Re: [PATCH v4 11/11] scsi: ufs-exynos: add UFS host support for Exynos SoCs

From: Alim Akhtar
Date: Tue Nov 03 2015 - 06:44:34 EST


Hi Kishon,
Thanks for your time.

On 10/28/2015 06:23 PM, Kishon Vijay Abraham I wrote:
Hi,

On Sunday 25 October 2015 05:34 PM, Alim Akhtar wrote:
Hi Kishon
Thanks again for you review.

On Fri, Oct 23, 2015 at 8:48 PM, Kishon Vijay Abraham I <kishon@xxxxxx> wrote:
Hi,

On Thursday 15 October 2015 08:38 AM, Alim Akhtar wrote:
+CCing kishon Vijay,

On 10/14/2015 06:25 PM, Alim Akhtar wrote:
From: Seungwon Jeon <essuuj@xxxxxxxxx>

This patch introduces Exynos UFS host controller driver,
which mainly handles vendor-specific operations including
link startup, power mode change and hibernation/unhibernation.

Signed-off-by: Seungwon Jeon <essuuj@xxxxxxxxx>
Signed-off-by: Alim Akhtar <alim.akhtar@xxxxxxxxxxx>
---
drivers/scsi/ufs/Kconfig | 12 +
drivers/scsi/ufs/Makefile | 1 +
drivers/scsi/ufs/ufs-exynos-hw.c | 131 ++++
drivers/scsi/ufs/ufs-exynos-hw.h | 43 ++
drivers/scsi/ufs/ufs-exynos.c | 1317
++++++++++++++++++++++++++++++++++++++
drivers/scsi/ufs/ufs-exynos.h | 247 +++++++
drivers/scsi/ufs/ufshci.h | 26 +-
drivers/scsi/ufs/unipro.h | 47 ++
8 files changed, 1823 insertions(+), 1 deletion(-)
create mode 100644 drivers/scsi/ufs/ufs-exynos-hw.c
create mode 100644 drivers/scsi/ufs/ufs-exynos-hw.h
create mode 100644 drivers/scsi/ufs/ufs-exynos.c
create mode 100644 drivers/scsi/ufs/ufs-exynos.h

.
.
<snip>
.
.
diff --git a/drivers/scsi/ufs/ufs-exynos-hw.c
b/drivers/scsi/ufs/ufs-exynos-hw.c
new file mode 100644
index 000000000000..be6c61541a8f
--- /dev/null
+++ b/drivers/scsi/ufs/ufs-exynos-hw.c
@@ -0,0 +1,131 @@
.
.
<snip>
.
.
+
+#define PWR_MODE_STR_LEN 64
+static int exynos_ufs_post_pwr_mode(struct ufs_hba *hba,
+ struct ufs_pa_layer_attr *pwr_max,
+ struct ufs_pa_layer_attr *pwr_req)
+{
+ struct exynos_ufs *ufs = to_exynos_ufs(hba);
+ struct exynos_ufs_phy_info *phy_info = phy_get_drvdata(ufs->phy);

This is abusing the interface. phy_get_drvdata is meant to be used only
by the PHY driver.
+ struct exynos_ufs_phy_specific_ops *phy_ops =
+ phy_info->phy_specific_ops;

I'm really not happy about having platform specific ops for PHY. We have
to see if existing PHY ops can be used for this or in worst case add new
PHY ops.
Well you said you like the controller driver to use only PHY ops[1], I
am sorry If I misunderstood that point, can you please help me to
understand that?

I meant PHY generic ops and not PHY ops.
Ok, got it, will use only generic phy here in controller driver.
- Will remove the platform specific PHY ops from phy driver introduce in this series (patch 02/11)
[1]-> https://lkml.org/lkml/2015/9/18/29

+ struct uic_pwr_mode *pwr = &ufs->pwr_act;
+ char pwr_str[PWR_MODE_STR_LEN] = "";
+ int ret = 0;
+
+ if (ufs->drv_data->post_pwr_change)
+ ufs->drv_data->post_pwr_change(ufs, pwr);
+
+ if (IS_UFS_PWR_MODE_HS(pwr->mode)) {
+ switch (pwr->hs_series) {
+ case PA_HS_MODE_A:
+ case PA_HS_MODE_B:
+ phy_ops->calibrate_phy(ufs->phy, CFG_POST_PWR_HS,
+ PWR_MODE_HS(pwr->gear, pwr->hs_series));
+ break;
+ }
+
+ ret = phy_ops->wait_for_lock_acq(ufs->phy);
+ snprintf(pwr_str, sizeof(pwr_str), "Fast%s series_%s G_%d L_%d",
+ pwr->mode == FASTAUTO_MODE ? "_Auto" : "",
+ pwr->hs_series == PA_HS_MODE_A ? "A" : "B",
+ pwr->gear, pwr->lane);
+ } else if (IS_UFS_PWR_MODE_PWM(pwr->mode)) {
+ snprintf(pwr_str, sizeof(pwr_str), "Slow%s G_%d L_%d",
+ pwr->mode == SLOWAUTO_MODE ? "_Auto" : "",
+ pwr->gear, pwr->lane);
+ }
+
+ dev_info(hba->dev, "Power mode change %d : %s\n", ret, pwr_str);
+ return ret;
+}
+
+static void exynos_ufs_specify_nexus_t_xfer_req(struct ufs_hba *hba,
+ int tag, struct scsi_cmnd *cmd)
+{
+ struct exynos_ufs *ufs = to_exynos_ufs(hba);
+ u32 type;
+
+ type = hci_readl(ufs, HCI_UTRL_NEXUS_TYPE);
+
+ if (cmd)
+ hci_writel(ufs, type | (1 << tag), HCI_UTRL_NEXUS_TYPE);
+ else
+ hci_writel(ufs, type & ~(1 << tag), HCI_UTRL_NEXUS_TYPE);
+}
+
+static void exynos_ufs_specify_nexus_t_tm_req(struct ufs_hba *hba,
+ int tag, u8 func)
+{
+ struct exynos_ufs *ufs = to_exynos_ufs(hba);
+ u32 type;
+
+ type = hci_readl(ufs, HCI_UTMRL_NEXUS_TYPE);
+
+ switch (func) {
+ case UFS_ABORT_TASK:
+ case UFS_QUERY_TASK:
+ hci_writel(ufs, type | (1 << tag), HCI_UTMRL_NEXUS_TYPE);
+ break;
+ case UFS_ABORT_TASK_SET:
+ case UFS_CLEAR_TASK_SET:
+ case UFS_LOGICAL_RESET:
+ case UFS_QUERY_TASK_SET:
+ hci_writel(ufs, type & ~(1 << tag), HCI_UTMRL_NEXUS_TYPE);
+ break;
+ }
+}
+
+static void exynos_ufs_phy_init(struct exynos_ufs *ufs)
+{
+ struct ufs_hba *hba = ufs->hba;
+ struct exynos_ufs_phy_info *phy_info = phy_get_drvdata(ufs->phy);
+ struct exynos_ufs_phy_specific_ops *phy_ops =
+ phy_info->phy_specific_ops;
+
+ if (ufs->avail_ln_rx == 0 || ufs->avail_ln_tx == 0) {
+ ufshcd_dme_get(hba, UIC_ARG_MIB(PA_AVAILRXDATALANES),
+ &ufs->avail_ln_rx);
+ ufshcd_dme_get(hba, UIC_ARG_MIB(PA_AVAILTXDATALANES),
+ &ufs->avail_ln_tx);
+ WARN(ufs->avail_ln_rx != ufs->avail_ln_tx,
+ "available data lane is not equal(rx:%d, tx:%d)\n",
+ ufs->avail_ln_rx, ufs->avail_ln_tx);
+ }
+
+ phy_ops->set_lane_cnt(ufs->phy, ufs->avail_ln_rx);

can't bus_width attribute in phy core be reused for this?

I will take a look on it.

+ phy_ops->calibrate_phy(ufs->phy, CFG_PRE_INIT, PWR_MODE_ANY);

Why can't calibrate PHY be directly done in phy_init?

This is just one instance, need to calibrate PHY when the ufs pwr mode changes.

Then maybe we should check when the power mode changes and see if it
makes sense to calibrate PHY in other generic PHY ops like phy_power_on
etc..

I did explore this but looks like in current situation it is not the best option.
Will post next revision after I am done with modifications, PTAL.

Thanks
Kishon

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/