Skip to content

Frequently Asked Questions

This page collects common questions and answers encountered when using Gthulhu and Qumun.

Q: How can I confirm my kernel supports sched_ext?

A: You can check using the following methods:

# Method 1: Check kernel configuration
grep -r "CONFIG_SCHED_CLASS_EXT" /boot/config-$(uname -r)

# Method 2: Check /proc/config.gz
zcat /proc/config.gz | grep "CONFIG_SCHED_CLASS_EXT"

# Method 3: Check sched_ext directory
ls /sys/kernel/sched_ext/ 2>/dev/null

If the output contains CONFIG_SCHED_CLASS_EXT=y, your kernel supports sched_ext.

Q: What should I do when getting "libbpf not found" error during compilation?

A: This is usually because libbpf is not properly installed. Please follow these steps to resolve:

# Ubuntu/Debian
sudo apt install libbpf-dev

# CentOS/RHEL/Fedora
sudo dnf install libbpf-devel

# Or manually compile libbpf
git clone https://github.com/libbpf/libbpf.git
cd libbpf/src
make
sudo make install

Q: Why is Clang 17+ required?

A: Clang 17+ provides more complete BPF support, including:

  • Better BPF CO-RE (Compile Once, Run Everywhere) support
  • Latest BPF instruction set support
  • More stable BPF program compilation

If your system doesn't have Clang 17+, you can install it like this:

# Ubuntu/Debian
sudo apt install clang-17

# Set environment variables
export CC=clang-17
export CXX=clang++-17

Q: Getting "Operation not permitted" error when running

A: This is a permission issue. BPF program loading requires root privileges:

# Correct way to run
sudo ./main

# Or use Docker
docker run --privileged=true --pid host --rm gthulhu:latest /gthulhu/main

Q: What to do if the system becomes slow after starting the scheduler?

A: This might be due to the following reasons:

  1. Scheduling parameters not suitable for your workload: 

    # Check system load
top
htop

# Check context switch frequency
vmstat 1

  2. Insufficient memory: 

    # Check memory usage
free -h
cat /proc/meminfo

  3. BPF program performance issues: 

    # Check BPF program statistics
sudo bpftool prog show
sudo bpftool prog profile

Solutions: - Stop the scheduler: sudo pkill -f "./main"
- Check system logs: dmesg | tail -50 - Adjust scheduling parameters or report the issue

Q: How to stop the scheduler?

A: You can stop the scheduler using the following methods:

# Method 1: Ctrl+C (if running in foreground)
^C

# Method 2: Send SIGTERM signal
sudo pkill -TERM -f "./main"

# Method 3: Send SIGINT signal
sudo pkill -INT -f "./main"

# Method 4: Force kill (not recommended)
sudo pkill -KILL -f "./main"

Q: How to monitor scheduler performance?

A: You can use various tools to monitor scheduler performance:

  1. System tools: 

    # Monitor CPU usage
htop

# Monitor context switches
vmstat 1

# Monitor scheduling latency
perf sched record -- sleep 10
perf sched latency

  2. BPF tools: 

    # Check BPF program status
sudo bpftool prog list | grep sched

# Check BPF map contents
sudo bpftool map dump name task_info_map

  3. Built-in scheduler monitoring: 

    # View scheduler logs
journalctl -f -u gthulhu

# View BPF trace messages
sudo cat /sys/kernel/debug/tracing/trace_pipe

Q: What advantages does the scheduler have compared to CFS?

A: Main advantages of Gthulhu scheduler:

Feature CFS Gthulhu
Latency Optimization Basic Specialized
Task Classification Unified processing Automatic classification
CPU Topology Awareness Limited Complete support
Dynamic Adjustment Static parameters Real-time adjustment
User-space Extension Not supported Fully supported

Q: How to adjust scheduler parameters?

A: Currently supported adjustment methods:

  1. Environment variables: 

    export GTHULHU_DEBUG=true
export GTHULHU_LOG_LEVEL=DEBUG
sudo -E ./main

  2. Compile-time parameters (modify main.bpf.c): 

    // Adjust base time slice
#define BASE_SLICE_NS    3000000ULL  // 3ms instead of 5ms

  3. Runtime API (planned): 

    // Future support for dynamic adjustment
params := &SchedulingParams{
    BaseSliceNs: 3000000,
    LatencyFactor: 1.5,
}
UpdateSchedulingParams(params)

Q: How to enable debug mode?

A: You can enable debugging through the following methods:

  1. Environment variables: 

    export GTHULHU_DEBUG=true
export GTHULHU_LOG_LEVEL=DEBUG
sudo -E ./main

  2. BPF tracing: 

    # Terminal 1: Start scheduler
sudo ./main

# Terminal 2: View BPF traces
sudo cat /sys/kernel/debug/tracing/trace_pipe

  3. System logs: 

    # View kernel logs
dmesg -w

# View systemd logs
journalctl -f

Q: What to do when encountering BPF verifier errors?

A: BPF verifier errors usually indicate program issues:

  1. Check error messages: 

    # View detailed errors
dmesg | grep -i bpf

  2. Common issues:

  3. Unbounded loops: Ensure all loops have clear exit conditions
  4. Memory out of bounds: Check array accesses are within range

  5. Pointer usage: Ensure pointers are NULL-checked before use

  6. Verify BPF program: 

    # Use bpftool to verify
sudo bpftool prog load main.bpf.o /sys/fs/bpf/test_prog

Q: How to report issues?

A: If you encounter problems, please follow these steps:

  1. Collect system information: 

    # System information
uname -a
cat /etc/os-release

# Kernel version and configuration
uname -r
grep CONFIG_SCHED_CLASS_EXT /boot/config-$(uname -r)

# Go version
go version

# Clang version
clang --version

  2. Collect error logs: 

    # Scheduler logs
sudo ./main 2>&1 | tee gthulhu.log

# System logs
dmesg > dmesg.log
journalctl --since "1 hour ago" > journal.log

  3. Submit GitHub Issue:

  4. Go to Gthulhu Issues
  5. Choose appropriate issue template
  6. Attach system information and error logs
  7. Describe reproduction steps

Q: How to participate in development?

A: Welcome to participate in development! Please refer to:

  1. View contributing guide: contributing.md

  2. Understand code structure: 

    Gthulhu/
├── main.go              # Main program
├── main.bpf.c          # BPF program
├── internal/sched/     # Scheduling logic
└── api/               # API services

  3. Set up development environment: 

    git clone https://github.com/Gthulhu/Gthulhu.git
cd Gthulhu
make dep
make build
make test

Q: How to add custom scheduling policies?

A: You can customize through the following methods:

  1. Modify BPF program (main.bpf.c): 

    // Add custom CPU selection logic
s32 custom_select_cpu(struct task_struct *p, s32 prev_cpu, u64 wake_flags) {
    // Your logic
    return selected_cpu;
}

  2. Modify Go program (main.go): 

    // Add custom task handling logic
func handleCustomTask(taskInfo *TaskInfo) {
    // Your logic
}

  3. Use Qumun API: 

    // Implement CustomScheduler interface
type MyScheduler struct{}

func (s *MyScheduler) ScheduleTask(task *Task) *ScheduleDecision {
    // Your scheduling logic
    return decision
}

Compatibility Issues

Q: Which Linux distributions are supported?

A: Theoretically supports all distributions with the following conditions:

  • Kernel version: 6.12+
  • sched_ext support: Enabled
  • Architecture: x86_64

Tested distributions: - Ubuntu 24.04+ - Fedora 39+ - Arch Linux (latest)

Planned support: - CentOS/RHEL 9+ - openSUSE Tumbleweed - Debian 13+

Q: Can it run in containers?

A: Yes, but requires special permissions:

# Docker execution
docker run --privileged=true --pid host --rm gthulhu:latest

# Podman execution
podman run --privileged --pid host --rm gthulhu:latest

# Kubernetes execution (requires special configuration)
# Please refer to examples/kubernetes/ directory

Q: Does it conflict with other schedulers?

A: Gthulhu will replace the system default scheduler, therefore:

  • Cannot run simultaneously with other sched_ext schedulers
  • Will not affect real-time scheduling classes (SCHED_FIFO, SCHED_RR)
  • Will replace CFS scheduler functionality

Problem not resolved?

If your problem is not answered here, please:

  1. Check GitHub Issues
  2. Search existing problems and solutions
  3. If not found, please create a new issue