[v2 PATCH] arm64: kprobes: call set_memory_rox() for kprobe page

[Yang Shi]

On 9/18/25 10:26 AM, Will Deacon wrote:
> On Thu, Sep 18, 2025 at 09:23:49AM -0700, Yang Shi wrote:
>> The kprobe page is allocated by execmem allocator with ROX permission.
>> It needs to call set_memory_rox() to set proper permission for the
>> direct map too. It was missed.
>>
>> Fixes: 10d5e97c1bf8 ("arm64: use PAGE_KERNEL_ROX directly in alloc_insn_page")
>> Cc: <stable at vger.kernel.org>
>> Signed-off-by: Yang Shi <yang at os.amperecomputing.com>
>> Reviewed-by: Catalin Marinas <catalin.marinas at arm.com>
>> ---
>> v2: Separated the patch from BBML2 series since it is an orthogonal bug
>>      fix per Ryan.
>>      Fixed the variable name nit per Catalin.
>>      Collected R-bs from Catalin.
>>
>>   arch/arm64/kernel/probes/kprobes.c | 12 ++++++++++++
>>   1 file changed, 12 insertions(+)
>>
>> diff --git a/arch/arm64/kernel/probes/kprobes.c b/arch/arm64/kernel/probes/kprobes.c
>> index 0c5d408afd95..8ab6104a4883 100644
>> --- a/arch/arm64/kernel/probes/kprobes.c
>> +++ b/arch/arm64/kernel/probes/kprobes.c
>> @@ -10,6 +10,7 @@
>>   
>>   #define pr_fmt(fmt) "kprobes: " fmt
>>   
>> +#include <linux/execmem.h>
>>   #include <linux/extable.h>
>>   #include <linux/kasan.h>
>>   #include <linux/kernel.h>
>> @@ -41,6 +42,17 @@ DEFINE_PER_CPU(struct kprobe_ctlblk, kprobe_ctlblk);
>>   static void __kprobes
>>   post_kprobe_handler(struct kprobe *, struct kprobe_ctlblk *, struct pt_regs *);
>>   
>> +void *alloc_insn_page(void)
>> +{
>> +	void *addr;
>> +
>> +	addr = execmem_alloc(EXECMEM_KPROBES, PAGE_SIZE);
>> +	if (!addr)
>> +		return NULL;
>> +	set_memory_rox((unsigned long)addr, 1);
>> +	return addr;
>> +}
> Why isn't execmem taking care of this? It looks to me like the
> execmem_cache_alloc() path calls set_memory_rox() but the
> execmem_vmalloc() path doesn't?

execmem_cache_alloc() is just called if execmem ROX cache is enabled, 
but it currently just supported by x86. Included Mike to this thread who 
is the author of execmem ROX cache.

>
> It feels a bit bizarre to me that we have to provide our own wrapper
> (which is identical to what s390 does). Also, how does alloc_insn_page()
> handle the direct map alias on x86?

x86 handles it via execmem ROX cache.

Thanks,
Yang

>
> Will