Changeset [7296]: Avoid filling the log with CACHEEX messages.
Code:Changeset [7295]: reader-videoguard: Add support for Indovision (09C1). Patch by sat_hunter @ sb.
Changeset [7296]: Avoid filling the log with CACHEEX messages.
Code:Changeset [7295]: reader-videoguard: Add support for Indovision (09C1). Patch by sat_hunter @ sb.
Changeset [7314]:
Code:dvbapi: Sent data as one packet. Some time ago I've sent a patch for generic dvbapi under linux (boxtype=pc). It is working fine but I think there is need for one important enhancement. Currently it is sending each CA_SET_PID and CA_SET_DESCR as two separate packets which are: the request integer and the data. I think it will be much better to sent it in one packet. At first it is not elegant to sent it this way, besides the UDP packet order is not the think that we should rely on. About client implementation: as far as I know there is currently two clients using the api: vdr-plugin-dvbapi and ostryck MuMuDVB fork (maybe tvheadend will also use it). I contacted ostryck and he said there will be no problem with this small api change. Obviously also no problem on vdr-plugin-dvbapi side. Patch by manio posted in ticket #2719. Changeset [7313]: config: Update OS X SDK detection to handle 10.7+ versions. In 10.7 the SDK location have moved. Thanks to stefansp for providing the new paths. Changeset [7312]: config: Add -O / --detect-osx-sdk-version option to help text.
Changeset [7365]: small error correction about debug, now debug for mac not disabled anymore
Code:Changeset [7364]: Getting rid off the irritating /usr/bin/ranlib has no symbols warning on mac 10.7 and 10.8 Changeset [7363]: Added possibility to compile on mac whitout pcsc with -DHAVE_PCSC=0 Cleaned up some unneeded messages Set for now 10.6 like before and not forced to x86_64 Changeset [7362]: Added new Xcode location for mac 10.7 (lion)10.8(mountain lion) Added static libusb support for mac Set 10.6 to complile x86_64 as well Note : On 10.7 Xcode needs to be updated to 4.3
Changeset [7374]: Try for sdk 10.6 fixed x86_64
Code:Changeset [7373]: Remove extra/unneeded parentheses. clang 3.x warns about these and it is right. Changeset [7372]: Fix wrong memset sizes. These bugs were uncovered by compiling with clang 3.x (make CC=clang). Changeset [7371]: adding architecture message for mac Changeset [7370]: Mac setting exact osx architecture should solve older hardware bugs Changeset [7369]: Fix IP access when IPv6 support is enabled. Patch by manio, fixes ticket #2723. Changeset [7368]: Possible pcsc error on sdk 10.6 Changeset [7367]: Forgot the smargo utils in previous patch Changeset [7366]: Added SDK location for use off Xcode version less then 4.3 on osx10.7 Correct compile error If build in libcrypto is used 10.6 now also compiles x86_64, I tested it with 10.6 sdk and seems to work good.
Changeset [7385]: Prevent OSX compilation failures when IPv6 support is enabled.
Code:Changeset [7384]: cfg: Switch oscam.conf parsing to new world order. Extras: A lot less code. Shows warnings if certain section is not loaded because there is no compiled in support. Warnings for unknown settings. Now every module when disabled, have no config data/code in the binary. Changeset [7383]: cfg: Add basic functions dealing with config sections. Changeset [7382]: cfg: Move len4caid() and write_config() functions. This is done in prepration for cleanup of oscam-config.c Just code movement, no code changes. Changeset [7381]: cfg: Add format string checking for fprintf_conf(). Add format string ... cfg: Add format string checking for fprintf_conf() and fix the errors detected by the compiler. The fixed warnings were: oscam-config.c: In function 'camd33_key_fn': oscam-config.c:384:2: warning: zero-length gnu_printf format string [-Wformat-zero-length] oscam-config.c: In function 'newcamd_key_fn': oscam-config.c:495:2: warning: zero-length gnu_printf format string [-Wformat-zero-length] oscam-config.c: In function 'write_services': oscam-config.c:1374:3: warning: zero-length gnu_printf format string [-Wformat-zero-length] oscam-config.c:1380:3: warning: zero-length gnu_printf format string [-Wformat-zero-length] oscam-config.c:1386:3: warning: zero-length gnu_printf format string [-Wformat-zero-length] oscam-config.c: In function 'write_server': oscam-config.c:1782:5: warning: zero-length gnu_printf format string [-Wformat-zero-length] oscam-config.c:1892:5: warning: zero-length gnu_printf format string [-Wformat-zero-length] oscam-config.c:2054:5: warning: format '%lX' expects argument of type 'long unsigned int', but argument 4 has type 'uint32_t' [-Wformat] Changeset [7380]: cfg: Move chk funcs from oscam-conf.c to oscam-conf-chk.{c,h} Just code movement, no code changes. Changeset [7379]: cfg: Move basic config parsing code to oscam-conf.{c,h} Just code movement, no code changes. Changeset [7378]: cfg: Move search_boxkey() function to reader-cryptoworks.c The only place where this function is used is in reader-cryptoworks. It is better move it out oscam-config.c Changeset [7377]: cfg: Move show_sidtab() function closer to where it is used. Just code movement, no code changes. Changeset [7376]: ipv6: Fix setting of client address. This patch fixes regression introduced with full IPv6 support. Patch by manio. Changeset [7375]: cfg: Fix setting of serverip config variables. Fixes ticket #2730 and #2733.
Changeset [7396]: cfg: Remove workaround for "httptpl" with missing ending slash. There are ...
cfg: Remove workaround for "httptpl" with missing ending slash.
There are a lot of paths described in the config and it makes little sense to fixup (add / at the end) only for "httptpl" and not for the others.
The users must set the directory with slash (/) at the end in order for that directory to be correctly prepended to file names.
Code:Changeset [7395]: cfg: Move functions dealing with config sections in oscam-conf.c Changeset [7394]: cfg: Set default values using config tables. Changeset [7393]: cfg: Add default values for string options. Changeset [7392]: cfg: Introduce union to hold default values. Changeset [7391]: cfg: Move mk_t_XX) funcs from oscam-config.c to oscam-conf-mk.c Just code movement, no code changes. Changeset [7390]: cfg: Introduce OPT_SAVE_FUNC. Move the logic that decides if a certain ... cfg: Introduce OPT_SAVE_FUNC. Move the logic that decides if a certain section is written next to configuration varaibles and cleanup config writing code. Changeset [7389]: cfg: Introduce union in struct config_list to save some space. Changeset [7388]: Allow use of bool type by including stdbool.h Changeset [7387]: Fix Cmake ticket 2735 Removed misleading messages when compiling on mac Removal from some /ranlib no symbols error on mac Changeset [7386]: config: Do not use "echo -n" use printf instead. OS X's /bin/sh do not support echo -n which leads to a "funny" bug when using ./config --gui. When Readers menu is entered the last selected reader is lost when the config is saved.
Changeset [7410]: cfg/webif: Refactor common code into webif_save_config().
Code:Changeset [7409]: Fix handling of aulow, hideclient_to and appendchaninfo settings. These settings were loaded from [monitor] section in oscam.conf but they are global in nature, used in monitor, webif, oscam logging and other places. Also when the new config parser was introduced, in case [monitor] was disabled these values didn't get their default values which causes strange behaviour of the webif that depended on them. This commit fixes the issues by making the variables not depend on [monitor] only. They are loaded from [monitor] and [webif] and saved in both places (the last one wins). That way, old configs that have set the values in [monitor] would continue to work. This commit fixes tickets #2725 and #2739. Changeset [7408]: webif: Allow cs357x (camd35), cs378x (camd35tcp) and csp sections to be disabled. Without this once any of cs357x (camd35), cs378x (camd35tcp) or csp sections are enabled (by setting the port) they can not be disabled anymore from the WebIf. Changeset [7407]: webif: Use correct section name for camd35 (cs357x). Without this the section was not saved when WebIf was used. Changeset [7406]: cfg: Allow config functions to be trully generic. This config allows functions to be useful when the variable storing configration data is dynamically allocated (like accounts and readers). Changeset [7405]: cfg: Rename function s/caid_fn/caidtab_fn/. This makes it easier to search for are using caidtab api. Changeset [7404]: cfg: Change config filenames to defines. That way we should not get ... cfg: Change config filenames to defines. That way we should not get unused warnings when some feature is disabled and the strings would not be in the final binary resulting in space savings. Fixes ticket #2738. Changeset [7403]: csctapi/ifd_pcsc: Fixup previous commit. Umm, that'll teach me to test more... Changeset [7402]: csctapi/ifd_pcsc: Fix two harmless warnings. The warnings were: csctapi/ifd_pcsc.c:197:15: warning: comparison of integers of different signs: 'LONG' (aka 'int') and 'unsigned int' [-Wsign-compare] csctapi/ifd_pcsc.c:207:21: warning: comparison of integers of different signs: 'LONG' (aka 'int') and 'unsigned int' [-Wsign-compare] Changeset [7401]: build: build: Fix libusb build on OS X when libusb + pcsc build is requested. Setting -isysroot causes the compiler to stop looking in /usr/local/{include,lib} which results in the following error, when make pcsc-libusb is run on OS X. In file included from csctapi/icc_async.c:37: csctapi/ifd_smartreader.h:14:31: error: libusb-1.0/libusb.h: No such file or directory make[2]: * [lib/libcsctapi-i686-apple-darwin10-libusb-pcsc.a(csctapi/icc_async.o)] Error 1 make[1]: * [all] Error 2 Changeset [7400]: cfg: Add open_config_file() and use it. Changeset [7399]: cfg: Implement {create,flush}_config_file(). Changeset [7398]: cfg: Add config fixups in configuration tables. Changeset [7397]: config: Prefer newest OS X SDK.
Changeset [7425]:
Different approach to solving httpttl ending slash problem.
With this, the variable coming from the config file is not modifed, instead when it is used and the ending slash is missing, the problem is corrected in place.
Code:Changeset [7424]: Fix use of possibly uninitialised values in readers. ATR_GetHistoricalBytes() can return without initializing length parameter so valgrind correctly detects this for each reader: Conditional jump or move depends on uninitialised value(s) at 0x80A8A30: conax_card_init (reader-conax.c:117) by 0x80A6306: reader_get_cardsystem (reader-common.c:210) by 0x80A640B: reader_reset (reader-common.c:264) by 0x80563A9: work_thread (oscam.c:4051) by 0x4060AE9: start_thread (in /lib/libpthread-2.15.so) by 0x415B69D: clone (in /lib/libc-2.15.so) Changeset [7423]: Force valgrind to stop complaining about possibly uninitialized values. The valgrind is wrong here but annoying. The problem is that it doesn't know what ioctl call does and assumes the structure is left not initialized. Conditional jump or move depends on uninitialised value(s) at 0x81062D7: IO_Serial_SetBitrate (io_serial.c:203) by 0x810642A: IO_Serial_SetParams (io_serial.c:233) by 0x810741F: IO_Serial_InitPnP (io_serial.c:747) by 0x81014EB: Phoenix_Init (ifd_phoenix.c:59) by 0x80FC78E: ICC_Async_Device_Init (icc_async.c:185) by 0x80A65AF: reader_device_init (reader-common.c:313) by 0x80D7CDC: reader_init (oscam-reader.c:844) by 0x805640B: work_thread (oscam.c:4065) by 0x4060AE9: start_thread (in /lib/libpthread-2.15.so) by 0x415B69D: clone (in /lib/libc-2.15.so) Conditional jump or move depends on uninitialised value(s) at 0x81062F3: IO_Serial_SetBitrate (io_serial.c:205) by 0x810642A: IO_Serial_SetParams (io_serial.c:233) by 0x810741F: IO_Serial_InitPnP (io_serial.c:747) by 0x81014EB: Phoenix_Init (ifd_phoenix.c:59) by 0x80FC78E: ICC_Async_Device_Init (icc_async.c:185) by 0x80A65AF: reader_device_init (reader-common.c:313) by 0x80D7CDC: reader_init (oscam-reader.c:844) by 0x805640B: work_thread (oscam.c:4065) by 0x4060AE9: start_thread (in /lib/libpthread-2.15.so) by 0x415B69D: clone (in /lib/libc-2.15.so) Changeset [7422]: Adding ending slash fixup for emmdir - for consistency with other configuration parameters pointing to folders Changeset [7421]: REVERT: cfg: Remove workaround for "httptpl" with missing ending slash. This one reverts r7396 because my analysis was wrong. All other variables that hold paths are working fine when slash (/) is missing at the end. Restore the needed fixup. Changeset [7420]: Some clean up in message ouput
Changeset [7435]:
Add account_set_defaults() and use it where needed.
This fixes at least one bug in WebIf (the new accounts were added with cccignorereshare = 0 when they should be using -1).
Code:Changeset [7434]: cfg/accounts: Change account->dyndns to char * This saves 60 bytes for each account in most cases. Changeset [7433]: cfg/account: Convert simple variables. This increases struct s_auth by 30 bytes, but we'll get some of that back once config parser start supporting boolean values. Also dyndns, usr and pwd fields when converted to char * would save at least 180 (60+60+60-8+ bytes per struct s_account. Changeset [7432]: webif: Do not display caid and srvid when they are set to internal values. Fixes ticket #2751 and partially ticket #1544. Changeset [7431]: webif: Do not return empty page when password authentication fails. Lets not be antisocial and say to the poor user at least that the access was denied. Changeset [7430]: cfg: Free config values allocated from the parser. Changeset [7429]: module-pandora: Do not kill OSCam for stupid reasons. The functions that are patched with this are all allowed to returing failure. Instead they were killing the whole OSCam, which is crazy. Changeset [7428]: module-constcw: Do not kill the whole oscam if init fails. Changeset [7427]: Added the possibility to compile libusb dynamic when LIBUSBDIR is used Changeset [7426]: cacheex: small stability enhancement cacheex: debuglevel 1024 (ECM logging) now logs oscam+csp-hashs
Changeset [7438]:
cfg: Fix build error when webif is disabled.
Fix an error introduced in the previous commit. Fixes ticket #2758.
Code:Changeset [7437]: cacheex: fixed full running queue Changeset [7436]: cfg: Prevent crash when freeing account values. Fixes ticket #2754.
Posting Permissions
Reply With Quote