Re: [PATCH] arch/x86/kernel/acpi/boot: fix buffer overflow on negative index in mp_config_acpi_gsi()

[Rafael J. Wysocki]

On Fri, Feb 17, 2023 at 2:17 PM Alexander Sapozhnikov <alsp705@xxxxxxxxx> wrote:
>
> Variable 'ioapic', which may receive negative value by calling
> function 'mp_find_ioapic' at boot.c:465, is used at
> io_apic.c:128 by passing as 1st parameter to function 'mpc_ioapic_id'
> at boot.c:466

And that function uses the given value for array indexing without
checking it, right?

I would rather call this "out of bounds access"

> Found by Linux Verification Center (linuxtesting.org) with SVACE.
>
> Signed-off-by: Alexander Sapozhnikov <alsp705@xxxxxxxxx>
> ---
>  arch/x86/kernel/acpi/boot.c | 6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)
>
> diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
> index 907cc98b1938..abb78822f164 100644
> --- a/arch/x86/kernel/acpi/boot.c
> +++ b/arch/x86/kernel/acpi/boot.c
> @@ -463,8 +463,10 @@ static void mp_config_acpi_gsi(struct device *dev, u32 gsi, int trigger,
>         mp_irq.srcbus = number;
>         mp_irq.srcbusirq = (((devfn >> 3) & 0x1f) << 2) | ((pin - 1) & 3);
>         ioapic = mp_find_ioapic(gsi);
> -       mp_irq.dstapic = mpc_ioapic_id(ioapic);
> -       mp_irq.dstirq = mp_find_ioapic_pin(ioapic, gsi);
> +       if (ioapic >= 0) {
> +               mp_irq.dstapic = mpc_ioapic_id(ioapic);
> +               mp_irq.dstirq = mp_find_ioapic_pin(ioapic, gsi);
> +       }
>
>         mp_save_irq(&mp_irq);

Is it still valid to call this if ioapic is negative?

>  #endif
> --
> 2.34.1
>