Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

Ignoring EPOLLET will generate more wakeups, but should work with most
applications.

Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

Large partitions set head=254, not 255.  Also be more accurate calculating
the exact point of overflow.

Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

More realistic.

Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

It's a superset of epoll_create(), so implement the latter in terms of the
former.

Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

Needed by boost.asio.

Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

std::streambuf<char> is defined as an "extern template", which means it is
instantiated in the library, not at the point-of-use, in order to reduce
object size and compilation time.

One of its member functions, std::basic_streambuf::seekpos(...) has a
parameter whose type depends on mbstate_t; this embeds the type name into
the mangled function name.

On Linux, mbstate_t is a typedef for __mbstate_t, so the latter type name
is embedded into the mangled function name.  On osv, there is no such typedef,
so mbstate_t is embedded.  This causes lookup to fail when loading a shared
object that uses streambufs.

Fix by defining mbstate_t in the same way as glibc.

Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

optind is a global variable that is expected to be zero-initialized by
getopt().  Running one object after another violates this, if both use
getopt().

Fix by saving the value of optind across calls.

Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

The 'define' form with '=' was introduced in GNU Make 3.82
and has the same meaning as without.

The default 'make' in Ubuntu 13.04 is stil on 3.81. Let's
drop '=' for better compatibility.

Symptoms:
 make[1]: Entering directory `~/src/osv/build/release'
      CC bsd/cddl/contrib/opensolaris/cmd/zpool/zpool_iter.o
    ../../bsd/cddl/contrib/opensolaris/cmd/zpool/zpool_iter.c:28:21:\
 fatal error: solaris.h: No such file or     directory
     #include <solaris.h>
                         ^
    compilation terminated.

Signed-off-by: Tomasz Grabiec <tgrabiec@cloudius-systems.com>

Much easier than constructing a traditional argc/argv.

Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

Unlike tls for the main module, which simply offsets fs:0, shared object
use an indirection scheme since each module gets its own tls base address,
which needs to be found (and initialized) via __get_tls_addr()

Implement __get_tls_addr() and the associated relocation and storage
management.

Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

Currently we search the entire vma list while looking for a range.  This
is less than optimal.

Define a vma::addr_compare function object that can be used as comparator
with boost::rbtree::equal_range(), and use that to search for both bounds
simultaneously.

Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

Signed-off-by: Tomasz Grabiec <tgrabiec@cloudius-systems.com>

Related changes previously made in java/build.xml
were reverted and moved to java/runjava/build.xml

Signed-off-by: Tomasz Grabiec <tgrabiec@cloudius-systems.com>

The system classloader should have classes which were
specified in the command line using -cp/-classpath/-jar.
Before this change only runjava.jar was included.

System class loader is queried explicitly by
java.util.logging framework to lookup classes
configred in logging properties file.

To solve this problem a custom system classloader
is set when starting JVM. It delegates to the application
classloader created when java program was started.

It also handles the case when a new java application
is started within the same JVM. Such application
should have its own view of system classes. However
we are able to install only one system classloader
per JVm. To solve this we start each java application
in a separate thread and use InheritableThreadLocal
to hold the application class loader which should be
used as system class loader view for the thread family.

Signed-off-by: Tomasz Grabiec <tgrabiec@cloudius-systems.com>

Building with ant requires no changes in the build scripts
when new classes are added. It is a standrd tool to build
java programs.

Signed-off-by: Tomasz Grabiec <tgrabiec@cloudius-systems.com>

Add a convenience function, osv::run(), to run an ELF executable.

The patch includes extensive Doxygen documentation for the new function,
also detailing the how osv::run() differs from the familiar Linux
functions like exec(2).

Signed-off-by: Nadav Har'El <nyh@cloudius-systems.com>

Fix mincore() to follow Linux ABI and return ENOMEM only if (part of
the) memory range is unmapped even if it's not resident.  This is needed
for mincore() to work with demand paged mmaps.

Signed-off-by: Pekka Enberg <penberg@cloudius-systems.com>

Fixes the following compilation error:

  ../../bsd/cddl/contrib/opensolaris/lib/libzpool/common/kernel.c: In function ‘rw_enter’:
  ../../bsd/cddl/contrib/opensolaris/lib/libzpool/common/kernel.c:205:3: error: pointer targets in passing argument 1 of ‘atomic_add_int’ differ in signedness [-Werror=pointer-sign]
     atomic_add_int(&rwlp->rw_count, 1);
     ^
  In file included from ../../bsd/sys/cddl/compat/opensolaris/sys/atomic.h:33:0,
                   from ../../bsd/cddl/contrib/opensolaris/lib/libzpool/common/sys/zfs_context.h:66,
                   from ../../bsd/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/spa.h:31,
                   from ../../bsd/cddl/contrib/opensolaris/lib/libzpool/common/kernel.c:32:
  ../../bsd/machine/atomic.h:285:1: note: expected ‘volatile u_int *’ but argument is of type ‘int *’
   ATOMIC_ASM(add,      int,   "addl %1,%0",  "ir",  v);
   ^

Signed-off-by: Pekka Enberg <penberg@cloudius-systems.com>

Fixes the following build breakage:

    CC bsd/cddl/contrib/opensolaris/lib/libzpool/common/kernel.o
  ../../bsd/cddl/contrib/opensolaris/lib/libzpool/common/kernel.c: In function ‘cv_timedwait’:
  ../../bsd/cddl/contrib/opensolaris/lib/libzpool/common/kernel.c:322:2: error: implicit declaration of function ‘gettimeofday’ [-Werror=implicit-function-declaration]
    if (gettimeofday(&tv, NULL) != 0)
  ^
  cc1: all warnings being treated as errors

Signed-off-by: Pekka Enberg <penberg@cloudius-systems.com>

I forgot to commit the change to dlfcn.cc in the previous patch. Sorry...

Signed-off-by: Nadav Har'El <nyh@cloudius-systems.com>

This is patch v2, incorporating most of the comments from the previous round.

Solves issue #47:

elf::program's API - add_object() and remove_object() was problematic in
two respects:

1. It did not do reference-counting on the loaded object, so if add_object()
   was done 5 times, a single remove_object() would unmap the object and
  its 4 other users will crash.

2. It is un-C++-like.

This patch replaces these two functions by a single function get_library():

  std::shared_ptr<elf::object>
   get_library(std::string lib, std::vector<std::string> extra_path = {});

get_library() returns a shared_ptr, which is reference counting and does
proper C++-style RAII. For example in the code:

	auto lib = elf::get_program()->get_library(path);
	auto main = lib->lookup<int (int, char**)>("main");
	int rc = main(argc, argv);

once lib out of scope, the reference count of the elf::object automatically
goes down, and if nobody else holds another shared-pointer to the same
library, the object is destroyed (causing the shared library to be unloaded).

This patch also documents, with Doxygen, all the functions it touches.

IMPORTANT NOTE: This reference count is completely unrelated to the issue
of concurrent use of dlopen()/dlclose()/dl_iterate_phdr(), which is still
buggy and we need to fix (I have a patch for that, but it's too long to fit
in the margin).

Signed-off-by: Nadav Har'El <nyh@cloudius-systems.com>

More mkfs infrastructure.

Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

root mounts accept both a device and a pool name, while non-root mounts
only accept the pool name (the device(s) already known while mounting the pool
during the root mount).

Adjust the mount protocol for this.

Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

A zpool has a zfs as its userspace interface; but the call to create it was
disabled as part of the porting effort.

Re-enable it so osv can create zpools.

Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

To build is_whole_disk() we need to pull in a lot of partition code.

Ignore it instead.

Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

MNTTAB points at /dev/zero, which doesn't exist, and is too long anyway.

Point it at the correct place.

Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

Signed-off-by: Avi Kivity <avi@cloudius-systems.com>