2024-05-06 02:23:41

by yang.zhang

Subject: [PATCH V2] riscv: set trap vector earlier

From: "yang.zhang" <[email protected]>

If trap earlier, trap vector doesn't yet set properly, current
value maybe set by previous firmwire, typically it's the _start of kernel,
it's confused and difficult to debuge, so set it earlier.
---
v1 -> v2:
As Alex commented, remove the patch for supporting hugesize kernek image
Add the omissive logic of set trap vector earlier

Signed-off-by: yang.zhang <[email protected]>
---
arch/riscv/kernel/head.S | 3 +++
1 file changed, 3 insertions(+)

diff --git a/arch/riscv/kernel/head.S b/arch/riscv/kernel/head.S
index 4236a69c35cb..03dc440e643e 100644
--- a/arch/riscv/kernel/head.S
+++ b/arch/riscv/kernel/head.S
@@ -300,6 +300,9 @@ SYM_CODE_START(_start_kernel)
#else
mv a0, a1
#endif /* CONFIG_BUILTIN_DTB */
+ /* Set trap vector to spin forever to help debug */
+ la a3, .Lsecondary_park
+ csrw CSR_TVEC, a3
call setup_vm
#ifdef CONFIG_MMU
la a0, early_pg_dir
--
2.25.1

2024-05-07 05:29:54

by Alexandre Ghiti

Subject: Re: [PATCH V2] riscv: set trap vector earlier

Hi Yang,

On 06/05/2024 04:22, yang.zhang wrote:
> From: "yang.zhang" <[email protected]>
>
> If trap earlier, trap vector doesn't yet set properly, current
> value maybe set by previous firmwire, typically it's the _start of kernel,
> it's confused and difficult to debuge, so set it earlier.


What about:

"The exception vector of the booting hart is not set before enabling the
mmu and then still points to the value of the previous firmware,
typically _start_kernel. That makes it hard to debug setup_vm() when bad
things happen. So fix that by setting the exception vector earlier."


> ---
> v1 -> v2:
> As Alex commented, remove the patch for supporting hugesize kernek image
> Add the omissive logic of set trap vector earlier
>
> Signed-off-by: yang.zhang <[email protected]>
> ---
> arch/riscv/kernel/head.S | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git a/arch/riscv/kernel/head.S b/arch/riscv/kernel/head.S
> index 4236a69c35cb..03dc440e643e 100644
> --- a/arch/riscv/kernel/head.S
> +++ b/arch/riscv/kernel/head.S
> @@ -300,6 +300,9 @@ SYM_CODE_START(_start_kernel)
> #else
> mv a0, a1
> #endif /* CONFIG_BUILTIN_DTB */
> + /* Set trap vector to spin forever to help debug */
> + la a3, .Lsecondary_park
> + csrw CSR_TVEC, a3
> call setup_vm
> #ifdef CONFIG_MMU
> la a0, early_pg_dir

Note that this patch will allow to catch any exception happening in
setup_vm() but as we reset stvec in relocate_enable_mmu. But that's ok,
it will help catching weird bugs more easily :)

With the commit message fixed, you can add:

Reviewed-by: Alexandre Ghiti <[email protected]>

Thanks,

Alex

2024-05-08 01:58:25

by yang.zhang

Subject: Re:Re: [PATCH V2] riscv: set trap vector earlier

At 2024-05-07 13:28:58, "Alexandre Ghiti" <[email protected]> wrote:
>Hi Yang,
>
>On 06/05/2024 04:22, yang.zhang wrote:
>> From: "yang.zhang" <[email protected]>
>>
>> If trap earlier, trap vector doesn't yet set properly, current
>> value maybe set by previous firmwire, typically it's the _start of kernel,
>> it's confused and difficult to debuge, so set it earlier.
>
>
>What about:
>
>"The exception vector of the booting hart is not set before enabling the
>mmu and then still points to the value of the previous firmware,
>typically _start_kernel. That makes it hard to debug setup_vm() when bad
>things happen. So fix that by setting the exception vector earlier."

Hi alex:
Maybe "typically _start" not "typically _start_kernel", right?
Thanks for your help??
>> ---
>> v1 -> v2:
>> As Alex commented, remove the patch for supporting hugesize kernek image
>> Add the omissive logic of set trap vector earlier
>>
>> Signed-off-by: yang.zhang <[email protected]>
>> ---
>> arch/riscv/kernel/head.S | 3 +++
>> 1 file changed, 3 insertions(+)
>>
>> diff --git a/arch/riscv/kernel/head.S b/arch/riscv/kernel/head.S
>> index 4236a69c35cb..03dc440e643e 100644
>> --- a/arch/riscv/kernel/head.S
>> +++ b/arch/riscv/kernel/head.S
>> @@ -300,6 +300,9 @@ SYM_CODE_START(_start_kernel)
>> #else
>> mv a0, a1
>> #endif /* CONFIG_BUILTIN_DTB */
>> + /* Set trap vector to spin forever to help debug */
>> + la a3, .Lsecondary_park
>> + csrw CSR_TVEC, a3
>> call setup_vm
>> #ifdef CONFIG_MMU
>> la a0, early_pg_dir
>
>Note that this patch will allow to catch any exception happening in
>setup_vm() but as we reset stvec in relocate_enable_mmu. But that's ok,
>it will help catching weird bugs more easily :)
>
>With the commit message fixed, you can add:
>
>Reviewed-by: Alexandre Ghiti <[email protected]>
>
>Thanks,
>
>Alex