Re: [PATCH v2] bus: mhi: core: Add unique qrtr node id support

From: Bhaumik Bhatt
Date: Fri Feb 26 2021 - 12:32:33 EST


On 2021-02-26 06:52 AM, Manivannan Sadhasivam wrote:
On Fri, Feb 26, 2021 at 04:12:49PM +0530, Gokul Sriram Palanisamy wrote:
On platforms with two or more identical mhi
devices, qmi service will run with identical
qrtr-node-id. Because of this identical ID,
host qrtr-lookup cannot register more than one
qmi service with identical node ID. Ultimately,
only one qmi service will be avilable for the
underlying drivers to communicate with.

On QCN9000, it implements a unique qrtr-node-id
and qmi instance ID using a unique instance ID
written to a debug register from host driver
soon after SBL is loaded.

This change generates a unique instance ID from
PCIe domain number and bus number, writes to the
given debug register just after SBL is loaded so
that it is available for FW when the QMI service
is spawned.

sample:
root@OpenWrt:/# qrtr-lookup
Service Version Instance Node Port
15 1 0 8 1 Test service
69 1 8 8 2 ATH10k WLAN firmware service
15 1 0 24 1 Test service
69 1 24 24 2 ATH10k WLAN firmware service

Here 8 and 24 on column 3 (QMI Instance ID)
and 4 (QRTR Node ID) are the node IDs that
is unique per mhi device.

Signed-off-by: Gokul Sriram Palanisamy <gokulsri@xxxxxxxxxxxxxx>
---
drivers/bus/mhi/core/boot.c | 14 ++++++++++++++
1 file changed, 14 insertions(+)

diff --git a/drivers/bus/mhi/core/boot.c b/drivers/bus/mhi/core/boot.c
index c2546bf..5e5dad5 100644
--- a/drivers/bus/mhi/core/boot.c
+++ b/drivers/bus/mhi/core/boot.c
@@ -16,8 +16,12 @@
#include <linux/random.h>
#include <linux/slab.h>
#include <linux/wait.h>
+#include <linux/pci.h>
#include "internal.h"

+#define QRTR_INSTANCE_MASK 0x000000FF
+#define QRTR_INSTANCE_SHIFT 0
+
/* Setup RDDM vector table for RDDM transfer and program RXVEC */
void mhi_rddm_prepare(struct mhi_controller *mhi_cntrl,
struct image_info *img_info)
@@ -391,6 +395,9 @@ void mhi_fw_load_handler(struct mhi_controller *mhi_cntrl)
const struct firmware *firmware = NULL;
struct image_info *image_info;
struct device *dev = &mhi_cntrl->mhi_dev->dev;
+ struct pci_dev *pci_dev = to_pci_dev(mhi_cntrl->cntrl_dev);
+ struct pci_bus *bus = pci_dev->bus;
+ uint32_t instance;
const char *fw_name;
void *buf;
dma_addr_t dma_addr;
@@ -466,6 +473,13 @@ void mhi_fw_load_handler(struct mhi_controller *mhi_cntrl)
return;
}

+ instance = ((pci_domain_nr(bus) & 0xF) << 4) | (bus->number & 0xF);
+ instance &= QRTR_INSTANCE_MASK;
+
+ mhi_write_reg_field(mhi_cntrl, mhi_cntrl->bhi,
+ BHI_ERRDBG2, QRTR_INSTANCE_MASK,
+ QRTR_INSTANCE_SHIFT, instance);

You cannot not do this in MHI stack. Why can't you do this in the MHI controller
specific to QCN9000? And btw, is QCN9000 supported in mainline?

Thanks,
Mani

+
write_lock_irq(&mhi_cntrl->pm_lock);
mhi_cntrl->dev_state = MHI_STATE_RESET;
write_unlock_irq(&mhi_cntrl->pm_lock);
--
2.7.4


As others have stated, please refrain from adding protocol specific code (such as PCIe)
in the MHI core driver. Please have this change in your controller.

If there is access to BHI registers required prior to power up from MHI core, it is not
exposed right now. We can talk about how you can achieve that, so you can do this write
in your controller after mhi_prepare_for_power_up().

Thanks,
Bhaumik
---
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
a Linux Foundation Collaborative Project