It is possible that in a mult-threaded address space write-faults
(or faults in general) on the same address can occur. This is because
threads may become runnable during the handling of the first fault.
This causes duplicate faults on the same private page in the same address space.
On the case of write faulted private page, this causes a spurious page allocation
Currently this case is detected and handled, but in the future we need
a generic way of detecting duplicate faults (of any kind) and cease duplicate
IPC at a very early stage. This is not done yet as it requires knowledge of the
PTEs of physical pages in the pager (like reverse mapping in Linux).
- KIP's pointer to UTCB seems to work with existing l4lib ipc functions.
- Works up to clone()
- In clone we mmap() the same UTCB on each new thread - excessive.
- Generally during page fault handling, cloned threads may fault on the same page
multiple times even though a single handling would be enough for all of them.
Need to detect and handle this.
Added setting of utcb address to l4_thread_control.
This is going to be moved to exchange_registers() since we need to pass
both the utcb physical and virtual address and exregs fits such context
modification better than thread_control.
- Now libl4 has no references to utcb page or shmat etc.
- Pager does not deal with special case utcb page allocation.
It instead allocates a shared page from shm memory pool.
- All tasks working to original standard.
Next:
- Add per-thread utcb allocation from the kernel
- Add larger register file for standard ipc
- Add long ipc (up to 1Kb)
Previously a so-called utcb shared page was used for transfering
data between posix services. This was a special shmat/get/dt case
allocating from its own virtual pool. Now the term utcb is renamed
as a shared page and integrated with the shm* handling routines.
Generic l4 threads will use long-ipc and not this method. Posix
services will continue to communicate on a shared page for now.
modified: tasks/libl4/include/l4lib/ipcdefs.h
modified: tasks/libl4/src/init.c
new file: tasks/libposix/init.c
modified: tasks/mm0/include/shm.h
modified: tasks/mm0/include/task.h
deleted: tasks/mm0/include/utcb.h
modified: tasks/mm0/main.c
modified: tasks/mm0/src/boot.c
modified: tasks/mm0/src/clone.c
modified: tasks/mm0/src/execve.c
modified: tasks/mm0/src/exit.c
modified: tasks/mm0/src/init.c
modified: tasks/mm0/src/shm.c
modified: tasks/mm0/src/task.c
deleted: tasks/mm0/src/utcb.c
deleted: tools/l4-qemu
- Directory creation, file read/write is OK.
- Cannot reuse old task's fds. They are not recycled for some reason.
- Problems with fork/clone/exit. They fail for a reason.
Uncovered a mmap() bug that came along this far. file_offset
parameter of do_mmap() was assigned to the mapped vma as is, i.e.
as a byte offset. This caused problems since most page fault and
other internal code assumed this was a page frame number. This is now
fixed. This came along unnoticed since all mmaps until now started at
file offset 0.
Increased inode block pointers to 40. The current maximum allowed (and checked).
Updates to file size after every file write ensures subsequent writes can
correctly operate using updated file size information (i.e. not try to add
more pages that are already present). We cannot do this inside write() because
directory writes rely on byte-granularity updates on file buffers, whereas
file updates are by page-granularity (currently).
It turned out we used one version of kmalloc for malloc() and another for kfree()!
Now fixed.
Added parent-child relationship to tasks. Need to polish handling CLONE_PARENT and THREAD.
l4_unmap now returns -1 if given range was only partially unmapped.
do_munmap() now only unmaps address ranges that have correspondence in
the unmapped vmas. Trying to unmap regions with no correspondent vmas
causes problems in corner cases, e.g. mm0 that tries to mmap its own
address space during initialisation would unmap its whole address space
and fail to execute.
When sys_munmap() splits a vma, the new vma had no copy of the objects
in the original vma. Now we fixed that using a vma_copy_links() function
which can also be used as part of fork().
Still testing sys_munmap(). It now correctly spots and unmaps the overlapping vma.
The issue now is that if a split occurs, we forgot to add same objects to new vma.
File open was failing when using 2 files with same name. TODO: Look at it in the future.
Need to increase writeable file size in fs0. 16 pages don't work.
do_munmap currently shrinks, splits, destroys vmas and unmaps the given
virtual address range from the task. Unmapped pages may go completely unused
but page reclamation will be done in another part of the pager rather than
directly on the munmap instance.
Bahadir Balban