Re: [PATCH 14/14] scsi: ufs: qcom: Add support for scaling interconnects
From: Dmitry Baryshkov
Date: Wed Jul 12 2023 - 13:23:10 EST
On 12/07/2023 19:41, Manivannan Sadhasivam wrote:
On Wed, Jul 12, 2023 at 04:22:51PM +0300, Dmitry Baryshkov wrote:
On 12/07/2023 13:32, Manivannan Sadhasivam wrote:
Qcom SoCs require scaling the interconnect paths for proper working of the
peripherals connected through interconnects. Even for accessing the UFS
controller, someone should setup the interconnect paths. So far, the
bootloaders used to setup the interconnect paths before booting linux as
they need to access the UFS storage for things like fetching boot firmware.
But with the advent of multi boot options, bootloader nowadays like in
SA8540p SoC do not setup the interconnect paths at all.
So trying to configure UFS in the absence of the interconnect path
configuration, results in boot crash.
To fix this issue and also to dynamically scale the interconnects (UFS-DDR
and CPU-UFS), interconnect API support is added to the Qcom UFS driver.
With this support, the interconnect paths are scaled dynamically based on
the gear configuration.
During the early stage of ufs_qcom_init(), ufs_qcom_icc_init() will setup
the paths to max bandwidth to allow configuring the UFS registers. Touching
the registers without configuring the icc paths would result in a crash.
However, we don't really need to set max vote for the icc paths as any
minimal vote would suffice. But the max value would allow initialization to
be done faster. After init, the bandwidth will get updated using
ufs_qcom_icc_update_bw() based on the gear and lane configuration.
The bandwidth values defined in ufs_qcom_bw_table struct are taken from
Qcom downstream vendor devicetree source and are calculated as per the
UFS3.1 Spec, Section 6.4.1, HS Gear Rates. So it is fixed across platforms.
Cc: Brian Masney <bmasney@xxxxxxxxxx>
Signed-off-by: Manivannan Sadhasivam <manivannan.sadhasivam@xxxxxxxxxx>
---
drivers/ufs/host/ufs-qcom.c | 131 +++++++++++++++++++++++++++++++++++-
drivers/ufs/host/ufs-qcom.h | 3 +
2 files changed, 133 insertions(+), 1 deletion(-)
diff --git a/drivers/ufs/host/ufs-qcom.c b/drivers/ufs/host/ufs-qcom.c
index 8d6fd4c3324f..8a3132d45a65 100644
--- a/drivers/ufs/host/ufs-qcom.c
+++ b/drivers/ufs/host/ufs-qcom.c
@@ -7,6 +7,7 @@
#include <linux/time.h>
#include <linux/clk.h>
#include <linux/delay.h>
+#include <linux/interconnect.h>
#include <linux/module.h>
#include <linux/of.h>
#include <linux/platform_device.h>
@@ -46,6 +47,49 @@ enum {
TSTBUS_MAX,
};
+#define QCOM_UFS_MAX_GEAR 4
+#define QCOM_UFS_MAX_LANE 2
+
+enum {
+ MODE_MIN,
+ MODE_PWM,
+ MODE_HS_RA,
+ MODE_HS_RB,
+ MODE_MAX,
MODE_MIN and MODE_MAX seem to be unused
No, they are used in the driver.
Ack, missed that. Will comment inline.
+};
+
+struct __ufs_qcom_bw_table {
+ u32 bw1;
+ u32 bw2;
Please consider:
s/bw1/mem_bw/
s/bw2/cfg_bw/
Ok.
+} ufs_qcom_bw_table[MODE_MAX + 1][QCOM_UFS_MAX_GEAR + 1][QCOM_UFS_MAX_LANE + 1] = {
I'd say, these +1's are slightly confusing and unnecessary.
These max values itself are used in the table. Without adding 1 to them, we will
end up with out of bounds access.
+ [MODE_MIN][0][0] = { 0, 0 }, /* Bandwidth values are in KB/s */
I'd say, this becomes impossible to check. Can you please structure it?
What do you mean by "impossible to check"?
Either by inclusion:
[MODE_PWM] = {
[1] = {
[1] = { .. },
[2] = { .. },
},
// etc.
Now, this seems confusion to me. The existing table looks fine IMO.
Fine with me then, your call!
};
Also, do we have defines for gears? Can we use them instead of indices?
There are defines for the gears but not for lanes. So I ended up using numbers
for simplicity.
My suggestion would be to use them for gears at least. Then it becomes
cleaner (and maybe will solve some of my other comments).
- Mani
+ [MODE_PWM][1][1] = { 922, 1000 },
+ [MODE_PWM][2][1] = { 1844, 1000 },
+ [MODE_PWM][3][1] = { 3688, 1000 },
+ [MODE_PWM][4][1] = { 7376, 1000 },
+ [MODE_PWM][1][2] = { 1844, 1000 },
+ [MODE_PWM][2][2] = { 3688, 1000 },
+ [MODE_PWM][3][2] = { 7376, 1000 },
+ [MODE_PWM][4][2] = { 14752, 1000 },
+ [MODE_HS_RA][1][1] = { 127796, 1000 },
+ [MODE_HS_RA][2][1] = { 255591, 1000 },
+ [MODE_HS_RA][3][1] = { 1492582, 102400 },
+ [MODE_HS_RA][4][1] = { 2915200, 204800 },
+ [MODE_HS_RA][1][2] = { 255591, 1000 },
+ [MODE_HS_RA][2][2] = { 511181, 1000 },
+ [MODE_HS_RA][3][2] = { 1492582, 204800 },
+ [MODE_HS_RA][4][2] = { 2915200, 409600 },
+ [MODE_HS_RB][1][1] = { 149422, 1000 },
+ [MODE_HS_RB][2][1] = { 298189, 1000 },
+ [MODE_HS_RB][3][1] = { 1492582, 102400 },
+ [MODE_HS_RB][4][1] = { 2915200, 204800 },
+ [MODE_HS_RB][1][2] = { 298189, 1000 },
+ [MODE_HS_RB][2][2] = { 596378, 1000 },
+ [MODE_HS_RB][3][2] = { 1492582, 204800 },
+ [MODE_HS_RB][4][2] = { 2915200, 409600 },
+ [MODE_MAX][0][0] = { 7643136, 307200 },
+};
+
static struct ufs_qcom_host *ufs_qcom_hosts[MAX_UFS_QCOM_HOSTS];
static void ufs_qcom_get_default_testbus_cfg(struct ufs_qcom_host *host);
@@ -789,6 +833,51 @@ static void ufs_qcom_dev_ref_clk_ctrl(struct ufs_qcom_host *host, bool enable)
}
}
+static int ufs_qcom_icc_set_bw(struct ufs_qcom_host *host, u32 bw1, u32 bw2)
+{
+ struct device *dev = host->hba->dev;
+ int ret;
+
+ ret = icc_set_bw(host->icc_ddr, 0, bw1);
+ if (ret < 0) {
+ dev_err(dev, "failed to set bandwidth request: %d\n", ret);
+ return ret;
+ }
+
+ ret = icc_set_bw(host->icc_cpu, 0, bw2);
+ if (ret < 0) {
+ dev_err(dev, "failed to set bandwidth request: %d\n", ret);
+ return ret;
+ }
+
+ return 0;
+}
+
+static struct __ufs_qcom_bw_table ufs_qcom_get_bw_table(struct ufs_qcom_host *host)
+{
+ struct ufs_pa_layer_attr *p = &host->dev_req_params;
+ int gear = max_t(u32, p->gear_rx, p->gear_tx);
+ int lane = max_t(u32, p->lane_rx, p->lane_tx);
+
+ if (ufshcd_is_hs_mode(p)) {
+ if (p->hs_rate == PA_HS_MODE_B)
+ return ufs_qcom_bw_table[MODE_HS_RB][gear][lane];
+ else
+ return ufs_qcom_bw_table[MODE_HS_RA][gear][lane];
+ } else {
+ return ufs_qcom_bw_table[MODE_PWM][gear][lane];
+ }
+}
+
+static int ufs_qcom_icc_update_bw(struct ufs_qcom_host *host)
+{
+ struct __ufs_qcom_bw_table bw_table;
+
+ bw_table = ufs_qcom_get_bw_table(host);
+
+ return ufs_qcom_icc_set_bw(host, bw_table.bw1, bw_table.bw2);
+}
+
static int ufs_qcom_pwr_change_notify(struct ufs_hba *hba,
enum ufs_notify_change_status status,
struct ufs_pa_layer_attr *dev_max_params,
@@ -852,6 +941,8 @@ static int ufs_qcom_pwr_change_notify(struct ufs_hba *hba,
memcpy(&host->dev_req_params,
dev_req_params, sizeof(*dev_req_params));
+ ufs_qcom_icc_update_bw(host);
+
/* disable the device ref clock if entered PWM mode */
if (ufshcd_is_hs_mode(&hba->pwr_info) &&
!ufshcd_is_hs_mode(dev_req_params))
@@ -981,7 +1072,9 @@ static int ufs_qcom_setup_clocks(struct ufs_hba *hba, bool on,
switch (status) {
case PRE_CHANGE:
- if (!on) {
+ if (on) {
+ ufs_qcom_icc_update_bw(host);
+ } else {
if (!ufs_qcom_is_link_active(hba)) {
/* disable device ref_clk */
ufs_qcom_dev_ref_clk_ctrl(host, false);
@@ -993,6 +1086,9 @@ static int ufs_qcom_setup_clocks(struct ufs_hba *hba, bool on,
/* enable the device ref clock for HS mode*/
if (ufshcd_is_hs_mode(&hba->pwr_info))
ufs_qcom_dev_ref_clk_ctrl(host, true);
+ } else {
+ ufs_qcom_icc_set_bw(host, ufs_qcom_bw_table[MODE_MIN][0][0].bw1,
+ ufs_qcom_bw_table[MODE_MIN][0][0].bw2);
With MODE_MIN values being initialised to 0, can we use the value
directly instead? You are not defining the whole table for MODE_MIN anyway.
}
break;
}
@@ -1031,6 +1127,34 @@ static const struct reset_control_ops ufs_qcom_reset_ops = {
.deassert = ufs_qcom_reset_deassert,
};
+static int ufs_qcom_icc_init(struct ufs_qcom_host *host)
+{
+ struct device *dev = host->hba->dev;
+ int ret;
+
+ host->icc_ddr = devm_of_icc_get(dev, "ufs-ddr");
+ if (IS_ERR(host->icc_ddr))
+ return dev_err_probe(dev, PTR_ERR(host->icc_ddr),
+ "failed to acquire interconnect path\n");
+
+ host->icc_cpu = devm_of_icc_get(dev, "cpu-ufs");
+ if (IS_ERR(host->icc_cpu))
+ return dev_err_probe(dev, PTR_ERR(host->icc_cpu),
+ "failed to acquire interconnect path\n");
+
+ /*
+ * Set Maximum bandwidth vote before initializing the UFS controller and
+ * device. Ideally, a minimal interconnect vote would suffice for the
+ * initialization, but a max vote would allow faster initialization.
+ */
+ ret = ufs_qcom_icc_set_bw(host, ufs_qcom_bw_table[MODE_MAX][0][0].bw1,
+ ufs_qcom_bw_table[MODE_MAX][0][0].bw2);
If this is a static value, can we define it separately? You don't have a
full gear/lanes enumeration anyway.
+ if (ret < 0)
+ return dev_err_probe(dev, ret, "failed to set bandwidth request\n");
+
+ return 0;
+}
+
/**
* ufs_qcom_init - bind phy with controller
* @hba: host controller instance
@@ -1085,6 +1209,10 @@ static int ufs_qcom_init(struct ufs_hba *hba)
}
}
+ err = ufs_qcom_icc_init(host);
+ if (err)
+ goto out_variant_clear;
+
host->device_reset = devm_gpiod_get_optional(dev, "reset",
GPIOD_OUT_HIGH);
if (IS_ERR(host->device_reset)) {
@@ -1282,6 +1410,7 @@ static int ufs_qcom_clk_scale_notify(struct ufs_hba *hba,
dev_req_params->pwr_rx,
dev_req_params->hs_rate,
false);
+ ufs_qcom_icc_update_bw(host);
ufshcd_uic_hibern8_exit(hba);
}
diff --git a/drivers/ufs/host/ufs-qcom.h b/drivers/ufs/host/ufs-qcom.h
index 6289ad5a42d0..dc27395ecba1 100644
--- a/drivers/ufs/host/ufs-qcom.h
+++ b/drivers/ufs/host/ufs-qcom.h
@@ -206,6 +206,9 @@ struct ufs_qcom_host {
struct clk *tx_l1_sync_clk;
bool is_lane_clks_enabled;
+ struct icc_path *icc_ddr;
+ struct icc_path *icc_cpu;
+
#ifdef CONFIG_SCSI_UFS_CRYPTO
struct qcom_ice *ice;
#endif
--
With best wishes
Dmitry
--
With best wishes
Dmitry