2022-05-23 08:13:38

by Liao, Chang

Subject: [PATCH] riscv/kprobe: reclaim insn_slot on kprobe unregistration

On kprobe registration kernel allocate one insn_slot for new kprobe,
but it forget to reclaim the insn_slot on unregistration, leading to a
potential leakage.

This bug reported by Chen Guokai <[email protected]>.

Signed-off-by: Liao Chang <[email protected]>
---
arch/riscv/kernel/probes/kprobes.c | 4 ++++
1 file changed, 4 insertions(+)

diff --git a/arch/riscv/kernel/probes/kprobes.c b/arch/riscv/kernel/probes/kprobes.c
index e6e950b7cf32..f12eb1fbb52c 100644
--- a/arch/riscv/kernel/probes/kprobes.c
+++ b/arch/riscv/kernel/probes/kprobes.c
@@ -110,6 +110,10 @@ void __kprobes arch_disarm_kprobe(struct kprobe *p)

void __kprobes arch_remove_kprobe(struct kprobe *p)
{
+ if (p->ainsn.api.insn) {
+ free_insn_slot(p->ainsn.api.insn, 0);
+ p->ainsn.api.insn = NULL;
+ }
}

static void __kprobes save_previous_kprobe(struct kprobe_ctlblk *kcb)
--
2.17.1

2022-05-25 08:37:47

by Liao, Chang

Subject: Re: [PATCH] riscv/kprobe: reclaim insn_slot on kprobe unregistration

在 2022/5/24 23:12, Jisheng Zhang 写道:
> On Mon, May 23, 2022 at 09:51:24AM +0800, Liao Chang wrote:
>> On kprobe registration kernel allocate one insn_slot for new kprobe,
>> but it forget to reclaim the insn_slot on unregistration, leading to a
>> potential leakage.
>
> + Guo Ren
>
> Nice catch! I think csky needs the fix as well.
>
>>
>> This bug reported by Chen Guokai <[email protected]>.
>
> This needs a "Reported-by: ..." tag

Thanks for comment.

>
>>
>> Signed-off-by: Liao Chang <[email protected]>
>> ---
>> arch/riscv/kernel/probes/kprobes.c | 4 ++++
>> 1 file changed, 4 insertions(+)
>>
>> diff --git a/arch/riscv/kernel/probes/kprobes.c b/arch/riscv/kernel/probes/kprobes.c
>> index e6e950b7cf32..f12eb1fbb52c 100644
>> --- a/arch/riscv/kernel/probes/kprobes.c
>> +++ b/arch/riscv/kernel/probes/kprobes.c
>> @@ -110,6 +110,10 @@ void __kprobes arch_disarm_kprobe(struct kprobe *p)
>>
>> void __kprobes arch_remove_kprobe(struct kprobe *p)
>> {
>> + if (p->ainsn.api.insn) {
>> + free_insn_slot(p->ainsn.api.insn, 0);
>> + p->ainsn.api.insn = NULL;
>> + }
>> }
>>
>> static void __kprobes save_previous_kprobe(struct kprobe_ctlblk *kcb)
>> --
>> 2.17.1
>>
>>
>> _______________________________________________
>> linux-riscv mailing list
>> [email protected]
>> http://lists.infradead.org/mailman/listinfo/linux-riscv
> .

--
BR,
Liao, Chang

2022-05-25 09:03:46

by Jisheng Zhang

Subject: Re: [PATCH] riscv/kprobe: reclaim insn_slot on kprobe unregistration

On Mon, May 23, 2022 at 09:51:24AM +0800, Liao Chang wrote:
> On kprobe registration kernel allocate one insn_slot for new kprobe,
> but it forget to reclaim the insn_slot on unregistration, leading to a
> potential leakage.

+ Guo Ren

Nice catch! I think csky needs the fix as well.

>
> This bug reported by Chen Guokai <[email protected]>.

This needs a "Reported-by: ..." tag

>
> Signed-off-by: Liao Chang <[email protected]>
> ---
> arch/riscv/kernel/probes/kprobes.c | 4 ++++
> 1 file changed, 4 insertions(+)
>
> diff --git a/arch/riscv/kernel/probes/kprobes.c b/arch/riscv/kernel/probes/kprobes.c
> index e6e950b7cf32..f12eb1fbb52c 100644
> --- a/arch/riscv/kernel/probes/kprobes.c
> +++ b/arch/riscv/kernel/probes/kprobes.c
> @@ -110,6 +110,10 @@ void __kprobes arch_disarm_kprobe(struct kprobe *p)
>
> void __kprobes arch_remove_kprobe(struct kprobe *p)
> {
> + if (p->ainsn.api.insn) {
> + free_insn_slot(p->ainsn.api.insn, 0);
> + p->ainsn.api.insn = NULL;
> + }
> }
>
> static void __kprobes save_previous_kprobe(struct kprobe_ctlblk *kcb)
> --
> 2.17.1
>
>
> _______________________________________________
> linux-riscv mailing list
> [email protected]
> http://lists.infradead.org/mailman/listinfo/linux-riscv