Resend: Regression in 5467801f1fcbd

Resend: Regression in 5467801f1fcbd

[Daniel Harding]

[Apologies for the original HTML email]

The commit "Restrict usage of GPIO chip irq members before 
initialization" breaks suspend on a Dell Inspiron 5515 laptop in a very 
severe way.  Suspending with this commit present causes the machine to 
lock up hard.  The only way to recover is to disconnect mains power, 
open up the case, disconnect the battery, and hold down the power 
button.  Bisecting pointed to 2c1fa3614795e2b24da1ba95de0b27b8f6ea4537 
in 5.16.20.  Testing with the source commit, 
5467801f1fcbdc46bc7298a84dbf3ca1ff2a7320 confirmed that it was the one 
that introduced the problem. Unfortunately, this commit was backported 
to multiple stable kernels:  5.17.3, 5.16.20, 5.15.34, and 5.10.111.

I have not yet done any debugging to determine exactly why this commit 
causing things to break, but am happy to try out any fixes over the next 
couple of days until I put my laptop back together properly.

Regards,

Daniel Harding

Re: Resend: Regression in 5467801f1fcbd

[Shreeya Patel]

On 26/04/22 00:48, Daniel Harding wrote:
> [Apologies for the original HTML email]
>
> The commit "Restrict usage of GPIO chip irq members before 
> initialization" breaks suspend on a Dell Inspiron 5515 laptop in a 
> very severe way.  Suspending with this commit present causes the 
> machine to lock up hard.  The only way to recover is to disconnect 
> mains power, open up the case, disconnect the battery, and hold down 
> the power button.  Bisecting pointed to 
> 2c1fa3614795e2b24da1ba95de0b27b8f6ea4537 in 5.16.20.  Testing with the 
> source commit, 5467801f1fcbdc46bc7298a84dbf3ca1ff2a7320 confirmed that 
> it was the one that introduced the problem. Unfortunately, this commit 
> was backported to multiple stable kernels:  5.17.3, 5.16.20, 5.15.34, 
> and 5.10.111.
>
> I have not yet done any debugging to determine exactly why this commit 
> causing things to break, but am happy to try out any fixes over the 
> next couple of days until I put my laptop back together properly.
>
Hi Daniel,

FYI, The fix for the bug has already been sent by Mario and it has been 
picked up by Linus.
https://lore.kernel.org/lkml/20220422131452.20757-2-mario.limonciello@amd.com/


Thanks,
Shreeya Patel

> Regards,
>
> Daniel Harding

Re: Regression in 5467801f1fcbd

[Linus Walleij]

Hi Daniel,

On Mon, Apr 25, 2022 at 9:00 PM Daniel Harding <dharding@living180.net> wrote:

> The commit "Restrict usage of GPIO chip irq members before initialization" breaks suspend on a
>  Dell Inspiron 5515 laptop in a very severe way.

Does this commit, which is already upstream in Torvald's tree, solve the issue?
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/drivers/gpio?id=06fb4ecfeac7e00d6704fa5ed19299f2fefb3cc9

Yours,
Linus Walleij

Re: Regression in 5467801f1fcbd

[Daniel Harding]

On 4/25/22 23:09, Linus Walleij wrote:
> Hi Daniel,
>
> On Mon, Apr 25, 2022 at 9:00 PM Daniel Harding <dharding@living180.net> wrote:
>
>> The commit "Restrict usage of GPIO chip irq members before initialization" breaks suspend on a
>>   Dell Inspiron 5515 laptop in a very severe way.
> Does this commit, which is already upstream in Torvald's tree, solve the issue?
> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/drivers/gpio?id=06fb4ecfeac7e00d6704fa5ed19299f2fefb3cc9

Yes, cherry-picking that commit on top of 5467801f1fcbd fixes the issue 
for me.  Thank so much for the quick response from you and Shreeya!

Regards,

Daniel Harding