iru-/9p4

An implementation of the 9P protocol in Forth

9p4

9p4 is an implementation of the 9P protocol in gforth [1]. It provides routines for encoding and decoding 9P [2] [3] messages, along with auxiliary routines for implementing 9P clients and servers.

The remainder of this document assumes familiarity with 9P and its messages.

Data structures

The fields in a 9p4 data structure are named prefix-field, where prefix is the 9P data structure/concept name.

Qid

• qid-type ( 'qid -> 'qid-type )
• qid-version ( 'qid -> 'qid-version )
• qid-path ( 'qid -> 'qid-path )
• qid% ( -> qid-alignment #qid ): for use with gforth struct allocation routines
• /qid ( -> #qid ): size of qid structure

Stat

• stat-size ( 'stat -> 'stat-size )
• stat-type ( 'stat -> 'stat-type )
• stat-dev ( 'stat -> 'stat-dev )
• stat-qid ( 'stat -> 'stat-qid )
• stat-mode ( 'stat -> 'stat-mode )
• stat-atime ( 'stat -> 'stat-atime )
• stat-mtime ( 'stat -> 'stat-mtime )
• stat-length ( 'stat -> 'stat-length )
• stat-name ( 'stat -> 'stat-name )
• stat-uid ( 'stat -> 'stat-uid )
• stat-gid ( 'stat -> 'stat-gid )
• stat-muid ( 'stat -> 'stat-muid )
• stat-base% ( -> stat-alignment #stat ): size of the constant-sized part of stat; for use with gforth struct allocation routines
• /stat-base ( -> #stat ): size of the constant-sized part of stat

Encoding and decoding

Routines encoding T- messages always return a buffer and length containing the encoded message on the top of stack, i.e. their stack diagram has the form ( ... -> ... buf #buf ). On the other hand, all routines decoding R- messages expect the message length on the top of stack, i.e. ( #msg -> ... ).

A routine's name and stack diagram reflect its name and parameters as described in [2]. Whenever there are less items in the stack diagram than in the protocol documentation, 9p4 chooses sensible values for the missing parameters.

• Tversion ( -> buf #buf )
• Rversion ( #msg -> version #version msize )
• Tattach ( uname #uname aname #aname -> rootfid buf #buf )
• Rattach ( #msg -> 'qid )
• Twalk ( name #name ... #names fid -> newfid buf #buf )
• clonefid ( fid -> newfid buf #buf ): same as 0 fid Twalk
• Rwalk ( #msg -> 'qids #qids )
• Topen ( fid mode -> buf #buf )
• Ropen ( #msg -> 'qid iounit )
• Tcreate ( fid name #name perm mode -> buf #buf )
• Rcreate ( #msg -> 'qid iounit )
• Tread ( fid offset count -> buf #buf )
• Rread ( #msg -> data count )
• Twrite ( fid offset data count -> buf #buf )
• Rwrite ( #msg -> count )
• Tclunk ( fid -> buf #buf )
• Rclunk ( #msg -> )
• Tremove ( fid -> buf #buf )
• Rremove ( fid -> buf #buf )
• Tstat ( fid -> buf #buf )
• Rstat ( #msg -> 'stat len )
• Twstat ( 'stat fid -> len )
• Rwstat ( #msg -> )

References

[1] gforth

[2] introduction to the Plan 9 File Protocol, 9P

[3] A sane distributed file system

9p4's Issues

c@+ undefined

c@+ is used without being defined. it's pretty obvious what it's meant to do, especially with c!+ defined, but i thought you might want to tidy it up. c@+ is not present in Forth standard document DPANS94 nor in gforth.

Add documentation