Input paths are now parsed at the beginning and components
put into an ordered linked list headed by struct pathdata.
Lookup functions use these components to look up vnodes.
Factored out mapping of the physical page as the final generic code
after all fault-specific handling is done.
Fixed the error that zero page didn't have an owner (devzero).
Fixed the error that struct dirent did not have the record length
field as u16 as expected by userspace.
Dirbuf allocation was broken, fixed it.
Added more comments on reading directories.
Issues: fs0 ought to check buf address for sys_readdir(). Currently
test0 is passing an area on its stack and fs0 is writing the data to its own
address space (i.e. an area on its own stack.)
Separated vfs file as a specific file. vm file is not always a vfs file.
Updated the README
sys_open was not returning back to client, added that.
Added comments for future vfs additions.
Removed some commented out code.
Removed excessive printfs.
Fixed spid not initialising for mm0
Fixed some faults with fs0.
TODO:
- Need to store vfs files in a separate list.
- Need to define vnum as a vfs-file-specific data, i.e. in priv_data field of vm_file.
- Need to then fix vfs_receive_sys_open.
- fixed is_err(x), was evaluating x twice, resulting in calling a
function x twice.
- Divided task initialisation into multiple parts.
- MM0 now creates a tcb for itself and maintains memory regions of its own.
- MM0's tcb is used for mmapping other tasks' regions. MM0 mmaps and prefaults
those regions, instead of the typical mmap() and fault approach used by
non-pager tasks.
For example there's an internal shmget_shmat() path to map in other tasks'
shm utcbs. Those mappings are then prefaulted into mm0's address space using
the default fault handling path.
- FS0 now reads task data into its utcb from mm0 via a syscall.
FS0 shmat()s to utcbs of other tasks, e.g. mm0 and test0.
FS0 then crashes, that is to be fixed and where this commit is left last.
utcb as a shared page instead of the message registers.
Implemented the code that passes task information from mm0 to fs0
using the fs0 utcb. The code seems to work OK but:
There's an issue with anon pages that they end up on the same swapfile
and with same file offsets (e.g. utcb and stack at offset 0). Need to
fix this issue but otherwise this implementation seems to work.
TODO:
- Separate anon regions into separate vmfiles.
- Possibly map the stacks from virtual files so that they can be
read from userspace in the future for debugging.
- Possibly utcb could be created as a shared memory object using shmget/shmat
during startup.
Environment is backed by a special per-task file maintained by mm0 for each task.
This file is filled in by the env pager, by simple copying of env data into the
faulty page upon a fault. UTCB and all anon regions (stack) could use the same
scheme.
Fixed IS_ERR(x) to accept negative values that are above -1000 for errors. This
protects against false positives for pointers such as 0xE0000000.
modified: include/l4/generic/scheduler.h
modified: include/l4/macros.h
modified: src/arch/arm/exception.c
modified: tasks/fs0/include/linker.lds
modified: tasks/libl4/src/init.c
modified: tasks/libposix/shm.c
new file: tasks/mm0/include/env.h
modified: tasks/mm0/include/file.h
new file: tasks/mm0/include/lib/addr.h
deleted: tasks/mm0/include/lib/vaddr.h
modified: tasks/mm0/include/task.h
new file: tasks/mm0/include/utcb.h
new file: tasks/mm0/src/env.c
modified: tasks/mm0/src/fault.c
modified: tasks/mm0/src/file.c
modified: tasks/mm0/src/init.c
new file: tasks/mm0/src/lib/addr.c
modified: tasks/mm0/src/lib/idpool.c
deleted: tasks/mm0/src/lib/vaddr.c
modified: tasks/mm0/src/mmap.c
modified: tasks/mm0/src/shm.c
modified: tasks/mm0/src/task.c
new file: tasks/mm0/src/utcb.c
modified: tasks/test0/include/linker.lds
Added reading pages from the page cache into user buffer for sys_read.
Increases stack sizes to 4 pages.
Updated README to include more details about multi-pager environments.
This implements the infrastructure for read/write system calls where
file content is first searched in mm0's page cache and then read-in
or written via the vfs read/write functions.
modified: tasks/fs0/src/syscalls.c
modified: tasks/mm0/include/lib/bit.h
modified: tasks/mm0/include/lib/idpool.h
modified: tasks/mm0/include/task.h
modified: tasks/mm0/include/vm_area.h
modified: tasks/mm0/main.c
modified: tasks/mm0/src/devzero.c
modified: tasks/mm0/src/fault.c
new file: tasks/mm0/src/file.c
modified: tasks/mm0/src/init.c
modified: tasks/mm0/src/lib/bit.c
modified: tasks/mm0/src/lib/idpool.c
modified: tasks/mm0/src/task.c
We now have a single dirbuf of size PAGE_SIZE kept on the vnode. This is
to be used for directory contents only. The reason it's kept on FS0 is
because the contents are modified by calls such as mkdir or create, and otherwise
these would have been handled by mm0 on the page cache buffers, which wouldn't work.
Need to still decide whether content buffers are provided by mm0,
whether need to return them back to mm0, and the read count.
modified: tasks/fs0/src/memfs/vnode.c
modified: tasks/fs0/src/syscalls.c
Headers 3 headers related to message registers and utcbs are now merged under
utcb.h in libl4. Some message register definitions used by the kernel are now
moved into kernel's glue/message.h. This avoids the duplication of same
definitions. Also the total number of mregs are now determined by arch-specific
kernel header, which is good.
Boot files and tasks are now initialised together. Theads can ask for particular
space and thread ids, if they're unused. This enables us to get predefined ids for
known tasks such as the VFS task.
Fixes to README
Other minor fixes.
Previously we had changed the method of setting the ipc tag from l4_ipc() call
argument to being passed as a message register.
- This change was not reflected in l4_ipc() signature as it still had a 3rd argument,
even though ignored.
- l4_set_sender and _set_tag had their arguments wrong way around.
- Previously 5 mrs were passed onto utcb instead of 6, relying on the fact that
l4_ipc tag argument was being passed in r3 directly, this wasnt true anymore
with new convention, but wasn't catered for.
TODO:
- MM0 shouldn't really allocate tids itself, but use ones supplied by C0.
- Sender tid shouldn't really passed by the sender task, but rather by C0. Otherwise
security can be easily breached by user tasks pretending to be other tasks. This
would also save us a message register.
Changed l4id_t type to integer to recognise negative id values like L4_ANYTHREAD.
Added an extremely simple script that cleans and builds everything in right order.
Increased boot pmds by one:
This is due to the fact that if the 1MB initial allocation area of the kernel is
not 1MB-aligned, it is ought to be mapped from the middle of one MB to next,
which requires 2 pmds.
modified: .gdbinit
modified: README
new file: buildall.sh
modified: include/l4/arch/arm/types.h
modified: include/l4/generic/scheduler.h
modified: loader/kernel.S
modified: loader/main.c
modified: loader/mylink.lds
modified: loader/start.axf.S
modified: src/glue/arm/init.c
modified: src/glue/arm/memory.c
modified: tasks/fs0/src/bdev.c
modified: tasks/mm0/include/kdata.h
modified: tasks/mm0/include/vm_area.h
modified: tasks/mm0/src/init.c
modified: tasks/mm0/src/task.c
modified: tools/ksym_to_lds.py
modified: tools/l4-qemu
Previously python hex() would put an extra 'L' after printing out the
value and this would be trimmed in readelf.py. Now it doesn't seem to
do that so the lsd of the number was trimmed. This patch fixes that.
