use license macro in the spec file

Change-Id: I29eb21b7e528d92751ead52962f172cb1d757d1d

ws-server: restrict returned Sec-Websocket-Protocol to the chosen name only

https://libwebsockets.org/pipermail/libwebsockets/2016-November/002948.html

Updated to fix a problem with no protocol

https://github.com/warmcat/libwebsockets/issues/705

Change-Id: I06e9f9d5849e66291b246316008c6e2c5981a372

Remove win32port

Change-Id: Iafa286c46019f90eaef2b3260d6f571bdbc8ee1c

Add packaging code and LICENSE

Change-Id: If938a6255de53076acc584d88b16e7706ed49454
Signed-off-by: Dmitry Kovalenko <d.kovalenko@samsung.com>

client: avoid possible NULL deref on error path

https://github.com/warmcat/libwebsockets/issues/672

ws-server: restrict returned Sec-Websocket-Protocol to the chosen name only

https://libwebsockets.org/pipermail/libwebsockets/2016-November/002948.html

client: protect againt losing ah by lws_client_connect_2

Remove the cleanup functions with OpenSSL 1.1.

travis: explicitly point to openssl on osx

One day this started failing at CMake autofind. This forces it to look at the right place.

docs: explain lws_write handling of truncated sends better

windows: fix snprintf vs _snprintf

appveyor: get nsis from libwebsockets.org

appveyor: get windows openssl from libwebsockets.org

travis: update to trusty for CMake version

Added option to build the static library with PIC

CMakeLists: move config file generation to after LWS_HAVE_OPENSSL_ECDH_H

This is already fixed in v2.0 and master

v1.7.9

SONAME to 7.1 because of lws_snprintf()

lws_snprintf

Thanks to Fabrice Gilot for reporting the problem that led to uncovering this.

Due to a misunderstanding of the return value of snprintf (it is not truncated according
to the max size passed in) in places relying on snprintf to truncate the length
overflows are possible.

This patch wraps snprintf with a new lws_snprintf() which does truncate its length to allow
the buffer limiting scheme to work properly.

All users should update with these fixes.

In 1.7.x, there's no affected code in the library itself, just a couple on instances in the
test app code.

fix for https connection code

client CONNECTION_ERROR also allow in LWSS_CLIENT_UNCONNECTED

Signed-off-by: Andy Green <andy@warmcat.com>

Fix leak caused by undestroyed pthread mutex

v1.7.8

Signed-off-by: Andy Green <andy@warmcat.com>

check oom on lws_malloc

Signed-off-by: Andy Green <andy@warmcat.com>

appveyor openssl 1.0.2h

Signed-off-by: Andy Green <andy@warmcat.com>

fix %3d handling in path part and add attack.sh

https://github.com/warmcat/libwebsockets/issues/518

Signed-off-by: Andy Green <andy@warmcat.com>

recv revert treating zero as hangup

While checking with ab, I found
commit 30cdb3ac8f24a4c289a3b4e8a41d6c5f2e6ec959
Author: Justin Chen <justinchen00@github.invalid.com>
Date:   Thu Apr 14 21:40:53 2016 +0800

    recv treat zero return as error

    https://github.com/warmcat/libwebsockets/issues/475

turned ab performance to crap, reverting it made everything fast again.

recv manpage says there is three ways to get zero returned

1)       When a stream socket peer has performed an orderly shutdown, the return value will be 0  (the  traditional  "end-of-file"
       return).

2)       Datagram  sockets  in  various  domains  (e.g., the UNIX and Internet domains) permit zero-length datagrams.  When such a
       datagram is received, the return value is 0.

3)       The value 0 may also be returned if the requested number of bytes to receive from a stream socket was 0.

we can't just assume it means the peer shut down.

If the peer shut down, then the event loop should get an event on the socket like POLLHUP and deal with it that way.

So the patch mentioned above is simply reverted here.

Signed-off-by: Andy Green <andy@warmcat.com>

client provide user_space on LWS_CALLBACK_CLIENT_APPEND_HANDSHAKE_HEADER

https://github.com/warmcat/libwebsockets/issues/509

Signed-off-by: Andy Green <andy@warmcat.com>

v1.7.7

Signed-off-by: Andy Green <andy@warmcat.com>

client fix reaction to tls failure

https://github.com/warmcat/libwebsockets/issues/508

Signed-off-by: Andy Green <andy@warmcat.com>

Add warn_unused_result check, attribute only supported by GCC 3.4 or later

warn_unused_result was introduced in GCC version 3.4.

Change-Id: I6c2cc938d2b868ddfe0889cc41d7fa9d70e1b907

Android needs sys/resource.h

v1.7.6

Signed-off-by: Andy Green <andy@warmcat.com>

cruft remove sigusr2 handling

Signed-off-by: Andy Green <andy@warmcat.com>

client account for retries

Signed-off-by: Andy Green <andy@warmcat.com>

Update libwebsockets.h

Fix for FreeBSD

recv treat zero return as error

https://github.com/warmcat/libwebsockets/issues/475

windows snprintf is _snprintf

https://github.com/warmcat/libwebsockets/issues/411#issuecomment-207290650

android fix rlimit

https://github.com/warmcat/libwebsockets/issues/488

urldecode forbid malformed

And update attack.sh to confirm the new test cases

Signed-off-by: Andy Green <andy@warmcat.com>

libuv add idle processing to force service where needed

https://github.com/warmcat/libwebsockets/issues/485

Signed-off-by: Andy Green <andy@warmcat.com>

lws_service_adjust_timeout optimize

Make it exit quicker if something is pending

Signed-off-by: Andy Green <andy@warmcat.com>

revert 622d9f2 ssl pending handcrank

https://github.com/warmcat/libwebsockets/issues/483

Signed-off-by: Andy Green <andy@warmcat.com>

win32 libuv build notes

win32 needs strange strftime args

libuv win32 fixes 2

https://github.com/warmcat/libwebsockets/issues/411#issuecomment-204284368

libuv win32 fixes

Signed-off-by: Andy Green <andy@warmcat.com>

uri processing reject paths not starting with slash

https://github.com/warmcat/libwebsockets/issues/481

Return 403 Forbidden if we don't end up with a uri path starting with /

Test server already did this, but this makes it built into the
library.

Signed-off-by: Andy Green <andy@warmcat.com>

v1.7.5

Signed-off-by: Andy Green <andy@warmcat.com>

test server align rxbuf with permessage deflate rx buf size

Add a test html button that will send 9KB of junk to confirm it

https://github.com/warmcat/libwebsockets/issues/480

permessage-deflate now checks the protocol rx buffer size for being
>=128, if not, permessage-deflate is disabled on that connection.

If it is >=128 but less than the zlib decompress buffer size, the
zlib decompress buffer size for that connection is reduced to the
nearest power of two of the protocol rx buf size.

To test this, dumb_increment is left violating the >= 128 rx buffer
size and permessage-deflte can be seen to be disabled on his
connections in the test html.

Signed-off-by: Andy Green <andy@warmcat.com>

Fix build with musl libc

Fix building libwebsockets with the musl C libary.

<sys/cdefs.h> is an internal glibc header and should be avoided in user code.

__P() was used for compatibility with some old K&R C compilers, when there were
no prototypes (which were introduced to C with C89). As supporting legacy
non-ANSI compilers is nowadays not necessary anymore get rid of the unnecessary
function prototype using __P().

revert cmake remove targets from install path

Signed-off-by: Andy Green <andy@warmcat.com>

http2 build with alpn capable ssl no debug

Signed-off-by: Andy Green <andy@warmcat.com>

libuv: handle signals only if requested

Signed-off-by: Denis Osvald <denis.osvald@sartura.hr>

v1.7.4

Signed-off-by: Andy Green <andy@warmcat.com>

b64decode correct decode of some strings

https://github.com/warmcat/libwebsockets/issues/467

Signed-off-by: Andy Green <andy@warmcat.com>

Optimize payload exhaustion

https://github.com/warmcat/libwebsockets/pull/462

AG refactor and do loop unrolling

appveyor update 1.0.2g

Signed-off-by: Andy Green <andy.green@linaro.org>

clang fixes 1

https://github.com/warmcat/libwebsockets/issues/461

Signed-off-by: Andy Green <andy.green@linaro.org>

client perform WSI_CREATE callback

ssl ecdh adapt if missing ecdh.h include

https://github.com/warmcat/libwebsockets/issues/457

client ext hdr skip if no arg

https://github.com/warmcat/libwebsockets/issues/453

Signed-off-by: Andy Green <andy.green@linaro.org>

release checklist specfile install soname

Signed-off-by: Andy Green <andy.green@linaro.org>

rpm specfile so install list bump

Signed-off-by: Andy Green <andy.green@linaro.org>

libev set foreign loop properly

After gaby64

https://github.com/warmcat/libwebsockets/issues/455

Signed-off-by: Andy Green <andy.green@linaro.org>

windows deal with no snprintf

Signed-off-by: Andy Green <andy.green@linaro.org>

libuv.c: Service fd with LWS_POLLHUP on poll errors

wince minor adaptations

https://github.com/warmcat/libwebsockets/issues/444

v1.7.3

Signed-off-by: Andy Green <andy.green@linaro.org>

libuv when in use skip shutdown close phase

Signed-off-by: Andy Green <andy.green@linaro.org>

windows listen for pollhup

After Ondraco

https://github.com/warmcat/libwebsockets/issues/441

Signed-off-by: Andy Green <andy.green@linaro.org>

test server libuv support status protocol

Signed-off-by: Andy Green <andy.green@linaro.org>

libuv create 1Hz background timeout check

Signed-off-by: Andy Green <andy.green@linaro.org>

ah move more_rx_waiting to wsi scope

Originally this was alright in wsi->u.hdr, because ah implied header
processing.  But since we allowed ah to be held across http
keep-alive transactions if we saw we had more header data, it means
we were trying to read this union member out of scope after it had
transitioned.

Moving the more_rx_waiting member to be a 1-bit bifield in the wsi
solves it and lets us check the state any time later at http
transaction completion.

https://github.com/warmcat/libwebsockets/issues/441

Signed-off-by: Andy Green <andy.green@linaro.org>

Don't include <sys/cdefs.h> for NetBSD

We needed it for the BSD symbol to be defined, while __NetBSD__ is defined
with a compiler.

Thanks Andy Green for the initial fix.

Signed-off-by: Kamil Rytarowski <n54@gmx.com>

netbsd netinet include

https://github.com/warmcat/libwebsockets/issues/442

Signed-off-by: Andy Green <andy.green@linaro.org>

force service properly when unconsumed rxbuf in ah

Signed-off-by: Andy Green <andy.green@linaro.org>

close wsi must do detatch ah flow even if no ah

Signed-off-by: Andy Green <andy.green@linaro.org>

improve timeout and ah list comments

Signed-off-by: Andy Green <andy.green@linaro.org>

remove ah scan

Signed-off-by: Andy Green <andy.green@linaro.org>

fix missing callback return check

Signed-off-by: Andy Green <andy.green@linaro.org>

unix privs change group before user

Otherwise we no longer have privs to change the group after doing the user

Signed-off-by: Andy Green <andy.green@linaro.org>

test server log LWS_CALLBACK_HTTP

Signed-off-by: Andy Green <andy.green@linaro.org>

v1.7.2

Signed-off-by: Andy Green <andy.green@linaro.org>

lws status fix LWS_PRE

Signed-off-by: Andy Green <andy.green@linaro.org>

defeat POLLOUT if socket in shutdown wait

After andrejs.hanins@ubnt.com

Signed-off-by: Andy Green <andy.green@linaro.org>

ws union member must have actual struct at start not pointer

Signed-off-by: Andy Green <andy.green@linaro.org>

client callback closed if not upgraded also for server connection

https://github.com/warmcat/libwebsockets/issues/437
https://github.com/warmcat/libwebsockets/pull/440

client connect must init position_in_fds_table

windows doesnt have localtime_r

Signed-off-by: Andy Green <andy.green@linaro.org>

test server status no ssl include time header

Signed-off-by: Andy Green <andy.green@linaro.org>

test server add lws_status

Signed-off-by: Andy Green <andy.green@linaro.org>

LWS_BUILD_HASH improve

Signed-off-by: Andy Green <andy.green@linaro.org>

test html add tabs

Signed-off-by: Andy Green <andy.green@linaro.org>

libwebsockets.org url updates

Signed-off-by: Andy Green <andy.green@linaro.org>

lwsl stderr sink use formatted time

AG change to use localtime_r for threadsafety and leave old method as fallback

ssl get rid of build_cert_chain

It's not supported on major distro SSL (Fedora is 1.1) and
libressl (on 2.x) doesn't have it either.

https://github.com/warmcat/libwebsockets/issues/435

Signed-off-by: Andy Green <andy.green@linaro.org>

Stop uv loop in default signal handler, clean-ups

Make sure every handle is closed before destroying the uv loop

Don't destroy ev/uv loops if they haven't been created yet

v1.7.1

Signed-off-by: Andy Green <andy.green@linaro.org>