x86/acpi: fix potential NULL deref in acpi_wakeup_cpu()

[PATCH] x86/acpi: fix potential NULL deref in acpi_wakeup_cpu()

Posted by Chenyuan Yang 10 months ago

The result of memremap() may be NULL on failure, leading to a NULL
dereference. Add explicit checks after memremap() call: if the
MADT mailbox fails to map, return immediately.

This is similar to the commit 966d47e1f27c
("efi: fix potential NULL deref in efi_mem_reserve_persistent").

This is found by our static analysis tool KNighter.

Signed-off-by: Chenyuan Yang <chenyuan0y@gmail.com>
Fixes: 2b5e22afae07 ("x86/acpi: Extract ACPI MADT wakeup code into a separate file")
---
 arch/x86/kernel/acpi/madt_wakeup.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/arch/x86/kernel/acpi/madt_wakeup.c b/arch/x86/kernel/acpi/madt_wakeup.c
index f36f28405dcc..b386ec4b87c2 100644
--- a/arch/x86/kernel/acpi/madt_wakeup.c
+++ b/arch/x86/kernel/acpi/madt_wakeup.c
@@ -143,6 +143,10 @@ static int acpi_wakeup_cpu(u32 apicid, unsigned long start_ip)
 		acpi_mp_wake_mailbox = memremap(acpi_mp_wake_mailbox_paddr,
 						sizeof(*acpi_mp_wake_mailbox),
 						MEMREMAP_WB);
+		if (!acpi_mp_wake_mailbox) {
+			pr_err("Failed to remap MADT mailbox\n");
+			return -ENOMEM;
+		}
 	}
 
 	/*
-- 
2.34.1

Re: [PATCH] x86/acpi: fix potential NULL deref in acpi_wakeup_cpu()

Posted by Rafael J. Wysocki 8 months, 3 weeks ago

On Fri, Apr 11, 2025 at 9:48 PM Chenyuan Yang <chenyuan0y@gmail.com> wrote:
>
> The result of memremap() may be NULL on failure, leading to a NULL
> dereference. Add explicit checks after memremap() call: if the
> MADT mailbox fails to map, return immediately.

This mailbox is called Multiprocessor Wakeup Mailbox in the
specification, so please follow the established terminology.

> This is similar to the commit 966d47e1f27c
> ("efi: fix potential NULL deref in efi_mem_reserve_persistent").
>
> This is found by our static analysis tool KNighter.
>
> Signed-off-by: Chenyuan Yang <chenyuan0y@gmail.com>
> Fixes: 2b5e22afae07 ("x86/acpi: Extract ACPI MADT wakeup code into a separate file")
> ---
>  arch/x86/kernel/acpi/madt_wakeup.c | 4 ++++
>  1 file changed, 4 insertions(+)
>
> diff --git a/arch/x86/kernel/acpi/madt_wakeup.c b/arch/x86/kernel/acpi/madt_wakeup.c
> index f36f28405dcc..b386ec4b87c2 100644
> --- a/arch/x86/kernel/acpi/madt_wakeup.c
> +++ b/arch/x86/kernel/acpi/madt_wakeup.c
> @@ -143,6 +143,10 @@ static int acpi_wakeup_cpu(u32 apicid, unsigned long start_ip)
>                 acpi_mp_wake_mailbox = memremap(acpi_mp_wake_mailbox_paddr,
>                                                 sizeof(*acpi_mp_wake_mailbox),
>                                                 MEMREMAP_WB);
> +               if (!acpi_mp_wake_mailbox) {
> +                       pr_err("Failed to remap MADT mailbox\n");

Please follow Kirill's advice on using pr_err_once() here and use the
proper terminology in the message (also this is mapping, not
remapping, even though memremap() is used).

> +                       return -ENOMEM;
> +               }
>         }
>
>         /*
> --

Thanks!

Re: [PATCH] x86/acpi: fix potential NULL deref in acpi_wakeup_cpu()

Posted by Sathyanarayanan Kuppuswamy 9 months, 3 weeks ago

On 4/11/25 12:48 PM, Chenyuan Yang wrote:
> The result of memremap() may be NULL on failure, leading to a NULL
> dereference. Add explicit checks after memremap() call: if the
> MADT mailbox fails to map, return immediately.
>
> This is similar to the commit 966d47e1f27c
> ("efi: fix potential NULL deref in efi_mem_reserve_persistent").
>
> This is found by our static analysis tool KNighter.
>
> Signed-off-by: Chenyuan Yang <chenyuan0y@gmail.com>
> Fixes: 2b5e22afae07 ("x86/acpi: Extract ACPI MADT wakeup code into a separate file")
> ---

Looks fine to me

Reviewed-by: Kuppuswamy Sathyanarayanan 
<sathyanarayanan.kuppuswamy@linux.intel.com>

>   arch/x86/kernel/acpi/madt_wakeup.c | 4 ++++
>   1 file changed, 4 insertions(+)
>
> diff --git a/arch/x86/kernel/acpi/madt_wakeup.c b/arch/x86/kernel/acpi/madt_wakeup.c
> index f36f28405dcc..b386ec4b87c2 100644
> --- a/arch/x86/kernel/acpi/madt_wakeup.c
> +++ b/arch/x86/kernel/acpi/madt_wakeup.c
> @@ -143,6 +143,10 @@ static int acpi_wakeup_cpu(u32 apicid, unsigned long start_ip)
>   		acpi_mp_wake_mailbox = memremap(acpi_mp_wake_mailbox_paddr,
>   						sizeof(*acpi_mp_wake_mailbox),
>   						MEMREMAP_WB);
> +		if (!acpi_mp_wake_mailbox) {
> +			pr_err("Failed to remap MADT mailbox\n");
> +			return -ENOMEM;
> +		}
>   	}
>   
>   	/*

-- 
Sathyanarayanan Kuppuswamy
Linux Kernel Developer

Re: [PATCH] x86/acpi: fix potential NULL deref in acpi_wakeup_cpu()

Posted by Rafael J. Wysocki 9 months, 3 weeks ago

On Fri, Apr 11, 2025 at 9:48 PM Chenyuan Yang <chenyuan0y@gmail.com> wrote:
>
> The result of memremap() may be NULL on failure, leading to a NULL
> dereference. Add explicit checks after memremap() call: if the
> MADT mailbox fails to map, return immediately.
>
> This is similar to the commit 966d47e1f27c
> ("efi: fix potential NULL deref in efi_mem_reserve_persistent").
>
> This is found by our static analysis tool KNighter.
>
> Signed-off-by: Chenyuan Yang <chenyuan0y@gmail.com>
> Fixes: 2b5e22afae07 ("x86/acpi: Extract ACPI MADT wakeup code into a separate file")

Well, it's good to add the author of the commit you're trying to fix
to the CC list.

Kirill, what do you think about this?

> ---
>  arch/x86/kernel/acpi/madt_wakeup.c | 4 ++++
>  1 file changed, 4 insertions(+)
>
> diff --git a/arch/x86/kernel/acpi/madt_wakeup.c b/arch/x86/kernel/acpi/madt_wakeup.c
> index f36f28405dcc..b386ec4b87c2 100644
> --- a/arch/x86/kernel/acpi/madt_wakeup.c
> +++ b/arch/x86/kernel/acpi/madt_wakeup.c
> @@ -143,6 +143,10 @@ static int acpi_wakeup_cpu(u32 apicid, unsigned long start_ip)
>                 acpi_mp_wake_mailbox = memremap(acpi_mp_wake_mailbox_paddr,
>                                                 sizeof(*acpi_mp_wake_mailbox),
>                                                 MEMREMAP_WB);
> +               if (!acpi_mp_wake_mailbox) {
> +                       pr_err("Failed to remap MADT mailbox\n");
> +                       return -ENOMEM;
> +               }
>         }
>
>         /*
> --
> 2.34.1
>

Re: [PATCH] x86/acpi: fix potential NULL deref in acpi_wakeup_cpu()

Posted by Kirill A. Shutemov 9 months, 3 weeks ago

On Tue, Apr 15, 2025 at 03:39:24PM +0200, Rafael J. Wysocki wrote:
> On Fri, Apr 11, 2025 at 9:48 PM Chenyuan Yang <chenyuan0y@gmail.com> wrote:
> >
> > The result of memremap() may be NULL on failure, leading to a NULL
> > dereference. Add explicit checks after memremap() call: if the
> > MADT mailbox fails to map, return immediately.
> >
> > This is similar to the commit 966d47e1f27c
> > ("efi: fix potential NULL deref in efi_mem_reserve_persistent").
> >
> > This is found by our static analysis tool KNighter.
> >
> > Signed-off-by: Chenyuan Yang <chenyuan0y@gmail.com>
> > Fixes: 2b5e22afae07 ("x86/acpi: Extract ACPI MADT wakeup code into a separate file")
> 
> Well, it's good to add the author of the commit you're trying to fix
> to the CC list.
> 
> Kirill, what do you think about this?

Looks reasonable to me.

We fail to remap a single page. It is likely to be fatally broken system
if we get there. But okay, let's handle it.

Maybe use pr_err_once(). No need to spam an error for every CPU.

> > ---
> >  arch/x86/kernel/acpi/madt_wakeup.c | 4 ++++
> >  1 file changed, 4 insertions(+)
> >
> > diff --git a/arch/x86/kernel/acpi/madt_wakeup.c b/arch/x86/kernel/acpi/madt_wakeup.c
> > index f36f28405dcc..b386ec4b87c2 100644
> > --- a/arch/x86/kernel/acpi/madt_wakeup.c
> > +++ b/arch/x86/kernel/acpi/madt_wakeup.c
> > @@ -143,6 +143,10 @@ static int acpi_wakeup_cpu(u32 apicid, unsigned long start_ip)
> >                 acpi_mp_wake_mailbox = memremap(acpi_mp_wake_mailbox_paddr,
> >                                                 sizeof(*acpi_mp_wake_mailbox),
> >                                                 MEMREMAP_WB);
> > +               if (!acpi_mp_wake_mailbox) {
> > +                       pr_err("Failed to remap MADT mailbox\n");
> > +                       return -ENOMEM;
> > +               }
> >         }
> >
> >         /*
> > --
> > 2.34.1
> >

-- 
  Kiryl Shutsemau / Kirill A. Shutemov