Re: [PATCH v2] loongarch: Support RANDOMIZE_KSTACK_OFFSET

[Jinjie Ruan <ruanjinjie@huawei.com>]

On 2024/7/20 0:01, Kees Cook wrote:
> On Fri, Jul 19, 2024 at 11:14:27AM +0800, Jinjie Ruan wrote:
>> Add support of kernel stack offset randomization while handling syscall,
>> the offset is defaultly limited by KSTACK_OFFSET_MAX().
>>
>> In order to avoid trigger stack canaries (due to __builtin_alloca) and
>> slowing down the entry path, use __no_stack_protector attribute to
>> disable stack protector for do_syscall() at function level.
>>
>> With this patch, the REPORT_STACK test show that:
>> 	`loongarch64 bits of stack entropy: 7`
> 
> I suspect this will report the correct "6" after now that this commit
> has landed:
> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=872bb37f6829d4f7f3ed5afe2786add3d4384b4b

Hi, Kees

I noticed your patch, and I reconfirm that I have updated to the latest
mainline and that your patch is in the code.

However，the following REPORT_STACK test of your below script has the
same result (run multiple times).

And riscv64, arm64, x86 also has the 7 bit  of stack entropy.

for i in $(seq 1 1000); do
	echo "REPORT_STACK" >/sys/kernel/debug/provoke-crash/DIRECT
done
offsets=$(dmesg | grep 'Stack offset' | cut -d: -f3 | sort | uniq -c |
sort -n | wc -l)
echo "$(uname -m) bits of stack entropy: $(echo "obase=2; $offsets" | bc
| wc -L)"

> 
>>
>> Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com>
>> Suggested-by: Huacai Chen <chenhuacai@kernel.org>
> 
> Thanks for adding this and getting it tested!
> 
> Reviewed-by: Kees Cook <kees@kernel.org>
>