View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0004179 | fsimx_Linux | Display | public | 2020-01-29 10:13 | 2024-03-26 11:26 |
| Reporter | Keller | Assigned To | |||
| Priority | normal | Severity | minor | Reproducibility | always |
| Status | resolved | Resolution | fixed | ||
| Product Version | fsimx6-B2019.08 | ||||
| Target Version | fsimx8mx-Y2020.11-pre | Fixed in Version | fsimx6-Y2020.03 | ||
| Summary | 0004179: [armStoneA9r2 ] LVDS backlight enabled too early in boot process | ||||
| Description | A customer reports problems on armStoneA9r2 with the LVDS backlight that can be broken down to three observations: 1. When armStoneA9r2 is switched on, the backlight of the LVDS display has a short bright flash before the display contains any content. 2. Later when Linux starts, the backlight is enabled (VCLF_ON) before VLCD is anabled, i.e. again before any content is visible. 3. VLCD toggles several times when it is enabled in Linux before it is stable. | ||||
| Steps To Reproduce | Configure device tree for LVDS. Attach oscilloscope to VCFL_ON and VLCD signals. Switch on the board. Observe that VCFL_ON is enabled right after start and later in Linux before VLCD is valid and stable. | ||||
| Additional Information | 1. Unfortunately F&S has chosen the pin SD4_DAT0 for the VCFL_ON signal. This pin has a second function NAND_DQS. This signal is only needed for DDR NAND flashes, but as investigation has shown, it is nonetheless initialized by the CPU ROM Loader and used as output. So the reset state GPIO as shown in the reference manual is not true when booting from NAND. This can not be avoided by software. The pin is configured back to a GPIO in U-Boot, but that is rather late, so that a high signal can be seen on this pin for about 75ms. Unfortunately this pin is used on *all* F&S boards based on i.MX6. So we have this problem on all these boards. 2. The backlight is switched on when the backlight driver is started. Basically it should follow the VLCD value, but as the "names" entry in the device tree does not match the display driver name, it is never switched off again. Here we have to find a way how we can name the display driver in a consistent way so that we can refer to it in the backlight node in device tree for all our display combinations. Check that the backlight is not enabled by default, only after an unblank notification is issued. 3. When the driver is configured, it disables and enables VLCD for every single setting that is changed. This is way too much. It should be sufficient to enable VLCD once when the blanking is finally disabled. | ||||
| Forum Link | |||||
