linux-kernel - Re: [PATCH v3 2/4] riscv: add support for SBI Supervisor Software Events extension

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] [day] [month] [year] [list]

Message-ID: <8115698f-3cf7-4a6a-83cf-9dc288c84ee0@rivosinc.com>
Date: Thu, 20 Mar 2025 13:26:00 +0100
From: Clément Léger <cleger@...osinc.com>
To: Andrew Jones <ajones@...tanamicro.com>
Cc: Paul Walmsley <paul.walmsley@...ive.com>,
 Palmer Dabbelt <palmer@...belt.com>, linux-riscv@...ts.infradead.org,
 linux-kernel@...r.kernel.org, linux-arm-kernel@...ts.infradead.org,
 Himanshu Chauhan <hchauhan@...tanamicro.com>,
 Anup Patel <apatel@...tanamicro.com>, Xu Lu <luxu.kernel@...edance.com>,
 Atish Patra <atishp@...shpatra.org>
Subject: Re: [PATCH v3 2/4] riscv: add support for SBI Supervisor Software
 Events extension



On 20/03/2025 12:52, Andrew Jones wrote:
> On Thu, Mar 20, 2025 at 09:16:07AM +0100, Clément Léger wrote:
>>
>>
>> On 19/03/2025 18:08, Andrew Jones wrote:
>>> On Fri, Dec 06, 2024 at 05:30:58PM +0100, Clément Léger wrote:
>>> ...
>>>> +int arch_sse_init_event(struct sse_event_arch_data *arch_evt, u32 evt_id, int cpu)
>>>> +{
>>>> +	void *stack;
>>>> +
>>>> +	arch_evt->evt_id = evt_id;
>>>> +	stack = sse_stack_alloc(cpu, SSE_STACK_SIZE);
>>>> +	if (!stack)
>>>> +		return -ENOMEM;
>>>> +
>>>> +	arch_evt->stack = stack + SSE_STACK_SIZE;
>>>> +
>>>> +	if (sse_init_scs(cpu, arch_evt))
>>>> +		goto free_stack;
>>>> +
>>>> +	if (is_kernel_percpu_address((unsigned long)&arch_evt->interrupted)) {
>>>> +		arch_evt->interrupted_state_phys =
>>>> +				per_cpu_ptr_to_phys(&arch_evt->interrupted);
>>>> +	} else {
>>>> +		arch_evt->interrupted_state_phys =
>>>> +				virt_to_phys(&arch_evt->interrupted);
>>>> +	}
>>>> +
>>>> +	return 0;
>>>
>>> Hi Clément,
>>>
>>> Testing SSE support with tools/testing/selftests/kvm/riscv/sbi_pmu_test
>>> led to an opensbi sbi_trap_error because the output_phys_lo address passed
>>> to sbi_sse_read_attrs() wasn't a physical address. The reason is that
>>> is_kernel_percpu_address() can only be used on static percpu addresses,
>>> but local sse events get their percpu addresses with alloc_percpu(), so
>>> is_kernel_percpu_address() was returning false even for local events. I
>>> made the following changes to get things working.
>>
>> Hi Andrew,
>>
>> Did something changed recently ? Because I tested that when it was send
>> (PMU + some kernel internal testsuite) and didn't saw that. Anyway, I'll
>> respin it with your changes as well.
> 
> It depends on the kernel config. Configs that don't have many
> alloc_percpu() calls prior to the one made by sse can work, because,
> iiuc, alloc_percpu() will get its allocation from the percpu allocator's
> first chunk until that chunck fills up. The first chunck is shared with
> the static allocations.

Makes sense ! Thanks, I'll look at it.

> 
> Thanks,
> drew
> 
>>
>> Thanks !
>>
>> Clément
>>
>>>
>>> Thanks,
>>> drew
>>>
>>> diff --git a/arch/riscv/kernel/sse.c b/arch/riscv/kernel/sse.c
>>> index b48ae69dad8d..f46893946086 100644
>>> --- a/arch/riscv/kernel/sse.c
>>> +++ b/arch/riscv/kernel/sse.c
>>> @@ -100,12 +100,12 @@ int arch_sse_init_event(struct sse_event_arch_data *arch_evt, u32 evt_id, int cp
>>>         if (sse_init_scs(cpu, arch_evt))
>>>                 goto free_stack;
>>>
>>> -       if (is_kernel_percpu_address((unsigned long)&arch_evt->interrupted)) {
>>> +       if (sse_event_is_global(evt_id)) {
>>>                 arch_evt->interrupted_state_phys =
>>> -                               per_cpu_ptr_to_phys(&arch_evt->interrupted);
>>> +                               virt_to_phys(&arch_evt->interrupted);
>>>         } else {
>>>                 arch_evt->interrupted_state_phys =
>>> -                               virt_to_phys(&arch_evt->interrupted);
>>> +                               per_cpu_ptr_to_phys(&arch_evt->interrupted);
>>>         }
>>>
>>>         return 0;
>>> diff --git a/drivers/firmware/riscv/riscv_sse.c b/drivers/firmware/riscv/riscv_sse.c
>>> index 511db9ad7a9e..fef375046f75 100644
>>> --- a/drivers/firmware/riscv/riscv_sse.c
>>> +++ b/drivers/firmware/riscv/riscv_sse.c
>>> @@ -62,11 +62,6 @@ void sse_handle_event(struct sse_event_arch_data *arch_event,
>>>                         ret);
>>>  }
>>>
>>> -static bool sse_event_is_global(u32 evt)
>>> -{
>>> -       return !!(evt & SBI_SSE_EVENT_GLOBAL);
>>> -}
>>> -
>>>  static
>>>  struct sse_event *sse_event_get(u32 evt)
>>>  {
>>> diff --git a/include/linux/riscv_sse.h b/include/linux/riscv_sse.h
>>> index 16700677f1e8..06b757b036b0 100644
>>> --- a/include/linux/riscv_sse.h
>>> +++ b/include/linux/riscv_sse.h
>>> @@ -8,6 +8,7 @@
>>>
>>>  #include <linux/types.h>
>>>  #include <linux/linkage.h>
>>> +#include <asm/sbi.h>
>>>
>>>  struct sse_event;
>>>  struct pt_regs;
>>> @@ -16,6 +17,11 @@ struct ghes;
>>>
>>>  typedef int (sse_event_handler)(u32 event_num, void *arg, struct pt_regs *regs);
>>>
>>> +static inline bool sse_event_is_global(u32 evt)
>>> +{
>>> +       return !!(evt & SBI_SSE_EVENT_GLOBAL);
>>> +}
>>> +
>>>  #ifdef CONFIG_RISCV_SSE
>>>
>>>  struct sse_event *sse_event_register(u32 event_num, u32 priority,
>>