OScam x86_64-PC-Linux

Oscam1.20 svn8457

- Changeset [8457] by gf
newcamd: Add paranoid checks for per port newcamd data.
Better safe than sorry.
- Changeset [8456] by gf
Slim down struct s_port.
struct s_port contains filter array and couple of other fields that are
used only for newcamd and nothing else. They waste more than 2k in common
case.
To stop losing this much space, create private structure only for
newcamd fields and allocate it dynamically. This lowers struct s_port
size a lot and of course struct s_ptab even more because it contains
array of struct s_ports's.
New structure sizes (32-bit):
before after
struct s_port 2144 12
struct s_ptab 68612 388
struct s_config 141260 4812
Size report (32-bit):
text data bss dec hex filename
998222 1944 602344 1602510 1873ce before/oscam-1.20-unstable_svn8453-i486-slackware-linux
998174 1944 61800 1061918 10341e after/oscam-1.20-unstable_svn8453-i486-slackware-linux
bloat-o-meter report (32-bit):
add/remove: 0/1 grow/shrink: 13/16 up/down: 7897/-548383 (-540486)
function old new delta
modules 1840 9520 +7680
...
cfg 141260 4812 -136448
static.ptab 411672 - -411672
- Changeset [8455] by gf
Make ptab in struct s_module embedded not a pointer.
This blows up bss (static) memory nearly twice but we'll get that
back and lots more once we slim down struct s_port in the next
commit.

thanks LePouSAT

08.03.2012 Oscam1.20 svn8510

- Changeset [8510] by gf
nagra: Fix regression introduced in r8486. Converting to private data in commit r8486 missed the fact the there
is a case where private data is needed before it is initialized. This
happens with irdeto cards in nagra mode.
Thanks to mechi for testing and the report.

thanks LePouSAT

11.03.2012 Oscam1.20 svn8532

- Changeset [8532] by gf
webif: Stop http server thread realiably on exit.

thanks LePouSAT

Oscam1.20 svn8538

- Changeset [8538] by gf
webif+cccam: add missing serverip setting.
All network facing modules support serverip setting except cccam
and webif. They used only the global serverip. That patch adds support
for serverip setting in [cccam] and [webif] sections.
Patch by manu posted in ticket #3235.
- Changeset [8537] by stefansat
Modified CMakeLists.txt to use Flags It's not finished yet, but here the reworked file .
LIBCRYPTO,LIBUSB,PCSC are already adapted

thanks LePouSAT

15.03.2012 Oscam1.20 svn8552

- Changeset [8552] by gf
Revert "Reliably stop check() thread on exit." It seems some platforms implementation of pthread_cond_timedwait() is
not that good enough and using pthread_cond_timedwait() instead of
nanosleep increased response times a lot.
This reverts commits r8525 and r8551.
Thanks to wimpie007 for the report and testing.

thanks LePouSAT

19.03.2012 Oscam1.20 svn8559

- Changeset [8559] by stefansat
Cmake extra libusb messaging correction
- Changeset [8558] by stefansat
Cmake libusb messaging updated

thanks LePouSAT

22.03.2012 Oscam1.20 svn8561

- Changeset [8561] by gf
Resolve channel names in ECM log and monitor.
Patch by malakudi posted in ticket #3239.

thanks LePouSAT

Oscam1.20 svn8566

- Changeset [8566] by gf
dvbapi: Improve Irdeto/Beta? EMM tunnel.
EMM tunnel now also depends on the betatunnel service-ids.
It is possible to specify betatunnel with srvid=0000 to activate EMM-only tunnel.
Some minor code improvements as suggested by user depp0n.
Patch by canna posted in ticket #3258.
- Changeset [8565] by gf
ecm: Simplify locking in CW processing thread.
- Changeset [8564] by gf
ecm: Reliably stop cw_process() thread on exit. First attempt to reliably stop check()/cw_process() thread on exit commited
in r8525 was reverted in r8552 because it introduced latencies.
The latencies were introduced because waking up the thread from inside
ECM processing functions was done by a signal which wasn't working once
pthread_cond_timedwait() started being used instead of nanosleep().
This commit should be the proper fix.
Thanks to wimpie007 for the testing.
- Changeset [8563] by gf
ecm: Move CW sending thread code in oscam-ecm.c
The moved code that implements CW processing thread is very central
to ECM procesing, so it is a good idea to move it closer to other
ECM functions.

thanks LePouSAT

26.03.2013 Oscam1.20 svn8576

- Changeset [8576] by Admin
WebIf: Fix missing file closes when reading disk template files which are not valid for the compiled binary.

thanks LePouSAT

Oscam1.20 svn8578

- Changeset [8578] by theparasol
Tryfix2 VU+ boxes slow responses on internal readers for T0/T1 protocol
T14 protocol was broken and now is handled like before and communicated again with reader driver

thanks LePouSAT

31.03.2013 Oscam1.20 svn8580

- Changeset [8580] by gf
config: Invoke mktemp(1) without parameters.
This increases portability because some mktemp implementations do
not support -t flag. Calling mktemp without parameters creates temp
files in $TMPDIR or /tmp which was the reason -t was used in the
first place.

thanks LePouSAT

02.04.2013 Oscam1.20 svn8596

- Changeset [8596] by gf
oscam-net: Remove unneeded static.
The variable is only used in the function and it is not returned to
the caller so using static is most certanly wrong and buggy.

thanks LePouSAT

07.04.2013 Oscam1.20 svn8602

- Changeset [8602] by theparasol
Quickfix compiler issues without dvbapi module enabled
- Changeset [8601] by theparasol
Killing DVBAPI client triggers a segfault if emms are being processed.
A killed dvbapi client can only be restarted by starting oscam again, so it doesnt make any sense.
This protects users from accidentally killing a dvbapi client from the webif.

thanks LePouSAT

Oscam1.20 svn8606

- Changeset [8606] by theparasol
Revert recent changes to oscam-work (8598, 8599, 8603) -> need more testing!
- Changeset [8605] by stefansat
removed the quite about the openssl library used
Then users can see the real libray's they are using if
target root dir is set
- Changeset [8604] by gf
Initialize mutexes and cond variables.
Don't assume that clearing the phread_mutex_t and pthread_cond_t variables
to zero is enough to consider them initialized. Use proper initialization
functions.

thanks LePouSAT

10.04.2013 Oscam1.20 svn8608

- Changeset [8608] by stefansat
Corrected doc about use off cccreshare in oscam.user
- Changeset [8607] by stefansat
Corrected doc about use off reshare and cccreshare parameter

thanks LePouSAT

14.04.2013 Oscam1.20 svn8617

- Changeset [8617] by gf
Move messages about skipped readers to debug level 1.
If more than one reader exist in a system these messages are misleading,
the problem solved in r8616 is still there. If user have one reader
and client with the same username, when oscam is upgraded from old version
(< r7210) the reader would stop being matched and the user would have no
idea why, unless it turns on debug level 1.

thanks LePouSAT

Oscam1.20 svn8621

- Changeset [8621] by Deas
css.css - fix missing row for TD.usercol23. tnx to skyndas!
- Changeset [8620] by gf
log: Skip seconds when logging to syslog.
Logging to syslog tries to cut unneeded information from the log
line but misses the seconds, so actually the text that reaches
syslog looks like this:
After this patch the seconds are correctly skipped and logged lines
look like this:

thanks LePouSAT

Oscam1.20 svn8625

- Changeset [8625] by gf
dvbapi: Fix format string used in sscanf().
Use the correct format string for int16_t.
- Changeset [8624] by canna
Decoupled betatunnel functionality from module-stat (loadbalancer).
Small adjustment/improvement of the betatunnel service-id bypass introduced in r8577, r8597. Previous version caused problems on certain STBs.
- Changeset [8623] by depp0n
- fix compiler warning
- Changeset [8622] by depp0n
- fix for ignoring irdeto chids
- fix prioritizing/ignoring for irdeto chid 0

thanks LePouSAT

25.04.2013 Oscam1.20 svn8629

- Changeset [8629] by canna
Some improvements for betatunnel ecm & emm handling

thanks LePouSAT

Oscam1.20 svn8636

- Changeset [8636] by Admin
Small C&P fix
- Changeset [8635] by Admin
Allow socket reuse in more places to prevent "bind failed" on restarts.
Add TCP_USER_TIMEOUT on all connections for systems that implement it. This disconnects users who do not ACK packages within 60s (but they should be caught by the keepalive settings anyway; just to be on the safe side).

thanks LePouSAT