Apalis iMX6Q 2GB IT V1.1Y
Evaluation board Apalis
Linux BSP 2.8 stable + Ethernet driver fixes for newer revisions
Hello to all from Toradex Support and other souls.
I would like to post and describe quite serious trouble we have with our imx6 Apalis 1.1Y modules.
As first, let me write down dome points of our usage:
We use Apalis Modules imx6 for more than 2 years, on very lightly customized production board based on Apalis Evaluation board ( just some minor peripheral changes ).
We started on module version 1.1C, than continue with 1.1Z, now working with 1.1Y revision.
Modules are used in LCD transportation infotainment, or touch terminals.
In production, image and other settings are done from USB flash by Toradex Easy installer 1.8. Never experienced trouble.
Until we start to use revision 1.1Y, we deliver nearly thousand devices ( yes, that’s only founded guess ). Since delivery start until use of 1.1Y, we had problem with modules only few times.
Definitively not more than 10 times. We have been contacted by customer, than device don’t work, and screen is black.
In most cases, we just reflash memory in situ. Five or so modules just died, it happens.
This is OK and we can work with this.
Since we start to use module with revision 1.1Y, we started received reports from customers, that devices don’t work.
After service, we found that modules 1.1Y has corrupted eMMC memory, and modul itself is unusable.
Modules worked only for few weeks, placed in different vehicles, places, customers.
Only in last two weeks, I got 6 of them, and I guess more is to come. One justs died in front of me, only after two USB reflash and 3 hours of work.
The error is still the same.
Module linux boot with memory exceptions
USB easy installer failed to work with message
[ 39.849577] sdhci-esdhc-imx 2198000.usdhc: Card stuck in wrong state! card_busy_detect status: 0xd00
[ 39.858887] blk_update_request: I/O error, dev mmcblk0, sector 0 op 0x1:(WRITE) flags 0x0 phys_seg 1 prio class 0
[ 39.869203] Buffer I/O error on dev mmcblk0, logical block 0, lost async page write
[ 5.724416] mmcblk0: card_busy_detect: error sending status cmd, status 0x4080d00
[ 15.752546] mmc0: Timeout waiting for hardware interrupt. retries left=0 opcode=19
[ 15.761614] mmcblk0: error -110 transferring data, sector 0, nr 8, cmd response 0x0, card status 0x400d00
[ 15.774546] blk_update_request: I/O error, dev mmcblk0, sector 0
[ 15.780564] Buffer I/O error on dev mmcblk0, logical block 0, lost async page write
[ 15.793028] mmcblk0: error -84 transferring data, sector 0, nr 8, cmd response 0x900, card status 0xb00
[ 15.809490] mmcblk0: error -84 transferring data, sector 0, nr 8, cmd response 0x900, card status 0x0
[ 15.818739] blk_update_request: I/O error, dev mmcblk0, sector 0
Modul cannot be recovered, not with USB update, not on Apalis Evaluation board with tezi recovery, nothing.
To be clear, I tried all possible TEZI versions, uboots and other stuff.
Same behavior with new BSP 5.5
There is no excessive writing into memory while device operates ( logs are written into ram )
Modules dies after few weeks of work.
By my guess, we have around 50 1.1Y revision devices deployed and working. 6 Modules 1.1Y failed in lasts two weeks.
We have almost 300revision 1.1Y devices shipped and before deployment .
While memory cannot by reused, I assume there is some HW problem with memory itself.
So, here is my point
- What is the deal with new eMMC memory ?
What error means, Im not able to do anything with module.
What happened, we are not able to recover module. Is there any possible way ?
In revision description is stated, that no other SW modification is needed, in case of use module 1.1Y instead of 1.1Z. So it should work as usual.
According to pages, use of TEZI 1.8 and older 2. Stable BSP has no other requirements. If not, what changes and why ?
How to avoid this terrible error ?
Thanks in advance, hope this will help to resolve the issue.