Mmc_send_cmd_bounced: MMC Timeout on TX2

We have a custom product using TX2.
One of them is now showing this error and is not able to complete boot anymore.
We have been able to reflash, but after a while the error came back.
How can we confirm that internal emmc is damaged?

Logs: sales2 (24.5 KB)

Kind regards,

Looks something new.

Could you describe more detail about this problem?

For example, after reflash the board, how long will you hit this problem? Are you using devkit or custom board? Can you also put this module to devkit and test?

Hello WayneWWW,

It’s custom board.
We will put the module on devkit and let you know.

Kind regards,

1 Like

Hi WayneWWW,

After reflashing on devkit with Jetpack 4.6, we don’t see any errors.
The module doesn’t seem thus to be broken but we need now to understand why this error appeared in the first place.

Kind regards,

Hello WayneWWW,

After reflashing another TX2 module with jetpack 4.6 that also had emmc issues, it shows this error on boot on a devkit:

[  303.333817] Process 5501(apport) has RLIMIT_CORE set to 1
[  303.339240] Aborting core
[  303.448451] Process 5520(apport) has RLIMIT_CORE set to 1
[  303.453880] Aborting core
[  316.489713] EXT4-fs (mmcblk0p1): error count since last fsck: 1
[  316.495645] EXT4-fs (mmcblk0p1): initial error at time 1631916227: htree_dirblock_to_tree:1004: inode 525661: block 2105581
[  316.506793] EXT4-fs (mmcblk0p1): last error at time 1631916227: htree_dirblock_to_tree:1004: inode 525661: block 2105581

emmc-corrupted-rma.log (96.4 KB)

Any hints? Is it good also for RMA?

Kind regards,

Do you have Nvidia developer kit to do the test? It sounds like you are using custom board?

We are producing custom boards but for faulty TX2 modules we are testing with nvidia devkit and original nvidia rootfs sample with Jetpack 4.6.

Ok, then it sounds like hardware faulty. Please file RMA for these modules.

You better check whether this is carrier board issue or not. I mean sometimes it is the carrier board that has hardware problem.