clkdev: fix potential NULL pointer dereference

[PATCH] clkdev: fix potential NULL pointer dereference

Posted by Marek Szyprowski 1 week, 5 days ago

dev_fmt argument is optional, so avoid dereferencing it unconditionally.

Fixes: 4d11c62ca8d7 ("clkdev: report over-sized strings when creating clkdev entries")
Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
---

This fixes the following kernel panic observed on Samsung Exynos542x
SoCs family:

8<--- cut here ---
Unable to handle kernel NULL pointer dereference at virtual address 00000000 when read
[00000000] *pgd=00000000
Internal error: Oops: 5 [#1] PREEMPT SMP ARM
Modules linked in:
CPU: 0 PID: 0 Comm: swapper/0 Not tainted 6.8.0-00001-g4d11c62ca8d7 #14975
Hardware name: Samsung Exynos (Flattened Device Tree)
PC is at vsnprintf+0x4c/0x3c8
LR is at init_stack+0x1dd6/0x2000
pc : [<c0c09d3c>]    lr : [<c1301dd6>]    psr: 800000d3
...
Flags: Nzcv  IRQs off  FIQs off  Mode SVC_32  ISA ARM  Segment none
Control: 10c5387d  Table: 4000406a  DAC: 00000051
...
Process swapper/0 (pid: 0, stack limit = 0x(ptrval))
Stack: (0xc1301c70 to 0xc1302000)
...
 vsnprintf from va_format.constprop.0+0x70/0x160
 va_format.constprop.0 from pointer+0x454/0x670
 pointer from vsnprintf+0x228/0x3c8
 vsnprintf from vprintk_store+0x100/0x428
 vprintk_store from vprintk_emit+0x6c/0x328
 vprintk_emit from vprintk_default+0x24/0x2c
 vprintk_default from _printk+0x28/0x4c
 _printk from vclkdev_alloc+0xf0/0x110
 vclkdev_alloc from clkdev_hw_create+0x28/0x88
 clkdev_hw_create from clk_hw_register_clkdev+0x38/0x3c
 clk_hw_register_clkdev from samsung_clk_register_fixed_rate+0xa4/0xd4
 samsung_clk_register_fixed_rate from exynos5x_clk_init+0xf0/0x2b0
 exynos5x_clk_init from of_clk_init+0x15c/0x228
 of_clk_init from time_init+0x24/0x30
 time_init from start_kernel+0x4b8/0x620
 start_kernel from 0x0
Code: e09e9001 e1a05002 e28da01c 2a000086 (e5d50000)
---[ end trace 0000000000000000 ]---
Kernel panic - not syncing: Attempted to kill the idle task!
---[ end Kernel panic - not syncing: Attempted to kill the idle task! ]---

Best regards
Marek Szyprowski, PhD
Samsung R&D Institute Poland
---
 drivers/clk/clkdev.c | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)

diff --git a/drivers/clk/clkdev.c b/drivers/clk/clkdev.c
index ddacab7863d0..d2801ae70e34 100644
--- a/drivers/clk/clkdev.c
+++ b/drivers/clk/clkdev.c
@@ -194,10 +194,12 @@ vclkdev_alloc(struct clk_hw *hw, const char *con_id, const char *dev_fmt,
 	return &cla->cl;
 
 fail:
-	fmt.fmt = dev_fmt;
-	fmt.va = &ap_copy;
-	pr_err("%pV:%s: %s ID is greater than %zu\n",
-	       &fmt, con_id, failure, max_size);
+	if (dev_fmt) {
+		fmt.fmt = dev_fmt;
+		fmt.va = &ap_copy;
+		pr_err("%pV:%s: %s ID is greater than %zu\n",
+		       &fmt, con_id, failure, max_size);
+	}
 	va_end(ap_copy);
 	kfree(cla);
 	return NULL;
-- 
2.34.1

Re: [PATCH] clkdev: fix potential NULL pointer dereference

Posted by Sam Protsenko 1 week, 3 days ago

On Tue, May 7, 2024 at 1:45 AM Marek Szyprowski
<m.szyprowski@samsung.com> wrote:
>
> dev_fmt argument is optional, so avoid dereferencing it unconditionally.
>
> Fixes: 4d11c62ca8d7 ("clkdev: report over-sized strings when creating clkdev entries")
> Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
> ---
>

Thanks for fixing this, Marek! I tested it on the E850-96 board, and
it fixes the boot on current kernel-next for me (next-20240508). Feel
free to add:

Reviewed-by: Sam Protsenko <semen.protsenko@linaro.org>
Tested-by: Sam Protsenko <semen.protsenko@linaro.org>

[snip]

Re: [PATCH] clkdev: fix potential NULL pointer dereference

Posted by André Draszik 1 week, 4 days ago

Hi Marek,

On Tue, 2024-05-07 at 08:44 +0200, Marek Szyprowski wrote:
> diff --git a/drivers/clk/clkdev.c b/drivers/clk/clkdev.c
> index ddacab7863d0..d2801ae70e34 100644
> --- a/drivers/clk/clkdev.c
> +++ b/drivers/clk/clkdev.c
> @@ -194,10 +194,12 @@ vclkdev_alloc(struct clk_hw *hw, const char *con_id, const char *dev_fmt,
>  	return &cla->cl;
>  
>  fail:
> -	fmt.fmt = dev_fmt;
> -	fmt.va = &ap_copy;
> -	pr_err("%pV:%s: %s ID is greater than %zu\n",
> -	       &fmt, con_id, failure, max_size);
> +	if (dev_fmt) {
> +		fmt.fmt = dev_fmt;
> +		fmt.va = &ap_copy;
> +		pr_err("%pV:%s: %s ID is greater than %zu\n",
> +		       &fmt, con_id, failure, max_size);
> +	}

It might be nice to still print the rest of the error, so it's easier to see which
clock is causing trouble.

Cheers,
Andre'

Re: [PATCH] clkdev: fix potential NULL pointer dereference

Posted by Russell King (Oracle) 1 week, 4 days ago

On Tue, May 07, 2024 at 05:42:45PM +0100, André Draszik wrote:
> Hi Marek,
> 
> On Tue, 2024-05-07 at 08:44 +0200, Marek Szyprowski wrote:
> > diff --git a/drivers/clk/clkdev.c b/drivers/clk/clkdev.c
> > index ddacab7863d0..d2801ae70e34 100644
> > --- a/drivers/clk/clkdev.c
> > +++ b/drivers/clk/clkdev.c
> > @@ -194,10 +194,12 @@ vclkdev_alloc(struct clk_hw *hw, const char *con_id, const char *dev_fmt,
> >  	return &cla->cl;
> >  
> >  fail:
> > -	fmt.fmt = dev_fmt;
> > -	fmt.va = &ap_copy;
> > -	pr_err("%pV:%s: %s ID is greater than %zu\n",
> > -	       &fmt, con_id, failure, max_size);
> > +	if (dev_fmt) {
> > +		fmt.fmt = dev_fmt;
> > +		fmt.va = &ap_copy;
> > +		pr_err("%pV:%s: %s ID is greater than %zu\n",
> > +		       &fmt, con_id, failure, max_size);
> > +	}
> 
> It might be nice to still print the rest of the error, so it's easier to see which
> clock is causing trouble.

Good point. I'll fix the patch myself, merging the fix in.

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!

Re: [PATCH] clkdev: fix potential NULL pointer dereference

Posted by Russell King (Oracle) 1 week, 5 days ago

On Tue, May 07, 2024 at 08:44:34AM +0200, Marek Szyprowski wrote:
> dev_fmt argument is optional, so avoid dereferencing it unconditionally.
> 
> Fixes: 4d11c62ca8d7 ("clkdev: report over-sized strings when creating clkdev entries")
> Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>

Please put this in the patch system so it can be merged along with the
change that created the problem. Thanks.

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!