Re: [PATCH] riscv: ftrace: reject out-of-range non-direct targets

["Rui Qi" <qirui.001@bytedance.com>]

On 8/31/26 10:58 PM, Steven Rostedt wrote:
> On Mon, 31 Aug 2026 11:50:45 +0800
> "Rui Qi" <qirui.001@bytedance.com> wrote:
> 
>> Hi Steven,
>>
>> Gentle ping on this patch. It's been over a month since submission
>> and I'd appreciate your review when you have a chance.
>>
>> The patch fixes ftrace_modify_call() on RISC-V to properly validate
>> old_addr/addr instead of always treating sites as FTRACE_ADDR, and
>> rejects out-of-range non-direct targets that can't be dispatched
>> through the existing ftrace_caller path.
> 
> This patch looks very architecture specific. Is there a generic ftrace
> issue you have with this? "out-of-range" issues are specific to an
> architecture (x86 doesn't have this issue).
> 
> You need the RISC-V maintainers to review and accept this patch. I don't
> see anything specific to generic ftrace infrastructure here, which is what
> I maintain.
> 
> -- Steve
> 

Hi Steven,

Thanks for the clarification. You're right, this is entirely
RISC-V specific -- the out-of-range issue comes from RISC-V's
AUIPC/JALR relocation limits, and the fix only touches
arch/riscv/kernel/ftrace.c. There's no generic ftrace change here.

Palmer, Paul, could you take a look at this patch? The core fix is
making ftrace_modify_call() honor its old_addr/addr arguments instead
of always validating against FTRACE_ADDR, and rejecting out-of-range
non-direct targets that can't fall back through ftrace_caller.

Thanks,
Rui

_______________________________________________
linux-riscv mailing list
linux-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-riscv