Hi Shyam,
On 6/28/21 1:04 PM, Shyam Sundar S K wrote:
> The protocol to submit a job request to SMU is to wait for
> AMD_PMC_REGISTER_RESPONSE to return 1,meaning SMU is ready to take
> requests. PMC driver has to make sure that the response code is always
> AMD_PMC_RESULT_OK before making any command submissions.
>
> When we submit a message to SMU, we have to wait until it processes
> the request. Adding a read_poll_timeout() check as this was missing in
> the existing code.
>
> Also, add a mutex to protect amd_pmc_send_cmd() calls to SMU.
>
> Fixes: 156ec4731cb2 ("platform/x86: amd-pmc: Add AMD platform support for S2Idle")
> Signed-off-by: Shyam Sundar S K <Shyam-sundar.S-k@xxxxxxx>
> ---
> v2: no change
> v3:
> - use mutex to protect multiple calls to SMU
> - add a switch-case to handle smu responses.
I'm sorry, but this still needs a bit more work, see below.
> drivers/platform/x86/amd-pmc.c | 29 +++++++++++++++++++++++++++--
> 1 file changed, 27 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/platform/x86/amd-pmc.c b/drivers/platform/x86/amd-pmc.c
> index b9da58ee9b1e..f3f5f754f75c 100644
> --- a/drivers/platform/x86/amd-pmc.c
> +++ b/drivers/platform/x86/amd-pmc.c
> @@ -68,6 +68,7 @@ struct amd_pmc_dev {
> u32 base_addr;
> u32 cpu_id;
> struct device *dev;
> + struct mutex lock; /* generic mutex lock */
> #if IS_ENABLED(CONFIG_DEBUG_FS)
> struct dentry *dbgfs_dir;
> #endif /* CONFIG_DEBUG_FS */
> @@ -138,9 +139,10 @@ static int amd_pmc_send_cmd(struct amd_pmc_dev *dev, bool set)
> u8 msg;
> u32 val;
>
> + mutex_lock(&dev->lock);
> /* Wait until we get a valid response */
> rc = readx_poll_timeout(ioread32, dev->regbase + AMD_PMC_REGISTER_RESPONSE,
> - val, val > 0, PMC_MSG_DELAY_MIN_US,
> + val, val != 0, PMC_MSG_DELAY_MIN_US,
> PMC_MSG_DELAY_MIN_US * RESPONSE_REGISTER_LOOP_MAX);
> if (rc) {
> dev_err(dev->dev, "failed to talk to SMU\n");
> @@ -156,7 +158,28 @@ static int amd_pmc_send_cmd(struct amd_pmc_dev *dev, bool set)
> /* Write message ID to message ID register */
> msg = (dev->cpu_id == AMD_CPU_ID_RN) ? MSG_OS_HINT_RN : MSG_OS_HINT_PCO;
> amd_pmc_reg_write(dev, AMD_PMC_REGISTER_MESSAGE, msg);
> - return 0;
> + /* Wait until we get a valid response */
> + rc = readx_poll_timeout(ioread32, dev->regbase + AMD_PMC_REGISTER_RESPONSE,
> + val, val != 0, PMC_MSG_DELAY_MIN_US,
> + PMC_MSG_DELAY_MIN_US * RESPONSE_REGISTER_LOOP_MAX);
> + if (rc)
> + dev_err(dev->dev, "SMU response timed out\n");
If rc != 0 then val will still be 0, so this function will end up
returning 0 while it should return an error.
Please add a rc = -ETIME; or rc = -ETIMEDOUT + goto out_unlock here.
> +
> + switch (val) {
> + case AMD_PMC_RESULT_OK:
> + break;
> + case AMD_PMC_RESULT_CMD_REJECT_BUSY:
> + case AMD_PMC_RESULT_CMD_REJECT_PREREQ:
> + case AMD_PMC_RESULT_CMD_UNKNOWN:
> + case AMD_PMC_RESULT_FAILED:
> + dev_err(dev->dev, "SMU not ready. err: 0x%x\n", val);
> + rc = -EBUSY;
> + goto out_unlock;
The rc should different for different val values, e.g.
CMD_REJECT_BUSY -> -EBUSY, CMD_UNKNOWN -> EINVAL
and map the others to -EIO.
Also you should have a default case to handle the case where the hw
returns some other value (and also map that to -EIO.
Regards,
Hans
> + }
> +
> +out_unlock:
> + mutex_unlock(&dev->lock);
> + return rc;
> }
>
> static int __maybe_unused amd_pmc_suspend(struct device *dev)
> @@ -259,6 +282,7 @@ static int amd_pmc_probe(struct platform_device *pdev)
>
> amd_pmc_dump_registers(dev);
>
> + mutex_init(&dev->lock);
> platform_set_drvdata(pdev, dev);
> amd_pmc_dbgfs_register(dev);
> return 0;
> @@ -269,6 +293,7 @@ static int amd_pmc_remove(struct platform_device *pdev)
> struct amd_pmc_dev *dev = platform_get_drvdata(pdev);
>
> amd_pmc_dbgfs_unregister(dev);
> + mutex_destroy(&dev->lock);
> return 0;
> }
>
>
