locks.txt 2.7 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768
  1. File Locking Release Notes
  2. Andy Walker <andy@lysaker.kvaerner.no>
  3. 12 May 1997
  4. 1. What's New?
  5. --------------
  6. 1.1 Broken Flock Emulation
  7. --------------------------
  8. The old flock(2) emulation in the kernel was swapped for proper BSD
  9. compatible flock(2) support in the 1.3.x series of kernels. With the
  10. release of the 2.1.x kernel series, support for the old emulation has
  11. been totally removed, so that we don't need to carry this baggage
  12. forever.
  13. This should not cause problems for anybody, since everybody using a
  14. 2.1.x kernel should have updated their C library to a suitable version
  15. anyway (see the file "Documentation/Changes".)
  16. 1.2 Allow Mixed Locks Again
  17. ---------------------------
  18. 1.2.1 Typical Problems - Sendmail
  19. ---------------------------------
  20. Because sendmail was unable to use the old flock() emulation, many sendmail
  21. installations use fcntl() instead of flock(). This is true of Slackware 3.0
  22. for example. This gave rise to some other subtle problems if sendmail was
  23. configured to rebuild the alias file. Sendmail tried to lock the aliases.dir
  24. file with fcntl() at the same time as the GDBM routines tried to lock this
  25. file with flock(). With pre 1.3.96 kernels this could result in deadlocks that,
  26. over time, or under a very heavy mail load, would eventually cause the kernel
  27. to lock solid with deadlocked processes.
  28. 1.2.2 The Solution
  29. ------------------
  30. The solution I have chosen, after much experimentation and discussion,
  31. is to make flock() and fcntl() locks oblivious to each other. Both can
  32. exists, and neither will have any effect on the other.
  33. I wanted the two lock styles to be cooperative, but there were so many
  34. race and deadlock conditions that the current solution was the only
  35. practical one. It puts us in the same position as, for example, SunOS
  36. 4.1.x and several other commercial Unices. The only OS's that support
  37. cooperative flock()/fcntl() are those that emulate flock() using
  38. fcntl(), with all the problems that implies.
  39. 1.3 Mandatory Locking As A Mount Option
  40. ---------------------------------------
  41. Mandatory locking, as described in
  42. 'Documentation/filesystems/mandatory-locking.txt' was prior to this release a
  43. general configuration option that was valid for all mounted filesystems. This
  44. had a number of inherent dangers, not the least of which was the ability to
  45. freeze an NFS server by asking it to read a file for which a mandatory lock
  46. existed.
  47. From this release of the kernel, mandatory locking can be turned on and off
  48. on a per-filesystem basis, using the mount options 'mand' and 'nomand'.
  49. The default is to disallow mandatory locking. The intention is that
  50. mandatory locking only be enabled on a local filesystem as the specific need
  51. arises.