summaryrefslogtreecommitdiff
path: root/include/uapi/linux
diff options
context:
space:
mode:
authorPeter-Simon Dieterich <peter-simon.dieterich@vaillant-group.com>2026-09-17 15:00:42 +0200
committerMiquel Raynal <miquel.raynal@bootlin.com>2026-09-25 15:33:51 +0200
commite97c701c09d3d8f23397f68478c7e2c7eaf87528 (patch)
tree6df70de0d1854239fbc4c32e8ec7c175bb23ef2b /include/uapi/linux
parent94edcda45d888c3bb1310de11e10e4537c166f00 (diff)
downloadlinux-next-e97c701c09d3d8f23397f68478c7e2c7eaf87528.tar.gz
linux-next-e97c701c09d3d8f23397f68478c7e2c7eaf87528.zip
mtd: rawnand: esmt: Disable broken timings feature on GD9FU2G8F3A
GigaDevice and ESMT share JEDEC manufacturer ID 0xc8, so GigaDevice raw NAND chips are handled by the ESMT manufacturer code. The GD9FU2G8F3A advertises SET/GET_FEATURES support in its parameter page, but its implementation of the timing mode feature is broken: GET_FEATURES(TIMING_MODE) reads back a value that does not match the mode previously written, which causes the following error during boot: nand: timing mode 5 not acknowledged by the NAND chip and falls back to timing mode 0. GigaDevice confirmed that the chip is compatible with all timing modes advertised in its parameter page, that SET_FEATURE(TIMING_MODE) is not required to enter them, and that it may be ignored by the chip. The timing mode is therefore a host-side concern only, and the readback value carries no information. Unflag the feature for this model so the core configures the controller and skips both the SET and GET, mirroring what nand_macronix.c does for the MX30LF AC series. Tested on a custom board with an imx6ull using gpmi-nand, this restores timing mode 5 operation and sequential reads improve from 7 MiB/s to 20 MiB/s. Signed-off-by: Peter-Simon Dieterich <peter-simon.dieterich@vaillant-group.com> Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Diffstat (limited to 'include/uapi/linux')
0 files changed, 0 insertions, 0 deletions