The friendly folks at #musl again:

21:17 < dalias> if you init the pointers both from the ctor and the "lazy
   init", then the assertion-failure/abort would happen during construction
   (either your ctor or somebody else's ctor calling your functions)
21:17 < dalias> so the program would exit "before it even gets started"
21:17 < dalias> and this is not so bad

21:18 < dalias> normally i consider it really bad practice for a library to
   abort the program
21:18 < dalias> but this happens during ctors, and there truely is no way to
   proceed
21:18 < dalias> if you've already replaced the program's library functions
   (open, etc.) with your copies, and then find that you can't actually supply
   working ones

The friendly folks at #musl say:

20:52 < dalias> i think it's good to have both that and the ctor
20:52 < Feh> This should increase robustness if it is in every function.
20:52 < dalias> because...
20:52 < dalias> several of these functions are required to be async-signal-safe
20:53 < dalias> so if the first call happened to be from a signal handler,
   you'd be in trouble if you didn't already have the ctor
20:54 < dalias> in theory another ctor in another lib or the app could run
   before yours and install and run a signal handler :-p but that's pretty
   unlikely and pathological
20:54 < dalias> so i think having both the ctor and the lazy-init keeps you
   safe in situations that will arise in the real world

(This fixes issue #3.)

(See Issue #12.)

This reverts commit 0a214bdb.

Apparently, building nocache on kfreebsd still fails, and anyway the linker
warns that posix_fadvise will always fail. So so wait until kfreebsd is
up-to-date with the POSIX standard. (See issue #12.)

* 'master' of git://github.com/onlyjob/nocache:
  minor man pages corrections (lintianisation): "hyphen-used-as-minus-sign"

* 'behavior_mods' of git://github.com/noushi/nocache:
  added -n option to nocache.global added BASEDIR to nocache.global (modifiable by sed) simplified LD_PRELOAD declaration

added BASEDIR to nocache.global (modifiable by sed)
simplified LD_PRELOAD declaration

(Otherwise `prove` would complain about “out of sequence” tests.)

* n-times-fadvise:
  Update man pages to reflect "-n" option
  Only do multi-fadvise if previous call was successful
  Add explanation of -n flag / NOCACHE_NR_FADVISE env variable
  Implement "-n <n>" flag for repeated posix_fadvise()

There was a trivial-to-solve merge conflict in nocache.c

Reported-by: Reda Noushi <reda@logzilla.pro>

> one remark: what should nocache do if a non-zero value is returned by posix_fadvise()?
> 1. is it okay to continue executing it and returning the last value?
> 2. or stop execution and return the last non-zero value of fadv()?

* noushi/behavior_mods:
  nocache retrieves max fds from actual process limits.

* 'master' of git://github.com/paulharris/nocache:
  Adjusted script so it will preload the .so from the script's directory

Thanks again, kocmo, for noticing.

Thanks to kocmo for the patch.

man pages

* test:
  Add "ls" test
  Make it easy to add new tests
  Add make target “test”
  Add very basic test

This issue came up in the real world because apparently
libselinux.so.1 does an fopen in a constructor function.
Big Thanks to Mike Shal (http://sourceware.org/ml/libc-help/2009-06/msg00000.html )

[Made some cosmetic changes -JP]

Fixes: GitHub issues #1 and #2
Signed-off-by: Julius Plenz <julius@plenz.com>

Calls to open() from within glibc cannot be intercepted
(see <http://cygwin.com/ml/libc-alpha/2006-04/msg00094.html>);
thus, include wrappers for fopen() and fclose() as well.

This should possibly be done for fdopen() and freopen(),
too, if there's need for it.

[Patch corrected and original description translated by Julius Plenz]
Signed-off-by: Julius Plenz <plenz@cis.fu-berlin.de>

When you invoke "tar cfv <tarball> <files>", then tar will close all
active file descriptors it has, forks, then opens the target tarball
as fd 1 (stdout). Then, it'll exec "gzip -", which *inherits* a stdout
which is a file.

Now we duplicate this stdout if it's a regular file so that we can
advise it later. (The original will be closed via the _exit() routine,
probably, which uses direct syscall that we cannot intercept.)

Thus, also shell redirection -- which makes heavy use of dup2() --  is handled
correctly:

    $ cachestats /tmp/foo
    open: No such file or directory
    $ nocache sh -c 'echo foo > /tmp/foo'
    $ cachestats /tmp/foo
    pages in cache: 0/1 (0.0%)  [filesize=0.0K, pagesize=4K]