Recherche avancée

Sur d’autres sites (11151)

• Dreamcast Finds

  15 avril 2022, par Multimedia Mike — Sega Dreamcast

  Pursuant to my recent post about finally understanding how Sega Dreamcast GD-ROM rips are structured, I was able to prepare the contents of various demo discs in a manner that makes exploration easy via the Internet Archive. This is due to the way that IA makes it easy to browse archives such as ZIP or ISO files (anything that 7zip knows how to unpack), and also presents the audio tracks for native playback directly through the web browser.

  These are some of the interesting things I have found while perusing the various Dreamcast sampler discs.

  Multimedia Formats
  First and foremost : Multimedia-wise, SFD and ADX files abound on all the discs. SFD files are Sofdec, a middleware format used for a lot of FMV on Dreamcast games. These were little more than MPEG video files with a non-MPEG (ADPCM instead) audio codec. VLC will usually play the video portions of these files but has trouble detecting the audio. It’s not for lack of audio codec support because it can play the ADX files just fine.
  

  It should be noted that Dreamcast Magazine Disc 11 has an actual .mpg file (as opposed to a .sfd file) that has proper MPEG audio instead instead of ADX ADPCM.

  The only other multimedia format I know of that was used in any Dreamcast games was 4XM, used on Alone In The Dark : The New Nightmare. I wrote a simple C tool a long time to recover these files from a disc image I extracted myself. Rather than interpreting the ISO-9660 filesystem, the tool just crawled through the binary blob searching for ‘4XMV’ file signatures and using length data within the files for extraction.

  Also, there are plentiful PVR files (in reference to the PowerVR2 GPU hardware that the DC uses) which ‘file’ dutifully identifies as “Sega PVR image”. There are probably tools to view them. It doesn’t appear to be a complicated format.

  Scripting
  I was fascinated to see Lua files on at least one of the discs. It turns out that MDK 2 leverages the language, as several other games do. But it was still interesting to see the .lua files show up in the Dreamcast version as well.

  That Windows CE Logo
  Every Sega Dreamcast is famously emblazoned with a logo mentioning Microsoft Windows CE :

  
  
  

  It has confused many folks. It also confused me until this exploratory exercise. Many would wonder if the Dreamcast booted up into some Windows CE OS environment that then ran the game, but that certainly wasn’t it. Indeed, Dreamcast was one of the last consoles that really didn’t have any kind of hypervisor operating system managing everything.

  I found a file called rt2dc.exe on one sampler disc. At first, I suspected that this was a development utility for Windows to convert some “RT” graphical format into a format more suitable for the Dreamcast. Then, ‘file’ told me that it was actually a Windows EXE but compiled for the Hitachi SH-4 CPU (the brain inside the DC). Does the conversion utility run on the Dreamcast itself ? Then I analyzed the strings inside the binary and saw references to train stations. That’s when it started to click for me that this was the binary executable for the demo version of Railroad Tycoon 2 : Gold Edition, hence “rt2dc.exe”. Still, this provides some insight about whether Dreamcast “runs” Windows. This binary was built against a series of Windows CE libraries. The symbols also imply DirectX compatibility.

  Here is a page with more info about the WinCE/DirectX variant for the Sega Dreamcast. It seems that this was useful for closing the gap between PC and DC ports of games (i.e., being able to re-use more code between the 2 platforms). I guess this was part of what made Dreamcast a dry run for the DirectXbox (later Xbox).

  Here is a list of all the Dreamcast games that are known to use Windows CE.

  Suddenly, I am curious if tools such as IDA Pro or Ghidra can possibly open up Windows CE binaries that contain SH-4 code. Not that I’m particularly interested in reverse engineering any algorithms locked up in Dreamcast land.

  Tomb Raider Easter Egg
  The volume 6 sampler disc has a demo of Tomb Raider : The Last Revelation. While inspecting the strings, I found an Easter egg. I was far from the first person to discover it, though, as seen on this The Cutting Room Floor wiki page (look under “Developer Message”). It looks like I am the first person to notice it on the Dreamcast version. It shows up at offset 0xE3978 in the Dreamcast (demo version) binary, if anyone with permissions wants to update the page.

  Web Browser
  Then there’s the Web Browser for Sega Dreamcast. It seemed to be included on a lot of these sampler discs. But only mentioning the web browser undersells it– the thing also bundled an email client and an IRC client. It’s important to remember that the Dreamcast also had a keyboard peripheral.

  I need to check the timeline for when the web browser first became available vs. when the MIL-CD hack became known. My thinking is that there is no way that the web browser program didn’t have some security issues– buffer overflows and the like. It seems like this would have been a good method of breaking the security of the system.

  Ironically, I suddenly can think of a reason why one might want to use advanced reverse engineering tools on Dreamcast binaries, something I struggled with just a few paragraphs ago.

  Odds ‘n Ends
  It’s always fun to find plain text files among video game assets and speculating on the precise meaning… while also marveling how long people have been struggling to correctly spell “length”.

  Internationalization via plain text files.

  Another game (Slave Zero) saw fit to zip its assets. Maybe this was to save space in order to fit everything on the magazine sampler disc. Quizzically, this didn’t really save an appreciable amount of space.

  Finally, all the discs have an audio track 2 that advises that the disc must be played in a Dreamcast console. Not unusual. However, volume 4 also has a Japanese lady saying the same thing on track 4. This is odd because track 4 is one of the GD area audio tracks and is not accessible with normal CD hardware. Further, she identifies the disc as a “Windows CE disc”.

  The post Dreamcast Finds first appeared on Breaking Eggs And Making Omelettes.

• Anomalie #4516 (Nouveau) : Fatal error en php 8.0

  9 juillet 2020, par Franck D

  Hello, :)

  Windows 10 (1909)
  Firefox 78.0.2
  Easyphp
  Apache 2.4.43 x64
  PHP 8.0.0 alpha1 x64
  MySQL 8.0.20 x64
  PhpMyAdmin 5.0.2

  memory_limit 512M
  post_max_size 130M
  upload_max_filesize 64M
  max_execution_time 300
  max_input_time -1

  Comme php 8.0 sort le 26 novembre, https://wiki.php.net/todo/php80 et que spip 3.4 ne sortira sans doute pas avant 2022... Je commence à faire des tests avec php 8 (plus c’est tôt mieux c’est) histoire de voir si cela posera de gros problèmes avec spip 3.3 !

  Si je lance spip_loader version 3.0.9, j’ai une page blanche avec :
  Fatal error : Uncaught Error : Call to undefined function get_magic_quotes_gpc() in C :\Program Files (x86)\EasyPHP-Devserver-17\eds-www\test2\config\ecran_securite.php:350 Stack trace : #0 C :\Program Files (x86)\EasyPHP-Devserver-17\eds-www\test2\ecrire\inc_version.php(130) : include() #1 C :\Program Files (x86)\EasyPHP-Devserver-17\eds-www\test2\spip.php(17) : include_once(’C :\Program File...’) #2 C :\Program Files (x86)\EasyPHP-Devserver-17\eds-www\test2\index.php(3) : include(’C :\Program File...’) #3 main thrown in C :\Program Files (x86)\EasyPHP-Devserver-17\eds-www\test2\config\ecran_securite.php on line 350

  Je monte la priorité à "haut" pour voir, si cela ne serait pas possible pour spip 3.3.0, mais c’est discutable...

• FFMPEG fatal signal 11 (SIGSEGV) on Android 5.0.2

  14 juillet 2015, par Jerikc XIONG

  I use the idol347(Android 5.0.2) to broadcast streaming, but I got the following error :

  07-14 15:16:20.484 F/libc    (26905): invalid address or address of corrupt block 0x7490b06f passed to dlfree
  
  07-14 15:16:20.485 F/libc    (26905): Fatal signal 11 (SIGSEGV), code 1, fault addr 0xdeadbaad in tid 27282 (FFmpeg)
  
  07-14 15:16:20.567 E/mm-camera(  342): cpp_hardware_process_frame:1105, v4l2 ioctl() failed. rc:-1, trans_code:-11, frame_id=217, identity=0x10002
  
  07-14 15:16:20.587 I/DEBUG   (  292): *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** ***
  
  07-14 15:16:20.587 I/DEBUG   (  292): Build fingerprint: 'TCL/6039K/idol347:5.0.2/LRX22G/v1AEY-0:user/release-keys'
  
  07-14 15:16:20.587 I/DEBUG   (  292): Revision: '0'
  
  07-14 15:16:20.587 I/DEBUG   (  292): ABI: 'arm' 
  
  07-14 15:16:20.587 I/DEBUG   (  292): pid: 26905, tid: 27282, name: FFmpeg  >>> com.streaming.camera.demo <<<
  
  07-14 15:16:20.587 I/DEBUG   (  292): signal 11 (SIGSEGV), code 1 (SEGV_MAPERR), fault addr 0xdeadbaad
  
  07-14 15:16:20.601 I/DEBUG   (  292): Abort message: 'invalid address or address of corrupt block 0x7490b06f passed to dlfree'
  
  07-14 15:16:20.601 I/DEBUG   (  292):     r0 00000000  r1 b6f40dec  r2 deadbaad  r3 00000000
  
  07-14 15:16:20.601 I/DEBUG   (  292):     r4 7490b06f  r5 b6f420f4  r6 a33d4000  r7 7490b077
  
  07-14 15:16:20.601 I/DEBUG   (  292):     r8 7490b070  r9 00000000  sl 00000000  fp 12e9c140
  
  07-14 15:16:20.601 I/DEBUG   (  292):     ip 00000000  sp a4bb6970  lr b6f126b7  pc b6f126b8  cpsr 600f0030
  
  07-14 15:16:20.602 I/DEBUG   (  292): 
  
  07-14 15:16:20.602 I/DEBUG   (  292): backtrace:
  
  07-14 15:16:20.602 I/DEBUG   (  292):     #00 pc 000286b8  /system/lib/libc.so (dlfree+1239)
  
  07-14 15:16:20.602 I/DEBUG   (  292):     #01 pc 0000ef2b  /system/lib/libc.so (free+10)
  
  07-14 15:16:20.602 I/DEBUG   (  292):     #02 pc 000234bf  /data/app/com.streaming.camera.demo-1/lib/arm/ffmpegbridge.so (write_packet+146)
  
  07-14 15:16:20.602 I/DEBUG   (  292):     #03 pc 000b2b87  /data/dalvik-cache/arm/data@app@com.streaming.camera.demo-1@base.apk@classes.dex

  The related code snippet as following :

  write_packet(FFmpegBridgeContext *br_ctx, uint8_t *data, int data_size, long pts,
  
  int is_video, int is_video_keyframe) {
  
    AVPacket *packet;
  
    AVStream *st;
  
    AVCodecContext *c;
  
    uint8_t *filtered_data = NULL, *keyframe_data = NULL;
  
  
  
    packet = av_malloc(sizeof(AVPacket));
  
    if (!packet) {
  
      LOGE("ERROR: write_packet couldn't allocate memory for the AVPacket");
  
    }
  
    _init_packet(packet);
  
    if (is_video) {
  
      packet->stream_index = br_ctx->video_stream_index;
  
      if (is_video_keyframe) {
  
        packet->flags |= AV_PKT_FLAG_KEY;
  
      }
  
    } else {
  
      packet->stream_index = br_ctx->audio_stream_index;
  
    }
  
  
  
    packet->size = data_size;
  
    packet->pts = packet->dts = pts;
  
    packet->data = data;
  
    st = br_ctx->output_fmt_ctx->streams[packet->stream_index];
  
    c = st->codec;
  
    // filter the packet (if necessary)
  
    filtered_data = _filter_packet(br_ctx, st, packet);
  
  
  
    // rescale the timing information for the packet
  
    _rescale_packet(br_ctx, st, packet);
  
  
  
    // write the frame
  
    _write_packet(br_ctx, packet);
  
  
  
    // clean up
  
    if (filtered_data) {
  
      av_free(filtered_data);
  
    }
  
    if (keyframe_data) {
  
      av_free(keyframe_data);
  
    }
  
    av_free_packet(packet);
  
  }

  After a struggle, I found it crashes because of here :

  av_free(filtered_data);

  BTW, it works fine on Android 4.3/4.4.

  How to solve it ?