Skip to content
Commits on Jun 30, 2016
  1. @jrfonseca

    gltrace: std::free -> free.

    jrfonseca committed Jun 30, 2016
  2. @jrfonseca

    travis: Fix labels.

    jrfonseca committed Jun 30, 2016
  3. @jrfonseca

    glretrace: Sync after creating X window.

    Otherwise SwiftShader EGL crashes.
    jrfonseca committed Jun 30, 2016
  4. @jrfonseca
  5. @jrfonseca
  6. @jrfonseca

    egltrace: Don't intercept dlopen calls done by EGL/GL/GLES implementa…

    …tion.
    
    Certain EGL implementations do this.
    jrfonseca committed Jun 30, 2016
  7. @jrfonseca

    glretrace: Consistently use 32 bits visuals.

    Different APIs interpret RGBA/DS bits differently: some return a visual
    with the smallest number of bits, others return the largest number of
    bits.
    
    All modern GL implementations should support 32bits visuals.
    
    So use them for sake of consistency.
    jrfonseca committed Jun 30, 2016
  8. @jrfonseca
  9. @jrfonseca
  10. @jrfonseca

    dxgitrace: Never attempt to delete a wrapper.

    It's not safe, as we're not sure exactly which wrapper class we have.
    
    This fixes the "unexpected virtual method" errors introduced with
    earlier changes.
    jrfonseca committed Jun 30, 2016
  11. @jrfonseca

    trace: Add more COM debugging.

    To try to diagnose regression introduced by ealier commits, whereby
    map_buffer randomly fail, but only on special circunstances.
    jrfonseca committed Jun 30, 2016
Commits on Jun 29, 2016
  1. @jrfonseca

    guids: Add ID3D11DebugTest.

    jrfonseca committed Jun 24, 2016
Commits on Jun 28, 2016
  1. @jrfonseca

    d3dretrace: Don't presume that COM interface pointers are freed when …

    …refcount reaches zero.
    
    This is the retrace counterpart of the previous commit.
    
    #462
    jrfonseca committed Jun 28, 2016
  2. @jrfonseca

    trace: Don't presume that COM interface pointers are freed when refco…

    …unt reaches zero.
    
    They often aren't.  And worst of all, some apps rely on them sticking
    around.
    
    With this change, the destruction of wrapper objects is effectively
    deferred until a different interface is created on the same pointer
    address, or more likely leaked.  But this is unavoidable, given there's
    no generic mechanism to know for sure when the wrapped object is freed.
    
    Fixes #462
    jrfonseca committed Jun 28, 2016
Commits on Jun 20, 2016
  1. @jrfonseca

    khronos: Remove unnecessary Android patch.

    Android support has been included upstream.
    jrfonseca committed Jun 20, 2016
  2. @jrfonseca
Commits on Jun 7, 2016
  1. @nhaehnle @jrfonseca

    glretrace: Handle implicit buffer unmap on glDeleteBuffer.

    v2: Check for DSA support (Jose Fonseca)
    
    Fixes #450
    Fixes #460
    nhaehnle committed with jrfonseca Jun 7, 2016
  2. @jrfonseca
  3. @nhaehnle @jrfonseca
Commits on May 19, 2016
  1. @jrfonseca

    dxgiretrace: Handle ID3D11VideoContext properly.

    Essentially do the same thing done for DXVA's IDirectXVideoDecoder interface.
    jrfonseca committed May 19, 2016
  2. @jrfonseca

    dxgitrace: Handle ID3D11VideoContext::Get/ReleaseDecoderBuffer properly.

    Essentially do the same thing done for DXVA's
    IDirectXVideoDecoder::Get/ReleaseBuffer methods.
    
    Untested, but it should fare no worse than it does now.
    jrfonseca committed May 19, 2016
Commits on May 18, 2016
  1. @jrfonseca
  2. @jrfonseca

    specs: Add EnumFlags helper.

    For enums with flags.
    jrfonseca committed May 18, 2016
  3. @jrfonseca
  4. @jrfonseca
  5. @jrfonseca
  6. @jrfonseca
Commits on May 16, 2016
  1. @jrfonseca

    d3dretrace: Ensure DXVA2_DecodeExecuteParams::pExtensionData is alloc…

    …ated.
    
    The ValueAllocator/ValueDeserializer split needs to be rethought, as it
    is not a good fit: there are several cases where allocations happen
    besides the root value.
    jrfonseca committed May 16, 2016
  2. @jrfonseca
Commits on May 14, 2016
  1. @jrfonseca
Commits on May 13, 2016
  1. @jrfonseca
  2. @jrfonseca
  3. @jrfonseca

    d3d9trace: Capture IDirectXVideoDecoder buffer mappings.

    By emitting fake memcpys.
    jrfonseca committed May 13, 2016
  4. @jrfonseca
Commits on May 11, 2016
  1. @jrfonseca

    memtrace: Align blocks to 64 bytes.

    WARP doesn't align memory mapping to pages, causing tests to fail, as
    blocks get different offsets.
    jrfonseca committed May 11, 2016
Something went wrong with that request. Please try again.