Smatch reports:
drivers/platform/x86/pmc_atom.c:496
pmc_setup_dev() warn: 'pmc->regmap' not released on lines: 496.
Fix this by deallocating pm->regmap when pmc_setup_clks fails.
Fixes: 282a4e4ce5f9("platform/x86: Enable Atom PMC platform clocks")
Signed-off-by: Dongliang Mu <[email protected]>
---
drivers/platform/x86/pmc_atom.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/drivers/platform/x86/pmc_atom.c b/drivers/platform/x86/pmc_atom.c
index a9d2a4b98e57..e338c1572237 100644
--- a/drivers/platform/x86/pmc_atom.c
+++ b/drivers/platform/x86/pmc_atom.c
@@ -488,9 +488,11 @@ static int pmc_setup_dev(struct pci_dev *pdev, const struct pci_device_id *ent)
/* Register platform clocks - PMC_PLT_CLK [0..5] */
ret = pmc_setup_clks(pdev, pmc->regmap, data);
- if (ret)
+ if (ret) {
+ iounmap(pmc->regmap);
dev_warn(&pdev->dev, "platform clocks register failed: %d\n",
ret);
+ }
pmc->init = true;
return ret;
--
2.25.1
On Sun, Dec 05, 2021 at 04:44:19PM +0800, Dongliang Mu wrote:
> Smatch reports:
>
> drivers/platform/x86/pmc_atom.c:496
> pmc_setup_dev() warn: 'pmc->regmap' not released on lines: 496.
>
> Fix this by deallocating pm->regmap when pmc_setup_clks fails.
>
> Fixes: 282a4e4ce5f9("platform/x86: Enable Atom PMC platform clocks")
> Signed-off-by: Dongliang Mu <[email protected]>
> ---
It says v2 in the subject, what is in v2 exactly in comparison to v1?
...
> ret = pmc_setup_clks(pdev, pmc->regmap, data);
> - if (ret)
> + if (ret) {
> + iounmap(pmc->regmap);
> dev_warn(&pdev->dev, "platform clocks register failed: %d\n",
> ret);
In this case it makes sense to move to
dev_err(&pdev->dev, "platform clocks register failed: %d\n", ret);
> + }
--
With Best Regards,
Andy Shevchenko
On Mon, Dec 06, 2021 at 04:00:13PM +0200, Andy Shevchenko wrote:
> On Sun, Dec 05, 2021 at 04:44:19PM +0800, Dongliang Mu wrote:
> > Smatch reports:
> >
> > drivers/platform/x86/pmc_atom.c:496
> > pmc_setup_dev() warn: 'pmc->regmap' not released on lines: 496.
> >
> > Fix this by deallocating pm->regmap when pmc_setup_clks fails.
> >
> > Fixes: 282a4e4ce5f9("platform/x86: Enable Atom PMC platform clocks")
> > Signed-off-by: Dongliang Mu <[email protected]>
> > ---
>
> It says v2 in the subject, what is in v2 exactly in comparison to v1?
>
> ...
>
> > ret = pmc_setup_clks(pdev, pmc->regmap, data);
> > - if (ret)
> > + if (ret) {
> > + iounmap(pmc->regmap);
> > dev_warn(&pdev->dev, "platform clocks register failed: %d\n",
> > ret);
>
> In this case it makes sense to move to
>
> dev_err(&pdev->dev, "platform clocks register failed: %d\n", ret);
>
> > + }
Also, what about cleaning up debugfs?
--
With Best Regards,
Andy Shevchenko
On Mon, Dec 6, 2021 at 10:01 PM Andy Shevchenko
<[email protected]> wrote:
>
> On Sun, Dec 05, 2021 at 04:44:19PM +0800, Dongliang Mu wrote:
> > Smatch reports:
> >
> > drivers/platform/x86/pmc_atom.c:496
> > pmc_setup_dev() warn: 'pmc->regmap' not released on lines: 496.
> >
> > Fix this by deallocating pm->regmap when pmc_setup_clks fails.
> >
> > Fixes: 282a4e4ce5f9("platform/x86: Enable Atom PMC platform clocks")
> > Signed-off-by: Dongliang Mu <[email protected]>
> > ---
>
> It says v2 in the subject, what is in v2 exactly in comparison to v1?
Sorry to forget to add this information
v1->v2: move iounmap(pmc->regmap); from success path to the error handling code.
>
> ...
>
> > ret = pmc_setup_clks(pdev, pmc->regmap, data);
> > - if (ret)
> > + if (ret) {
> > + iounmap(pmc->regmap);
> > dev_warn(&pdev->dev, "platform clocks register failed: %d\n",
> > ret);
>
> In this case it makes sense to move to
>
> dev_err(&pdev->dev, "platform clocks register failed: %d\n", ret);
>
> > + }
>
> --
> With Best Regards,
> Andy Shevchenko
>
>
On Sun, Dec 05, 2021 at 04:44:19PM +0800, Dongliang Mu wrote:
> Smatch reports:
>
> drivers/platform/x86/pmc_atom.c:496
> pmc_setup_dev() warn: 'pmc->regmap' not released on lines: 496.
>
> Fix this by deallocating pm->regmap when pmc_setup_clks fails.
>
> Fixes: 282a4e4ce5f9("platform/x86: Enable Atom PMC platform clocks")
> Signed-off-by: Dongliang Mu <[email protected]>
Okay, this driver is two in one, i.e. library for PMC access and some other
stuff. When you do iounmap() you ruin library routines with that.
So, my other comments are fine in general, but not applicable here.
NAK.
--
With Best Regards,
Andy Shevchenko