- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi Cypress,
We have a design that is using the CYPD5225 with TGL and a burnside bridge retimer on each port, however after the system exits S4/S5 after the first boot we cannot enumerate usb3/DP alt mode/thunderbolt unless we physically unplug/replug.
This is based on the CYPD5225-96BZXI_notebook_tbt project based on sdk 3.4.0.
After some discussion, we have a workaround of issuing a port reset command based on the system power state changes written to SYS_PWR_STATE register from the EC. (This was disabled on the project due to BC1.2 being disabled due to size requirements.
So I shortened the callback handler for this in app.c app_update_sys_pwr_state to only trigger dpm_typec_command (i, DPM_CMD_TYPEC_ERR_RECOVERY, NULL);
This fixes the issue of devices not being recognized after resume, however it will reset power to the device - which we want to avoid, as we cannot resume/hibernate etc without the device resetting.
Is there a way to trigger only re-configuring the retimer/SOC on resume?
I have noticed the following:
It seems the SDK code monitors VSYS - and will trigger retimer_set_evt(param_1,RT_EVT_VSYS_ADDED); for each retimer when VSYS is restored. However in our design the retimer is on a different supply than the CCG5225 (which is always on). So I do not think this will get triggered.
I tried to manually call this from our power state transition handler:
retimer_set_evt(TYPEC_PORT_0_IDX,RT_EVT_VSYS_ADDED); and RT_EVT_VSYS_REMOVED when we would enter/exit S0 (using the EC command handler app_update_sys_pwr_state) for each port, however this does not seem to enable the alternate modes through the retimer.
I did some additional digging and thinking, and I also wondered if the PD controller needs to notify the SOC to configure alt modes as well? perhaps through the call ridge_force_status_update or some other method?
- Labels:
-
USB EZ-PD Type-C
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi,
Recommend you have a define for SYS_PWR_STATE register,
#define SYSTEM_STATE_S0 (0)
#define SYSTEM_STATE_S3 (1)
#define SYSTEM_STATE_S4 (2)
#define SYSTEM_STATE_S5 (3)
And then, Adding below work around for case APP_EVT_TYPE_C_ERROR_RECOVERY: and under if (!typec_only)
#if BB_RETIMER_ENABLE
/* Turn the retimer off after it's been written to */
CyDelayUs(10);
if(gl_system_state != SYSTEM_STATE_S0)
timer_start(port, APP_RETIMER_DISABLE_WAIT_TIMER, APP_RETIMER_DISABLE_DELAY, app_retimer_disable_cbk);
else
{
is_system_state_changed[port] = false;
retimer_disable(port);
}
#if INTEL_BB_RETIMER_HPI_CTRL_ENABLE
is_retimer_configured = false;
#endif
#endif /* BB_RETIMER_ENABLE */
Best Regards,
Lisa
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi Lisa,
Thanks for your response,
I tried this, however none of these three items are defined in the SDK I have.
app_retimer_disable_cbk
is_system_state_changed
Also the symbol is_retimer_configured only builds for the backup firmware, it is undefined for the main firmware.
--------------- Build Started: 03/31/2021 17:53:52 Project: CYPD5225-96BZXI_notebook_tbt, Configuration: ARM MDK Generic Release ---------------
ERROR: L6218E: Undefined symbol is_retimer_configured (referred from app.o).
Maybe I should do what was done elsewhere in the firmware?
retimer_disable (port, (gl_system_state != NB_SYS_PWR_STATE_S0));
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
@ShifangZ_26 do you have any other suggestions for resolving this? Since the current SDK is missing components you suggested to modify.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hello Kyle,
Recommend you contact local FAE/AE to get further support. Otherwise, you have to customize the firmware for S3 and S4 state based on SDK example code.
Best Regards,
Lisa