Re: [Patch v6 3/7] slimbus: qcom: Add Qualcomm Slimbus controller driver

From: Srinivas Kandagatla
Date: Wed Oct 18 2017 - 12:39:52 EST


Thanks for Review Comments.

On 18/10/17 08:27, Bjorn Andersson wrote:
On Fri 06 Oct 08:51 PDT 2017, srinivas.kandagatla@xxxxxxxxxx wrote:

From: Sagar Dharia <sdharia@xxxxxxxxxxxxxx>

This controller driver programs manager, interface, and framer
devices for Qualcomm's slimbus HW block.
Manager component currently implements logical address setting,
and messaging interface.
Interface device reports bus synchronization information, and framer
device clocks the bus from the time it's woken up, until clock-pause
is executed by the manager device.

Signed-off-by: Sagar Dharia <sdharia@xxxxxxxxxxxxxx>
Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@xxxxxxxxxx>
---
.../devicetree/bindings/slimbus/slim-qcom-ctrl.txt | 43 ++
drivers/slimbus/Kconfig | 9 +
drivers/slimbus/Makefile | 3 +
drivers/slimbus/slim-qcom-ctrl.c | 594 +++++++++++++++++++++
drivers/slimbus/slim-qcom.h | 63 +++
5 files changed, 712 insertions(+)
create mode 100644 Documentation/devicetree/bindings/slimbus/slim-qcom-ctrl.txt
create mode 100644 drivers/slimbus/slim-qcom-ctrl.c
create mode 100644 drivers/slimbus/slim-qcom.h

diff --git a/Documentation/devicetree/bindings/slimbus/slim-qcom-ctrl.txt b/Documentation/devicetree/bindings/slimbus/slim-qcom-ctrl.txt
new file mode 100644
index 0000000..081110d
--- /dev/null
+++ b/Documentation/devicetree/bindings/slimbus/slim-qcom-ctrl.txt
@@ -0,0 +1,43 @@
+Qualcomm SLIMBUS controller
+This controller is used if applications processor driver controls slimbus
+master component.
+
+Required properties:
+
+ - #address-cells - refer to Documentation/devicetree/bindings/slimbus/bus.txt
+ - #size-cells - refer to Documentation/devicetree/bindings/slimbus/bus.txt
+
+ - reg : Offset and length of the register region(s) for the device
+ - reg-names : Register region name(s) referenced in reg above
+ Required register resource entries are:
+ "ctrl": Physical adderess of controller register blocks
+ - compatible : should be "qcom,<SOC-NAME>-slim" for SOC specific compatible or

As you implementation is only compatible with "qcom,slim" this should be
"and" not "or".
yep,


+ "qcom,slim" if using generic qcom SLIM IP.
+ - interrupts : Interrupt number used by this controller
+ - clocks : Interface and core clocks used by this slimbus controller
+ - clock-names : Required clock-name entries are:
+ "iface_clk" : Interface clock for this controller
+ "core_clk" : Interrupt for controller core's BAM
+
+
+Optional property:
+ - reg entry for slew rate : If slew rate control register is provided, this
+ entry should be used.
+ - reg-name for slew rate: "slew"
+
+Example:
+ slim@28080000 {
+ compatible = "qcom,slim";
+ #address-cells = <4>;
+ #size-cells = <0>;
+ reg = <0x28080000 0x2000>, <0x80207C 4>;
+ reg-names = "ctrl", "slew";
+ interrupts = <0 33 0>;
+ clocks = <&lcc SLIMBUS_SRC>, <&lcc AUDIO_SLIMBUS_CLK>;
+ clock-names = "iface_clk", "core_clk";
+
+ codec: wcd9310@1{
+ compatible = "slim217,60";
+ reg = <1 0>;
+ };
+ };
diff --git a/drivers/slimbus/Kconfig b/drivers/slimbus/Kconfig
index f0b118a..438773e 100644
--- a/drivers/slimbus/Kconfig
+++ b/drivers/slimbus/Kconfig
@@ -9,3 +9,12 @@ menuconfig SLIMBUS
If unsure, choose N.
yep.

+if SLIMBUS
+config SLIM_QCOM_CTRL
+ tristate "Qualcomm Slimbus Manager Component"
+ depends on SLIMBUS
+ help
+ Select driver if Qualcomm's Slimbus Manager Component is
+ programmed using Linux kernel.
+
+endif
diff --git a/drivers/slimbus/Makefile b/drivers/slimbus/Makefile
index bd1d35e..ed8521a 100644
--- a/drivers/slimbus/Makefile
+++ b/drivers/slimbus/Makefile
@@ -3,3 +3,6 @@
#
obj-$(CONFIG_SLIMBUS) += slimbus.o
slimbus-y := slim-core.o slim-messaging.o
+
+#Controllers
+obj-$(CONFIG_SLIM_QCOM_CTRL) += slim-qcom-ctrl.o
diff --git a/drivers/slimbus/slim-qcom-ctrl.c b/drivers/slimbus/slim-qcom-ctrl.c
new file mode 100644
index 0000000..d0574e1
--- /dev/null
+++ b/drivers/slimbus/slim-qcom-ctrl.c
..
+#include "slim-qcom.h"
+
+#define MSM_SLIM_NAME "msm_slim_ctrl"

Just put this string in the driver struct.

yes..


+
+/* Manager registers */
+#define MGR_CFG 0x200
...

+
+static int msm_slim_queue_tx(struct msm_slim_ctrl *dev, u32 *buf, u8 len,
+ u32 tx_reg)

Use void * for buf.

okay.

+{
+ int i;
+
+ for (i = 0; i < (len + 3) >> 2; i++) {

If len is in bytes then this looks like you will read outside the
buffer. If buf is guaranteed to be 4-byte aligned make "len" number of
32-bit entries in the array.

+ dev_dbg(dev->dev, "AHB TX data:0x%x\n", buf[i]);

Drop the debug print, if anything use print_hex_dump to get the whole
chunk.

makes sense.
+ writel_relaxed(buf[i], dev->base + tx_reg + (i * 4));
+ }

Replace this loop with:

__iowrite32_copy(dev->base + tx_reg, buf, len / sizeof(u32));

+ /* Guarantee that message is sent before returning */

"Ensure ordering of subsequent writes", there's no guarantees that the
write is done at the end of this function.

will update the comment!

+ mb();
+ return 0;

Why return an int when there are not other return paths?

I agree.
+}
+
+static irqreturn_t msm_slim_interrupt(int irq, void *d)
+{
+ struct msm_slim_ctrl *dev = d;
+ u32 stat = readl_relaxed(dev->base + MGR_INT_STAT);
+ int err = 0, ret = IRQ_NONE;
+

If you split the two large parts of this function into two functions
things will be cleaner and you'll save yourself one level of
indentation.
Will give it a try.


+ if (stat & MGR_INT_TX_MSG_SENT || stat & MGR_INT_TX_NACKED_2) {
+ if (stat & MGR_INT_TX_MSG_SENT)
+ writel_relaxed(MGR_INT_TX_MSG_SENT,
+ dev->base + MGR_INT_CLR);
+ if (stat & MGR_INT_TX_NACKED_2) {
+ u32 mgr_stat = readl_relaxed(dev->base + MGR_STATUS);
+ u32 mgr_ie_stat = readl_relaxed(dev->base +
+ MGR_IE_STAT);
+ u32 frm_stat = readl_relaxed(dev->base + FRM_STAT);
+ u32 frm_cfg = readl_relaxed(dev->base + FRM_CFG);
+ u32 frm_intr_stat = readl_relaxed(dev->base +
+ FRM_INT_STAT);
+ u32 frm_ie_stat = readl_relaxed(dev->base +
+ FRM_IE_STAT);
+ u32 intf_stat = readl_relaxed(dev->base + INTF_STAT);
+ u32 intf_intr_stat = readl_relaxed(dev->base +
+ INTF_INT_STAT);
+ u32 intf_ie_stat = readl_relaxed(dev->base +
+ INTF_IE_STAT);
+
+ writel_relaxed(MGR_INT_TX_NACKED_2, dev->base +
+ MGR_INT_CLR);
+ dev_err(dev->dev, "TX Nack MGR:int:0x%x, stat:0x%x\n",
+ stat, mgr_stat);
+ dev_err(dev->dev, "TX Nack MGR:ie:0x%x\n", mgr_ie_stat);
+ dev_err(dev->dev, "TX Nack FRM:int:0x%x, stat:0x%x\n",
+ frm_intr_stat, frm_stat);
+ dev_err(dev->dev, "TX Nack FRM:cfg:0x%x, ie:0x%x\n",
+ frm_cfg, frm_ie_stat);
+ dev_err(dev->dev, "TX Nack INTF:intr:0x%x, stat:0x%x\n",
+ intf_intr_stat, intf_stat);
+ dev_err(dev->dev, "TX Nack INTF:ie:0x%x\n",
+ intf_ie_stat);
+ err = -ENOTCONN;
+ }
+ /**
+ * Guarantee that interrupt clear bit write goes through before
+ * signalling completion/exiting ISR
+ */

I'm not sure why this is necessary, but this is not "guaranteed" by the
mb().

Will update the comment.


+ mb();
+ slim_return_tx(&dev->ctrl, err);
+ ret = IRQ_HANDLED;
+ }
+ if (stat & MGR_INT_RX_MSG_RCVD) {
+ u8 mc, mt;
+ u8 len, i;
+ u32 *rx_buf, pkt[10];
+ bool q_rx = false;
+
+ pkt[0] = readl_relaxed(dev->base + MGR_RX_MSG);
+ mt = (pkt[0] >> 5) & 0x7;
+ mc = (pkt[0] >> 8) & 0xff;
+ len = pkt[0] & 0x1F;

The way this function deals with cryptic variable names and cryptic
bit shifts is quite hard to follow. Perhaps it makes sense if you know
exactly what's going on?

A comment describing the fields in the read information would be quite
helpful, at least.

I will try to add some notes here to make it easy for readers.

+ dev_dbg(dev->dev, "RX-IRQ: MC: %x, MT: %x\n", mc, mt);
+
+ /**
+ * this message cannot be handled by ISR, so
+ * let work-queue handle it
+ */
+ if (mt == SLIM_MSG_MT_CORE &&
+ mc == SLIM_MSG_MC_REPORT_PRESENT)
+ rx_buf = (u32 *)slim_get_rx(&dev->ctrl);
+ else
+ rx_buf = pkt;
+
+ if (rx_buf == NULL) {

if (!rx_buf)

+ dev_err(dev->dev, "dropping RX:0x%x due to RX full\n",
+ pkt[0]);

Why do we need to drop this incoming message? Will things recover from
that?

Caller might timeout due to this.



+ goto rx_ret_irq;
+ }
+
+ rx_buf[0] = pkt[0];
+ for (i = 1; i < ((len + 3) >> 2); i++) {
+ rx_buf[i] = readl_relaxed(dev->base + MGR_RX_MSG +
+ (4 * i));
+ dev_dbg(dev->dev, "reading data: %x\n", rx_buf[i]);

Drop this debug message, if you really want to show this use
print_hex_dump() after reading the entire thing.


And you should be able to replace this loop with __ioread32_copy()

yep.

+ }
+
+ switch (mc) {
+ u8 *buf, la;
+ u16 ele;

What?!
It does not look very good!!

Its should be legal as it is within the block with automatic storage duration. But it would never be evaluated if we try to initialize these.

I will move them to better place in next version.


+
+ case SLIM_MSG_MC_REPORT_PRESENT:
+ q_rx = true;
+ break;
+ case SLIM_MSG_MC_REPLY_INFORMATION:
+ case SLIM_MSG_MC_REPLY_VALUE:
+ slim_msg_response(&dev->ctrl, (u8 *)(rx_buf + 1),
+ (u8)(*rx_buf >> 24), (len - 4));
+ break;
+ case SLIM_MSG_MC_REPORT_INFORMATION:
+ buf = (u8 *)rx_buf;
+ la = buf[2];
+ ele = (u16)buf[4] << 4;
+
+ ele |= ((buf[3] & 0xf0) >> 4);
+ /**
+ * report information is most likely loss of
+ * sync or collision detected in data slots
+ */
+ dev_err(dev->dev, "LA:%d report inf ele:0x%x\n",
+ la, ele);
+ for (i = 0; i < len - 5; i++)
+ dev_err(dev->dev, "bit-mask:%x\n",
+ buf[i+5]);

To whom is this cryptic dump valuable? Please use print_hex_dump()
rather than rolling your own print-loops.

Yep.. Will cleanup all such instances.
+ break;
+ default:
+ dev_err(dev->dev, "unsupported MC,%x MT:%x\n",
+ mc, mt);
+ break;
+ }
+rx_ret_irq:
+ writel_relaxed(MGR_INT_RX_MSG_RCVD, dev->base +
+ MGR_INT_CLR);
+ /**
+ * Guarantee that CLR bit write goes through
+ * before exiting
+ */

"Ensure ordering of CLR bit write with subsequent writes" although I'm
wondering how useful this is...

+ mb();
+ if (q_rx)
+ queue_work(dev->rxwq, &dev->wd);
+
+ ret = IRQ_HANDLED;
+ }
+ return ret;
+}
+
+static int msm_xfer_msg(struct slim_controller *ctrl, struct slim_msg_txn *txn,
+ void *pbuf)
+{
+ struct msm_slim_ctrl *dev = slim_get_ctrldata(ctrl);
+ u32 *head = (u32 *)pbuf;
+ u8 *puc = (u8 *)pbuf;
+ u8 la = txn->la;
+
+ /* HW expects length field to be excluded */
+ txn->rl--;
+
+ if (txn->dt == SLIM_MSG_DEST_LOGICALADDR)
+ *head = SLIM_MSG_ASM_FIRST_WORD(txn->rl, txn->mt, txn->mc, 0,
+ la);
+ else
+ *head = SLIM_MSG_ASM_FIRST_WORD(txn->rl, txn->mt, txn->mc, 1,
+ la);

Put "dt" (whatever that is?!) in a variable and then do this line once,
saves you from the line break.


+
+ if (txn->dt == SLIM_MSG_DEST_LOGICALADDR)
+ puc += 3;
+ else
+ puc += 2;
+
+ if (txn->mt == SLIM_MSG_MT_CORE && slim_tid_txn(txn->mt, txn->mc))
+ *(puc++) = txn->tid;
+
+ if ((txn->mt == SLIM_MSG_MT_CORE) &&
+ ((txn->mc >= SLIM_MSG_MC_REQUEST_INFORMATION &&
+ txn->mc <= SLIM_MSG_MC_REPORT_INFORMATION) ||
+ (txn->mc >= SLIM_MSG_MC_REQUEST_VALUE &&
+ txn->mc <= SLIM_MSG_MC_CHANGE_VALUE))) {
+ *(puc++) = (txn->ec & 0xFF);
+ *(puc++) = (txn->ec >> 8) & 0xFF;
+ }
+
+ if (txn->msg && txn->msg->wbuf)
+ memcpy(puc, txn->msg->wbuf, txn->msg->num_bytes);
+
+ return msm_slim_queue_tx(dev, head, txn->rl, MGR_TX_MSG);
+}
+
+static int msm_set_laddr(struct slim_controller *ctrl,
+ struct slim_eaddr *ead, u8 laddr)
+{
+ struct msm_slim_ctrl *dev = slim_get_ctrldata(ctrl);
+ u8 buf[7];
+ int ret;
+ struct slim_val_inf msg = {0};
+
+ DEFINE_SLIM_EDEST_TXN(txn, SLIM_MSG_MC_ASSIGN_LOGICAL_ADDRESS,
+ 10, laddr, &msg);
+
+ /* Enumeration address */
+ buf[0] = (u8)(ead->manf_id >> 8);
+ buf[1] = (u8)(ead->manf_id & 0xFF);
+ buf[2] = (u8) (ead->prod_code >> 8);
+ buf[3] = (u8) (ead->prod_code & 0xFF);
+ buf[4] = ead->dev_index;
+ buf[5] = ead->instance;
+
+ /* Logical address for this EA */
+ buf[6] = laddr;

This would be easier to deal with in the form of:

struct foo {
__le16 manf_id;
__le16 prod_code;
u8 dev_idx;
u8 instance;
u8 laddr;
};

And it doesn't look like a "buf", it looks like a packet of some
specific type, so it could be given a better name.
I agree.

+
+ /**
+ * Retries are needed since bus may lose sync when multiple devices
+ * are coming up and reporting present
+ */

Is this a todo? I'm not sure how this relates to the code.

not sure. but can be removed from here.

+ msg.wbuf = buf;
+ msg.num_bytes = 7;
+
+ ret = slim_processtxn(&dev->ctrl, &txn);
+
+ if (ret)
+ dev_err(dev->dev, "set LA:0x%x failed:ret:%d\n",
+ laddr, ret);
+ return ret;
+}
+
+static void msm_slim_rxwq(struct work_struct *work)
+{
+ u8 buf[40];

40? Are you perhaps trying to say SLIM_MSGQ_BUF_LEN?

yes.

+ u8 mc, mt, len;
+ int i, ret;
+ struct msm_slim_ctrl *dev = container_of(work, struct msm_slim_ctrl,
+ wd);
+
+ while ((slim_return_rx(&dev->ctrl, buf)) != -ENODATA) {
+ len = buf[0] & 0x1F;
+ mt = (buf[0] >> 5) & 0x7;
+ mc = buf[1];

How about describing this like:

struct some_pkt {
u8 len:5;
u8 mt:3;
u8 mc;

__le16 manf_id;
__le16 prod_code;
u8 dev_idx;
u8 instance;
};

But perhaps manf_id and prod_code are big endian here? I'm getting
confused from this code.
Its little endian.

+ if (mt == SLIM_MSG_MT_CORE &&
+ mc == SLIM_MSG_MC_REPORT_PRESENT) {
+ u8 laddr;
+ struct slim_eaddr ea;
+ u8 e_addr[6];>> +/* MAX message size over control channel */
+#define SLIM_MSGQ_BUF_LEN 40
+#define MSM_TX_MSGS 2
+#define MSM_RX_MSGS 8
+
+#define CFG_PORT(r, v) ((v) ? CFG_PORT_V2(r) : CFG_PORT_V1(r))
+
+/* V2 Component registers */
+#define CFG_PORT_V2(r) ((r ## _V2))
+#define COMP_CFG_V2 4
+#define COMP_TRUST_CFG_V2 0x3000
+
+/* V1 Component registers */
+#define CFG_PORT_V1(r) ((r ## _V1))
+#define COMP_CFG_V1 0
+#define COMP_TRUST_CFG_V1 0x14
+
+/* Resource group info for manager, and non-ported generic device-components */
+#define EE_MGR_RSC_GRP (1 << 10)
+#define EE_NGD_2 (2 << 6)
+#define EE_NGD_1 0
+
+struct msm_slim_ctrl {
+ struct slim_controller ctrl;
+ struct slim_framer framer;
+ struct device *dev;
+ void __iomem *base;
+ struct resource *slew_mem;
+
+ for (i = 0; i < 6; i++)
+ e_addr[i] = buf[7-i];
+
+ ea.manf_id = (u16)(e_addr[5] << 8) | e_addr[4];
+ ea.prod_code = (u16)(e_addr[3] << 8) | e_addr[2];
+ ea.dev_index = e_addr[1];
+ ea.instance = e_addr[0];
+ ret = slim_assign_laddr(&dev->ctrl, &ea, &laddr, false);
+ if (ret)
+ dev_err(dev->dev, "assign laddr failed:%d\n",
+ ret);
+ } else {
+ dev_err(dev->dev, "unexpected message:mc:%x, mt:%x\n",
+ mc, mt);

The idiomatic way is to deal with errors first and then you don't have
to indent the normal code path.

+
+ }
+
+ }
+}
+
+static void msm_slim_prg_slew(struct platform_device *pdev,
+ struct msm_slim_ctrl *dev)
+{
+ void __iomem *slew_reg;
+
+ /* SLEW RATE register for this slimbus */
+ dev->slew_mem = platform_get_resource_byname(pdev, IORESOURCE_MEM,

This is a local variable...

+ "slew");
+ if (!dev->slew_mem) {
+ dev_warn(&pdev->dev, "no slimbus slew resource\n");
+ return;
+ }
+

It's idiomatic to do:

res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "slew");
slew_reg = devm_ioremap_resource(&pdev->dev, res);
if (IS_ERR(slew_reg))
...

yep.

+ slew_reg = devm_ioremap(&pdev->dev, dev->slew_mem->start,
+ resource_size(dev->slew_mem));
+ if (!slew_reg) {
+ dev_err(dev->dev, "slew register mapping failed");

Is this not a problem? Is it okay to silently discard this error.

+ release_mem_region(dev->slew_mem->start,
+ resource_size(dev->slew_mem));
+ dev->slew_mem = NULL;
+ return;
+ }
+ writel_relaxed(1, slew_reg);
+ /* Make sure slimbus-slew rate enabling goes through */

This is not what wmb() does. If you want to ensure that it's written do
a readback.

+ wmb();

There's no reason to keep the slew register mapped beyond this function,
so you could drop the devm_ and iounmap() slew_reg here again.


If there's a reason to touch this in the future, then keep track of
slew_reg rather than the slew_mem...

I agree, we can clean this part up.

+}
+
+static int msm_slim_probe(struct platform_device *pdev)
+{
+ struct msm_slim_ctrl *dev;

"dev" is not the best name for this (dev->dev?!)

+ struct slim_controller *ctrl;
+ struct resource *slim_mem;
+ struct resource *irq;
+ struct clk *hclk, *rclk;

Why the odd names?

yes, I was bit confused when i started reading this part
Will fix this.
+ int ret;
+
+ hclk = devm_clk_get(&pdev->dev, "iface_clk");
+ if (IS_ERR(hclk))
+ return PTR_ERR(hclk);
+
+ rclk = devm_clk_get(&pdev->dev, "core_clk");
+ if (IS_ERR(rclk)) {
+ /* unlikely that this is probe-defer */

Do this properly.

this is a carry over.. i will remove unnecessary commetns.

+ dev_err(&pdev->dev, "rclk get failed:%ld\n", PTR_ERR(rclk));
+ return PTR_ERR(rclk);
+ }
+
+ ret = clk_set_rate(rclk, SLIM_ROOT_FREQ);
+ if (ret) {
+ dev_err(&pdev->dev, "ref-clock set-rate failed:%d\n", ret);
+ return ret;
+ }
+
+ slim_mem = platform_get_resource_byname(pdev, IORESOURCE_MEM, "ctrl");

Move this down to the devm_ioremap() and skip the error check.
yep.

+ if (!slim_mem) { >> + dev_err(&pdev->dev, "no slimbus physical memory resource\n");
+ return -ENODEV;
+ }
+
+ irq = platform_get_resource(pdev, IORESOURCE_IRQ, 0);
+ if (!irq) {
+ dev_err(&pdev->dev, "no slimbus IRQ resource\n");
+ return -ENODEV;
+ }
+
+ dev = devm_kzalloc(&pdev->dev, sizeof(*dev), GFP_KERNEL);

If you allocate this earlier you can reduce the number of local
variables.

possibly..
+ if (!dev)
+ return -ENOMEM;
+
+ dev->hclk = hclk;
+ dev->rclk = rclk;
+ ctrl = &dev->ctrl;
+ dev->dev = &pdev->dev;
+ platform_set_drvdata(pdev, dev);
+ slim_set_ctrldata(&dev->ctrl, dev);
+ dev->base = devm_ioremap(dev->dev, slim_mem->start,
+ resource_size(slim_mem));

devm_ioremap_resource()

+ if (!dev->base) {
+ dev_err(&pdev->dev, "IOremap failed\n");
+ return -ENOMEM;
+ }
+
+ dev->ctrl.set_laddr = msm_set_laddr;
+ dev->ctrl.xfer_msg = msm_xfer_msg;
+ dev->ctrl.tx.n = MSM_TX_MSGS;
+ dev->ctrl.rx.n = MSM_RX_MSGS;
+ dev->ctrl.tx.sl_sz = SLIM_MSGQ_BUF_LEN;
+ dev->ctrl.rx.sl_sz = SLIM_MSGQ_BUF_LEN;
+
+ dev->irq = irq->start;

Use platform_get_irq() instead and you don't need to do this.

yep,

+
+ INIT_WORK(&dev->wd, msm_slim_rxwq);
+ dev->rxwq = create_singlethread_workqueue("msm_slim_rx");
+ if (!dev->rxwq) {
+ dev_err(dev->dev, "Failed to start Rx WQ\n");
+ return -ENOMEM;
+ }
+
+ dev->framer.rootfreq = SLIM_ROOT_FREQ >> 3;

Dividing by 8 makes the code readable, compilers will make sure it's
still super fast.

yep..
+ dev->framer.superfreq =
+ dev->framer.rootfreq / SLIM_CL_PER_SUPERFRAME_DIV8;
+ dev->ctrl.a_framer = &dev->framer;
+ dev->ctrl.clkgear = SLIM_MAX_CLK_GEAR;
+ dev->ctrl.dev.parent = &pdev->dev;
+ dev->ctrl.dev.of_node = pdev->dev.of_node;
+
+ msm_slim_prg_slew(pdev, dev);
+
+ ret = devm_request_irq(&pdev->dev, dev->irq, msm_slim_interrupt,
+ IRQF_TRIGGER_HIGH, "msm_slim_irq", dev);
+ if (ret) {
+ dev_err(&pdev->dev, "request IRQ failed\n");
+ goto err_request_irq_failed;
+ }

At this point you might get interrupts, which I presume will try to
access the registers which you start clocking on the next line. Please
reorder these.

makes sense..
+
+ ret = clk_prepare_enable(hclk);
+ if (ret)
+ goto err_hclk_enable_failed;
+
+ ret = clk_prepare_enable(rclk);
+ if (ret)
+ goto err_rclk_enable_failed;

Use the clk_bulk api instead.

yep.
+
+
+ ctrl->tx.base = dma_alloc_coherent(&pdev->dev,
+ (ctrl->tx.sl_sz * ctrl->tx.n),
+ &ctrl->tx.phy, GFP_KERNEL);
+ if (!ctrl->tx.base) {
+ ret = -ENOMEM;
+ goto tx_alloc_failed;
+ }
+
+ ctrl->rx.base = dma_alloc_coherent(&pdev->dev,
+ (ctrl->rx.sl_sz * ctrl->rx.n),
+ &ctrl->rx.phy, GFP_KERNEL);
+ if (!ctrl->rx.base) {
+ ret = -ENOMEM;
+ goto rx_alloc_failed;
+ }
+
+
+ /* Register with framework before enabling frame, clock */
+ ret = slim_register_controller(&dev->ctrl);
+ if (ret) {
+ dev_err(dev->dev, "error adding controller\n");
+ goto err_ctrl_failed;
+ }
+
+ dev->ver = readl_relaxed(dev->base);
+ /* Version info in 16 MSbits */
+ dev->ver >>= 16;
+ /* Component register initialization */
+ writel_relaxed(1, dev->base + CFG_PORT(COMP_CFG, dev->ver));

Use non-relaxed writel, unless there's a good reason not to.

+ writel_relaxed((EE_MGR_RSC_GRP | EE_NGD_2 | EE_NGD_1),
+ dev->base + CFG_PORT(COMP_TRUST_CFG, dev->ver));
+
+ writel_relaxed((MGR_INT_TX_NACKED_2 |
+ MGR_INT_MSG_BUF_CONTE | MGR_INT_RX_MSG_RCVD |
+ MGR_INT_TX_MSG_SENT), dev->base + MGR_INT_EN);
+ writel_relaxed(1, dev->base + MGR_CFG);
+ /*
+ * Framer registers are beyond 1K memory region after Manager and/or
+ * component registers. Make sure those writes are ordered
+ * before framer register writes
+ */

Drop this comment and just use a non-relaxed writel for the framer
registration initialization.

+ wmb();
+
+ /* Framer register initialization */
+ writel_relaxed((1 << INTR_WAKE) | (0xA << REF_CLK_GEAR) |
+ (0xA << CLK_GEAR) | (1 << ROOT_FREQ) | (1 << FRM_ACTIVE) | 1,
+ dev->base + FRM_CFG);
+ /*
+ * Make sure that framer wake-up and enabling writes go through
+ * before any other component is enabled. Framer is responsible for
+ * clocking the bus and enabling framer first will ensure that other
+ * devices can report presence when they are enabled
+ */
+ mb();

Drop this mb and use writel()

+
+ writel_relaxed(MGR_CFG_ENABLE, dev->base + MGR_CFG);
+ /*
+ * Make sure that manager-enable is written through before interface
+ * device is enabled

Dito

+ */
+ mb();
+ writel_relaxed(1, dev->base + INTF_CFG);
+ /*
+ * Make sure that interface-enable is written through before enabling
+ * ported generic device inside MSM manager
+ */

Dito

+ mb();
+
+ writel_relaxed(1, dev->base + CFG_PORT(COMP_CFG, dev->ver));
+ /*
+ * Make sure that all writes have gone through before exiting this
+ * function
+ */

Dito
I agree, we could just do writel here.


+ mb();
+
+ dev_dbg(dev->dev, "MSM SB controller is up:ver:0x%x!\n", dev->ver);
+ return 0;
+
+err_ctrl_failed:
+ dma_free_coherent(&pdev->dev, (ctrl->rx.sl_sz * ctrl->rx.n),
+ ctrl->rx.base, ctrl->rx.phy);
+rx_alloc_failed:
+ dma_free_coherent(ctrl->dev.parent, (ctrl->tx.sl_sz * ctrl->tx.n),
+ ctrl->tx.base, ctrl->tx.phy);
+tx_alloc_failed:
+ clk_disable_unprepare(dev->rclk);
+err_rclk_enable_failed:
+ clk_disable_unprepare(dev->hclk);
+
+err_hclk_enable_failed:
+err_request_irq_failed:
+ destroy_workqueue(dev->rxwq);
+ return ret;
+}
+
+static int msm_slim_remove(struct platform_device *pdev)
+{
+ struct msm_slim_ctrl *dev = platform_get_drvdata(pdev);
+ struct slim_controller *ctrl = to_slim_controller(&pdev->dev);
+
+ dma_free_coherent(&pdev->dev, (ctrl->rx.sl_sz * ctrl->rx.n),
+ ctrl->rx.base, ctrl->rx.phy);
+ dma_free_coherent(&pdev->dev, (ctrl->tx.sl_sz * ctrl->tx.n),
+ ctrl->tx.base, ctrl->tx.phy);

Don't you at least want to disable irqs before you free this memory?

yes.. makes sense..
+
+ disable_irq(dev->irq);
+ clk_disable_unprepare(dev->rclk);
+ clk_disable_unprepare(dev->hclk);
+ slim_del_controller(&dev->ctrl);
+ destroy_workqueue(dev->rxwq);
+ return 0;
+}
+
+static const struct of_device_id msm_slim_dt_match[] = {
+ {
+ .compatible = "qcom,slim",
+ },
+ {}
+};
+
+static struct platform_driver msm_slim_driver = {
+ .probe = msm_slim_probe,
+ .remove = msm_slim_remove,
+ .driver = {
+ .name = MSM_SLIM_NAME,
+ .owner = THIS_MODULE,

No "owner"


y

+ .of_match_table = msm_slim_dt_match,
+ },
+};
+module_platform_driver(msm_slim_driver);
+
+MODULE_LICENSE("GPL v2");
+MODULE_VERSION("0.1");

Skip the version, we'll never update it anyways.

Yep.

+MODULE_DESCRIPTION("Qualcomm Slimbus controller");
+MODULE_ALIAS("platform:qcom-slim");

No-one will request the module by this alias, so skip it.

yep.

diff --git a/drivers/slimbus/slim-qcom.h b/drivers/slimbus/slim-qcom.h
new file mode 100644
index 0000000..0ad59c3
--- /dev/null
+++ b/drivers/slimbus/slim-qcom.h

As far as I can see this information is all local to lim-qcom-ctrl.c, so
put it there.

Yes, as of today, but its possible that this header can be shared across other slim controller in Qcom B Family.



@@ -0,0 +1,63 @@
+/* Copyright (c) 2011-2016, The Linux Foundation. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ */
+
+#ifndef _SLIM_QCOM_H
+#define _SLIM_QCOM_H
+
+#include <linux/irq.h>
+#include <linux/workqueue.h>
+
+#define QC_MFGID_LSB 0x2
+#define QC_MFGID_MSB 0x17

Unused.

Will remove this in next version.


+
+#define SLIM_MSG_ASM_FIRST_WORD(l, mt, mc, dt, ad) \
+ ((l) | ((mt) << 5) | ((mc) << 8) | ((dt) << 15) | ((ad) << 16))
+
+#define SLIM_ROOT_FREQ 24576000
+
..

+ int irq;
+ struct workqueue_struct *rxwq;
+ struct work_struct wd;
+ struct clk *rclk;
+ struct clk *hclk;
+ u32 ver;

"ver" is only used within probe(), use a local variable instead.
yep.


+};
+
+#endif

Regards,
Bjorn