Timeline

Table of Contents Office hours Project Plan May April March

Office hours

3 - 10 May
- try to figure out how to print filename of the traced function.
- patch functions code to introduce additional information
- analyze define_trace initial commit
- can we add accounting information to kmemleak to keep track of allocated/freed memory

26 - 3 May
- place a tracepoint inside kmalloc and record stats (filename, size).
19 - 26 April
- Analyze CONFIG_SLAB_DEBUG and do the same for CONFIG_MEM_DEBUG
- Try to tag memory by hooking in kmalloc code with an inline function that getters information about the context.
- Analyze if tracepoints can do the same as above.
- Is there any memory path allocation in kernel which does not appear in /proc/meminfo ?
12 - 19 April
- Propose solution for memory tagging
- Add -f option for meminfo script
5 - 12 April
- Add wiki page with all memory allocation sources
- To search for link with slub/slob/slab config
- Understand and find the code for the fields in cat /proc/meminfo output
- Check where top program gets its information (probably its /proc/meminfo)
- Analyze if we can do slabtop per process
- create a script for diff-ing two cat /proc/meminfo outputs.

22 - 29 March
- On demand call for dump_tasks (procfs or sysfs)
- Tracepoints (Daniel to provide more info)
- Push code for modules on repo.
- Check for mm subsystem mailing list.
- Determine all available sources for memory allocation (use dump_stack in __alloc_pages()).
15 - 22 March
- create a kernel module that generates an OOM
- analyze the code for OOM Killer (check if possible to dynamically deactivate it).
- check if OOM killer can be activated on demand to print memory info on all existing processes.
- determine all available sources for memory allocation.
8 - 15 March
- identify sources for memory allocation in kernel
- analyze source code for kmemleak
- analyze ExAllocatePoolWithTag Windows function. Can we borrow some idea?