legacy/libburn/branches/ZeroFourTwo/doc/ddlp.txt

389 lines
18 KiB
Plaintext
Raw Normal View History

-------------------------------------------------------------------------------
Users of modern desktop Linux installations report misburns with CD/DVD
recording due to concurrency problems.
This text describes two locking protocols which have been developed by our
best possible effort. But finally they rather serve as repelling example of
what would be needed in user space to achieve an insufficient partial solution.
Ted Ts'o was so friendly to help as critic with his own use cases. It turned
out that we cannot imagine a way in user space how to cover reliably the needs
of callers of libblkid and the needs of our burn programs.
-------------------------------------------------------------------------------
Content:
The "Delicate Device Locking Protocol" shall demonstrate our sincere
consideration of the problem.
"What are the Stumble Stones ?" lists reasons why the effort finally failed.
-----------------------------------------------------------------------------
Delicate Device Locking Protocol
(a joint sub project of cdrkit and libburnia)
(contact: scdbackup@gmx.net )
Our projects provide programs which allow recording of data on CD or DVD.
We encounter an increasing number of bug reports about spoiled burn runs and
wasted media which obviously have one common cause: interference by other
programs which access the drive's device files.
There is some riddling about which gestures exactly are dangerous for
ongoing recordings or can cause weirdly misformatted drive replies to MMC
commands.
We do know, nevertheless, that these effects do not occur if no other program
accesses a device file of the drive while our programs use it.
DDLP shall help to avoid collisions between programs in the process of
recording to a CD or DVD drive and other programs which access that drive.
The protocol intends to provide advisory locking. So any good-willing program
has to take some extra precautions to participate.
If a program does not feel vulnerable to disturbance, then the precautions
impose much less effort than if the program feels the need for protection.
Two locking strategies are specified:
DDLP-A operates on device files only. It is very Linux specific.
DDLP-B adds proxy lock files, inspired by FHS /var/lock standard.
DDLP-A
This protocol relies on the hardly documented feature open(O_EXCL | O_RDWR)
with Linux device files and on POSIX compliant fcntl(F_SETLK).
Other than the original meaning of O_EXCL with creating regular files, the
effect on device files is mutual exclusion of access. I.e. if one
filedescriptor is open on that combination of major-minor device number, then
no other open(O_EXCL) will succeed. But open() without O_EXCL would succeed.
So this is advisory and exclusive locking.
With kernel 2.6 it seems to work on all device drivers which might get used
to access a CD/DVD drive.
The vulnerable programs shall not start their operation before they occupied a
wide collection of drive representations.
Non-vulnerable programs shall take care to detect the occupation of _one_ such
representation.
So for Friendly Programs
A program which does not feel vulnerable to disturbance is urged to access
CD/DVD drives by opening a file descriptor which will uphold the lock
as long as it does not get closed. There are two alternative ways to achieve
this.
Very reliable is
open( some_path , O_EXCL | ...)
But O_EXCL imposes restrictions and interferences:
- O_EXCL | O_RDONLY does not succeed with /dev/sg* !
- O_EXCL cannot provide shared locks for programs which only want to lock
against burn programs but not against their own peers.
- O_EXCL keeps from obtaining information by harmless activities.
- O_EXCL already has a meaning with devices which are mounted as filesystems.
This priority meaning is more liberal than the one needed for CD/DV recording
protection.
So it may be necessary to use a cautious open() without O_EXCL and to aquire
a POSIX lock via fcntl(). "Cautious" means to add O_NDELAY to the flags of
open(), because this is declared to avoid side effects within open().
With this gesture it is important to use the paths expected by our burn
programs: /dev/sr[0..255] /dev/scd[0..255] /dev/sg[0..255] /dev/hd[a..z]
because fcntl(F_SETLK) does not lock the device but only a device-inode.
std_path = one of the standard device files:
/dev/sr[0..255] /dev/scd[0..255] /dev/sg[0..255] /dev/hd[a..z]
or a symbolic link pointing to one of them.
open( std_path , ... | O_NDELAY)
fcntl(F_SETLK) and close() on failure
... eventually disable O_NDELAY by fcntl(F_SETFL) ...
There is a pitfall mentioned in man 2 fcntl :
"locks are automatically released [...] if it closes any file descriptor
referring to a file on which locks are held. This is bad [...]"
So you may have to re-lock after some temporary fd got closed.
Vulnerable Programs
For programs which do feel vulnerable, O_EXCL would suffice for the /dev/hd*
device file family and their driver. But USB and SATA recorders appear with
at least two different major-minor combinations simultaneously.
One as /dev/sr* alias /dev/scd*, the other as /dev/sg*.
The same is true for ide-scsi or recorders attached to SCSI controllers.
So, in order to lock any access to the recorder, one has to open(O_EXCL)
not only the device file that is intended for accessing the recorder but also
a device file of any other major-minor representation of the recorder.
This is done via the SCSI address parameter vector (Host,Channel,Id,Lun)
and a search on standard device file paths /dev/sr* /dev/scd* /dev/sg*.
In this text the alternative device representations are called "siblings".
For finding them, it is necessary to apply open() to many device files which
might be occupied by delicate operations. On the other hand it is very
important to occupy all reasonable representations of the drive.
So the reading of the (Host,Channel,Id,Lun) parameters demands an
open(O_RDONLY | O_NDELAY) _without_ fcntl() in order to find the outmost
number of representations among the standard device files. Only ioctls
SCSI_IOCTL_GET_IDLUN and SCSI_IOCTL_GET_BUS_NUMBER are applied.
Hopefully this gesture is unable to cause harmful side effects on kernel 2.6.
At least one file of each class sr, scd and sg should be found to regard
the occupation as satisfying. Thus corresponding sr-scd-sg triplets should have
matching ownerships and access permissions.
One will have to help the sysadmins to find those triplets.
A spicy detail is that sr and scd may be distinct device files for the same
major-minor combination. In this case fcntl() locks on both are needed
but O_EXCL can only be applied to one of them.
An open and free implementation ddlpa.[ch] is provided as
http://libburnia.pykix.org/browser/libburn/trunk/libburn/ddlpa.h?format=txt
http://libburnia.pykix.org/browser/libburn/trunk/libburn/ddlpa.c?format=txt
The current version of this text is
http://libburnia.pykix.org/browser/libburn/trunk/doc/ddlp.txt?format=txt
Put ddlpa.h and ddlpa.c into the same directory and compile as test program by
cc -g -Wall -DDDLPA_C_STANDALONE -o ddlpa ddlpa.c
Use it to occupy a drive's representations for a given number of seconds
./ddlpa /dev/sr0 300
It should do no harm to any of your running activities.
If it does: Please, please alert us.
Your own programs should not be able to circumvent the occupation if they
obey above rules for Friendly Programs.
Of course ./ddlpa should be unable to circumvent itself.
A successfull occupation looks like
DDLPA_DEBUG: ddlpa_std_by_rdev("/dev/scd0") = "/dev/sr0"
DDLPA_DEBUG: ddlpa_collect_siblings() found "/dev/sr0"
DDLPA_DEBUG: ddlpa_collect_siblings() found "/dev/scd0"
DDLPA_DEBUG: ddlpa_collect_siblings() found "/dev/sg0"
DDLPA_DEBUG: ddlpa_occupy() : '/dev/scd0'
DDLPA_DEBUG: ddlpa_occupy() O_EXCL : '/dev/sg0'
DDLPA_DEBUG: ddlpa_occupy() O_EXCL : '/dev/sr0'
---------------------------------------------- Lock gained
ddlpa: opened /dev/sr0
ddlpa: opened siblings: /dev/scd0 /dev/sg0
slept 1 seconds of 300
Now an attempt via device file alias /dev/NEC must fail:
DDLPA_DEBUG: ddlpa_std_by_rdev("/dev/NEC") = "/dev/sg0"
DDLPA_DEBUG: ddlpa_collect_siblings() found "/dev/sr0"
DDLPA_DEBUG: ddlpa_collect_siblings() found "/dev/scd0"
DDLPA_DEBUG: ddlpa_collect_siblings() found "/dev/sg0"
Cannot exclusively open '/dev/sg0'
Reason given : Failed to open O_RDWR | O_NDELAY | O_EXCL : '/dev/sr0'
Error condition : 16 'Device or resource busy'
With hdc, of course, things are trivial
DDLPA_DEBUG: ddlpa_std_by_rdev("/dev/hdc") = "/dev/hdc"
DDLPA_DEBUG: ddlpa_occupy() O_EXCL : '/dev/hdc'
---------------------------------------------- Lock gained
ddlpa: opened /dev/hdc
slept 1 seconds of 1
Ted Ts'o provided program open-cd-excl which allows to explore open(2) on
device files with combinations of read-write, O_EXCL, and fcntl().
(This does not mean that Ted endorsed our project yet. He helps exploring.)
Friendly in the sense of DDLP-A would be any run which uses at least one of
the options -e (i.e. O_EXCL) or -f (i.e. F_SETLK, applied to a file
descriptor which was obtained from a standard device file path).
The code is available under GPL at
http://libburnia.pykix.org/browser/libburn/trunk/test/open-cd-excl.c?format=txt
To be compiled by
cc -g -Wall -o open-cd-excl open-cd-excl.c
Options:
-e : open O_EXCL
-f : aquire lock by fcntl(F_SETLK) after sucessful open
-i : do not wait in case of success but exit 0 immediately
-r : open O_RDONLY , with -f use F_RDLCK
-w : open O_RDWR , with -f use F_WRLCK
plus the path of the devce file to open.
Friendly Programs would use gestures like:
./open-cd-excl -e -r /dev/sr0
./open-cd-excl -e -w /dev/sg1
./open-cd-excl -e -w /dev/black-drive
./open-cd-excl -f -r /dev/sg1
./open-cd-excl -e -f -w /dev/sr0
Ignorant programs would use and cause potential trouble by:
./open-cd-excl -r /dev/sr0
./open-cd-excl -w /dev/sg1
./open-cd-excl -f -w /dev/black-drive
where "/dev/black-drive" is _not_ a symbolic link to
any of /dev/sr* /dev/scd* /dev/sg* /dev/hd*, but has an own inode.
Prone to failure without further reason is:
./open-cd-excl -e -r /dev/sg1
----------------------------------------------------------------------------
DDLP-B
This protocol relies on proxy lock files in some filesystem directory. It can
be embedded into DDLP-A or it can be used be used standalone, outside DDLP-A.
DDLP-A shall be kept by DDLP-B from trying to access any device file which
might already be in use. There is a problematic gesture in DDLP-A when SCSI
address parameters are to be retrieved. For now this gesture seems to be
harmless. But one never knows.
Vice versa DDLP-B may get from DDLP-A the service to search for SCSI device
file siblings. So they are best as a couple.
But they are not perfect. Not even as couple. fcntl() locking is flawed.
There is a proxy file locking protocol described in FHS:
http://www.pathname.com/fhs/pub/fhs-2.3.html#VARLOCKLOCKFILES
But it has shortcommings (see below). Decisive obstacle for its usage are the
possibility for stale locks and the lack of shared locks.
DDLP-B rather defines a "path prefix" which is advised to be
/tmp/ddlpb-lock-
This prefix will get appended "device specific suffixes" and then form the path
of a "lockfile".
Not the existence of a lockfile but its occupation by an fcntl(F_SETLK) will
constitute a lock. Lockfiles may get prepared by the sysadmin in directories
where normal users are not allowed to create new files. Their rw-permissions
then act as additional access restriction to the device files.
The use of fcntl(F_SETLK) will prevent any stale locks after the process ended.
It will also allow to obtain shared locks as well as exclusive locks.
There are two classes of device specific suffixes:
- Device file path suffix. Absolute paths only. "/" gets replaced by "_-".
Eventual "_-" in path gets replaced by "_-_-". The leading group of "_-"
is always interpreted as a group of "/", though. E.g.:
/dev/sr0 <-> "_-dev_-sr0"
/mydevs/burner/nec <-> "_-mydevs_-burners_-nec"
/dev/rare_-name <-> "_-dev_-rare_-_-name"
///strange/dev/x <-> "_-_-_-strange_-dev_-x"
- st_rdev suffix. A hex representation of struct stat.st_rdev. Capital letters.
The number of characters is pare with at most one leading 0. I.e. bytewise
printf("%2.2X") beginning with the highest order byte that is not zero.
E.g. : "0B01", "2200", "01000000000004001"
If a lockfile does not exist and cannot be created then this shall not keep
a program from working on a device. But if a lockfile exists and if permissions
or locking state do not allow to obtain a lock of the appropirate type, then
this shall prevent any opening of device file in question resp. shall cause
immediate close(2) of an already opened device file.
The vulnerable programs shall not start their operation before they locked a
wide collection of drive representations.
Non-vulnerable programs shall take care to lock the suffix resulting from the
path they will be using and the suffix from the st_rdev from that path.
The latter is to be obtained by call stat(2).
Locks get upheld as long as their file descriptor is not closed or no other
incident as described in man 2 fcntl releases the lock.
So with shared locks there are no imandatory further activities after they
have been obtained.
In case of exclusive locks, the file has to have been opened for writing and
must be truncated to 0 bytes length immediately after obtaining the lock.
When releasing an exclusive lock it is a nice gesture to
already do this truncation.
Then a /var/lock/ compatible first line has to be written.
E.g. by: printf("%10u\n",(unsigned) getpid()) yielding " 1230\n".
Any further lines are optional. They shall have the form Name=Value and must
be printable cleartext. If such further lines exist, then the last one must
have the name "endmark".
Defined Names are:
hostid =hostname of the machine where the process number of line 1 is valid
start =start time of lock in seconds since 1970. E.g: 1177147634.592410
program =self chosen name of the program which obtained the lock
argv0 =argv[0] of that program
mainpath =device file path which will be used for operations by that program
path =device file path which lead to the lock
st_rdev =st_rdev suffix which is associated with path
scsi_hcil=eventual SCSI parameters Host,Channel,Id,Lun
scsi_bus =eventual SCSI parameter Bus
endmark =declares the info as complete.
Any undefined name or a line without "=" shall be handled as comment.
"=" in the value is allowed. Any line beginning with an "=" character is an
extension of the previous value.
If programs encounter an exclusive lock, they are invited to read the content
of the lockfile anyway. But they should be aware that the info might be in the
progress of emerging. There is a race condition possible in the short time
between obtaining the exclusive lock and erasing the file content.
If it is not crucial to obtain most accurate info then one may take the newline
of the first line as indicator of a valid process number and the "endmark"
name as indicator that the preceding lines are valid.
Very cautious readers should obtain the info twice with a decent waiting period
inbetween. Only if both results are identical they should be considered valid.
There is no implementation of DDLP-B yet.
----------------------------------------------------------------------------
What are the Stumble Stones ?
----------------------------------------------------------------------------
Any of the considered locking mechanisms has decisive shortcommings
which keeps it from being the solution to all known legitimate use cases.
The attempt has failed to compose a waterproof locking mechanism from means of
POSIX, FHS and from hardly documented Linux open(O_EXCL) on device files.
The resulting mechanisms would need about 1000 lines of code and still do
not close all gaps resp. cover the well motivated use cases.
This attempt you see above: DDLP-A and DDLP-B.
Summary of the reasons why the established locking mechanisms do not suffice:
None of the mechanisms can take care of the double device driver identity
sr versus sg. To deduce the one device file from the other involves the need
to open many other (possibly unrelated) device files with the risk to disturb
them.
This hard to solve problem is aggravated by the following facts.
Shortcommings of Linux specific open(O_EXCL) :
- O_EXCL | O_RDONLY does not succeed with /dev/sg*
- O_EXCL cannot provide shared locks for programs which only want to lock
against burn programs but not against their own peers.
- O_EXCL keeps from obtaining information by harmless activities.
- O_EXCL already has a meaning with devices which are mounted as filesystems.
This priority meaning is more liberal than the one needed for CD/DV recording
protection.
Shortcommings of POSIX fcntl(F_SETLK) :
- fcntl() demands an open file descriptor. open(2) might have side effects.
- fcntl() locks can be released inadvertedly by submodules which just open and
close the same file (inode ?) without refering to fcntl locks in any way.
See man 2 fcntl "This is bad:".
Stacking of software modules is a widely used design pattern. But fcntl()
cannot cope with that.
Shortcommings of FHS /var/lock/ :
- Stale locks are possible.
- It is necessary to create a file (using the _old_ meaning of O_EXCL flag ?)
but /var/lock/ might not be available early during system start and it often
has restrictive permission settings.
- There is no way to indicate a difference between exclusive and shared locks.
- The FHS prescription relies entirely on the basename of the device file path.