First couple of frames always come from trace logging functions. I
think we should not show them because they do not add any information.

There was already a list of function names which should be filtered
out. This is not appropriate in this case because tracing goes through
a function named 'log' which is a pretty generic name and could as
well be an application symbol, so we cannot just filter out all
functions named 'log. This patch adds support for knocking off a
leading sequence of frames form the backtrace which is more accurate
and seems good enough.

Reviewed-by: Nadav Har'El <nyh@cloudius-systems.com>
Signed-off-by: Tomasz Grabiec <tgrabiec@cloudius-systems.com>
Signed-off-by: Pekka Enberg <penberg@cloudius-systems.com>

Sampler will need to set and later restore value of this option.

Reviewed-by: Nadav Har'El <nyh@cloudius-systems.com>
Signed-off-by: Tomasz Grabiec <tgrabiec@cloudius-systems.com>
Signed-off-by: Pekka Enberg <penberg@cloudius-systems.com>

Tracepoints can be enbaled not only via enable_tracepoint(std::string)
but via tracepoint_base::enable() also. This change also makes the
initialization thread-safe as it may be called from aribtrary thread.

Reviewed-by: Nadav Har'El <nyh@cloudius-systems.com>
Signed-off-by: Tomasz Grabiec <tgrabiec@cloudius-systems.com>
Signed-off-by: Pekka Enberg <penberg@cloudius-systems.com>

Some functionality from the zfs and zpool commands require some functions
contained in zfs_prop.c and zpool_prop.c. This patch solves this problem
by adding the files as part of the libzfs compilation process.
Currently, OSv fall into the 'failed to lookup symbol' scenario when using
certain parameters from the aforementioned commands, e.g.:
scripts/run.py -e 'zfs.so list osv'

_OPENSOLARIS_SYS_UIO_H_ was added to CFLAGS to fix a confliction where
'struct uio' was being defined twice (Glauber Costa).

Reviewed-by: Glauber Costa <glommer@cloudius-systems.com>
Signed-off-by: Raphael S. Carvalho <raphaelsc@cloudius-systems.com>
Signed-off-by: Pekka Enberg <penberg@cloudius-systems.com>

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

Resolve and demangle symbol name of RIP, show it.

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

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

lookup_name_demangled() lookups a symbol name, demangle it, then
snprintf onto preallocated buffer.

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

Signed-off-by: KANATSU Minoru <icc.pot.tyew272@gmail.com>
Signed-off-by: Pekka Enberg <penberg@cloudius-systems.com>

Signed-off-by: KANATSU Minoru <icc.pot.tyew272@gmail.com>
Signed-off-by: Pekka Enberg <penberg@cloudius-systems.com>

add symbol to libc for port CRuby

Signed-off-by: KANATSU Minoru <icc.pot.tyew272@gmail.com>
Signed-off-by: Pekka Enberg <penberg@cloudius-systems.com>

Signed-off-by: KANATSU Minoru <icc.pot.tyew272@gmail.com>
Signed-off-by: Pekka Enberg <penberg@cloudius-systems.com>

an effect of commit 9bbbe9dc is that no output is possible
before prio 'console' initializers have been run.
This change allows to have at least one API available
really early (from boot code and premain).
Document the requirements for the early console class
regarding the write() method.

Signed-off-by: Claudio Fontana <claudio.fontana@huawei.com>
Signed-off-by: Pekka Enberg <penberg@cloudius-systems.com>

While running one of the redis benchmarks, I saw around 23k calls to
malloc_large.  Among those, ~10 - 11k were 2-page sized. I managed to track it
down to the creation of net channels. The problem here is that the net channel
structure is slightly larger than half a page - the maximum size for small
object pools. That will throw all allocations into malloc_large. Besides being
slow, it also wastes a page for every net channel created, since malloc_large
will include an extra page in the beginning of each allocation.

This patch fixes this by overloading the operators new and delete for the
netchannel structure so that we use the more efficient and less wasteful
alloc_page.

Reviewed-by: Nadav Har'El <nyh@cloudius-systems.com>
Signed-off-by: Glauber Costa <glommer@cloudius-systems.com>
Signed-off-by: Pekka Enberg <penberg@cloudius-systems.com>

This change aims to reduce the likelihood of running into issue
merged stream may not be. If samples cross CPUs and the CPU of the
second sample has its earliest sample after the first sample, we
ignore that pair as the second sample has a high chance of belonging
to other pair.

This change does not solve the problem completly because there may be
a situation when sample which was recorded on another CPU was
discarded and the next sample will be again on the first CPU.

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

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

This command shows a duration profile like 'prof-wait' or 'prof-lock'
but is more generic, it can work with any timed traces.

'trace prof-wait' is equivalent of 'trace prof-timed -t sched_wait'
'trace prof-lock' is equivalent of 'trace prof-timed -t mutex_lock_wait'

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

The algorithm was duplicated between 'prof-wait','prof-lock' and 'list
timed','summary --timed'. This change unifies all paths.

As a side effect, mutex_lock_wait/mutex_lock_wake pair is now
recognized as timed sample and appears in 'summary --timed' and
'trace list-timed'

This refactoring will also make it easy to declare timed samples which
are correlated by data rather than calling thread.

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

Broken by commit 1e53a52f.

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

Export the shrinker interface to C users.

Signed-off-by: Glauber Costa <glommer@cloudius-systems.com>
Reviewed-by: Raphael S. Carvalho <raphaelsc@cloudius-systems.com>
Signed-off-by: Pekka Enberg <penberg@cloudius-systems.com>

Due to a global variable in image type definition code
small instance directive becomes sticky and applies automatically
to all further images of the batch.

Fixed by making the variable local.

Signed-off-by: Dmitry Fleytman <dmitry@daynix.com>
Signed-off-by: Pekka Enberg <penberg@cloudius-systems.com>

implement a simple interrupt table with descriptors
which record interrupt IDs, handlers and interrupt type.

Signed-off-by: Claudio Fontana <claudio.fontana@huawei.com>
Signed-off-by: Pekka Enberg <penberg@cloudius-systems.com>

Fixes #282

Signed-off-by: Jaspal Singh Dhillon <jaspal.iiith@gmail.com>
Signed-off-by: Pekka Enberg <penberg@cloudius-systems.com>

Traceback (most recent call last):
  File "/home/tgrabiec/src/osv/scripts/loader.py", line 1165, in invoke
    show_leak()
  File "/home/tgrabiec/src/osv/scripts/loader.py", line 1104, in show_leak
    si = syminfo(f)
  File "/home/tgrabiec/src/osv/scripts/loader.py", line 117, in syminfo
    return symbol_formatter(syminfo_resolver(addr))
TypeError: object() takes no parameters
Error occurred in Python command: object() takes no parameters

Reported-by: Gleb Natapov <gleb@cloudius-systems.com>
Signed-off-by: Tomasz Grabiec <tgrabiec@cloudius-systems.com>
Signed-off-by: Pekka Enberg <penberg@cloudius-systems.com>

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

pthread_mutex uses lazy_indirect object to allocate actual object
on a first use. The lazily allocated object is not destroyed unless
pthread_mutex_destroy is called.

Reviewed-by: Nadav Har'El <nyh@cloudius-systems.com>
Signed-off-by: Gleb Natapov <gleb@cloudius-systems.com>
Signed-off-by: Pekka Enberg <penberg@cloudius-systems.com>

That formatter is used for ad-hoc printing of samples, eg.
in error messages.

This fixes the following error message:

Traceback (most recent call last):
  File "scripts/trace.py", line 576, in <module>
    args.func(args)
  File "scripts/trace.py", line 244, in prof_lock
    show_profile(args, get_profile)
  File "scripts/trace.py", line 208, in show_profile
    max_levels=args.max_levels)
  File "/home/tgrabiec/src/osv/scripts/osv/prof.py", line 170, in print_profile
    for sample in samples:
  File "/home/tgrabiec/src/osv/scripts/osv/prof.py", line 108, in get_duration_profile
    raise Exception('Double entry:\n%s\n%s\n' % (str(old), str(trace)))
  File "/home/tgrabiec/src/osv/scripts/osv/trace.py", line 125, in __str__
    return self.format()
  File "/home/tgrabiec/src/osv/scripts/osv/trace.py", line 122, in format
    bt_formatter(self.backtrace))
  File "/home/tgrabiec/src/osv/scripts/osv/trace.py", line 32, in __call__
    frames = list(debug.resolve_all(self.resolver, (x - 1 for x in backtrace if x)))
  File "/home/tgrabiec/src/osv/scripts/osv/debug.py", line 107, in resolve_all
    return itertools.chain.from_iterable(map(resolver, raw_addresses))
TypeError: __init__() takes exactly 1 argument (2 given)

When this should actually be printed:

Traceback (most recent call last):
  File "scripts/trace.py", line 576, in <module>
    args.func(args)
  File "scripts/trace.py", line 244, in prof_lock
    show_profile(args, get_profile)
  File "scripts/trace.py", line 208, in show_profile
    max_levels=args.max_levels)
  File "/data/tgrabiec/osv/scripts/osv/prof.py", line 170, in print_profile
    for sample in samples:
  File "/data/tgrabiec/osv/scripts/osv/prof.py", line 108, in get_duration_profile
    raise Exception('Double entry:\n%s\n%s\n' % (str(old), str(trace)))
Exception: Double entry:
0xffffc00017668010                  3 1399542568.505835772 mutex_lock_wait (...)
0xffffc00017668010                  1 1399542574.668642044 mutex_lock_wait (...)

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

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

This patch fixes the case of silently hanging of OSv when >64 cpus are
provided.

Signed-off-by: Jaspal Singh Dhillon <jaspal.iiith@gmail.com>
Signed-off-by: Pekka Enberg <penberg@cloudius-systems.com>

OSv is currently limited to 64 vCPUs, because we use a 64-bit bitmask for
wakeups (see max_cpus in sched.cc). Having exactly 64 CPUs *should* work,
but unfortunately didn't because of a bug:

cpu_set::operator++ first incremented the index, and then called advance()
to find the following one-bit. We had a bug when the index was 63: we then
expect operator++ to return 64 (end(), signaling the end of the iteration),
but what happened was that after it incremented the index to 64, advance()
wrongly handled the case idx=64 (1<<64 returns 1, unexpectedly) and moved
it back to idx=63.

The patch fixes operator++ to not call advance when idx=64 is reached,
so now it works correctly also for idx=63, and booting with 64 CPUs now
works.

Fixes #234.

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

add include <osv/prio.hh>, this time at the beginning of the file.

Signed-off-by: Claudio Fontana <claudio.fontana@huawei.com>
Signed-off-by: Pekka Enberg <penberg@cloudius-systems.com>

Implement three of the Linux functions we're still missing (see issue #77
for a list of more things to do).

These are pretty easy to add, because we just need to correctly call the
lower layer functions, sys_*, which already exist.

This patch also adds tests for the three functions to tst-readdir.cc.

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

Our readdir_r() and friends were implemented in libc/dir.cc, calling
a function "ll_readdir()" from fs/vfs/main.cc. The name of this function
was confusing ("ll_" seemed to imply lock-less, which it isn't), and the way
it was "extern"ed was ugly, as a comment even admitted.

Moreover, when we want to implement more missing functions like rewinddir()
(see issue #77), we need to write more of these silly "ll_*" wrapper functions.

Instead, let's just move the few dir.cc functions into main.cc. The latter
already defines most of our filesystem system calls, so feels the right
place anyway. This change allows us to get rid of the silly ll_readdir()
layer, and makes implementing rewinddir() trivial (see next patch).

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

remove a spurious include <osv/prio.hh> which seems to be
a merge artifact.

Signed-off-by: Claudio Fontana <claudio.fontana@huawei.com>
Signed-off-by: Pekka Enberg <penberg@cloudius-systems.com>

This patch changes the definition of __assert_fail() in api/assert.h which
would allow it and other header files which include it (such as debug.hh) to
be used in mgmt submodules. Fixes conflict with declaration of
__assert_fail() in external/x64/glibc.bin/usr/include/assert.h

Signed-off-by: Jaspal Singh Dhillon <jaspal.iiith@gmail.com>
Signed-off-by: Pekka Enberg <penberg@cloudius-systems.com>

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

This change moves syncache_head construction to its constructor so that the
timer can be nicely embedded in the structure.

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

Now we enabled both VGA and serial, we nolonger need to add --vga cmdline for VMW/VBox.

Reviewed-by: Asias He <asias@cloudius-systems.com>
Signed-off-by: Takuya ASADA <syuu@cloudius-systems.com>
Signed-off-by: Pekka Enberg <penberg@cloudius-systems.com>

According to the manual page, the nsec field from both it_value
and it_interval must be checked, if wrong, return EINVAL.

Reviewed-by: Nadav Har'El <nyh@cloudius-systems.com>
Signed-off-by: Raphael S. Carvalho <raphaelsc@cloudius-systems.com>
Signed-off-by: Pekka Enberg <penberg@cloudius-systems.com>

The return values of LOCK() and UNLOCK() are never used. Besides, in the
current Musl version these macros do not have return values at all.

Reviewed-by: Nadav Har'El <nyh@cloudius-systems.com>
Signed-off-by: Pawel Dziepak <pdziepak@quarnos.org>
Signed-off-by: Pekka Enberg <penberg@cloudius-systems.com>