On Thu, Feb 19, 2026 at 11:01:26AM -0500, Frank Li wrote:

On Thu, Feb 19, 2026 at 04:11:49PM +0100, max.oss.09@gmail.com wrote:

quoted

From: Max Krummenacher <redacted>

This reverts commit 8124b4a4a96b57d6cc3705a9df9623c52baa047b.

The change introduced a regression: at least Colibri iMX6ULL and
Colibri iMX7 no longer boot with that commit applied, while they boot
again after reverting it.

Although this has only been verified on these two modules, the issue
is expected to affect all device trees using the gpmi-nand driver.

[    0.876938] Creating 5 MTD partitions on "gpmi-nand":
[    0.876974] 0x000000000000-0x000000080000 : "mx7-bcb"
[    0.879860] 0x000000080000-0x000000200000 : "u-boot1"
[    0.884761] 0x000000200000-0x000000380000 : "u-boot2"
[    0.886993] 0x000000380000-0x000000400000 : "u-boot-env"
[    0.894686] 0x000000400000-0x000020000000 : "ubi"
[    0.899054] gpmi-nand 33002000.nand-controller: driver registered.
...
[    0.960443] ubi0: default fastmap pool size: 200
[    0.960476] ubi0: default fastmap WL pool size: 100
[    0.960500] ubi0: attaching mtd4
[    1.636355] ubi0 error: scan_peb: bad image sequence number 1588722158 in PEB 4060, expected 1574791632
...
[    1.649889] ubi0 error: ubi_attach_mtd_dev: failed to attach mtd4, error -22
[    1.650029] UBI error: cannot attach mtd4
...
[    1.670262] Kernel panic - not syncing: VFS: Unable to mount root fs on unknown-block(0,253)

Can you help found the real reason why not boot?
nand@0 is preferred format for MTD raw nand.

I already raised the issue when you sent this patch series, we need to
consider the other users of this DT blob, for example U-Boot.

Please see my comments and Krzysztof one here [1]

I agree on investigating this, but I see no other option than reverting
for now.

[1] https://lore.kernel.org/all/20251105115538.GA17091@francesco-nb/ (local)

Francesco