🐛 fix bus error on smaller readonly file in unix

[xgdgsc]

fix #28245 in unix

[xgdgsc]

This seems clearer to me. But don' t know if it would have any performance issue.

[giordano]

Tests are extremely unhappy on almost all platforms

[vtjnash]

We may be able to emulate (aka greatly accelerate) the filesize test by doing a seekend instead and checking the position?

[xgdgsc]

As grow! does nothing to Anonymous. I did the above commit. What do you think of the macOS failure? I looked at the c code and have no idea. Would it be fine to keep the old behaviour for macOS? Should SharedArrays.jl be changed?

[KristofferC]

@vtjnash, could you take another look at this?

[vtjnash]

Tests are still failing, so there is not much more I can really do here for review. My previous suggestion for a fix seems to have been unsuccessful, since the seekend is failing.

Error in testset SharedArrays:
Error During Test at /Users/julia/.julia/scratchspaces/a66863c6-20e8-4ff4-8a62-49f30b1f605e/agent-cache/default-macmini-x64-2.0/build/default-macmini-x64-2-0/julialang/julia-master/julia-926f1dde71/share/julia/test/testdefs.jl:21
  Got exception outside of a @test
  LoadError: IOError: filesize: Illegal seek for <fd 21>
  Stacktrace:
    [1] filesize(s::IOStream)
      @ Base ./iostream.jl:226

[vtjnash]

Actually, IIUC, this is a very weird file on macOS (from shm_open). I can read and write, but not seek the fd. And stat is a bit awkward, as Julia cannot return it:

julia> Base.StatStruct(stat_buf)
StatStruct for ""
   size: 4096 bytes
 device: 0
  inode: 0
   mode: 0o000600 (-rw-------)
  nlink: 0
    uid: 501 (jameson)
    gid: 20 (staff)
   rdev: 0
  blksz: 0
 blocks: 0
  mtime: 1969-12-31T19:00:00-0500 (19401 days ago)
  ctime: 1969-12-31T19:00:00-0500 (19401 days ago)

julia> stat(s)
ERROR: stat returned zero type for a valid path

implemented here: https://github.com/apple/darwin-xnu/blob/a1babec6b135d1f35b2590a1990af3c5c5393479/bsd/kern/posix_shm.c#L695-L700

This seems to be an implementation quality issue with IOStream, where we attempt to use seekend as a fast-path for filesize, but we should fall back to attempting fstat when that fails.

Unrelatedly, there appears to be a minor issue in SharedArrays forgetting to close the fd_mem handle in this functions caller.

[xgdgsc]

Not sure whether the InteractiveUtils test failure is related. Would it be fine to fix it only for linux first like this and wait for macOS solution later? Guess not many would encounter this bug on macOS.

[DilumAluthge]

CI is all green on this.

@vtjnash Is this good to merge?

[mkitti]

As reported in #54203 this fix introduced another issue since filesize seeks which may not be appropriate for certain kinds of IO.