i.MX6SoloLite kernel hang following TLB invalidate

[JohnKaye]

Hi All,

We're migrating our Android 4.2 platform from an iMX6Solo design to an iMXSoloLite design, and we're experiencing a hang while flushing the TLBs (local_flush_tlb_all) during Linux (mfg-tool configured) boot.

Specifics:

• Our SoloLite design is based on the FSL EVK design, which the major difference being that DDR3 is used rather than LPDDR2.
• DDR3 was properly calibrated using the FSL DDR stress/calibration tool. The results were successfully used for the initial IOMUX and MMDC configuration.
• u-boot runs without any problems.
• Linux/Android release based on FSL Android JB4.2.2_1.0.0-ga release.
• HW specifics: i.MXSoloLite T01.9 @ 800MHZ; Silicon rev 1.2; 512MB DDR3
• PAGE_OFFSET and PHYS_OFFSET point at 0x80000000

The hang occurs while the mfg tool version of the kernel is booting: devicemaps_init() --> local_flush_tlb_all() --> invalidate all TLBs.

Any help in understanding or troubleshooting this problem would be greatly appreciated.

Cheers,

-John

[lytkinanton]

So. here is working config for 3.0.35

https://yadi.sk/d/W5YMECttg2cdY

Default configs for imx6sl are not working for me.

[justin_jiang]

Can you send all the mfgtool u-boot/kernel log to us?

[JohnKaye]

Thank you for the reply. I have resolved this issue. I have a new problem! During boot, the CPU is stuck in an infinite loop waiting for the GPU to complete its reset -- it's stuck in _ResetGPU. I've validated that the LDOs are setup properly (we're using internal LDOs). Any pointers to solve this issue would be greatly appreciated.

It's an im6sl design using the jb4.3_1.1.0-ga version of the kernel.

Thanks,

-John

[sanjayshekar]

Hello John Kaye,

Nice to hear that you resolved the TBL flush issue. We are migrating from 3.0.35 Linux to 3.11 linux are during bootup it hangs in devicemaps_init() --> local_flush_tlb_all() --> invalidate all TLBs.

Please suggest us the solution which helped you to solve this problem.

Thank you

Regards,

Sanju.

[lytkinanton]

I assume that something wrong with kernel configuration. It means that I have this problem too... hope John will help us with this issue :smileyhappy: