SQLite-based Hibernate Metadata Management

[pbalduino]

Description

Implement SQLite database integration for storing hibernation metadata, providing a robust and queryable way to manage process information, memory mappings, file handles, and system state during hibernate/resume cycles.

Overview

SQLite provides the perfect solution for hibernate metadata:

• Reliable persistence across reboots
• ACID transactions ensure consistency
• SQL queries for complex state reconstruction
• Embedded (no external dependencies)
• Well-tested and battle-hardened

Database Schema

Tables

hibernate_snapshot - Overall snapshot metadata

• snapshot_id (primary key)
• timestamp
• kernel_version
• total_processes
• total_memory_kb
• blob_file_path

hibernate_processes - Per-process information

• pid
• parent_pid
• name
• state (running, sleeping, etc)
• registers_offset (in binary blob)
• memory_start, memory_size
• cwd_path
• snapshot_id (foreign key)

hibernate_memory_map - Memory page mappings

• pid
• virtual_addr
• physical_addr (optional)
• page_size
• permissions (rwx)
• flags (dirty, shared, etc)
• blob_offset (where in binary file)
• snapshot_id (foreign key)
• Primary key: (pid, virtual_addr, snapshot_id)

hibernate_files - Open file descriptors

• pid
• fd (file descriptor number)
• path
• position (file offset)
• flags (O_RDONLY, O_WRONLY, etc)
• snapshot_id (foreign key)

hibernate_signals - Signal handlers

• pid
• signal_number
• handler_address
• flags
• snapshot_id (foreign key)

Indexes

Create indexes on commonly queried columns:

• (snapshot_id) on all tables
• (pid) on process-related tables
• (pid, virtual_addr) on memory_map

Implementation

Save Phase

1. Create new snapshot record with metadata
2. Begin transaction
3. For each process, insert into hibernate_processes
4. Insert memory mappings into hibernate_memory_map
5. Insert open files into hibernate_files
6. Insert signal handlers into hibernate_signals
7. Commit transaction

Restore Phase

1. Open SQLite database from hibernate file
2. Query latest snapshot_id
3. Reconstruct process list from hibernate_processes
4. For each process:
   • Query memory mappings
   • Query open files
   • Query signal handlers
5. Use metadata to guide memory and state restoration

SQLite Integration

Userspace Library

• Port SQLite3 to meniOS (or use minimal build)
• Create hibernate-specific wrapper API
• Handle file I/O through meniOS VFS

Kernel Integration Options

Option A: Userspace Daemon (Recommended)

• Hibernate daemon runs in userspace
• Uses libc + SQLite
• Coordinates with kernel via syscalls
• Kernel provides memory snapshot, daemon manages metadata

Option B: Kernel SQLite

• Embed minimal SQLite in kernel
• More complex, kernel can't easily use standard SQLite
• Not recommended unless absolutely necessary

API

High-level API for hibernate metadata operations.

Tasks

• Port SQLite3 to meniOS userspace
• Create database schema
• Implement snapshot creation functions
• Implement query/restore functions
• Add transaction management
• Create helper functions for common queries
• Add database validation and repair
• Implement version migration (if schema changes)
• Add database compression (optional)
• Create debugging/inspection tools

Acceptance Criteria

• SQLite library integrated into meniOS
• Database schema created and documented
• Can save complete system state to database
• Can query and restore state from database
• Transactions ensure consistency
• Database file validates correctly
• Performance acceptable (metadata operations under 1 second)
• Error handling for database corruption
• Tools for inspecting hibernate database

Priority

High - Required for hibernation metadata

Estimated Effort

1-2 weeks

Notes

• SQLite is ~150KB library, reasonable for meniOS
• Use in-memory mode during save, flush to disk at end
• Consider using WAL mode for better crash resistance
• This is the perfect use case for SQLite - NOT performance critical