Added P12-log

[Amber-Agarwal]

IMPORTANT:- Note that in the implementation I have committed I have taken the MKFS from the C directory only so my makefile assumes MKFS to be already present in a specified folder. This was done as I didn't have the code for MKFS in rust. Once that code is added we can change the mkfs dependency in the Makefile.

I have implemented the redo logging functionality analogous to the one done in the C code although the file layer is not yet added. I built it on the write commit which I made

• Added src/log.rs implementing:
  • initlog(dev)
  • begin_op()
  • end_op()
  • log_write(buf_idx)
• Internal log mechanics:
  • read_head()
  • write_head()
  • write_log()
  • install_trans()
  • recover_from_log()
  • commit()

Behavior matches the simple single-transaction model:

• begin_op() is a no-op
• end_op() commits immediately
• log header and logged blocks are written in commit order
• recovery replays committed log entries at boot

2. Added log sizing constants

• Updated src/param.rs:
  • MAXOPBLOCKS
  • LOGSIZE = MAXOPBLOCKS * 3

3. Integrated log into filesystem writes
   Updated src/fs.rs so mutating operations use log_write() instead of direct bwrite():

• block zeroing/allocation path (bzero, balloc)
• inode allocation/update (ialloc, iupdate)
• indirect mapping updates in bmap
• inode data writes in writei

This ensures modified blocks are tracked in the transaction and written through the log commit path.

4. Boot-time initialization and transaction boundaries

• Updated src/lib.rs:
  • call log::initlog(ROOTDEV) after iinit(ROOTDEV)
  • wrap mutation paths in welcome() with begin_op() / end_op()

Validation

• make xv6.img fs.img succeeds
• make qemu boots and runs correctly
• Fresh image boot confirms creation paths still work
• Reboot confirms persisted state can be read correctly