Support PVS visibility changes for entities

[makindotcc]

In lately introduced update two extra bits that represents:

1. is "preserve" or "update" event - we already know that
2. is entity visible in PVS (valve's exact name: m_bVisibleinPVS) - a new thing i discovered now

Part of an decompiled CClientState::ReadPacketEntities:

        *(_DWORD *)updateType = 2;              // delta update
        (*(void (__fastcall **)(_DWORD *, CEntityReadInfo *))(*(_QWORD *)clientState + 1040i64))(
          clientState,
          entityReadInfo);
        *(_DWORD *)(entityReadInfo->qword18 + 12i64) = *(_DWORD *)pNewEntity;
        v95 = (__int64 *)qword_5742A0;
        *(_DWORD *)(entityReadInfo->qword18 + 4 * ((__int64)*(int *)pNewEntity >> 5) + 32) |= 1 << (*(_DWORD *)pNewEntity & 0x1F);
        hasPvsBitSet_1 = (*(_DWORD *)&entityReadInfo->m_UpdateFlags & 0x10) != 0; // m_UpdateFlags is an our "cmd"
        has_vis_bits = *p_has_vis_bits; // has_vis_bits from packet
        if ( v95 )
        {
          v98 = *v95;
          v99 = *(unsigned int *)&entityReadInfo->m_nNewEntity;
          v147 = 0i64;
          entityInstance = 0i64;
          v149 = 0i64;
          v150 = 0i64;
          v151 = 0i64;
          v152 = 0;
          v153 = -1i64;
          if ( (*(unsigned __int8 (__fastcall **)(__int64 *, __int64, __int128 *))(v98 + 208))(v95, v99, &v147) ) // get entity or smth
          {
            visibleInPVS = 1;
            if ( has_vis_bits )
              visibleInPVS = hasPvsBitSet_1;
            *(_BYTE *)(entityInstance + 0x30) = visibleInPVS;// m_bVisibleinPVS
          }
        }
        v101 = entityReadInfo->qword10;

Similar code is in PreserveEnt handler.

Looks like server does not use PVS for entities yet.

Im requesting this as a draft because I dont know what to do with m_bVisibleinPVS yet - there is active field in Entity parser struct, but should we use it as representation of m_bVisibleinPVS? I don't understand why there is additional bit for it when there already is updateFlagLeave and updateFlagDelete.

[codecov]

Codecov Report

❌ Patch coverage is 70.21277% with 14 lines in your changes missing coverage. Please review.
✅ Project coverage is 80.50%. Comparing base (359934e) to head (cba32f0).
⚠️ Report is 190 commits behind head on master.

Files with missing lines	Patch %	Lines
pkg/demoinfocs/sendtables2/entity.go	70.21%	9 Missing and 5 partials ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##           master     #507   +/-   ##
=======================================
  Coverage   80.49%   80.50%           
=======================================
  Files          47       47           
  Lines        6367     6375    +8     
=======================================
+ Hits         5125     5132    +7     
- Misses       1029     1030    +1     
  Partials      213      213           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[makindotcc]

to reviewdog: it was already "complex" 😭

I can "flatten" it or extract handlers to other functions, that was on my mind first, but I thought people in go don't like too many functions.

[markus-wa]

I will take a look when I'm on my PC 🙂

Personally I like having lots of functions in Go (I know this project doesn't look like it as a lot of it is REd and kinda done 1:1) but let's leave it for now until I had a chance to take a proper look 😄

[markus-wa]

PS: reviewdog is more of a guideline than hard rules, so we don't always have to listen to it 😄

[makindotcc]

Oh, I already made this way - inspired by game code.

[makindotcc]

More about m_bVisibleinPVS:

1. it is set to 1 when new entity is created
2. it is set to packet_has_pvs_vis_bits == 0 when entity left (not deleted)

Still no idea why they introduced these extra bits, since there already were soft "entity left".

[spheenik]

@makindotcc

Hmm. So it is set to 1 on creation. Fine.
In case of update, the value from the PVS bits is taken. Also fine.
But in case the entity leaves, what happens exactly?
In case the packet has PVS bits, we'll set it to 1, and 0 otherwise?
Why not on delete? It is a leave and delete after all.

[makindotcc]

In case the packet has PVS bits, we'll set it to 1, and 0 otherwise?

In case the packet has PVS bits, we'll set it to 0 and 1 otherwise. I guess this is done for backwards compatibility.

really raw decompilation of cs2 engine2.dll
ProcessPacketEntities:

// [...]
no_pvs_vis_bits = packetEntities->_impl_.has_pvs_vis_bits_ == 0;
entityReadInfo->has_pvs_vis_bits1 = !no_pvs_vis_bits;
// [...]
ReadPacketEntities(clientStateIGuess, entityReadInfo);
// [...]

ReadPacketEntities:

    // [...]
    p_has_pvs_vis_bits1 = &entityReadInfo->has_pvs_vis_bits1
    // [...]
    p_UpdateFlags = &entityReadInfo->m_UpdateFlags;
    *(_DWORD *)updateType = 3;
    pNewEntity = &entityReadInfo->m_nNewEntity;
    while ( 1 )
    {
      packet_has_pvs_vis_bits = *p_has_pvs_vis_bits1;
      // [...]
      cmd = *(_DWORD *)&entityReadInfo->m_UpdateFlags;
      if ( (cmd & 4) == 0 )
      {
        if ( (cmd & 1) != 0 )
        {
          isNotDelete = (*p_UpdateFlags & 2) == 0;// https://github.com/markus-wa/demoinfocs-golang/blob/cba32f0669f0b9f055d711547ece1ea4f703d9f6/pkg/demoinfocs/sendtables2/entity.go#L578
          *(_DWORD *)updateType = 1;            // leave entity
          if ( isNotDelete )
          {
            if ( qword_5742A0 )
            {
              v71 = (unsigned int)*v9;
              v169 = 0;
              v164 = 0i64;
              entity = 0i64;
              v166 = 0i64;
              v167 = 0i64;
              v168 = 0i64;
              v170 = -1i64;
              if ( (*(unsigned __int8 (__fastcall **)(__int64, __int64, __int128 *))(*(_QWORD *)qword_5742A0 + 208i64))(
                     qword_5742A0,
                     v71,
                     &v164) )                   // get entity or smth
              {
                entity->m_bVisibleinPVS = packet_has_pvs_vis_bits == 0;
              }
            }
          }
          else
          {
            v70 = qword_5742A0;
            *(&entityReadInfo->unsigned_intE0 + ((__int64)*v9 >> 5)) |= 1 << (*v9 & 0x1F);
            memset(v161, 0, sizeof(v161));
            v162 = 0;
            v163 = -1i64;
            if ( v70
              && (*(unsigned __int8 (__fastcall **)(__int64, _QWORD, __int128 *))(*(_QWORD *)v70 + 208i64))(
                   v70,
                   (unsigned int)*v9,
                   v161) )
            {
              *(_DWORD *)(*((_QWORD *)&v161[0] + 1) + 0x30i64) |= 0x8000u; // I didn't reverse this flag, but it hasn't changed in the "arms" update, so I assume this marks entity as deleted
            }
          }
      // [...]