Needed by boost.asio.

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

Signed-off-by: Avi Kivity <avi@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>

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

Our default signal handler (for, e.g., ^C) used abort(), which results
in a hung VM, and is not very useful. Change this to osv::poweroff(),
which gives more useful behavior. Now one can, for example, run a
test with run.py and interrupt it in the middle with ^C causing the
VM to power off and run.py to complete.

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

Noticed a minor bug in the OSv dlfcn implementation.
According to the linux man pages, dladdr returns non-zero on success
and zero on error -> http://linux.die.net/man/3/dladdr



Signed-off-by: Dinakar Guniguntala <dino@linux.vnet.ibm.com>
Signed-off-by: Pekka Enberg <penberg@cloudius-systems.com>

We had a different Linux version compiled into uname() (3.7.0) than
we had compiled into other code (via <linux/version.h>).

This patch makes them both pretend to be 3.7.0 - arbitrarily chosen
as the current Linux version at the time OSv was created. The new code
verifies with a static assertion that both files contain the same version,
so if they diverge, uname.c will fail compilation.

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

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

Signed-off-by: Benoit Canet <benoit@irqsave.net>
Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

Also convert it's header and change it's caller includes.

This patch is done so the libc resolver will be able to retrieve the dns from
OSv dns API.

Signed-off-by: Benoit Canet <benoit@irqsave.net>
Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

Debug build was broken by it since boost::asio was used.

Signed-off-by: Benoit Canet <benoit@irqsave.net>
Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

boost's asio.hpp module init() function was calling pthread_key_create which
tried to manipulate some libc/pthread.cc unitialized objects.

pthread_key_create then returned -ENOMEM so an exception was thrown by boost
causing the loader's premain() function to fail and the boot to loop.

This patch lower the priority of the initialization of the libc/pthread.cc
objects so they are ready when premain is calling the boost module init()
function.

Signed-off-by: Benoit Canet <benoit@irqsave.net>
Signed-off-by: Avi Kivity <avi@cloudius-systems.com>

Used somewhere within libcurses, so provide an implementation.

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

This patch adds a few simple terminal-related libc functions that ncurses
requires (needed to run curses-based applications on OSv).

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

Exit tracepoints are useful to record errors, and also for timing the
mmap operation.

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

This patch adds support for Linux's alarm(2) system call.

alarm() is needed in some ancient Unix software, like netperf :(

This implementation is fully compatible with Linux's alarm(), but please
note that what alarm() does when the alarm times out is a kill(SIGALRM) -
so all the caveats mentioned in the previous patch regarding kill(),
will also apply here.

Alarm() is implemented by running one "alarm thread", started on the
first alarm() call. This alarm-thread waits for the alarm to expire,
or for instructions to change the alarm. When an alarm expires the
alarm thread does a kill(0,SIGALRM).

tst-kill.cc (from the previous patch) includes a test to see that
alarm() really sends a SIGALRM signal on time, and also that we can
cancel a pending alarm and not receive a signal.

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

This patch adds support for the Linux kill(2) function.
The next patch will add alarm(2) support, which uses kill(2).

To be honest, we sort-of implement kill(). This implementation is
compatible with the API, but the semantics are somewhat different:
While in Linux kill() causes the signal handler to run on one of the
existing threads, in this implementation, the signal handler is run in a
*new* thread.

Implementing the exact Linux semantics in OSv would require tracking when
OSv runs kernel code (i.e., code in the main executable, not a shared
object) so we can delay running the signal handler until returning to the
user code. Moreover, we'll need to be able to interrupt sleeping kernel
code. This is complicated and adds overhead even if signals aren't used
(and they aren't used in most modern code).

I expect that this code will be "good enough" in many use cases.
This code will *not* be good in enough in programs that expect one of the
following:

1. A program that by using Posix Thread's "signal masks" tried to ensure
   that the signal is delivered to one specific thread, and not to an
   arbitrary thread.

2. A program that used kill() or alarm() not intending to run a signal
   handler, but rather intending to interrupt a sleeping system call
   like sleep() or read(). Our kill() does not interrupt sleeping OSv
   function calls, which will continue to sleep on the thread they run
   on.

The support in this patch (and see next patch, for alarm()) is good
enough for netperf's use of alarm().

P.S. kill() can be used only to send a signal to the current process, the
only process we have in OSv (you can also use pid=0 and pid=-1 to achieve
the same results).

This patch also adds a test for kill() and alarm(). The alarm() test
will fail until the next patch :-)

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

signal() verified that its argument is not >nsigs, but it should be >=nsigs
because nsigs itself is already illegal (0...nsigs-1 are valid signals).

Discovered by Coverity.

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

Add to libc/user.cc stub setresgid(), setresgid(), which as usual in
OSv only allow user and group 0. These functions are used by Rogue ;-)

Also move getuid() and friends from runtime.cc to libc/user.cc, where
they feel more at home.

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

We used to return the POLLRDHUP event when polling a write-side of a pipe
whose reader is closed, but Linux returns POLLERR|POLLOUT, so let's do
the same.

Linux's behavior can be explained in that when the reader is closed, a write
will return immediately (and hence POLLOUT) but moreover, will return with
an error (EPIPE). And since a read will also cause an error (EBADF when
reading on the write side of the pipe), a POLLERR makes sense and says
that any operation on this fd will now result in an error.

This patch also adds a test for this case.

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

The code in pipe_buffer.cc had the meaning of POLLHUP and POLLRDHUP
reversed, and as a result POLLHUP wasn't correctly returned when polling
a read-end of the pipe whose write-end is closed.

This fixes the mixup, and as a result the test added to tst-pipe.cc in
the previous commit, and also another one added in this patch, all pass.

We are still incompatible with Linux with what we do regarding POLLRDHUP.
This will be a separate patch.

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

Add Cloudius copyright statements in libc/.

One unfortunate problem we have in libc/ is that most of the code there was copied
from Musl, and Musl is copyrighted and licensed, but individual files do not carry
any statements - Rich Felker explains in Musl's COPYRIGHT file: "The decision to
exclude such comments is intentional, as it should be possible to carry around the
complete source code on tiny storage media."

In the future, we should find all the Musl files and add to each one the line
Copyright (C) 2005-2013 Rich Felker. But we'll need to do it carefully so as not
to wrongly attribute non-Musl code to Musl. In the meantime, if we just have
documentation/LICENSE-musl, and don't claim our own copyright on the musl files,
I think we're playing fair.

the musl implementation immidiately issues a DNS query, instead of looking
first at the /etc/hosts file, this patch fixes it.

As a side effect, the long boot duration is shortened by 3-4 seconds.

If "-javaagent" option is enabled, the JVM page faults:

  (gdb) bt
  #0  0x000000000033d432 in halt_no_interrupts () at ../../arch/x64/processor.hh:232
  #1  osv::halt () at ../../core/power.cc:20
  #2  0x0000000000214d82 in abort (msg=msg@entry=0x57ea90 "page fault outside application") at ../../runtime.cc:98
  #3  0x00000000002fd4d8 in page_fault (ef=0xffffc0003ffe6008) at ../../core/mmu.cc:943
  #4  <signal handler called>
  #5  0x0000000000373169 in __stpncpy (d=d@entry=0x2000001fe7f0 "dlsym: symbol JVM_begin_signal_setting not found", s=0x0, n=1023)
      at ../../libc/string/stpncpy.c:19
  #6  0x0000000000373ad1 in strncpy (d=0x2000001fe7f0 "dlsym: symbol JVM_begin_signal_setting not found", s=<optimized out>,
      n=<optimized out>) at ../../libc/string/strncpy.c:7
  #7  0x0000100000b0ab67 in os::dll_load (filename=filename@entry=0x2000001febf0 "libinstrument.so",
      ebuf=ebuf@entry=0x2000001fe7f0 "dlsym: symbol JVM_begin_signal_setting not found", ebuflen=ebuflen@entry=1024)
      at /usr/src/debug/java-1.7.0-openjdk-1.7.0.25-2.3.12.3.fc19.x86_64/openjdk/hotspot/src/os/linux/vm/os_linux.cpp:1841
  #8  0x0000100000c247eb in lookup_on_load (agent=agent@entry=0xffffc0003976ccc0,
      on_load_symbols=on_load_symbols@entry=0x2000001ffd40, num_symbol_entries=1)
      at /usr/src/debug/java-1.7.0-openjdk-1.7.0.25-2.3.12.3.fc19.x86_64/openjdk/hotspot/src/share/vm/runtime/thread.cpp:3585
  #9  0x0000100000c2a64f in lookup_agent_on_load (agent=0xffffc0003976ccc0)
      at /usr/src/debug/java-1.7.0-openjdk-1.7.0.25-2.3.12.3.fc19.x86_64/openjdk/hotspot/src/share/vm/runtime/thread.cpp:3617
  #10 create_vm_init_agents ()
      at /usr/src/debug/java-1.7.0-openjdk-1.7.0.25-2.3.12.3.fc19.x86_64/openjdk/hotspot/src/share/vm/runtime/thread.cpp:3656
  #11 Threads::create_vm (args=<optimized out>, canTryAgain=canTryAgain@entry=0x2000001ffdb0)
      at /usr/src/debug/java-1.7.0-openjdk-1.7.0.25-2.3.12.3.fc19.x86_64/openjdk/hotspot/src/share/vm/runtime/thread.cpp:3177
  #12 0x000010000094d7b0 in JNI_CreateJavaVM (vm=0x2000001ffe58, penv=0x2000001ffe60, args=<optimized out>)
      at /usr/src/debug/java-1.7.0-openjdk-1.7.0.25-2.3.12.3.fc19.x86_64/openjdk/hotspot/src/share/vm/prims/jni.cpp:5127
  #13 0x0000100000007b3b in main (argc=<optimized out>, argv=0xffffc0003fff2008) at ../../java/java.cc:73
  #14 0x0000000000208ac8 in run_main (prog=<optimized out>, args=args@entry=0xffffc0003fe9bfa0) at ../../loader.cc:196
  #15 0x0000000000208c68 in do_main_thread (_args=0xffffc0003fe9bfa0) at ../../loader.cc:217
  #16 0x0000000000376d36 in operator() (__closure=0xffffc0003b5e2a00) at ../../libc/pthread.cc:59
  #17 std::_Function_handler<void(), pthread_private::pthread::pthread(void* (*)(void*), void*, sigset_t, const pthread_private::thread_attr*)::__lambda0>::_M_invoke(const std::_Any_data &) (__functor=...) at ../../external/gcc.bin/usr/include/c++/4.8.1/functional:2071
  #18 0x000000000032ed6b in main (this=0xffffc0003a807010) at ../../core/sched.cc:536
  #19 sched::thread_main_c (t=0xffffc0003a807010) at ../../arch/x64/arch-switch.hh:133
  #20 0x000000000031873e in thread_main () at ../../arch/x64/entry.S:101
  Backtrace stopped: frame did not save the PC

This is caused by the JVM expecting dlerror() to return an error string
if dlopen() fails. Fix that up.

Use the new mmu::msync() function to make sure file-backed memory maps
are written out to disk in munmap().

This adds simple msync() implementation for file-backed memory maps. It
uses the newly added 'file_vma' data structure to write out and fsync
the msync'd region as suggested by Avi Kivity.

Add a new 'file_vma' class that extends 'vma'. This is needed to keep
track of fileref and offset for file-backed VMAs for msync().

Fixes the following build brekage caused by commit a7d6b269 ("mman: Use
libc_error() in mprotect()"):

../../libc/mman.cc: In function ‘int mprotect(void*, size_t, int)’:
../../libc/mman.cc:36:33: error: ‘libc_error’ was not declared in this scope
         return libc_error(EINVAL);
                                 ^
../../libc/mman.cc:41:33: error: ‘libc_error’ was not declared in this scope
         return libc_error(ENOMEM);
                                 ^
  CXX libc/pipe_buffer.o
  CXX libc/pipe.o
make[1]: *** [libc/mman.o] Error 1

sched.hh included elf.hh, just so it can refer to the elf::tls_data
type. But now that we have rcu.hh which includes sched.hh and therefore
elf.hh, if we wish to use rcu in elf.hh (we'll do this in a later patch),
we have an include loop mess.

So better not include elf.hh from sched.hh, and just declare the one
struct we need.

After sched.hh no longer includes elf.hh and the dozen includes that
it further included, we need to add missing includes to some of the
code that included sched.hh and relied on its implict includes.

The dependency on sse4.1 crashes on older cpus, use the generic musl
implementation.

libstdc++ switched to relying on this symbol for the gthread machinery,
define it so it works.

Fixes std::thread used in mpsc queue tests.

Needed for statically linking the entirety of libgcc_eh.a.

This patch fixes the exception handling bug seen in tst-except.so.

The callback given dl_iterate_phdr (such as _Unwind_IteratePhdrCallback
in libgcc_eh.a used to implement exceptions) may decide to cache previous
information we gave it, as long as the "adds" and "subs" fields are
unchanged.

The bug was that we passed to the callback a on-stack *copy* of the
obj->_phdrs vector, and if the callback saved pointers to that in its
cache, they became invalid on the next call. We need the pointers to
remain valid as long as adds/subs do not change. So we need to pass
the actual obj->_phdrs (which doesn't change after the object's load),
NOT a copy.

Note there's a locking issue remaining here - if someone dlclose()s
an object while the callback is running (and already checked adds/subs)
it can use a stale pointer. This should be fixed separately, probably
by using reference counting on objects.

The callback function passed to dl_iterate_phdr, such as
_UnWind_IteratePhdrCallback (used in libgcc_eh.a to implement exceptions),
may want to cache previous lookups, and wants to know when the list of
iterated modules hasn't changed since the last call to dl_iterate_phdr.
For this, dl_iterate_phdr() is supposed to fill two fields, dlpi_adds
and dlpi_subs, counting the number of times objects were loaded or
unloaded from the program. If both dlpi_subs and dlpi_adds are unchanged,
the callback is guaranteed the list of objects is unchanged.

In the existing code, we forgot to set these two fields, so they got
random values which caused the exception unwinding code to sometimes
cache, and sometime not cache, depending on the phase of the moon.

This patch adds the counting of the correct "subs" and "adds" counters,
and after it exception unwinding will always use its cache (as long
as the list of objects doesn't change).

Note that this does NOT fix the crash in tst-except.so. That is a bug
which appears when caching is enabled (which before this patch happend
randomly), and will be fixed by the next patch.

dlerror is needed when an application is started on the JVM with the
"-javaagent" option defined.

We want to use backtrace() (and therefore dl_allocate_phdr(), and therefore
program::with_modules()) in interrupt disabled context; change with_modules()
to avoid any allocations.