linux-6.19-rc4/kernel/kexec_core.c:425: Possible loss of data ?

linux-6.19-rc4/kernel/kexec_core.c:425: Possible loss of data ?

[David Binderman]

Hello there,

Static analyser cppcheck says:

linux-6.19-rc4/kernel/kexec_core.c:425:7: style: int result is assigned to long variable. If the variable is long to avoid loss of information, then you have loss of information. [truncLongCastAssignment]

Source code is

    size = (1 << order) << PAGE_SHIFT;

I have no idea how large order gets, but if it gets over about
30 or so, then IMHO there will be a loss of bits.

Perhaps some sanity check on the value of order would
be a good idea ? Either that or

    size = (1UL << order) << PAGE_SHIFT;

might be better code.

Regards

David Binderman

Re: linux-6.19-rc4/kernel/kexec_core.c:425: Possible loss of data ?

[bhe]

On 01/07/26 at 02:32pm, David Binderman wrote:
> Hello there,
> 
> Static analyser cppcheck says:
> 
> linux-6.19-rc4/kernel/kexec_core.c:425:7: style: int result is assigned to long variable. If the variable is long to avoid loss of information, then you have loss of information. [truncLongCastAssignment]
> 
> Source code is
> 
>     size = (1 << order) << PAGE_SHIFT;
> 
> I have no idea how large order gets, but if it gets over about
> 30 or so, then IMHO there will be a loss of bits.

We may not need to worry about it. In all places where
kimage_alloc_control_pages is called, the order is 0 or
get_order(KEXEC_CONTROL_PAGE_SIZE) on mips which is 2. 

arch/mips/include/asm/kexec.h:#define KEXEC_CONTROL_PAGE_SIZE (4096 + 3*4096)

[~]$ git grep kimage_alloc_control_pages
arch/arm64/kernel/machine_kexec.c:      struct page *page = kimage_alloc_control_pages(kimage, 0);
arch/x86/kernel/machine_kexec_64.c:     page = kimage_alloc_control_pages(image, 0);
include/linux/kexec.h:extern struct page *kimage_alloc_control_pages(struct kimage *image,
kernel/crash_core.c:    vmcoreinfo_page = kimage_alloc_control_pages(image, 0);
kernel/kexec.c: image->control_code_page = kimage_alloc_control_pages(image,
kernel/kexec.c:         image->swap_page = kimage_alloc_control_pages(image, 0);
kernel/kexec_core.c:struct page *kimage_alloc_control_pages(struct kimage *image,
kernel/kexec_file.c:    image->control_code_page = kimage_alloc_control_pages(image,
kernel/kexec_file.c:            image->swap_page = kimage_alloc_control_pages(image, 0);

> 
> Perhaps some sanity check on the value of order would
> be a good idea ? Either that or
> 
>     size = (1UL << order) << PAGE_SHIFT;
> 
> might be better code.

That is fine if people is worried, a patch is welcome.

Re: linux-6.19-rc4/kernel/kexec_core.c:425: Possible loss of data ?

[David Binderman]

Hello there,

>We may not need to worry about it. In all places where
>kimage_alloc_control_pages is called, the order is 0 or
>get_order(KEXEC_CONTROL_PAGE_SIZE) on mips which is 2.

Thanks for the update.

>That is fine if people is worried, a patch is welcome.

Since my kernel patch success rate is 0.0 % (approx) I will avoid doing this.
Might be worth an assert that the value is in range [0..30].

Regards

David Binderman