- Removed virtual functions for all of the stat stuff.
This did however introduce some issues, mainly with /proc
becoming out of sync if you changed your ID. I propose we
do the linux thing and just have a stat update function
which is optional, but allows dynamic updates of stat fields
for cases such as those in uid/gid in /proc.
- Simplified the API, although still kind of annoying
it is a bit simpler.
- Moved some of the FS structure from having the FS inode inside
the in memory inode to a Serialise <-> Deserialise model where
Inodes are deserialised from disk into in memory ones and then
back into on disk ones when it comes time for syncing.
This makes it semantically better in my opinion, as it explicitly
separates disk and non-disk functionality.
This moves locking to the inodes themselves which allows reducing lock
times significantly. Main inodes (ext2 and tmpfs) still do contain a
single big mutex that gets locked during operations but now we have the
architecture to optimize these.
If multiple threads were waiting for more block buffers without anyone
releasing them, they ended up in a deadlock.
Now we store 6 blocks for 8 threads. If a thread already has a block
buffer, it will not have to wait for a new one. Only if there are more
than 8 threads using blocks, will it block until there are free slots
for a thread available.
Each allocated inode used to call sync(). Each sync reads and writes
a block from the filesystem. Doing a 1 MiB write ended up syncing around
257 times
Small symlinks store their target in inline storage of data block
indices. Freeing blocks did not account for this and tried to interpret
the link target as block inidices to delete
This also removes the now old recvfrom and sendto syscalls. These are
now implemented as wrappers around recvmsg and sendmsg.
Also replace unnecessary spinlocks from unix socket with mutexes
I had a hardlimit of 10 block buffers and if they ran out, the kernel
would crash. this patchs increases the number of buffers to 16 and
removes the crash condition when they run out :D
Kernel can just use raw threads, pretty muchs the only thing that
process provides is syscalls which kernel threads of course don't
need.
Also this makes init process have pid 1 :D
This allows banan-os to boot on hardware where we don't have working
storage driver or the storage driver fails (pretty common with my usb
mass storage drivers...)
This makes creating files and appending to then A LOT faster. Some code
I tested took 40 seconds in the previous implementation and less than a
second on the new one!
This code is really sketcy, I hope I'll never have to touch it again :)
