CL-CLINT Register Handling to Support Multiple Clusters

[mridul-tosil]

The existing implementation of the CL-CLINT register handling in banshee-pulp is limited to a single cluster. Specifically, the code directly checks the x value against a single base address (self.engine.config.address.cl_clint) and its offset (+ 0x8) to determine whether it is a "set" or "clear" operation.

banshee/src/engine.rs

Line 837 in 0e14490

x if x >= self.engine.config.address.cl_clint

banshee/src/engine.rs

Line 1012 in 0e14490

x if x == self.engine.config.address.cl_clint => {

This implementation does not support systems with multiple clusters where each cluster has its own CLINT register mapped to different addresses.

Existing implementation

banshee/src/engine.rs

Line 837 in 0e14490

x if x >= self.engine.config.address.cl_clint

 x if x >= self.engine.config.address.cl_clint
 && x < self.engine.config.address.cl_clint + 0x8 =>
 {
 0
 }

Proposed implementation

 x if (0..self.engine.num_clusters).any(|i| {
 x >= (self.engine.config.address.cl_clint
 + self.engine.config.memory.periphs.offset * i as u32)
 && x <= (self.engine.config.address.cl_clint
 + self.engine.config.memory.periphs.offset * i as u32
 + 0x8) // Original size for cl_clint
 }) =>
 {
 0
 }

Existing implementation

banshee/src/engine.rs

Line 1012 in 0e14490

x if x == self.engine.config.address.cl_clint => {

 x if x == self.engine.config.address.cl_clint => {
 // clint set register
 let old_entry = self
 .cl_clint
 .fetch_or((value & mask) as usize, Ordering::SeqCst);
 // wake cores affected by this write
 let hart_base = (self.cluster_id * self.num_cores) as u32;
 for i in 0..32 {
 if ((!old_entry & (value & mask) as usize) & (1 << i)) != 0 {
 trace!(
 " wakeup_wus.req[{}] from cluster-local CLINT",
 (hart_base + i) as usize
 );
 self.wake((hart_base + i) as u32);
 }
 }
 }
 x if x == self.engine.config.address.cl_clint + 0x8 => {
 // clint clear register
 self.cl_clint
 .fetch_and(!(value & mask) as usize, Ordering::SeqCst);
 }

Proposed implementation

 x if (0..self.engine.num_clusters).any(|i| {
 let base = self.engine.config.address.cl_clint
 + self.engine.config.memory.periphs.offset * i as u32;
 x >= base && x <= base + 0x8
 }) =>
 {
 let id = (0..self.engine.num_clusters)
 .position(|i| {
 let base = self.engine.config.address.cl_clint
 + self.engine.config.memory.periphs.offset * i as u32;
 addr >= base && addr <= base + 0x8
 })
 .unwrap();

 let offset = addr
 - (self.engine.config.address.cl_clint
 + self.engine.config.memory.periphs.offset * id as u32);

 match offset {
 0 => {
 // clint set register
 let old_entry = self
 .cl_clint
 .fetch_or((value & mask) as usize, Ordering::SeqCst);
 // wake cores affected by this write
 let hart_base = (id * self.num_cores) as u32;
 for i in 0..32 {
 if ((!old_entry & (value & mask) as usize) & (1 << i)) != 0 {
 trace!(
 " wakeup_wus.req[{}] from cluster-local CLINT",
 (hart_base + i) as usize
 );
 self.wake((hart_base + i) as u32);
 }
 }
 }
 0x8 => {
 // clint clear register
 self.cl_clint
 .fetch_and(!(value & mask) as usize, Ordering::SeqCst);
 }
 _ => (),
 }
 }

Additional Note:
I would like to mention that I am new to Rust programming and still learning the best practices. I apologize in advance if the code provided or the approach taken is not optimal.

@SamuelRiedel @paulsc96