brintos

brintos / linux-shallow public Read only

0
0
Text · 43.4 KiB · bdde7e4 Raw
1671 lines · plain
1.. SPDX-License-Identifier: GPL-2.02 3===========================4Coda Kernel-Venus Interface5===========================6 7.. Note::8 9   This is one of the technical documents describing a component of10   Coda -- this document describes the client kernel-Venus interface.11 12For more information:13 14  http://www.coda.cs.cmu.edu15 16For user level software needed to run Coda:17 18  ftp://ftp.coda.cs.cmu.edu19 20To run Coda you need to get a user level cache manager for the client,21named Venus, as well as tools to manipulate ACLs, to log in, etc.  The22client needs to have the Coda filesystem selected in the kernel23configuration.24 25The server needs a user level server and at present does not depend on26kernel support.27 28  The Venus kernel interface29 30  Peter J. Braam31 32  v1.0, Nov 9, 199733 34  This document describes the communication between Venus and kernel35  level filesystem code needed for the operation of the Coda file sys-36  tem.  This document version is meant to describe the current interface37  (version 1.0) as well as improvements we envisage.38 39.. Table of Contents40 41  1. Introduction42 43  2. Servicing Coda filesystem calls44 45  3. The message layer46 47     3.1 Implementation details48 49  4. The interface at the call level50 51     4.1 Data structures shared by the kernel and Venus52     4.2 The pioctl interface53     4.3 root54     4.4 lookup55     4.5 getattr56     4.6 setattr57     4.7 access58     4.8 create59     4.9 mkdir60     4.10 link61     4.11 symlink62     4.12 remove63     4.13 rmdir64     4.14 readlink65     4.15 open66     4.16 close67     4.17 ioctl68     4.18 rename69     4.19 readdir70     4.20 vget71     4.21 fsync72     4.22 inactive73     4.23 rdwr74     4.24 odymount75     4.25 ody_lookup76     4.26 ody_expand77     4.27 prefetch78     4.28 signal79 80  5. The minicache and downcalls81 82     5.1 INVALIDATE83     5.2 FLUSH84     5.3 PURGEUSER85     5.4 ZAPFILE86     5.5 ZAPDIR87     5.6 ZAPVNODE88     5.7 PURGEFID89     5.8 REPLACE90 91  6. Initialization and cleanup92 93     6.1 Requirements94 951. Introduction96===============97 98  A key component in the Coda Distributed File System is the cache99  manager, Venus.100 101  When processes on a Coda enabled system access files in the Coda102  filesystem, requests are directed at the filesystem layer in the103  operating system. The operating system will communicate with Venus to104  service the request for the process.  Venus manages a persistent105  client cache and makes remote procedure calls to Coda file servers and106  related servers (such as authentication servers) to service these107  requests it receives from the operating system.  When Venus has108  serviced a request it replies to the operating system with appropriate109  return codes, and other data related to the request.  Optionally the110  kernel support for Coda may maintain a minicache of recently processed111  requests to limit the number of interactions with Venus.  Venus112  possesses the facility to inform the kernel when elements from its113  minicache are no longer valid.114 115  This document describes precisely this communication between the116  kernel and Venus.  The definitions of so called upcalls and downcalls117  will be given with the format of the data they handle. We shall also118  describe the semantic invariants resulting from the calls.119 120  Historically Coda was implemented in a BSD file system in Mach 2.6.121  The interface between the kernel and Venus is very similar to the BSD122  VFS interface.  Similar functionality is provided, and the format of123  the parameters and returned data is very similar to the BSD VFS.  This124  leads to an almost natural environment for implementing a kernel-level125  filesystem driver for Coda in a BSD system.  However, other operating126  systems such as Linux and Windows 95 and NT have virtual filesystem127  with different interfaces.128 129  To implement Coda on these systems some reverse engineering of the130  Venus/Kernel protocol is necessary.  Also it came to light that other131  systems could profit significantly from certain small optimizations132  and modifications to the protocol. To facilitate this work as well as133  to make future ports easier, communication between Venus and the134  kernel should be documented in great detail.  This is the aim of this135  document.136 1372.  Servicing Coda filesystem calls138===================================139 140  The service of a request for a Coda file system service originates in141  a process P which accessing a Coda file. It makes a system call which142  traps to the OS kernel. Examples of such calls trapping to the kernel143  are ``read``, ``write``, ``open``, ``close``, ``create``, ``mkdir``,144  ``rmdir``, ``chmod`` in a Unix ontext.  Similar calls exist in the Win32145  environment, and are named ``CreateFile``.146 147  Generally the operating system handles the request in a virtual148  filesystem (VFS) layer, which is named I/O Manager in NT and IFS149  manager in Windows 95.  The VFS is responsible for partial processing150  of the request and for locating the specific filesystem(s) which will151  service parts of the request.  Usually the information in the path152  assists in locating the correct FS drivers.  Sometimes after extensive153  pre-processing, the VFS starts invoking exported routines in the FS154  driver.  This is the point where the FS specific processing of the155  request starts, and here the Coda specific kernel code comes into156  play.157 158  The FS layer for Coda must expose and implement several interfaces.159  First and foremost the VFS must be able to make all necessary calls to160  the Coda FS layer, so the Coda FS driver must expose the VFS interface161  as applicable in the operating system. These differ very significantly162  among operating systems, but share features such as facilities to163  read/write and create and remove objects.  The Coda FS layer services164  such VFS requests by invoking one or more well defined services165  offered by the cache manager Venus.  When the replies from Venus have166  come back to the FS driver, servicing of the VFS call continues and167  finishes with a reply to the kernel's VFS. Finally the VFS layer168  returns to the process.169 170  As a result of this design a basic interface exposed by the FS driver171  must allow Venus to manage message traffic.  In particular Venus must172  be able to retrieve and place messages and to be notified of the173  arrival of a new message. The notification must be through a mechanism174  which does not block Venus since Venus must attend to other tasks even175  when no messages are waiting or being processed.176 177  **Interfaces of the Coda FS Driver**178 179  Furthermore the FS layer provides for a special path of communication180  between a user process and Venus, called the pioctl interface. The181  pioctl interface is used for Coda specific services, such as182  requesting detailed information about the persistent cache managed by183  Venus. Here the involvement of the kernel is minimal.  It identifies184  the calling process and passes the information on to Venus.  When185  Venus replies the response is passed back to the caller in unmodified186  form.187 188  Finally Venus allows the kernel FS driver to cache the results from189  certain services.  This is done to avoid excessive context switches190  and results in an efficient system.  However, Venus may acquire191  information, for example from the network which implies that cached192  information must be flushed or replaced. Venus then makes a downcall193  to the Coda FS layer to request flushes or updates in the cache.  The194  kernel FS driver handles such requests synchronously.195 196  Among these interfaces the VFS interface and the facility to place,197  receive and be notified of messages are platform specific.  We will198  not go into the calls exported to the VFS layer but we will state the199  requirements of the message exchange mechanism.200 201 2023.  The message layer203=====================204 205  At the lowest level the communication between Venus and the FS driver206  proceeds through messages.  The synchronization between processes207  requesting Coda file service and Venus relies on blocking and waking208  up processes.  The Coda FS driver processes VFS- and pioctl-requests209  on behalf of a process P, creates messages for Venus, awaits replies210  and finally returns to the caller.  The implementation of the exchange211  of messages is platform specific, but the semantics have (so far)212  appeared to be generally applicable.  Data buffers are created by the213  FS Driver in kernel memory on behalf of P and copied to user memory in214  Venus.215 216  The FS Driver while servicing P makes upcalls to Venus.  Such an217  upcall is dispatched to Venus by creating a message structure.  The218  structure contains the identification of P, the message sequence219  number, the size of the request and a pointer to the data in kernel220  memory for the request.  Since the data buffer is re-used to hold the221  reply from Venus, there is a field for the size of the reply.  A flags222  field is used in the message to precisely record the status of the223  message.  Additional platform dependent structures involve pointers to224  determine the position of the message on queues and pointers to225  synchronization objects.  In the upcall routine the message structure226  is filled in, flags are set to 0, and it is placed on the *pending*227  queue.  The routine calling upcall is responsible for allocating the228  data buffer; its structure will be described in the next section.229 230  A facility must exist to notify Venus that the message has been231  created, and implemented using available synchronization objects in232  the OS. This notification is done in the upcall context of the process233  P. When the message is on the pending queue, process P cannot proceed234  in upcall.  The (kernel mode) processing of P in the filesystem235  request routine must be suspended until Venus has replied.  Therefore236  the calling thread in P is blocked in upcall.  A pointer in the237  message structure will locate the synchronization object on which P is238  sleeping.239 240  Venus detects the notification that a message has arrived, and the FS241  driver allow Venus to retrieve the message with a getmsg_from_kernel242  call. This action finishes in the kernel by putting the message on the243  queue of processing messages and setting flags to READ.  Venus is244  passed the contents of the data buffer. The getmsg_from_kernel call245  now returns and Venus processes the request.246 247  At some later point the FS driver receives a message from Venus,248  namely when Venus calls sendmsg_to_kernel.  At this moment the Coda FS249  driver looks at the contents of the message and decides if:250 251 252  *  the message is a reply for a suspended thread P.  If so it removes253     the message from the processing queue and marks the message as254     WRITTEN.  Finally, the FS driver unblocks P (still in the kernel255     mode context of Venus) and the sendmsg_to_kernel call returns to256     Venus.  The process P will be scheduled at some point and continues257     processing its upcall with the data buffer replaced with the reply258     from Venus.259 260  *  The message is a ``downcall``.  A downcall is a request from Venus to261     the FS Driver. The FS driver processes the request immediately262     (usually a cache eviction or replacement) and when it finishes263     sendmsg_to_kernel returns.264 265  Now P awakes and continues processing upcall.  There are some266  subtleties to take account of. First P will determine if it was woken267  up in upcall by a signal from some other source (for example an268  attempt to terminate P) or as is normally the case by Venus in its269  sendmsg_to_kernel call.  In the normal case, the upcall routine will270  deallocate the message structure and return.  The FS routine can proceed271  with its processing.272 273 274  **Sleeping and IPC arrangements**275 276  In case P is woken up by a signal and not by Venus, it will first look277  at the flags field.  If the message is not yet READ, the process P can278  handle its signal without notifying Venus.  If Venus has READ, and279  the request should not be processed, P can send Venus a signal message280  to indicate that it should disregard the previous message.  Such281  signals are put in the queue at the head, and read first by Venus.  If282  the message is already marked as WRITTEN it is too late to stop the283  processing.  The VFS routine will now continue.  (-- If a VFS request284  involves more than one upcall, this can lead to complicated state, an285  extra field "handle_signals" could be added in the message structure286  to indicate points of no return have been passed.--)287 288 289 2903.1.  Implementation details291----------------------------292 293  The Unix implementation of this mechanism has been through the294  implementation of a character device associated with Coda.  Venus295  retrieves messages by doing a read on the device, replies are sent296  with a write and notification is through the select system call on the297  file descriptor for the device.  The process P is kept waiting on an298  interruptible wait queue object.299 300  In Windows NT and the DPMI Windows 95 implementation a DeviceIoControl301  call is used.  The DeviceIoControl call is designed to copy buffers302  from user memory to kernel memory with OPCODES. The sendmsg_to_kernel303  is issued as a synchronous call, while the getmsg_from_kernel call is304  asynchronous.  Windows EventObjects are used for notification of305  message arrival.  The process P is kept waiting on a KernelEvent306  object in NT and a semaphore in Windows 95.307 308 3094.  The interface at the call level310===================================311 312 313  This section describes the upcalls a Coda FS driver can make to Venus.314  Each of these upcalls make use of two structures: inputArgs and315  outputArgs.   In pseudo BNF form the structures take the following316  form::317 318 319	struct inputArgs {320	    u_long opcode;321	    u_long unique;     /* Keep multiple outstanding msgs distinct */322	    u_short pid;                 /* Common to all */323	    u_short pgid;                /* Common to all */324	    struct CodaCred cred;        /* Common to all */325 326	    <union "in" of call dependent parts of inputArgs>327	};328 329	struct outputArgs {330	    u_long opcode;331	    u_long unique;       /* Keep multiple outstanding msgs distinct */332	    u_long result;333 334	    <union "out" of call dependent parts of inputArgs>335	};336 337 338 339  Before going on let us elucidate the role of the various fields. The340  inputArgs start with the opcode which defines the type of service341  requested from Venus. There are approximately 30 upcalls at present342  which we will discuss.   The unique field labels the inputArg with a343  unique number which will identify the message uniquely.  A process and344  process group id are passed.  Finally the credentials of the caller345  are included.346 347  Before delving into the specific calls we need to discuss a variety of348  data structures shared by the kernel and Venus.349 350 351 352 3534.1.  Data structures shared by the kernel and Venus354----------------------------------------------------355 356 357  The CodaCred structure defines a variety of user and group ids as358  they are set for the calling process. The vuid_t and vgid_t are 32 bit359  unsigned integers.  It also defines group membership in an array.  On360  Unix the CodaCred has proven sufficient to implement good security361  semantics for Coda but the structure may have to undergo modification362  for the Windows environment when these mature::363 364	struct CodaCred {365	    vuid_t cr_uid, cr_euid, cr_suid, cr_fsuid; /* Real, effective, set, fs uid */366	    vgid_t cr_gid, cr_egid, cr_sgid, cr_fsgid; /* same for groups */367	    vgid_t cr_groups[NGROUPS];        /* Group membership for caller */368	};369 370 371  .. Note::372 373     It is questionable if we need CodaCreds in Venus. Finally Venus374     doesn't know about groups, although it does create files with the375     default uid/gid.  Perhaps the list of group membership is superfluous.376 377 378  The next item is the fundamental identifier used to identify Coda379  files, the ViceFid.  A fid of a file uniquely defines a file or380  directory in the Coda filesystem within a cell [1]_::381 382	typedef struct ViceFid {383	    VolumeId Volume;384	    VnodeId Vnode;385	    Unique_t Unique;386	} ViceFid;387 388  .. [1] A cell is agroup of Coda servers acting under the aegis of a single389	 system control machine or SCM. See the Coda Administration manual390	 for a detailed description of the role of the SCM.391 392  Each of the constituent fields: VolumeId, VnodeId and Unique_t are393  unsigned 32 bit integers.  We envisage that a further field will need394  to be prefixed to identify the Coda cell; this will probably take the395  form of a Ipv6 size IP address naming the Coda cell through DNS.396 397  The next important structure shared between Venus and the kernel is398  the attributes of the file.  The following structure is used to399  exchange information.  It has room for future extensions such as400  support for device files (currently not present in Coda)::401 402 403	struct coda_timespec {404		int64_t         tv_sec;         /* seconds */405		long            tv_nsec;        /* nanoseconds */406	};407 408	struct coda_vattr {409		enum coda_vtype va_type;        /* vnode type (for create) */410		u_short         va_mode;        /* files access mode and type */411		short           va_nlink;       /* number of references to file */412		vuid_t          va_uid;         /* owner user id */413		vgid_t          va_gid;         /* owner group id */414		long            va_fsid;        /* file system id (dev for now) */415		long            va_fileid;      /* file id */416		u_quad_t        va_size;        /* file size in bytes */417		long            va_blocksize;   /* blocksize preferred for i/o */418		struct coda_timespec va_atime;  /* time of last access */419		struct coda_timespec va_mtime;  /* time of last modification */420		struct coda_timespec va_ctime;  /* time file changed */421		u_long          va_gen;         /* generation number of file */422		u_long          va_flags;       /* flags defined for file */423		dev_t           va_rdev;        /* device special file represents */424		u_quad_t        va_bytes;       /* bytes of disk space held by file */425		u_quad_t        va_filerev;     /* file modification number */426		u_int           va_vaflags;     /* operations flags, see below */427		long            va_spare;       /* remain quad aligned */428	};429 430 4314.2.  The pioctl interface432--------------------------433 434 435  Coda specific requests can be made by application through the pioctl436  interface. The pioctl is implemented as an ordinary ioctl on a437  fictitious file /coda/.CONTROL.  The pioctl call opens this file, gets438  a file handle and makes the ioctl call. Finally it closes the file.439 440  The kernel involvement in this is limited to providing the facility to441  open and close and pass the ioctl message and to verify that a path in442  the pioctl data buffers is a file in a Coda filesystem.443 444  The kernel is handed a data packet of the form::445 446	struct {447	    const char *path;448	    struct ViceIoctl vidata;449	    int follow;450	} data;451 452 453 454  where::455 456 457	struct ViceIoctl {458		caddr_t in, out;        /* Data to be transferred in, or out */459		short in_size;          /* Size of input buffer <= 2K */460		short out_size;         /* Maximum size of output buffer, <= 2K */461	};462 463 464 465  The path must be a Coda file, otherwise the ioctl upcall will not be466  made.467 468  .. Note:: The data structures and code are a mess.  We need to clean this up.469 470 471**We now proceed to document the individual calls**:472 473 4744.3.  root475----------476 477 478  Arguments479     in480 481	empty482 483     out::484 485		struct cfs_root_out {486		    ViceFid VFid;487		} cfs_root;488 489 490 491  Description492    This call is made to Venus during the initialization of493    the Coda filesystem. If the result is zero, the cfs_root structure494    contains the ViceFid of the root of the Coda filesystem. If a non-zero495    result is generated, its value is a platform dependent error code496    indicating the difficulty Venus encountered in locating the root of497    the Coda filesystem.498 4994.4.  lookup500------------501 502 503  Summary504    Find the ViceFid and type of an object in a directory if it exists.505 506  Arguments507     in::508 509		struct  cfs_lookup_in {510		    ViceFid     VFid;511		    char        *name;          /* Place holder for data. */512		} cfs_lookup;513 514 515 516     out::517 518		struct cfs_lookup_out {519		    ViceFid VFid;520		    int vtype;521		} cfs_lookup;522 523 524 525  Description526    This call is made to determine the ViceFid and filetype of527    a directory entry.  The directory entry requested carries name 'name'528    and Venus will search the directory identified by cfs_lookup_in.VFid.529    The result may indicate that the name does not exist, or that530    difficulty was encountered in finding it (e.g. due to disconnection).531    If the result is zero, the field cfs_lookup_out.VFid contains the532    targets ViceFid and cfs_lookup_out.vtype the coda_vtype giving the533    type of object the name designates.534 535  The name of the object is an 8 bit character string of maximum length536  CFS_MAXNAMLEN, currently set to 256 (including a 0 terminator.)537 538  It is extremely important to realize that Venus bitwise ors the field539  cfs_lookup.vtype with CFS_NOCACHE to indicate that the object should540  not be put in the kernel name cache.541 542  .. Note::543 544     The type of the vtype is currently wrong.  It should be545     coda_vtype. Linux does not take note of CFS_NOCACHE.  It should.546 547 5484.5.  getattr549-------------550 551 552  Summary Get the attributes of a file.553 554  Arguments555     in::556 557		struct cfs_getattr_in {558		    ViceFid VFid;559		    struct coda_vattr attr; /* XXXXX */560		} cfs_getattr;561 562 563 564     out::565 566		struct cfs_getattr_out {567		    struct coda_vattr attr;568		} cfs_getattr;569 570 571 572  Description573    This call returns the attributes of the file identified by fid.574 575  Errors576    Errors can occur if the object with fid does not exist, is577    unaccessible or if the caller does not have permission to fetch578    attributes.579 580  .. Note::581 582     Many kernel FS drivers (Linux, NT and Windows 95) need to acquire583     the attributes as well as the Fid for the instantiation of an internal584     "inode" or "FileHandle".  A significant improvement in performance on585     such systems could be made by combining the lookup and getattr calls586     both at the Venus/kernel interaction level and at the RPC level.587 588  The vattr structure included in the input arguments is superfluous and589  should be removed.590 591 5924.6.  setattr593-------------594 595 596  Summary597    Set the attributes of a file.598 599  Arguments600     in::601 602		struct cfs_setattr_in {603		    ViceFid VFid;604		    struct coda_vattr attr;605		} cfs_setattr;606 607 608 609 610     out611 612	empty613 614  Description615    The structure attr is filled with attributes to be changed616    in BSD style.  Attributes not to be changed are set to -1, apart from617    vtype which is set to VNON. Other are set to the value to be assigned.618    The only attributes which the FS driver may request to change are the619    mode, owner, groupid, atime, mtime and ctime.  The return value620    indicates success or failure.621 622  Errors623    A variety of errors can occur.  The object may not exist, may624    be inaccessible, or permission may not be granted by Venus.625 626 6274.7.  access628------------629 630 631  Arguments632     in::633 634		struct cfs_access_in {635		    ViceFid     VFid;636		    int flags;637		} cfs_access;638 639 640 641     out642 643	empty644 645  Description646    Verify if access to the object identified by VFid for647    operations described by flags is permitted.  The result indicates if648    access will be granted.  It is important to remember that Coda uses649    ACLs to enforce protection and that ultimately the servers, not the650    clients enforce the security of the system.  The result of this call651    will depend on whether a token is held by the user.652 653  Errors654    The object may not exist, or the ACL describing the protection655    may not be accessible.656 657 6584.8.  create659------------660 661 662  Summary663    Invoked to create a file664 665  Arguments666     in::667 668		struct cfs_create_in {669		    ViceFid VFid;670		    struct coda_vattr attr;671		    int excl;672		    int mode;673		    char        *name;          /* Place holder for data. */674		} cfs_create;675 676 677 678 679     out::680 681		struct cfs_create_out {682		    ViceFid VFid;683		    struct coda_vattr attr;684		} cfs_create;685 686 687 688  Description689    This upcall is invoked to request creation of a file.690    The file will be created in the directory identified by VFid, its name691    will be name, and the mode will be mode.  If excl is set an error will692    be returned if the file already exists.  If the size field in attr is693    set to zero the file will be truncated.  The uid and gid of the file694    are set by converting the CodaCred to a uid using a macro CRTOUID695    (this macro is platform dependent).  Upon success the VFid and696    attributes of the file are returned.  The Coda FS Driver will normally697    instantiate a vnode, inode or file handle at kernel level for the new698    object.699 700 701  Errors702    A variety of errors can occur. Permissions may be insufficient.703    If the object exists and is not a file the error EISDIR is returned704    under Unix.705 706  .. Note::707 708     The packing of parameters is very inefficient and appears to709     indicate confusion between the system call creat and the VFS operation710     create. The VFS operation create is only called to create new objects.711     This create call differs from the Unix one in that it is not invoked712     to return a file descriptor. The truncate and exclusive options,713     together with the mode, could simply be part of the mode as it is714     under Unix.  There should be no flags argument; this is used in open715     (2) to return a file descriptor for READ or WRITE mode.716 717  The attributes of the directory should be returned too, since the size718  and mtime changed.719 720 7214.9.  mkdir722-----------723 724 725  Summary726    Create a new directory.727 728  Arguments729     in::730 731		struct cfs_mkdir_in {732		    ViceFid     VFid;733		    struct coda_vattr attr;734		    char        *name;          /* Place holder for data. */735		} cfs_mkdir;736 737 738 739     out::740 741		struct cfs_mkdir_out {742		    ViceFid VFid;743		    struct coda_vattr attr;744		} cfs_mkdir;745 746 747 748 749  Description750    This call is similar to create but creates a directory.751    Only the mode field in the input parameters is used for creation.752    Upon successful creation, the attr returned contains the attributes of753    the new directory.754 755  Errors756    As for create.757 758  .. Note::759 760     The input parameter should be changed to mode instead of761     attributes.762 763  The attributes of the parent should be returned since the size and764  mtime changes.765 766 7674.10.  link768-----------769 770 771  Summary772    Create a link to an existing file.773 774  Arguments775     in::776 777		struct cfs_link_in {778		    ViceFid sourceFid;          /* cnode to link *to* */779		    ViceFid destFid;            /* Directory in which to place link */780		    char        *tname;         /* Place holder for data. */781		} cfs_link;782 783 784 785     out786 787	empty788 789  Description790    This call creates a link to the sourceFid in the directory791    identified by destFid with name tname.  The source must reside in the792    target's parent, i.e. the source must be have parent destFid, i.e. Coda793    does not support cross directory hard links.  Only the return value is794    relevant.  It indicates success or the type of failure.795 796  Errors797    The usual errors can occur.798 799 8004.11.  symlink801--------------802 803 804  Summary805    create a symbolic link806 807  Arguments808     in::809 810		struct cfs_symlink_in {811		    ViceFid     VFid;          /* Directory to put symlink in */812		    char        *srcname;813		    struct coda_vattr attr;814		    char        *tname;815		} cfs_symlink;816 817 818 819     out820 821	none822 823  Description824    Create a symbolic link. The link is to be placed in the825    directory identified by VFid and named tname.  It should point to the826    pathname srcname.  The attributes of the newly created object are to827    be set to attr.828 829  .. Note::830 831     The attributes of the target directory should be returned since832     its size changed.833 834 8354.12.  remove836-------------837 838 839  Summary840    Remove a file841 842  Arguments843     in::844 845		struct cfs_remove_in {846		    ViceFid     VFid;847		    char        *name;          /* Place holder for data. */848		} cfs_remove;849 850 851 852     out853 854	none855 856  Description857    Remove file named cfs_remove_in.name in directory858    identified by   VFid.859 860 861  .. Note::862 863     The attributes of the directory should be returned since its864     mtime and size may change.865 866 8674.13.  rmdir868------------869 870 871  Summary872    Remove a directory873 874  Arguments875     in::876 877		struct cfs_rmdir_in {878		    ViceFid     VFid;879		    char        *name;          /* Place holder for data. */880		} cfs_rmdir;881 882 883 884     out885 886	none887 888  Description889    Remove the directory with name 'name' from the directory890    identified by VFid.891 892  .. Note:: The attributes of the parent directory should be returned since893	    its mtime and size may change.894 895 8964.14.  readlink897---------------898 899 900  Summary901    Read the value of a symbolic link.902 903  Arguments904     in::905 906		struct cfs_readlink_in {907		    ViceFid VFid;908		} cfs_readlink;909 910 911 912     out::913 914		struct cfs_readlink_out {915		    int count;916		    caddr_t     data;           /* Place holder for data. */917		} cfs_readlink;918 919 920 921  Description922    This routine reads the contents of symbolic link923    identified by VFid into the buffer data.  The buffer data must be able924    to hold any name up to CFS_MAXNAMLEN (PATH or NAM??).925 926  Errors927    No unusual errors.928 929 9304.15.  open931-----------932 933 934  Summary935    Open a file.936 937  Arguments938     in::939 940		struct cfs_open_in {941		    ViceFid     VFid;942		    int flags;943		} cfs_open;944 945 946 947     out::948 949		struct cfs_open_out {950		    dev_t       dev;951		    ino_t       inode;952		} cfs_open;953 954 955 956  Description957    This request asks Venus to place the file identified by958    VFid in its cache and to note that the calling process wishes to open959    it with flags as in open(2).  The return value to the kernel differs960    for Unix and Windows systems.  For Unix systems the Coda FS Driver is961    informed of the device and inode number of the container file in the962    fields dev and inode.  For Windows the path of the container file is963    returned to the kernel.964 965 966  .. Note::967 968     Currently the cfs_open_out structure is not properly adapted to969     deal with the Windows case.  It might be best to implement two970     upcalls, one to open aiming at a container file name, the other at a971     container file inode.972 973 9744.16.  close975------------976 977 978  Summary979    Close a file, update it on the servers.980 981  Arguments982     in::983 984		struct cfs_close_in {985		    ViceFid     VFid;986		    int flags;987		} cfs_close;988 989 990 991     out992 993	none994 995  Description996    Close the file identified by VFid.997 998  .. Note::999 1000     The flags argument is bogus and not used.  However, Venus' code1001     has room to deal with an execp input field, probably this field should1002     be used to inform Venus that the file was closed but is still memory1003     mapped for execution.  There are comments about fetching versus not1004     fetching the data in Venus vproc_vfscalls.  This seems silly.  If a1005     file is being closed, the data in the container file is to be the new1006     data.  Here again the execp flag might be in play to create confusion:1007     currently Venus might think a file can be flushed from the cache when1008     it is still memory mapped.  This needs to be understood.1009 1010 10114.17.  ioctl1012------------1013 1014 1015  Summary1016    Do an ioctl on a file. This includes the pioctl interface.1017 1018  Arguments1019     in::1020 1021		struct cfs_ioctl_in {1022		    ViceFid VFid;1023		    int cmd;1024		    int len;1025		    int rwflag;1026		    char *data;                 /* Place holder for data. */1027		} cfs_ioctl;1028 1029 1030 1031     out::1032 1033 1034		struct cfs_ioctl_out {1035		    int len;1036		    caddr_t     data;           /* Place holder for data. */1037		} cfs_ioctl;1038 1039 1040 1041  Description1042    Do an ioctl operation on a file.  The command, len and1043    data arguments are filled as usual.  flags is not used by Venus.1044 1045  .. Note::1046 1047     Another bogus parameter.  flags is not used.  What is the1048     business about PREFETCHING in the Venus code?1049 1050 1051 10524.18.  rename1053-------------1054 1055 1056  Summary1057    Rename a fid.1058 1059  Arguments1060     in::1061 1062		struct cfs_rename_in {1063		    ViceFid     sourceFid;1064		    char        *srcname;1065		    ViceFid destFid;1066		    char        *destname;1067		} cfs_rename;1068 1069 1070 1071     out1072 1073	none1074 1075  Description1076    Rename the object with name srcname in directory1077    sourceFid to destname in destFid.   It is important that the names1078    srcname and destname are 0 terminated strings.  Strings in Unix1079    kernels are not always null terminated.1080 1081 10824.19.  readdir1083--------------1084 1085 1086  Summary1087    Read directory entries.1088 1089  Arguments1090     in::1091 1092		struct cfs_readdir_in {1093		    ViceFid     VFid;1094		    int count;1095		    int offset;1096		} cfs_readdir;1097 1098 1099 1100 1101     out::1102 1103		struct cfs_readdir_out {1104		    int size;1105		    caddr_t     data;           /* Place holder for data. */1106		} cfs_readdir;1107 1108 1109 1110  Description1111    Read directory entries from VFid starting at offset and1112    read at most count bytes.  Returns the data in data and returns1113    the size in size.1114 1115 1116  .. Note::1117 1118     This call is not used.  Readdir operations exploit container1119     files.  We will re-evaluate this during the directory revamp which is1120     about to take place.1121 1122 11234.20.  vget1124-----------1125 1126 1127  Summary1128    instructs Venus to do an FSDB->Get.1129 1130  Arguments1131     in::1132 1133		struct cfs_vget_in {1134		    ViceFid VFid;1135		} cfs_vget;1136 1137 1138 1139     out::1140 1141		struct cfs_vget_out {1142		    ViceFid VFid;1143		    int vtype;1144		} cfs_vget;1145 1146 1147 1148  Description1149    This upcall asks Venus to do a get operation on an fsobj1150    labelled by VFid.1151 1152  .. Note::1153 1154     This operation is not used.  However, it is extremely useful1155     since it can be used to deal with read/write memory mapped files.1156     These can be "pinned" in the Venus cache using vget and released with1157     inactive.1158 1159 11604.21.  fsync1161------------1162 1163 1164  Summary1165    Tell Venus to update the RVM attributes of a file.1166 1167  Arguments1168     in::1169 1170		struct cfs_fsync_in {1171		    ViceFid VFid;1172		} cfs_fsync;1173 1174 1175 1176     out1177 1178	none1179 1180  Description1181    Ask Venus to update RVM attributes of object VFid. This1182    should be called as part of kernel level fsync type calls.  The1183    result indicates if the syncing was successful.1184 1185  .. Note:: Linux does not implement this call. It should.1186 1187 11884.22.  inactive1189---------------1190 1191 1192  Summary1193    Tell Venus a vnode is no longer in use.1194 1195  Arguments1196     in::1197 1198		struct cfs_inactive_in {1199		    ViceFid VFid;1200		} cfs_inactive;1201 1202 1203 1204     out1205 1206	none1207 1208  Description1209    This operation returns EOPNOTSUPP.1210 1211  .. Note:: This should perhaps be removed.1212 1213 12144.23.  rdwr1215-----------1216 1217 1218  Summary1219    Read or write from a file1220 1221  Arguments1222     in::1223 1224		struct cfs_rdwr_in {1225		    ViceFid     VFid;1226		    int rwflag;1227		    int count;1228		    int offset;1229		    int ioflag;1230		    caddr_t     data;           /* Place holder for data. */1231		} cfs_rdwr;1232 1233 1234 1235 1236     out::1237 1238		struct cfs_rdwr_out {1239		    int rwflag;1240		    int count;1241		    caddr_t     data;   /* Place holder for data. */1242		} cfs_rdwr;1243 1244 1245 1246  Description1247    This upcall asks Venus to read or write from a file.1248 1249 1250  .. Note::1251 1252    It should be removed since it is against the Coda philosophy that1253    read/write operations never reach Venus.  I have been told the1254    operation does not work.  It is not currently used.1255 1256 1257 12584.24.  odymount1259---------------1260 1261 1262  Summary1263    Allows mounting multiple Coda "filesystems" on one Unix mount point.1264 1265  Arguments1266     in::1267 1268		struct ody_mount_in {1269		    char        *name;          /* Place holder for data. */1270		} ody_mount;1271 1272 1273 1274     out::1275 1276		struct ody_mount_out {1277		    ViceFid VFid;1278		} ody_mount;1279 1280 1281 1282  Description1283    Asks Venus to return the rootfid of a Coda system named1284    name.  The fid is returned in VFid.1285 1286  .. Note::1287 1288     This call was used by David for dynamic sets.  It should be1289     removed since it causes a jungle of pointers in the VFS mounting area.1290     It is not used by Coda proper.  Call is not implemented by Venus.1291 1292 12934.25.  ody_lookup1294-----------------1295 1296 1297  Summary1298    Looks up something.1299 1300  Arguments1301     in1302 1303	irrelevant1304 1305 1306     out1307 1308	irrelevant1309 1310 1311  .. Note:: Gut it. Call is not implemented by Venus.1312 1313 13144.26.  ody_expand1315-----------------1316 1317 1318  Summary1319    expands something in a dynamic set.1320 1321  Arguments1322     in1323 1324	irrelevant1325 1326     out1327 1328	irrelevant1329 1330  .. Note:: Gut it. Call is not implemented by Venus.1331 1332 13334.27.  prefetch1334---------------1335 1336 1337  Summary1338    Prefetch a dynamic set.1339 1340  Arguments1341 1342     in1343 1344	Not documented.1345 1346     out1347 1348	Not documented.1349 1350  Description1351    Venus worker.cc has support for this call, although it is1352    noted that it doesn't work.  Not surprising, since the kernel does not1353    have support for it. (ODY_PREFETCH is not a defined operation).1354 1355 1356  .. Note:: Gut it. It isn't working and isn't used by Coda.1357 1358 1359 13604.28.  signal1361-------------1362 1363 1364  Summary1365    Send Venus a signal about an upcall.1366 1367  Arguments1368     in1369 1370	none1371 1372     out1373 1374	not applicable.1375 1376  Description1377    This is an out-of-band upcall to Venus to inform Venus1378    that the calling process received a signal after Venus read the1379    message from the input queue.  Venus is supposed to clean up the1380    operation.1381 1382  Errors1383    No reply is given.1384 1385  .. Note::1386 1387     We need to better understand what Venus needs to clean up and if1388     it is doing this correctly.  Also we need to handle multiple upcall1389     per system call situations correctly.  It would be important to know1390     what state changes in Venus take place after an upcall for which the1391     kernel is responsible for notifying Venus to clean up (e.g. open1392     definitely is such a state change, but many others are maybe not).1393 1394 13955.  The minicache and downcalls1396===============================1397 1398 1399  The Coda FS Driver can cache results of lookup and access upcalls, to1400  limit the frequency of upcalls.  Upcalls carry a price since a process1401  context switch needs to take place.  The counterpart of caching the1402  information is that Venus will notify the FS Driver that cached1403  entries must be flushed or renamed.1404 1405  The kernel code generally has to maintain a structure which links the1406  internal file handles (called vnodes in BSD, inodes in Linux and1407  FileHandles in Windows) with the ViceFid's which Venus maintains.  The1408  reason is that frequent translations back and forth are needed in1409  order to make upcalls and use the results of upcalls.  Such linking1410  objects are called cnodes.1411 1412  The current minicache implementations have cache entries which record1413  the following:1414 1415  1. the name of the file1416 1417  2. the cnode of the directory containing the object1418 1419  3. a list of CodaCred's for which the lookup is permitted.1420 1421  4. the cnode of the object1422 1423  The lookup call in the Coda FS Driver may request the cnode of the1424  desired object from the cache, by passing its name, directory and the1425  CodaCred's of the caller.  The cache will return the cnode or indicate1426  that it cannot be found.  The Coda FS Driver must be careful to1427  invalidate cache entries when it modifies or removes objects.1428 1429  When Venus obtains information that indicates that cache entries are1430  no longer valid, it will make a downcall to the kernel.  Downcalls are1431  intercepted by the Coda FS Driver and lead to cache invalidations of1432  the kind described below.  The Coda FS Driver does not return an error1433  unless the downcall data could not be read into kernel memory.1434 1435 14365.1.  INVALIDATE1437----------------1438 1439 1440  No information is available on this call.1441 1442 14435.2.  FLUSH1444-----------1445 1446 1447 1448  Arguments1449    None1450 1451  Summary1452    Flush the name cache entirely.1453 1454  Description1455    Venus issues this call upon startup and when it dies. This1456    is to prevent stale cache information being held.  Some operating1457    systems allow the kernel name cache to be switched off dynamically.1458    When this is done, this downcall is made.1459 1460 14615.3.  PURGEUSER1462---------------1463 1464 1465  Arguments1466    ::1467 1468	  struct cfs_purgeuser_out {/* CFS_PURGEUSER is a venus->kernel call */1469	      struct CodaCred cred;1470	  } cfs_purgeuser;1471 1472 1473 1474  Description1475    Remove all entries in the cache carrying the Cred.  This1476    call is issued when tokens for a user expire or are flushed.1477 1478 14795.4.  ZAPFILE1480-------------1481 1482 1483  Arguments1484    ::1485 1486	  struct cfs_zapfile_out {  /* CFS_ZAPFILE is a venus->kernel call */1487	      ViceFid CodaFid;1488	  } cfs_zapfile;1489 1490 1491 1492  Description1493    Remove all entries which have the (dir vnode, name) pair.1494    This is issued as a result of an invalidation of cached attributes of1495    a vnode.1496 1497  .. Note::1498 1499     Call is not named correctly in NetBSD and Mach.  The minicache1500     zapfile routine takes different arguments. Linux does not implement1501     the invalidation of attributes correctly.1502 1503 1504 15055.5.  ZAPDIR1506------------1507 1508 1509  Arguments1510    ::1511 1512	  struct cfs_zapdir_out {   /* CFS_ZAPDIR is a venus->kernel call */1513	      ViceFid CodaFid;1514	  } cfs_zapdir;1515 1516 1517 1518  Description1519    Remove all entries in the cache lying in a directory1520    CodaFid, and all children of this directory. This call is issued when1521    Venus receives a callback on the directory.1522 1523 15245.6.  ZAPVNODE1525--------------1526 1527 1528 1529  Arguments1530    ::1531 1532	  struct cfs_zapvnode_out { /* CFS_ZAPVNODE is a venus->kernel call */1533	      struct CodaCred cred;1534	      ViceFid VFid;1535	  } cfs_zapvnode;1536 1537 1538 1539  Description1540    Remove all entries in the cache carrying the cred and VFid1541    as in the arguments. This downcall is probably never issued.1542 1543 15445.7.  PURGEFID1545--------------1546 1547 1548  Arguments1549    ::1550 1551	  struct cfs_purgefid_out { /* CFS_PURGEFID is a venus->kernel call */1552	      ViceFid CodaFid;1553	  } cfs_purgefid;1554 1555 1556 1557  Description1558    Flush the attribute for the file. If it is a dir (odd1559    vnode), purge its children from the namecache and remove the file from the1560    namecache.1561 1562 1563 15645.8.  REPLACE1565-------------1566 1567 1568  Summary1569    Replace the Fid's for a collection of names.1570 1571  Arguments1572    ::1573 1574	  struct cfs_replace_out { /* cfs_replace is a venus->kernel call */1575	      ViceFid NewFid;1576	      ViceFid OldFid;1577	  } cfs_replace;1578 1579 1580 1581  Description1582    This routine replaces a ViceFid in the name cache with1583    another.  It is added to allow Venus during reintegration to replace1584    locally allocated temp fids while disconnected with global fids even1585    when the reference counts on those fids are not zero.1586 1587 15886.  Initialization and cleanup1589==============================1590 1591 1592  This section gives brief hints as to desirable features for the Coda1593  FS Driver at startup and upon shutdown or Venus failures.  Before1594  entering the discussion it is useful to repeat that the Coda FS Driver1595  maintains the following data:1596 1597 1598  1. message queues1599 1600  2. cnodes1601 1602  3. name cache entries1603 1604     The name cache entries are entirely private to the driver, so they1605     can easily be manipulated.   The message queues will generally have1606     clear points of initialization and destruction.  The cnodes are1607     much more delicate.  User processes hold reference counts in Coda1608     filesystems and it can be difficult to clean up the cnodes.1609 1610  It can expect requests through:1611 1612  1. the message subsystem1613 1614  2. the VFS layer1615 1616  3. pioctl interface1617 1618     Currently the pioctl passes through the VFS for Coda so we can1619     treat these similarly.1620 1621 16226.1.  Requirements1623------------------1624 1625 1626  The following requirements should be accommodated:1627 1628  1. The message queues should have open and close routines.  On Unix1629     the opening of the character devices are such routines.1630 1631    -  Before opening, no messages can be placed.1632 1633    -  Opening will remove any old messages still pending.1634 1635    -  Close will notify any sleeping processes that their upcall cannot1636       be completed.1637 1638    -  Close will free all memory allocated by the message queues.1639 1640 1641  2. At open the namecache shall be initialized to empty state.1642 1643  3. Before the message queues are open, all VFS operations will fail.1644     Fortunately this can be achieved by making sure than mounting the1645     Coda filesystem cannot succeed before opening.1646 1647  4. After closing of the queues, no VFS operations can succeed.  Here1648     one needs to be careful, since a few operations (lookup,1649     read/write, readdir) can proceed without upcalls.  These must be1650     explicitly blocked.1651 1652  5. Upon closing the namecache shall be flushed and disabled.1653 1654  6. All memory held by cnodes can be freed without relying on upcalls.1655 1656  7. Unmounting the file system can be done without relying on upcalls.1657 1658  8. Mounting the Coda filesystem should fail gracefully if Venus cannot1659     get the rootfid or the attributes of the rootfid.  The latter is1660     best implemented by Venus fetching these objects before attempting1661     to mount.1662 1663  .. Note::1664 1665     NetBSD in particular but also Linux have not implemented the1666     above requirements fully.  For smooth operation this needs to be1667     corrected.1668 1669 1670 1671