon.c b/drivers/hwmon/intel-m10-bmc-hwmon.c
> index 2f0323c14bab..92900ce7986b 100644
> --- a/drivers/hwmon/intel-m10-bmc-hwmon.c
> +++ b/drivers/hwmon/intel-m10-bmc-hwmon.c
> @@ -794,3 +794,4 @@ MODULE_DEVICE_TABLE(platform, intel_m10bmc_hwmon_ids);
> MODULE_AUTHOR("Intel Corporation");
> MODULE_DESCRIPTION("Intel MAX 10 BMC hardware monitor");
> MODULE_LICENSE("GPL");
> +MODULE_IMPORT_NS(INTEL_M10_BMC_CORE);
> diff --git a/drivers/mfd/intel-m10-bmc-core.c b/drivers/mfd/intel-m10-bmc-core.c
> index 879d98b9b14d..4a1bfe135293 100644
> --- a/drivers/mfd/intel-m10-bmc-core.c
> +++ b/drivers/mfd/intel-m10-bmc-core.c
> @@ -12,6 +12,20 @@
> #include <linux/mfd/intel-m10-bmc.h>
> #include <linux/module.h>
>
> +/*
> + * This function helps to simplify the accessing of the system registers.
> + *
> + * The base of the system registers is configured through the struct
> + * csr_map.
> + */
> +int m10bmc_sys_read(struct intel_m10bmc *m10bmc, unsigned int offset, unsigned int *val)
> +{
> + const struct m10bmc_csr_map *csr_map = m10bmc->info->csr_map;
> +
> + return m10bmc_raw_read(m10bmc, csr_map->base + offset, val);
> +}
> +EXPORT_SYMBOL_NS_GPL(m10bmc_sys_read, INTEL_M10_BMC_CORE);
> +
> int m10bmc_sys_update_bits(struct intel_m10bmc *m10bmc, unsigned int offset,
> unsigned int msk, unsigned int val)
> {
> diff --git a/include/linux/mfd/intel-m10-bmc.h b/include/linux/mfd/intel-m10-bmc.h
> index 5418f7279ed0..252644fa61be 100644
> --- a/include/linux/mfd/intel-m10-bmc.h
> +++ b/include/linux/mfd/intel-m10-bmc.h
> @@ -267,22 +267,7 @@ m10bmc_raw_read(struct intel_m10bmc *m10bmc, unsigned int addr,
> return ret;
> }
>
> -/*
> - * The base of the system registers could be configured by HW developers, and
> - * in HW SPEC, the base is not added to the addresses of the system registers.
> - *
> - * This function helps to simplify the accessing of the system registers. And if
> - * the base is reconfigured in HW, SW developers could simply change the
> - * csr_map's base accordingly.
> - */
> -static inline int m10bmc_sys_read(struct intel_m10bmc *m10bmc, unsigned int offset,
> - unsigned int *val)
> -{
> - const struct m10bmc_csr_map *csr_map = m10bmc->info->csr_map;
> -
> - return m10bmc_raw_read(m10bmc, csr_map->base + offset, val);
> -}
> -
> +int m10bmc_sys_read(struct intel_m10bmc *m10bmc, unsigned int offset, unsigned int *val);
> int m10bmc_sys_update_bits(struct intel_m10bmc *m10bmc, unsigned int offset,
> unsigned int msk, unsigned int val);
>
On 2023-04-05 at 11:01:52 +0300, Ilpo J?rvinen wrote:
> On some MAX 10 cards, the BMC firmware is not available to service
> handshake registers during secure update erase and write phases at
> normal speeds. This problem affects at least hwmon driver. When the MAX
> 10 hwmon driver tries to read the sensor values during a secure update,
> the reads are slowed down (e.g., reading all D5005 sensors takes ~24s
> which is magnitudes worse than the normal <0.02s).
>
> Manage access to the handshake registers using a rw semaphore and a FW
> state variable to prevent accesses during those secure update phases
> and return -EBUSY instead.
>
> Co-developed-by: Russ Weight <[email protected]>
> Signed-off-by: Russ Weight <[email protected]>
> Co-developed-by: Xu Yilun <[email protected]>
> Signed-off-by: Xu Yilun <[email protected]>
> Signed-off-by: Ilpo J?rvinen <[email protected]>
> ---
> drivers/fpga/intel-m10-bmc-sec-update.c | 17 +++++--
> drivers/mfd/intel-m10-bmc-core.c | 63 ++++++++++++++++++++++++-
> drivers/mfd/intel-m10-bmc-pmci.c | 4 ++
> drivers/mfd/intel-m10-bmc-spi.c | 14 ++++++
> include/linux/mfd/intel-m10-bmc.h | 27 +++++++++++
> 5 files changed, 120 insertions(+), 5 deletions(-)
>
[...]
>
> +static const struct regmap_range null_fw_handshake_regs[0];
> +
> static const struct m10bmc_csr_map m10bmc_n6000_csr_map = {
> .base = M10BMC_N6000_SYS_BASE,
> .build_version = M10BMC_N6000_BUILD_VER,
> @@ -375,6 +377,8 @@ static const struct m10bmc_csr_map m10bmc_n6000_csr_map = {
> static const struct intel_m10bmc_platform_info m10bmc_pmci_n6000 = {
> .cells = m10bmc_pmci_n6000_bmc_subdevs,
> .n_cells = ARRAY_SIZE(m10bmc_pmci_n6000_bmc_subdevs),
> + .handshake_sys_reg_ranges = null_fw_handshake_regs,
> + .handshake_sys_reg_nranges = 0,
Not sure why a zero length array is needed? Could we just remove
these 2 lines?
Thanks,
Yilun
On 2023-04-05 at 11:01:52 +0300, Ilpo J?rvinen wrote:
> On some MAX 10 cards, the BMC firmware is not available to service
> handshake registers during secure update erase and write phases at
> normal speeds. This problem affects at least hwmon driver. When the MAX
> 10 hwmon driver tries to read the sensor values during a secure update,
> the reads are slowed down (e.g., reading all D5005 sensors takes ~24s
> which is magnitudes worse than the normal <0.02s).
>
> Manage access to the handshake registers using a rw semaphore and a FW
> state variable to prevent accesses during those secure update phases
> and return -EBUSY instead.
>
> Co-developed-by: Russ Weight <[email protected]>
> Signed-off-by: Russ Weight <[email protected]>
> Co-developed-by: Xu Yilun <[email protected]>
> Signed-off-by: Xu Yilun <[email protected]>
> Signed-off-by: Ilpo J?rvinen <[email protected]>
> ---
> drivers/fpga/intel-m10-bmc-sec-update.c | 17 +++++--
> drivers/mfd/intel-m10-bmc-core.c | 63 ++++++++++++++++++++++++-
> drivers/mfd/intel-m10-bmc-pmci.c | 4 ++
> drivers/mfd/intel-m10-bmc-spi.c | 14 ++++++
> include/linux/mfd/intel-m10-bmc.h | 27 +++++++++++
> 5 files changed, 120 insertions(+), 5 deletions(-)
>
[...]
>
> +void m10bmc_fw_state_set(struct intel_m10bmc *m10bmc, enum m10bmc_fw_state new_state)
> +{
> + down_write(&m10bmc->bmcfw_lock);
> + m10bmc->bmcfw_state = new_state;
> + up_write(&m10bmc->bmcfw_lock);
Could we also skip this if no handshake is possible like for PMCI?
Thanks,
Yilun
On Fri, 7 Apr 2023, Xu Yilun wrote:
> On 2023-04-05 at 11:01:52 +0300, Ilpo J?rvinen wrote:
> > On some MAX 10 cards, the BMC firmware is not available to service
> > handshake registers during secure update erase and write phases at
> > normal speeds. This problem affects at least hwmon driver. When the MAX
> > 10 hwmon driver tries to read the sensor values during a secure update,
> > the reads are slowed down (e.g., reading all D5005 sensors takes ~24s
> > which is magnitudes worse than the normal <0.02s).
> >
> > Manage access to the handshake registers using a rw semaphore and a FW
> > state variable to prevent accesses during those secure update phases
> > and return -EBUSY instead.
> >
> > Co-developed-by: Russ Weight <[email protected]>
> > Signed-off-by: Russ Weight <[email protected]>
> > Co-developed-by: Xu Yilun <[email protected]>
> > Signed-off-by: Xu Yilun <[email protected]>
> > Signed-off-by: Ilpo J?rvinen <[email protected]>
> > ---
> > drivers/fpga/intel-m10-bmc-sec-update.c | 17 +++++--
> > drivers/mfd/intel-m10-bmc-core.c | 63 ++++++++++++++++++++++++-
> > drivers/mfd/intel-m10-bmc-pmci.c | 4 ++
> > drivers/mfd/intel-m10-bmc-spi.c | 14 ++++++
> > include/linux/mfd/intel-m10-bmc.h | 27 +++++++++++
> > 5 files changed, 120 insertions(+), 5 deletions(-)
> >
>
> [...]
>
> >
> > +static const struct regmap_range null_fw_handshake_regs[0];
> > +
> > static const struct m10bmc_csr_map m10bmc_n6000_csr_map = {
> > .base = M10BMC_N6000_SYS_BASE,
> > .build_version = M10BMC_N6000_BUILD_VER,
> > @@ -375,6 +377,8 @@ static const struct m10bmc_csr_map m10bmc_n6000_csr_map = {
> > static const struct intel_m10bmc_platform_info m10bmc_pmci_n6000 = {
> > .cells = m10bmc_pmci_n6000_bmc_subdevs,
> > .n_cells = ARRAY_SIZE(m10bmc_pmci_n6000_bmc_subdevs),
> > + .handshake_sys_reg_ranges = null_fw_handshake_regs,
> > + .handshake_sys_reg_nranges = 0,
>
> Not sure why a zero length array is needed? Could we just remove
> these 2 lines?
It seems to be safe to remove them so I dropped it.
--
i.
On Fri, 7 Apr 2023, Xu Yilun wrote:
> On 2023-04-05 at 11:01:52 +0300, Ilpo J?rvinen wrote:
> > On some MAX 10 cards, the BMC firmware is not available to service
> > handshake registers during secure update erase and write phases at
> > normal speeds. This problem affects at least hwmon driver. When the MAX
> > 10 hwmon driver tries to read the sensor values during a secure update,
> > the reads are slowed down (e.g., reading all D5005 sensors takes ~24s
> > which is magnitudes worse than the normal <0.02s).
> >
> > Manage access to the handshake registers using a rw semaphore and a FW
> > state variable to prevent accesses during those secure update phases
> > and return -EBUSY instead.
> >
> > Co-developed-by: Russ Weight <[email protected]>
> > Signed-off-by: Russ Weight <[email protected]>
> > Co-developed-by: Xu Yilun <[email protected]>
> > Signed-off-by: Xu Yilun <[email protected]>
> > Signed-off-by: Ilpo J?rvinen <[email protected]>
> > ---
> > drivers/fpga/intel-m10-bmc-sec-update.c | 17 +++++--
> > drivers/mfd/intel-m10-bmc-core.c | 63 ++++++++++++++++++++++++-
> > drivers/mfd/intel-m10-bmc-pmci.c | 4 ++
> > drivers/mfd/intel-m10-bmc-spi.c | 14 ++++++
> > include/linux/mfd/intel-m10-bmc.h | 27 +++++++++++
> > 5 files changed, 120 insertions(+), 5 deletions(-)
> >
>
> [...]
>
> >
> > +void m10bmc_fw_state_set(struct intel_m10bmc *m10bmc, enum m10bmc_fw_state new_state)
> > +{
> > + down_write(&m10bmc->bmcfw_lock);
> > + m10bmc->bmcfw_state = new_state;
> > + up_write(&m10bmc->bmcfw_lock);
>
> Could we also skip this if no handshake is possible like for PMCI?
Did you mean guarding it with !m10bmc->info->handshake_sys_reg_nranges ?
If yes, it's doable (+ I'd add comment mentioning it since it's slightly
trappy to not always have that state updated).
--
i.
On 2023-04-11 at 14:54:58 +0300, Ilpo J?rvinen wrote:
> On Fri, 7 Apr 2023, Xu Yilun wrote:
>
> > On 2023-04-05 at 11:01:52 +0300, Ilpo J?rvinen wrote:
> > > On some MAX 10 cards, the BMC firmware is not available to service
> > > handshake registers during secure update erase and write phases at
> > > normal speeds. This problem affects at least hwmon driver. When the MAX
> > > 10 hwmon driver tries to read the sensor values during a secure update,
> > > the reads are slowed down (e.g., reading all D5005 sensors takes ~24s
> > > which is magnitudes worse than the normal <0.02s).
> > >
> > > Manage access to the handshake registers using a rw semaphore and a FW
> > > state variable to prevent accesses during those secure update phases
> > > and return -EBUSY instead.
> > >
> > > Co-developed-by: Russ Weight <[email protected]>
> > > Signed-off-by: Russ Weight <[email protected]>
> > > Co-developed-by: Xu Yilun <[email protected]>
> > > Signed-off-by: Xu Yilun <[email protected]>
> > > Signed-off-by: Ilpo J?rvinen <[email protected]>
> > > ---
> > > drivers/fpga/intel-m10-bmc-sec-update.c | 17 +++++--
> > > drivers/mfd/intel-m10-bmc-core.c | 63 ++++++++++++++++++++++++-
> > > drivers/mfd/intel-m10-bmc-pmci.c | 4 ++
> > > drivers/mfd/intel-m10-bmc-spi.c | 14 ++++++
> > > include/linux/mfd/intel-m10-bmc.h | 27 +++++++++++
> > > 5 files changed, 120 insertions(+), 5 deletions(-)
> > >
> >
> > [...]
> >
> > >
> > > +void m10bmc_fw_state_set(struct intel_m10bmc *m10bmc, enum m10bmc_fw_state new_state)
> > > +{
> > > + down_write(&m10bmc->bmcfw_lock);
> > > + m10bmc->bmcfw_state = new_state;
> > > + up_write(&m10bmc->bmcfw_lock);
> >
> > Could we also skip this if no handshake is possible like for PMCI?
>
> Did you mean guarding it with !m10bmc->info->handshake_sys_reg_nranges ?
Yes. My concern is, the handshake_sys_reg_nranges is the constant value
for a device, so if the device doesn't have handshake registers, we
could save the locking costs.
Thanks,
Yilun
> If yes, it's doable (+ I'd add comment mentioning it since it's slightly
> trappy to not always have that state updated).
>
>
> --
> i.