linux-kernel - Re: [PATCH v1 RESEND] spmi: hisi-spmi-controller: Fix kernel panic on rmmod

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date: Fri, 2 Feb 2024 17:29:32 +0530
From: VAMSHI GAJJELA <vamshigajjela@...gle.com>
To: Johan Hovold <johan@...nel.org>
Cc: Mauro Carvalho Chehab <mchehab+huawei@...nel.org>, Stephen Boyd <sboyd@...nel.org>, 
	Johan Hovold <johan+linaro@...nel.org>, Caleb Connolly <caleb.connolly@...aro.org>, 
	linux-kernel@...r.kernel.org, manugautam@...gle.com
Subject: Re: [PATCH v1 RESEND] spmi: hisi-spmi-controller: Fix kernel panic on rmmod

On Mon, Jan 29, 2024 at 1:01 PM Johan Hovold <johan@...nel.org> wrote:
>
> On Mon, Jan 29, 2024 at 10:13:22AM +0530, VAMSHI GAJJELA wrote:
> > On Fri, Jan 26, 2024 at 1:48 PM Johan Hovold <johan@...nel.org> wrote:
> > >
> > > On Fri, Jan 26, 2024 at 11:41:53AM +0530, Vamshi Gajjela wrote:
> > > > Ensure consistency in spmi_controller pointers between
> > > > spmi_controller_remove/put and driver spmi_del_controller functions.
> > > > The former requires a pointer to struct spmi_controller, while the
> > > > latter passes a pointer of struct spmi_controller_dev, leading to a
> > > > "Null pointer exception".
> > > >
> > > > 'nr' member of struct spmi_controller, which serves as an identifier
> > > > for the controller/bus. This value is assigned a dynamic ID in
> > > > spmi_controller_alloc, and overriding it from the driver results in an
> > > > ida_free error "ida_free called for id=xx which is not allocated".
> > >
> > > No Fixes tag?
>
> > There isn't a bug, I will remove word "Fix"
>
> Both of the issues you point out above sounds like bugs that deserve a
> Fixes tag.
ACK
>
> > > > Signed-off-by: Vamshi Gajjela <vamshigajjela@...gle.com>
> > > > ---
> > > >  drivers/spmi/hisi-spmi-controller.c | 4 ++--
> > > >  1 file changed, 2 insertions(+), 2 deletions(-)
> > > >
> > > > diff --git a/drivers/spmi/hisi-spmi-controller.c b/drivers/spmi/hisi-spmi-controller.c
> > > > index 9cbd473487cb..af51ffe24072 100644
> > > > --- a/drivers/spmi/hisi-spmi-controller.c
> > > > +++ b/drivers/spmi/hisi-spmi-controller.c
> > > > @@ -303,7 +303,6 @@ static int spmi_controller_probe(struct platform_device *pdev)
> > > >
> > > >       spin_lock_init(&spmi_controller->lock);
> > > >
> > > > -     ctrl->nr = spmi_controller->channel;
>
> > This remains applicable, however, it could lead to a failure in the
> > spmi_ctrl_release, I
> > will refactor the patch to address this.
> > also "spmi_del_controller" is removed from 6.7.2
>
> No, this has not changed in 6.7.2, it has been removed from 6.8-rc1.
ACK
>
> > > >       ctrl->dev.parent = pdev->dev.parent;
> > > >       ctrl->dev.of_node = of_node_get(pdev->dev.of_node);
> > > >
> > > > @@ -326,7 +325,8 @@ static int spmi_controller_probe(struct platform_device *pdev)
> > > >
> > > >  static void spmi_del_controller(struct platform_device *pdev)
> > >
> > > This function does not exist in mainline so presumably this is some bug
> > > you've introduced in your downstream driver that you're trying to fix.
> > >
> > > So this patch looks all bogus.
>
> > spmi_del_controller is present until in 6.7.2, I have made this patch
> > in last week of Dec,
> > I should have checked before resending, apologies.
>
> The bug you found was apparently accidentally fixed by commit
> 490d88ef548d ("spmi: hisi-spmi-controller: Use
> devm_spmi_controller_add()") in 6.8-rc1 but I don't see any record of it
> having been backported yet.
>
> As it depends on new helper function that will likely not happen either.
>
> Perhaps you can split your patch in two separate fixes and ask the stable
> team to backport the driver-data one.
Sure, I will split this into two as suggested, Thanks.
>
> > > >  {
> > > > -     struct spmi_controller *ctrl = platform_get_drvdata(pdev);
> > > > +     struct spmi_controller_dev *spmi_controller = platform_get_drvdata(pdev);
> > > > +     struct spmi_controller *ctrl = spmi_controller->controller;
> > > >
> > > >       spmi_controller_remove(ctrl);
> > > >       spmi_controller_put(ctrl);
>
> Johan