On Mon, Nov 08, 2021 at 02:36:42PM +0300, Dmitry Osipenko wrote:

quoted

08.11.2021 14:22, Jonathan Neuschäfer пишет:

quoted

On Sun, Nov 07, 2021 at 08:42:33PM +0300, Dmitry Osipenko wrote:
[...]

quoted

EC drivers tend to use higher priority in general. Jonathan, could you
please confirm that NTXEC driver is a more preferable restart method
than the watchdog?

Yes. The original firmware uses the NTXEC to restart, and it works well,
so I do think it's preferable.

Thank you, then I'll update the NTXEC patch like this:

https://github.com/grate-driver/linux/commit/22da3d91f1734d9a0ed036220ad4ea28465af988

I tested again, but sys_off_handler_reboot called a bogus pointer
(probably reboot_prepare_cb). I think it was left uninitialized in
ntxec_probe, which uses devm_kmalloc. I guess we could switch it to
devm_kzalloc:

diff --git a/drivers/mfd/ntxec.c b/drivers/mfd/ntxec.c
index 1f55dfce14308..30364beb4b1d0 100644
--- a/drivers/mfd/ntxec.c
+++ b/drivers/mfd/ntxec.c

@@ -144,7 +144,7 @@ static int ntxec_probe(struct i2c_client *client)
 	const struct mfd_cell *subdevs;
 	size_t n_subdevs;
 
-	ec = devm_kmalloc(&client->dev, sizeof(*ec), GFP_KERNEL);
+	ec = devm_kzalloc(&client->dev, sizeof(*ec), GFP_KERNEL);
 	if (!ec)
 		return -ENOMEM;
 

With that done, it works flawlessly.