Re: [PATCH libevl v2] build: add initial RISC-V support

Click on a legend element to toggle display of that core, ctrl-click inverts display, shift-click enables all.

Data to construct the above plot have been generated using the RT test utility cyclictest.
Unexpectedly long latencies may be caused by SMIs
Total number of samples: 100 million
Resolution of latency scale: normal
Duration: 5 hours, 33 minutes, lowest P state: performance

On 7/9/26 17:32, Tobias Schaffner wrote:
> 
> 
> On 7/9/26 17:17, Florian Bezdeka wrote:
>> On Thu, 2026-07-09 at 15:44 +0200, Tobias Schaffner wrote:
>>> Provide the basic build infrastructure and VDSO definitions
>>> required to build libevl on RISC-V.
>>>
>>> Signed-off-by: Tobias Schaffner <[email protected]>
>>> ---
>>>   lib/arch/riscv/include/asm/evl/vdso.h | 14 ++++++++++++++
>>>   meson.build                           |  4 ++++
>>>   meson/riscv64-linux-gnu               | 17 +++++++++++++++++
>>>   3 files changed, 35 insertions(+)
>>>   create mode 100644 lib/arch/riscv/include/asm/evl/vdso.h
>>>   create mode 100644 meson/riscv64-linux-gnu
>>>
>>> diff --git a/lib/arch/riscv/include/asm/evl/vdso.h b/lib/arch/riscv/ 
>>> include/asm/evl/vdso.h
>>> new file mode 100644
>>> index 0000000..5e331fb
>>> --- /dev/null
>>> +++ b/lib/arch/riscv/include/asm/evl/vdso.h
>>> @@ -0,0 +1,14 @@
>>> +/*
>>> + * SPDX-License-Identifier: MIT
>>> + *
>>> + * Copyright (C) 2021 Philippe Gerum  <[email protected]>
>>> + */
>>
>> ^^ Copy paste mistake? At least wrong year...

Click on a legend element to toggle display of that core, ctrl-click inverts display, shift-click enables all.

Data to construct the above plot have been generated using the RT test utility cyclictest.
Unexpectedly long latencies may be caused by SMIs
Total number of samples: 100 million
Resolution of latency scale: normal
Duration: 5 hours, 33 minutes, lowest P state: performance

> Shanmu directly copied this from one of the other vdso header files 
> without modifications to the copyright header and I did not touch it 
> further.
> 
> Would leaf it as is if it is ok for you even if the ifdef was slightly 
> changed.
> 
> Best,
> Tobias
> 
>>
> 
>