Differences between revisions 13 and 15 (spanning 2 versions)
Revision 13 as of 2007-04-06 07:49:45
Size: 10467
Comment:
Revision 15 as of 2007-04-06 08:41:12
Size: 10076
Comment:
Deletions are marked like this. Additions are marked like this.
Line 107: Line 107:
 * cdrskin: opens the devices with O_EXCL flag. Opening of /dev/sg is prevented by creation of bus/target/lun table and mapping the request to /dev/sr with it, even when dev=/dev/sgX is specified.  * growisofs: opens the block devices with O_EXCL flag. Opening /dev/sg was never encouraged and does not work on kernel 2.4 (not tested yet on 2.6).

 * cdrskin (via libburn): opens the devices with O_EXCL flag. It uses a unique device file path for serious operations on the drive. This is /dev/sg* on kernel 2.4, and recently has become /dev/sr* on kernel 2.6. Operations on other path representations of the same device are restricted to open(2) O_RDONLY and to obtaining SCSI parameters host,channel,id,lun.
Line 111: Line 113:
=== Proposed general locking algorithm === === Proposed general locking algorithms ===
Line 113: Line 115:
'''NOTE: all of this comes down to fall if the user creates custom device files and works with them. When we rely on file identity to represent the device then it won't work. The user (even root) shall not be allowed custom device files (in theory) unless there is some way for burning apps and unsafe reading apps to know about each other and the activity on the device. *HELP*. Kernel developers don't get it.''' === What can be done within the CD/DVD tool community ===
Line 117: Line 119:
The locking methods with additional lock files are identified as inappropriate because of inconsistent security settings on different applications, see above. Instead, we propose a two-step locking method: The locking methods with additional lock files are identified as very cumbersome because of inconsistent security settings on different applications, see above. Instead, this proposal relies on a two-step locking method and does not yet solve the ambiguity problem of sr-scd-sg paths:
Line 138: Line 140:
Normally, fcntl(2) imposes advisory locking. But the sysadmin can make this mandatory locking by a mount option. (But is /dev/ mount(8)ed at all ?)
Line 139: Line 142:
=== Unique file resolution === Unique path resolution is necessary so all interested processes come together at the same inode where this locking is guaranteed to collide. (Does F_SETLK work with all the implementations of directory /dev/ ?)
Line 141: Line 144:
The following applies only to applications using the delicate operation mode (burning).
Since there is still the problem with Linux' inability to share locks across the device driver borders, we need to make sure that the access happens through a unique device file.
Links or additional device paths created by mknod(1) can be translated into /dev/sg* resp. /dev/sr* by help call stat(2) and its result element .st_rdev. But the jump from /dev/sg* to /dev/sr* is not possible via stat(2). For that we need open(2) O_RDONLY, ioctl(SCSI_IOCTL_GET_IDLUN), close(2).
Line 144: Line 146:
For kernel 2.6, it is possible to retrieve physical address information from /dev/sgX which can be used to scan "/dev/sr%d" nodes quickly in this order unless the first node matching the physical address of /dev/sgX device in question is found. Thereafter this device file should be used for locking and further communication. Since no additional physical information is maintained by the kernel for /dev/hdX and maybe other types of devices, the locking shall happen on the device chosen by user. If some kind of fake SCSI hardware address emulation needs to be performed (for example for old cdrecord-using applications) then the scanning mechanism should always select the first appropriate device when scanning device files matching the "/dev/hd%c" pattern beginning "/dev/hda". The translation is done by obtaining info from the given path, by iterating over the desired device paths /dev/hd%c /dev/sg%d /dev/sr%d, and by comparing their info with the one we look for. Kernel 2.6: If the result is a /dev/sg%d then it has to be translated into a /dev/sr%d in another step.
Line 148: Line 150:
For kernel 2.4, we are almost out of luck. On the one hand, only /dev/sg* devices are recommended for burning, on the other hand there are no good ways to map the identities between /dev/{sg,sr,hd}* devices. The best thing that can be done is the use of fcntl locks (see above) applied on the device file specified by the user. When an alternative SCSI address (real or fake) is used for device specifications, the device scanning method shall happen in the way described above for kernel 2.6.x. When the old unofficial ATAPI interface is used with /dev/hdX nodes, the locking shall be performed on them directly.
Kernel 2.4 imposes the problem that ioctl(SG_IO) is not possible with sr, so most of the programs have to use sg. Only growisofs uses sr via ioctl(CDROM_SEND_PACKET). We will possibly not come to a completely sufficient agreement under these circumstances. Well, we 2.4ers are used to suffer neglect. (sob, hehe)

On Locking Schemes on Linux Device Drivers

Hello fellow application developer or maintainer,

recently we (cdrkit and libburnia developers) came accross increasing problems with reliable and safe device locking. This paper enlightens the issues behind the scenes and presents possible future solutions.

Introduction

Our original concern is the influence of even read-only operations on optical media drives (recorders) during their duty as recorders -- depending on the device model such read-only work may interrupt the process badly practically destroying the medium.

Since many programs already do act on such devices in an unsafe manner, either willingly (e.g. liblkid) or accidentally (e.g. hald, opening with O_EXCL but still clashing with cdr applications working on the competing sg driver), we see the need for reliable communication in order to ensure proper device locking where appropriate, in a way which is appropriate for the particular application. In the following document, first the currently possible mechanisms are itemized with their advantages and their problems, followed by a draft of a locking scheme which shall cope with the particular requirements.

State of the practice

There are various locking techniques used in other areas which are more or less applicable in our case.

General inter-process locking mechanisms

In general, all the mechanisms listed below are not optimally appropriate for our purpose. They lack on two places which make then not reliable when used alone:

  • they do not cope with multiple device file which imply the access to the same driver through different files
  • they do not automatically cope with multiple device drivers accessible through different user space interfaces, like with sg vs. sr drivers on Linux. No matter how many excuses some kernel developers do present to paper over this obvious shortcomings. Automatic use of /dev/sr instead of /dev/sg is not always possible or may not be wanted by the user.

Finally, they may be sufficient to lower the risk on inappropriate operation. Which exactly are available in the wild?

  • System V Semaphores Principle: a magic integer, "key" or "semid", identifies a set of state objects on which the atomic operations can be performed which are necessary for implementing a proper locking algorithm. See man semget(2), semop(2) SEM_UNDO. Pros: semaphores are originally designed for our purpose and they are very traditional Unix requisites. Cons:
    • the semaphore key must be systemwide unique for the set of lockable drives and all participating programs have to use the same key. This situation is prone to collisions with locking mechanisms for other system resources. Function ftok(3) is not a secure solution.
    • each device needs a fixely defined index number in the set of semaphores which are allocated system resources. So we can hardly span up a giant index space where we can map different device file classes to disjoint index intervals.
  • Lock files associated with target file Principle: an additional file is created during the action on the real target file. Pros: regular filesystem operation, no additional infrastructure required Cons:
    • Possible races unless OS mechanisms are used for exclusive operation on the lock file, see below
    • The location and name of the lock file need to be known and discussed upfront among all application developers, or be documented excessively
    • Permission problems may disallow the creation of lock files (security issues), especially for self-compiled applications and having no root permissions to install them in a required way
    • Special precautions are necessary against stale locks
  • fcntl(2) exclusive file locking Principle: lock applied on open file handles. Internally associated with a path, see fcntl(2) for details. Pros:
    • known (POSIX.1-2001), usually reliable mechanism
    Cons:
    • diverges from flock() implementation on Linux, see below. Results in independent locking.
    • possible problems on network file systems
  • flock(2) exclusive file locking Principle: similar to fcntl locks, applied with a different system function. Pros: see fcntl(2) locking above Cons: like flock(2), but less portable, not working over network file systems

Advanced Linux-specific locking mechanisms

  • O_EXCL locking
    • Principle: passing of the O_EXCL flag to the open call. The device is locked exclusively for the calling PID, the lock is maintained in the device driver to the particular major/minor combination.
    Pros:
    • reliable for a device accessible through one driver
    Cons:
    • requires kernel 2.6.x (x>=7 or so)

    • does not automagicaly make the device inaccessible, only applications using O_EXCL will know about the locked state when getting negative result with EBUSY errno value.

Applicability on CD/(HD)DVD/BD drives

As explained in the introduction, the locking is important on optical media recording due to the delicate operation mode during the recording. Ideally, no application should touch them, even reading from the media is an evil task. But how does the state of the practice look like?

  • mount: the block device is mounted with the O_EXCL flag BUT the mount executable also uses libblkid which opens the devices without locking and read magic data from it. This also provides no solution for operation through the sg driver.

  • hald (HAL daemon): periodically opens the cdrom block devices with O_EXCL flag. Clashes with operation on sg is possible.
  • wodim: opens the devices with O_EXCL flag. Opening /dev/sg is possible and happens more likely with versions prior to 1.1.4.
  • growisofs: opens the block devices with O_EXCL flag. Opening /dev/sg was never encouraged and does not work on kernel 2.4 (not tested yet on 2.6).
  • cdrskin (via libburn): opens the devices with O_EXCL flag. It uses a unique device file path for serious operations on the drive. This is /dev/sg* on kernel 2.4, and recently has become /dev/sr* on kernel 2.6. Operations on other path representations of the same device are restricted to open(2) O_RDONLY and to obtaining SCSI parameters host,channel,id,lun.
  • cdrecord: no locking. Author recommends to get rid of applications which may touch the device somehow.

Proposed general locking algorithms

What can be done within the CD/DVD tool community

The following method is proposed to create a midway between the limitations of kernel and the requirements of others, also unifying the way of dealing with the device locks.

The locking methods with additional lock files are identified as very cumbersome because of inconsistent security settings on different applications, see above. Instead, this proposal relies on a two-step locking method and does not yet solve the ambiguity problem of sr-scd-sg paths:

  1. Open the device. For applications that operate in a delicate way (burning tools), O_EXCL shall be set. For others, it may be omited.
  2. Set or check the additional fcntl lock on the device file. It must be exclusive! Sample code (by Thomas Schmitt)

        struct flock lockthing;
        ...
        f = open(device, mode|O_EXCL);
        if (f != -1) {
             memset(&lockthing, 0, sizeof(lockthing));
             lockthing.l_type = F_WRLCK;
             lockthing.l_whence = SEEK_SET;
             lockthing.l_start = 0;
             lockthing.l_len = 0;
             if (fcntl(f, F_SETLK, &lockthing)) {
                close(f);
                /* user feedback, report error, etc... */
                f = -1;
             }
        }

Normally, fcntl(2) imposes advisory locking. But the sysadmin can make this mandatory locking by a mount option. (But is /dev/ mount(8)ed at all ?)

Unique path resolution is necessary so all interested processes come together at the same inode where this locking is guaranteed to collide. (Does F_SETLK work with all the implementations of directory /dev/ ?)

Links or additional device paths created by mknod(1) can be translated into /dev/sg* resp. /dev/sr* by help call stat(2) and its result element .st_rdev. But the jump from /dev/sg* to /dev/sr* is not possible via stat(2). For that we need open(2) O_RDONLY, ioctl(SCSI_IOCTL_GET_IDLUN), close(2).

The translation is done by obtaining info from the given path, by iterating over the desired device paths /dev/hd%c /dev/sg%d /dev/sr%d, and by comparing their info with the one we look for. Kernel 2.6: If the result is a /dev/sg%d then it has to be translated into a /dev/sr%d in another step.

NOTE: there are sysfs symlinks that set up a path usable to map exactly. However, this depends on a mounted sysfs and the required symlinks have also been declared deprecated in the recent Linux kernel versions.

Kernel 2.4 imposes the problem that ioctl(SG_IO) is not possible with sr, so most of the programs have to use sg. Only growisofs uses sr via ioctl(CDROM_SEND_PACKET). We will possibly not come to a completely sufficient agreement under these circumstances. Well, we 2.4ers are used to suffer neglect. (sob, hehe)

Appendix

Why not lock files? As told above...

First: races

Second: unclear or unreliable cleanup technique, dangling bad lockfiles possible

Third: The most obvious problem is the usual permission setting of /var/lock :

SuSE 9.0 (kernel 2.4):
  drwxrwxr-x    3 root     uucp         4096 Apr  4     05:07 /var/lock
SuSE 9.3 (kernel 2.6):
  drwxrwxr-t    4 root     uucp         4096 2007-04-04 17:50 /var/lock
| Fedora Core 3.x:
|   drwxrwxr-x    5 root     lock         4096 Apr  4     04:03 /var/lock
| Debian gives rw-permission to anybody and thus would be no problem.

This system may work with the plain old UUCP program and few others programs with low device opening activity AND administrated by root but is a real PITA nowadays.