Re: [PATCH] rust: sync: optimize rust symbol generation for CondVar

[Kunwu Chan <kunwu.chan@linux.dev>]

On 2025/3/24 01:59, Boqun Feng wrote:
> On Mon, Mar 17, 2025 at 04:13:50PM +0800, Kunwu Chan wrote:
>> From: Kunwu Chan <kunwu.chan@hotmail.com>
>>
>> When build the kernel using the llvm-18.1.3-rust-1.85.0-x86_64
>> with ARCH=arm64, the following symbols are generated:
>>
>> $nm vmlinux | grep ' _R'.*CondVar | rustfilt
>> ... T <kernel::sync::condvar::CondVar>::notify_all
>> ... T <kernel::sync::condvar::CondVar>::notify_one
>> ... T <kernel::sync::condvar::CondVar>::notify_sync
>> ... T <kernel::sync::condvar::CondVar>::new::{closure#0}::{closure#0}::panic_cold_explicit
>> ... T <kernel::sync::condvar::CondVar>::new::{closure#0}::{closure#0}::panic_cold_explicit
>> ... T <kernel::sync::poll::PollCondVar>::new::{closure#0}::{closure#0}::panic_cold_explicit
>> ... T <kernel::sync::poll::PollCondVar as core::ops::drop::Drop>::drop
>>
>> These notify* symbols are trivial wrappers around the C functions
>> __wake_up and __wake_up_sync.
>> It doesn't make sense to go through a trivial wrapper for these
>> functions, so mark them inline.
>>
>> Link: https://github.com/Rust-for-Linux/linux/issues/1145
>> Suggested-by: Alice Ryhl <aliceryhl@google.com>
>> Co-developed-by: Grace Deng <Grace.Deng006@Gmail.com>
>> Signed-off-by: Grace Deng <Grace.Deng006@Gmail.com>
>> Signed-off-by: Kunwu Chan <kunwu.chan@hotmail.com>
>> ---
>>   rust/kernel/sync/condvar.rs | 4 ++++
>>   1 file changed, 4 insertions(+)
>>
>> diff --git a/rust/kernel/sync/condvar.rs b/rust/kernel/sync/condvar.rs
>> index 7df565038d7d..a826896ba3f0 100644
>> --- a/rust/kernel/sync/condvar.rs
>> +++ b/rust/kernel/sync/condvar.rs
>> @@ -181,6 +181,7 @@ pub fn wait_interruptible_timeout<T: ?Sized, B: Backend>(
>>       }
>>   
>>       /// Calls the kernel function to notify the appropriate number of threads.
>> +    #[inline]
>>       fn notify(&self, count: c_int) {
> Hmm.. I think CondVar::notify() gets inlined even without this
> `#[inline]` attribute, do we need this actually?

Actualy, after add '#[inline]', the build result is (compilecmd is 'make 
ARCH=arm64 LLVM=1-j8' ):

$nm vmlinux | grep ' _R'.*CondVar | rustfilt
ffff800080019e90 T 
<kernel::sync::condvar::CondVar>::new::{closure#0}::{closure#0}::panic_cold_ex
plicit
ffff800080019e90 T 
<kernel::sync::condvar::CondVar>::new::{closure#0}::{closure#0}::panic_cold_ex
plicit
ffff800080019e90 T 
<kernel::sync::poll::PollCondVar>::new::{closure#0}::{closure#0}::panic_cold_e
xplicit
ffff8000805b8c7c T <kernel::sync::poll::PollCondVar as 
core::ops::drop::Drop>::drop


And before add '#[inline]',the 'nm vmlinux | grep ' _R'.*CondVar | 
rustfilt' appear 'notify' function,

Seems like the LLVM didn't make it inline.


> Regards,
> Boqun
>
>>           // SAFETY: `wait_queue_head` points to valid memory.
>>           unsafe {
> [...]

-- 
Thanks,
   Kunwu.Chan(Tao.Chan)