linux-kernel - Re: [PATCH V2 2/2] LoongArch: KVM: fix "unreliable stack" issue

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <33541c5f-82ca-c86d-fcf9-437c4071c6b8@loongson.cn>
Date: Wed, 24 Dec 2025 10:40:08 +0800
From: lixianglai <lixianglai@...ngson.cn>
To: Bibo Mao <maobibo@...ngson.cn>, Huacai Chen <chenhuacai@...nel.org>,
 yangtiezhu@...ngson.cn
Cc: loongarch@...ts.linux.dev, linux-kernel@...r.kernel.org,
 kvm@...r.kernel.org, stable@...r.kernel.org, WANG Xuerui
 <kernel@...0n.name>, Tianrui Zhao <zhaotianrui@...ngson.cn>,
 Charlie Jenkins <charlie@...osinc.com>, Thomas Gleixner <tglx@...utronix.de>
Subject: Re: [PATCH V2 2/2] LoongArch: KVM: fix "unreliable stack" issue

Add yangtiezhu@...ngson.cn

Hi :
>
>
> On 2025/12/23 上午10:46, Huacai Chen wrote:
>> On Tue, Dec 23, 2025 at 9:27 AM Bibo Mao <maobibo@...ngson.cn> wrote:
>>>
>>>
>>>
>>> On 2025/12/22 下午7:34, Xianglai Li wrote:
>>>> Insert the appropriate UNWIND macro definition into the 
>>>> kvm_exc_entry in
>>>> the assembly function to guide the generation of correct ORC table 
>>>> entries,
>>>> thereby solving the timeout problem of loading the livepatch-sample 
>>>> module
>>>> on a physical machine running multiple vcpus virtual machines.
>>>>
>>>> While solving the above problems, we have gained an additional 
>>>> benefit,
>>>> that is, we can obtain more call stack information
>>>>
>>>> Stack information that can be obtained before the problem is fixed:
>>>> [<0>] kvm_vcpu_block+0x88/0x120 [kvm]
>>>> [<0>] kvm_vcpu_halt+0x68/0x580 [kvm]
>>>> [<0>] kvm_emu_idle+0xd4/0xf0 [kvm]
>>>> [<0>] kvm_handle_gspr+0x7c/0x700 [kvm]
>>>> [<0>] kvm_handle_exit+0x160/0x270 [kvm]
>>>> [<0>] kvm_exc_entry+0x100/0x1e0
>>>>
>>>> Stack information that can be obtained after the problem is fixed:
>>>> [<0>] kvm_vcpu_block+0x88/0x120 [kvm]
>>>> [<0>] kvm_vcpu_halt+0x68/0x580 [kvm]
>>>> [<0>] kvm_emu_idle+0xd4/0xf0 [kvm]
>>>> [<0>] kvm_handle_gspr+0x7c/0x700 [kvm]
>>>> [<0>] kvm_handle_exit+0x160/0x270 [kvm]
>>>> [<0>] kvm_exc_entry+0x100/0x1e0
>>>> [<0>] kvm_arch_vcpu_ioctl_run+0x260/0x488 [kvm]
>>>> [<0>] kvm_vcpu_ioctl+0x200/0xcd8 [kvm]
>>>> [<0>] sys_ioctl+0x498/0xf00
>>>> [<0>] do_syscall+0x94/0x190
>>>> [<0>] handle_syscall+0xb8/0x158
>>>>
>>>> Cc: stable@...r.kernel.org
>>>> Signed-off-by: Xianglai Li <lixianglai@...ngson.cn>
>>>> ---
>>>> Cc: Huacai Chen <chenhuacai@...nel.org>
>>>> Cc: WANG Xuerui <kernel@...0n.name>
>>>> Cc: Tianrui Zhao <zhaotianrui@...ngson.cn>
>>>> Cc: Bibo Mao <maobibo@...ngson.cn>
>>>> Cc: Charlie Jenkins <charlie@...osinc.com>
>>>> Cc: Xianglai Li <lixianglai@...ngson.cn>
>>>> Cc: Thomas Gleixner <tglx@...utronix.de>
>>>>
>>>>    arch/loongarch/kvm/switch.S | 2 ++
>>>>    1 file changed, 2 insertions(+)
>>>>
>>>> diff --git a/arch/loongarch/kvm/switch.S b/arch/loongarch/kvm/switch.S
>>>> index 93845ce53651..e3ecb24a3bc5 100644
>>>> --- a/arch/loongarch/kvm/switch.S
>>>> +++ b/arch/loongarch/kvm/switch.S
>>>> @@ -170,6 +170,7 @@ SYM_CODE_START(kvm_exc_entry)
>>>>        /* restore per cpu register */
>>>>        ld.d    u0, a2, KVM_ARCH_HPERCPU
>>>>        addi.d  sp, sp, -PT_SIZE
>>>> +     UNWIND_HINT_REGS
>>>>
>>>>        /* Prepare handle exception */
>>>>        or      a0, s0, zero
>>>> @@ -214,6 +215,7 @@ SYM_FUNC_START(kvm_enter_guest)
>>>>        addi.d  a2, sp, -PT_SIZE
>>>>        /* Save host GPRs */
>>>>        kvm_save_host_gpr a2
>>>> +     st.d    ra, a2, PT_ERA
>>> Had better add some comments here to show that it is special for unwind
>>> usage since there is "st.d ra, a2, PT_R1" already in macro
>>> kvm_save_host_gpr().
>> Then there is a new problem, why can unwinder not recognize the
>> instruction in  kvm_save_host_gpr()?
> maybe it need unwinder owner to answer this question.
>
kvm_save_host_gpr() is an assembler macro that has already been executed 
and is no longer normal on the stack.
Am I explaining correctly? @tiezhu

I guess you might be wondering why unwinder didn't recognize 
kvm_enter_guest().

There's something wrong with the logic that we're implementing here that 
we should put the current pc in era instead of ra.
This will allow unwind to identify the symbol kvm_enter_guest.

So I will fix it in the next version like this:

@@ -214,6 +215,7 @@ SYM_FUNC_START(kvm_enter_guest)
        addi.d  a2, sp, -PT_SIZE
        /* Save host GPRs */
        kvm_save_host_gpr a2

+    /*
+     * The csr_era member variable of the pt_regs structure is required
+     * for unwinding orc to perform stack traceback, so we need to put
+     * pc into csr_era member variable here.
+     */
+    pcaddi    t0, 0
+    st.d    t0, a2, PT_ERA
+

Thanks,
Xianglai.
>>
>> Huacai
>>>
>>> Regards
>>> Bibo Mao
>>>>
>>>>        addi.d  a2, a1, KVM_VCPU_ARCH
>>>>        st.d    sp, a2, KVM_ARCH_HSP
>>>>
>>>