CHANGES

2012/07/29 (wololo)
- renamed *.nids into *.nid, an additional step into corectly supporting FAT32 limitations

2012/05/15 (wololo)
- Additional hooks to fix some SC3 VHBL crashes (snes9xTYL, Lamecraft,...) 

2012/03/18 (wth)
- wth's additional hooks

2012/01/02 (wololo)
- Some tweaks to get Doom to work. Syscall estimation seems to half-work on the PSP, and not at all on the Vita...maybe I'm seeing random effects...

2011/12/27 (wololo)
- Some modifications to make HBL compatible with "project life" (HBL on the PS Vita). Hopefully this
doesn't break backwards compatibility
-- I made some of the changes from r117 happen only at compile time. These
changes were assuming that 6.60 == signed launcher, which is not the case here
-- the exploit we're using does not have access by default to the same cache
invalidate function, so I made that a macro that can easily be undef/defined in
exploit_config.h

This does not contain the new exploit folder itself, which will be submitted
later.

2011/08/14 (JJS)
- this version is only compatible with FW 6.60
- added all NIDs of the relevant libraries to the launcher
- to reduce the memory consumption of the NID table, the lowest syscall and gap are extracted from the launcher imports and FROM_LOWEST estimation is used
- added option of relocating the function addresses in buildsdk.rb, this is mainly useful when building an SDK from the launcher

2011/01/25 (JJS)
- added files missing in R113
- some changes for the "signed" HBL
-- no kernel memory dump is created
-- p5 stubs are not gathered
-- function hooking is performed as if g->syscall_known was set
-- HBL load address increased to 0x09F67000

2011/01/23 (JJS, wololo)
- added launcher to run directly from the XMB (requires minpspw 0.11 for compatibility reasons, also requires external tools for signing)
- Added "launcher-compatible" version of HBL, works on OFW 6.37 and below

2010/12/20 (JJS)
- added lowest syscall memory offsets for sukkiri on 6.35
- added code to close all files left open by the exploitet game

2010/12/20 (JJS)
- added Everybody's Sukkiri, exploit by Darxploit, Flyer, minomushi, some1 and HBL port by some1
- added code to reset the HOME screen language to the system default (only for sukkiri)
- made GAME_FREEMEM_ADDR an array for all exploits
- fixed bug in static ELF loading that led to HBL being overwritten in sukkiri

2010/11/23 (JJS)
- added kernel memory offsets for hotshots2/everybody2 on 6.20
- changed _hook_sceKernelExitDeleteThread to actually call sceKernelExitDeleteThread because it is available through p5 stubs now, this should fix freezing in JellyCar
- minor correction to subinterrupthandler_cleanup()

2010/11/15 (JJS)
- fixed  issue 342  (some homebrew crashing on exit)
- added code to track opened files by hooking sceIoOpen/sceIoClose and to close remaining files on homebrew exit
- now releasing all subinterrupt handler, should fix that no homebrew could run after gpsp

2010/11/14 (JJS)
- added alternative version of Hot Shots Golf: Open Tee

2010/11/13 (JJS)
- applied fix for savegame exploit code by neur0n (h.bin file handle not being closed, see http://wololo.net/talk/viewtopic.php?f=4&t=286). Thank you!
- added two missing addresses in exploit_config.h for Patapon 2
- added releasing of two additional sub interrupt handlers to FreeMem()
- added new nid files for 6.3x

2010/11/07 (J416)
- I fixed the crash when hbl erased the memory on minna no golf2.

2010/11/05 (JJS)
- checked in hotshots2, minna2 and minna2best config files
- fixed import stub parsing again by also relocating variable imports, they now pass the user memory pointer check

2010/10/29 (JJS)
- changed getFirmwareVersion to the code by neur0n (see http://wololo.net/talk/viewtopic.php?f=4&t=128&start=0). Thank you!
- removed all memory offsets for firmware detection from the exploit_config.h files
- added GO perfect syscall memory offsets for hotshots2/everybody2 on 6.3x
- added GO perfect syscall memory offsets for minna2/minna2best on 6.20 and 6.3x
- fixed bug that disabled perfect syscalls on 6.3x GOs (g->syscalls_known was being reset)
- fixed issue 333 (patapon2 broken on 5.03 and 5.51)

2010/10/19 (JJS)
- changed files in "minna2" folder to the original Minna No Golf 2 release version
- added "minna2best" folder for the Minna No Golf 2 budget "the best" release
- added more stubs to eLoaderconf.rb
- fixed regression introduced in R102 that prevented the p5 stubs from being parsed (stub addresses on p5 got rejected by the user memory pointer sanity check)
- fixed compiling and linking on MINPSPW 0.9.6+ (explicit typecast in modmgr.c and rebuilt sdk_loader.S for Patapon 2)

2010/10/15 (Wololo)
- initial checkin for Hotshots golf 2 / everybody's golf2 / minna no golf 2. Credits for the exploits/binary loaders go to mamosuke, J416, JJS, wololo.
- Added some sanity checks for stub entries.
- Added some generation tools in /tools (including a ruby version of the moskitool) (JJS, wololo)

2010/09/19 (Wololo)
- JJS's patch for perfect syscalls on the golf exploit
- Changed loading address for minna no golf exploit, per J416's recommendation
- moved some global variables into scratchpad, which should solve firmware detection issues (issue 323)

2010/09/08 (Wololo)
- Adpated the code to J416's Exploit in minna no golf, Freddy's Exploit in Everybody's golf, and Wololo's exploit in hotshots golf. Adapted by JJS and Wololo
-- This is majorly untested, but I'm afraid of losing the many code changes involed here. I basically tested this on the patapon exploit and the hotshots exploit on a psp go 6.20
-- This improves HBL portability, although lots remain to be done
-- This officially makes HBL compatible with OFW 6.30 and 6.31

2010/08/29 (JJS)
- Added Nymphaeas patch that allows to exit HBL to the XMB by using the HOME exit menu ( http://advancedpsp.tk/foro_es/viewtopic.php?f=37&t=1046 )
- Added code to get additional syscalls using the stubs of utility dialogs loaded into p5.
-- This means sceCtrlPeekBufferPositive can now always be used and the option to override it is now obsolete for all 6.20 users.
-- Also all relevant sceIO* functions should be perfectly estimated now.
- Fix for issue 43 (crash on start from MS on the GO). Kernel dump is now saved to ef0.
- Fix for issue 306 (control issues). Hopefully fixed by replacing the sceCtrlReadBufferPositive call with sceCtrlPeekBufferPositive.
- Corrected 5.70 nids files. Thanks n00b81!
- Corrected scesuspendforuser.nids for 5.xx and 6.xx.

2010/08/08 (wololo)
- fix for issue 296 (off by 1 error). No visible change
- fix for issue 137 (reset graphics). Fixes graphic issues with PSP Rhythm, Spider solitaire, FCE Ultra,...
- added an option for "force exit". Allows to come back to HBL menu for homebrews that don't have an exit menu. This is an attempt to mitigate issue 181 (exit button not working anymore)
- syscalls_known can now be overriden in the config file for tests purposes (untested)

2010/08/01 (wololo)
- JJS' patch for syscall estimation. Improves syscall estimation on psp2000 < 6.20, psp3000 < 6.20, and pspGo all firmwares

2010/07/18 (wololo)
- Fix for issue 270 Sound in picodrive (needs to be 44KHz though, check the settings in picodrive!)
- Fix a bug introduced in R94 with libraries unloading. This fixes issues 280 and 283 (goEar and pspkvm)
- Fix for issue 284 (can't save in snes9xTYL). Thanks JJS and SeanC994

I ran the following tests on a pspGo 6.20 with the distrib version:
Doom Runs, can exit, Network found
Wagic Runs with sound, can exit
CSPSP Runs, play exit
GPSP Runs, play NO Exit
Daedalus Runs (slow), play no exit (?)
Picodrive Runs with sound (no mp3 support yet I think) 
FCEUltra Runs, load rom, can exit
ScummVM Runs, loads game, can exit
TOME Starts, can exit
Snes9xTYL Starts (sometimes crash at startup?) play, can exit, can save 
kvm Starts. didn't find a way to quit (except "pause game" feature on the Go)
GoEar Starts, but I can't seem to connect.

2010/07/12 (wololo)
- Added JJS's patch for Static elfs loading. This improves compatibility with some games
- Fixed issue 269 (Exit from Doom crashes)
- override_sceCtrlPeekBufferPositive is now an option in hbl_config.txt, for those who want to take the risk and get more speed in emulators

2010/07/05 (wololo)
- added JJS's patch for network modules. This partially fixes issue 109. Thanks a lot man!
-- see the subject at http://advancedpsp.tk/foro_es/viewtopic.php?f=37&t=784
- removed the code that attempts to delete callbacks. Based on my experiments with psplink, I'm assuming that unloading the patapon module simply unloads the callbacks as well. This fixes issue 45

2010/06/27 (wololo)
- unload utility modules when exiting a homebrew (partially fixes issue 179)

2010/06/25 (wololo)
- Moved the menu outside of HBL, it is now a homebrew so that anybody respecting the API can code their own menu as a "standard" homebrew.
- Added possibility to go back to the menu after playing a homebrew
-- This feature is experimental and there are known bugs about it, please check the issues list!

2010/06/16 (m0skit0)
- Function to find the library exports of a loaded module by name (the real deal this time xD) (modmgr.c)
- Added more utility loading code (EXPERIMENTAL) (resolve.c, modmgr.c)
--> Network still not available :( Any further testing/feedback is appreciated
- Added a couple of functions that search for string and 32-bit value on memory (lib.c)
- Moved hook switches to hook.h
- Added switchable hook for sceUtilityLoadModule (hook.c)
- Added displaying real homebrew name (by MrShawman)

2010/06/14 (m0skit0)
- Function to find a loaded module by name (modmgr.c)
- PoC code to resolve utility imports (resolve.c, modmgr.c)
--> Only works for MP3 library over 5.00 M33, and it sounds absolutely awful with WTH (lag owns)
- Changed some defines from eloader.h to sdk.h
- Added data stucture to parse exports (elf.h)
- Added function to insert a library into library table (tables.c)

2010/05/26 (wololo)
- Put HBL back into partition 2. This fixes issue 133 (exit/sleep not working). Changes into malloc also fix issues with Wagic and some roms not loading in gpsp since rev 80
- Move all globals into a struct in scratchpad. This fixes issue 128 (HBL too big since rev 79)
- Fix issue 123 (devkitversion)...again :) Which should fix "firmware" errors in Daedalus
- Added a sanity "test" homebrew. If you code some changes and this homebrew reports errors : Don't submit your code!

2010/05/22 (m0skit0)
- Added scrolling menu for 40 homebrews max (thanks to MrShawman!)
- Separate mallocs for p5 and p2 (r80 and r81 malloc code was wrong)
- Included missing file lib.h
- Included some code for loading utility modules (not operating yet)
--> IMHO HBL needs a serious recoding effort...

2010/05/22 (wololo)
- Still considered as EXPERIMENTAL RELEASE
- Made the Makefile more strict
- Fixed a few issues detected by gcc
- Fixed a malloc bug that prevented Wagic from loading correctly in rev80
- Fixed issue 129

2010/05/19 (m0skit0)
- EXPERIMENTAL RELEASE
- Tested on PSP 2000 5.00 M33 & PSP Go 6.20
- Changed HBL to be loaded on memory partition 5 instead of the usual 2
--> This gives us whole parition 2 memory, and should get rid of stuff being overwritten
- I got rid of the HBL memory allocation since we actually inject HBL to partition 5
- HBL stubs are back to scratchpad (since we can't allocate p5 memory from p2)

2010/05/13 (wololo)
- added md5 functions (they can be removed and replaced by a _hook_generic_success if we are short on ram, I only added that for gpsp, but it works without it)
- added some more hooks for improved compatibility
- Added configuration file with possibility of local override on a per game basis
-- the config file allows to choose the folder in which the menu will start

- I'm seeing weird effects from time to time when compiling. I think we have a huge bug somewhere OR as m0skit0 said we are running out of memory, and the stack is meeting our data :(

2010/05/11 (m0skit0)
- Fixed an error on "make distrib" (again, thanks Cathesdus)
- Put back the tables memory allocation (seems to work fine now)
- Deleted all user mode libraries from the tables
--> It's useless to have them there when the user mode modules have been unloaded
- Fixed REESTIMATE_SYSCALL define to be useful (wasn't used)
- Added a new make target (nonids) to build without NID debug
- Fixed big stupid bug on new syscall estimation method (FROM_CLOSEST)

2010/05/10 (m0skit0)
- Reverted the tables memory allocation, made a lot of things to fail
- Fixed error on "make distrib" (thanks Cathesdus)
- Fixed bug on syscall reestimation
- Added a new reestimation method
- Reverted first estimation method (FROM_LOWEST)
- Fixed 5.03 FW detection (should be detected as 5.50)
- HBL tables are now allocated separately, this fixes a bug where they were overwritten when another module was loaded
- Changed reestimation method order (I don't see any improvement though)

2010/05/07 (wololo)
- More functions overrides
- Automatic SVN version number through makefile (tested on cygwin and dos command line, hopefully will work on linux too)
- Some cleanup. Removed some dead code.
- Fixed a nid bug in eLoaderconf.rb

2010/05/04 (m0skit0)
- Fixed a bug in syscall estimation for PSP Go
- Added 2 new methods for syscall estimation (also for reestimation)
- Added new C files for function testing and reestimation
- Now reestimation seems to work sometimes (compared to never working before xD)
--> Needs more heavy testing
- Modified data structure to hold NIDs (added number of nids)
- Now estimated syscalls are saved on nid_table so they don't have to be estimated again (until table fills)
--> This feature may need a more dynamic nid_table (linked list?)
- Added revision number to print on screen (please take care on updating this!)

2010/05/02 (wololo)
- more Functions overrides. Some of them are direct Aliases so they're 100% good
-- other are more borderline and we need a way to use them or not on a game-per-game and maybe firmware basis (config file)
- updated Makefile. make or make dev for debug, make distrib for no debug info
- Added sceReg.nids

2010/05/02 (m0skit0)
- Changed detection of FW 5.51 as 5.55 (previously set to 5.50) to load correct configuration

2010/04/30 (m0skit0)
- Fixed resolving functions (resolve.c) to accept other loading address than PRX_LOAD_ADDRESS
--> Now external modules seem to load/relocate/resolve fine, they do not crash but do not start either. PSPLink reports the external module thread is stopped
- Minor code/style fixing

2010/04/30 (wololo)
- Override a bunch of functions to avoid estimating them
-- Most of those can be deactivated with macros in eLoader.h (in the future we'll want a config for each game)
-- With these overrides, Wagic now loads on the PSPGo as long as you remove the mp3s from the Sound folder. (it crashes when attempting to save options though...)
- Fixed a bug that prevented big EBOOTS from loading (scummVM) because the path was overwritten

2010/04/30 (m0skit0)
- Added hooks for external module loading
- Fixed some bugs on module loading/starting
--> HBL crashes when resolving the imports for loaded module. Seems values have not been relocated.

2010/04/28 (m0skit0)
- Changed prx_load_program() to accept addresses other than PRX_LOAD_ADDRESS
- Added load_module and start_module functions in modmgr.c
- Moved some generic functions from utils.c to lib.c
- Changed EBOOT offset type to SceOff
- Added relevant data structures and functions for module loading
- To load external modules, hook sceKernelLoadModule and sceKernelStartModule (not tested), and uncomment LOAD_MODULE
- Big thanks to ab5000 for the memory allocation library!

2010/04/27 (m0skit0)
- Added a new member to tNIDResolver struct to relate each NID to a library
--> I think this is better than including a pointer/array for NID in tSceLibrary
- Added more debugging functions to dump ELF structures directly (more need to be done)
- Added a new data structure to handle module loading information
- New parsing user modules to free (thanks to Davee)

2010/04/26 (m0skit0)
- Fixed minor bug on menu (error is < 0, not <= 0)
- More eloader.c house cleaning

2010/04/24 (m0skit0)
- Added kernel memory dump for PSPGo at the start of the loader, because otherwise accessing kmem will fail (i dont know why)
--> This can be improved dumping only necessary offsets and not the whole partition
- Added finding lowest syscall from kernel memory for some libraries -only for PSP Go, and using the kmem dump-
- Got rid of intermediate static buffer for loading HBL into memory
- Changed tool "umem dumper" to do a kernel dump as well (only works on PSP Go) and changed the linker address to the new one on the new savedata (new!)

2010/04/25 (wololo)
- cleaned up Davee's free memory function
-- Also fixed issue 42 : 6.20 hangs on free_game_memory (as usual, syscall estimation was the culprit, I replaced the "find_thread" call with a direct address fetch like Davee does)
- Written own sceKernelMaxFreeMemSize() and sceKernelTotalFreeMemSize() functions to avoid 2 unecessary syscall estimates (avoids a crash in free_memory by ab5000)
- Fixed issue 41 (How to distinguish between Go and other PSPs ) : call getPSPModel() in utils.c (see eLoader.c which does that call). Note: impossible with this technique to differentiate other psp models though
- Fixed issue 39 ([Rev 49]Still no Free Ram). This was actually an issue with the game T.O.M.E and I'll post an updated Eboot for this game on my blog (http://wololo.net/wagic) ASAP
- Fixed issue 23 (List Only the Homebrew ) with a simple filter in menu.bin. Hopefully that should clean things a little bit.
- Added some more syscall boundaries, but I'm not super happy with the result (a.k.a. haven't seen any improvement), so maybe that update can be reverted :(
- removed the FAKEMEM flag as it is not needed anymore. I kept the hook for AllocPartitionMemory for debug purposes, but it should probabl be removed in the future. 

2010/04/23 (m0skit0)
- STILL DOESN'T WORK
- Fixed a few bugs, mostly concerning syscall estimations being wrongly calculated. Now they seem fine, but still no go :(
- Put back a loop at the end of loader (do not exit user_main)
- Davee's free_game_memory adjusted to free user_main
- Added sceHprm.nids from 6.20 FW (libs_6xx)
- Deleted NID table processor tool (not needed anymore)

2010/04/22 (m0skit0)
- NOT WORKING REVISION
- Splitted several stuff from eloader.c to tables.c and hook.c (can be splitted more)
- Added several functions to HBL imports 
- Updated eLoaderconf.rb tool too
- Added Davee's free memory function to HBL (you can choose between ab5000's and Davee's method in eloader.h)
- Fixed a bug where HBL would not check if own imports already exist in NID table
--> All seems to work fine, but HB doesn't execute

2010/04/18 (m0skit0)
- Added imports.config for 5.50 and 5.55 differences (thanks to ultimakillz)

2010/04/18 (wololo)
- Added firmware detection for 5.55 (thanks to Epel for the memory dump)
- Added basic "printf" functions for improved logging (thanks Fanjita & noobz for the initial code)

2010/04/17 (wololo)
- This change requires a new Save Game file (changed the address where h.bin is loaded). Regenerate the Save Game by yourself (codes in the folder SDK) or get it at http://wololo.net/wagic
- Code to unload the module "Labo". The code works, but it basically breaks compatibility with all the homebrews I tested, so right now I commented the code out.
To activate it, define the macro "UNLOAD_MODULE" in eLoader.h

2010/04/17 (wololo)
- Experimental!!! firmware detection mechanism. No need to rename libs_5xx and libs_6xx, as well as imports.cfg_* : the HBL automatically loads the "correct" one

2010/04/16 (ultimakillz)
- Added new imports.config files for 5.0x, 5.5x, 5.70 and 6.20 (PSP Go) versions (courtesy from ultimakillz)

2010/04/14 (m0skit0)
- Fixed nidtohuman tool Makefile
- Added a very very very simple syscall reestimation function

2010/04/13 (m0skit0)
- Fixed buffer overflow in estimate_syscall() (issue 14)
- Fixed ELF path (issue 13)
- Separate imports.config for 5XX and 6XX. Rename to imports.config
- Added 2 more tools: a processor to make NID table dumps more readable and a user memory dumper (useful for 6.20)

2010/04/12 (m0skit0)
- Added more .lib.stubs sections to imports.config
- Added code to HBL process any number of .lib.stubs defined in imports.config
--> eLoaderconf.rb broken, wololo please fix it
- Added more functions and a lot of order to sdk_loader.S
- Moved get_ent_top() from syscall.c to eloader.c
- Added a new directory with tools for HBL development
- Added NIDSPRX tool by arisma
- Added whole GPLv3 to LICENSE
- This changes should improve syscall estimation :)
--> Need more .nids files for the new libraries

2010/04/10 (wololo)
- All files except h.bin now NEED TO BE PUT in a "hbl" folder at the root of the memstick
-- this can be changed in the code by modifying HBL_ROOT
- simplified the "menu/no menu" choice from a user's perspective (since the menu seems to be not working for lots of people):
-- if a "ms0:/hbl/game/EBOOT.PBP" file exists, it will be launched, otherwise the menu is launched

2010/04/08 (m0skit0)
- Changed imports.config structure (check config.h for new structure)
- Changed config.c library to suit new file structure (also fixed some old bugs)
- Some code cleaning (i.e. moved reestimate_syscall() to syscall.c)
- Disabled return to HBL menu on homebrew exit
- Adjusted MAX_LIBRARY_EXPORTS to 160.

2010/04/07 (wololo)
- small menu improvements
- shortened the delay for free mem ("free memory" step goes faster)
- dblog is deleted everytime the HBL is started

2010/04/06 (wololo)
- adding output to the screen while HBL loads
- putting back my fakemem mechanism so that T.O.M.E runs. To deactivate it, comment out the "define FAKEMEM"
-- that fakemem system is of course temporary until we can actually free the ram

2010/04/04 (wololo)
- Adding a very basic menu
- Put back the exit_with_log if thread not found, feel free to revert that change

2010/04/04 (ab5000)
- Still very unstable (Wagic crashes and T.O.M.E. gives "core() error")
- Added malloc.c and malloc.h
- Now relocate_sections reads the whole section in a dinamically allocated buffer
--> Now relocation is A LOT faster! :)
--> T.O.M.E. takes about 15 seconds to start (debug enabled)
--> Wagic takes about 1 minute to start (debug enabled)
- Now loader allocates HBL stubs using PSP_SMEM_High because homebrew was overwriting them
--> We need to unload Labo and use PSP_SMEM_Low so they'll be BEFORE the homebrew!
--> We also need to make HBL relocatable!
- Added a bit of debug to find_thread
- Added a bit of debug to relocate_sections
- Moved hardcoded scratchpad addresses to scratchpad.h
--> They are still hardcoded in sdk_hbl.S, someone knows how to use a C defines in ASM?

2010/04/03
- Still doesn't work!!!
--> It crashes in sgx-psp-freq-thr thread when freeing memory
--> Threads aren't found (why???) and aren't terminated
--> So when HBL frees semaphores/event flags the thread crashes
- Now AddNID in sdk_hbl.S use $v0 instead of $fp
--> This fixes sceSystemMemoryManager crash in revision 10 :)

2010/04/03
- This revision doesn't work!!!
--> Crash in sceSystemMemoryManager (???) when resolving homebrew imports
- Added sceKernelAllocPartitionMemory
- Now the loader allocates memory for HBL
- Now HBL allocates memory for the homebrew
- Removed hooks from sceKernelAllocPartitionMemory, sceKernelGetBlockHeadAddr and sceKernelGetMaxFreeMemSize
--> Now the homebrew can use the whole free memory :-)
- Fixed a bug in loader.c that was causing crashes
--> In search_game_stubs, stub_list wasn't completely zeroed
- Fixed "CANNOT FIND THREAD TO DELETE" bug (HBL now doesn't exit)
--> We still need a proper fix!

2010/03/31
- Added sceKernelReferEventFlagStatus
- Added sceKernelDeleteEventFlag
--> imports.config and loader are changed, so remember to use the new ones when testing!
- Added find_evflag (thread.c)
- Added event flags, remaining semaphores and UserSbrk address to memory.c
--> Now we have 0x1410400 free bytes (20 Megabytes) :-)

2010/03/19
- Added code for stopping and unloading the game module, but it crashes on sceKernelStopModule

2010/03/13
- Attempt at fixing misaligned relocations

2010/03/07
- Fixed issue with relocated EBOOT.PBP :)

2010/03/07
- fixed the return value of prx_load_program in elf.c.
--> prx loading now works. To try, add BUILD_PRX=1 to the hello/Makefile, and copy eboot.prx to eboot.elf on the root of the memstick
--> Still doesn't work from EBOOT, there's a problem with the offset, needs investigation

04/03/2010
- Added a new config file (file structure on config.h) to move there some game information
- Implemented a library for easy interfacing with config file
- Added code for loader to handle the config file, thus resolving HBL imports that exist on game
- Added code for HBL to handle the config file, thus estimating own imports that the loader couldn't resolve

Last changes:
- Added a loader that loads HBL on scratchpad instead of game region
- HBL loader now also copies stubs used by HBL to scratchpad
- Added a new data structure (tSceLibrary) and subsituted estimate_syscall() with a new implementation