File diff 83c97a002998 → ea9a59c551f8
www/conservancy/static/linux-compliance/vmware-lawsuit-faq.html
Show inline comments
...
 
@@ -180,7 +180,6 @@ Code, and for which (at least some) source code is provided.
 
    whole was developed by modifying Linux code in tandem with
 
    modifications to &ldquo;vmkernel&rdquo; in a tightly coupled manner.</p>
 
    </dd>
 

	
 
   <dt id="shim-meaningless">Is Conservancy proposing a &ldquo;shim
 
      layer&rdquo; as a viable solution for GPL compliance?</dt>
 

	
...
 
@@ -190,7 +189,7 @@ Code, and for which (at least some) source code is provided.
 
        technological manipulation that changes the outcome of a
 
        combined/derivative work analysis.</dd>
 

	
 
    <dt>Can you give a <em>specific</em> example, with code, showing how
 
    <dt id="example">Can you give a <em>specific</em> example, with code, showing how
 
    VMware combined Linux source code with their binary-only components?</dt>
 

	
 
     <dd><p>There are numerous examples available that show this.  The
...
 
@@ -213,27 +212,39 @@ Code, and for which (at least some) source code is provided.
 
       Hellwig's copyrights; VMware incorporated Linux code copyrighted by
 
       many others as well into their kernel.</p>
 

	
 
       <h4>Example of &ldquo;vmkernel&rdquo;'s combination with Linux code</h4>
 
       <p>As one example, examine the file
 
       <h3>Example of &ldquo;vmkernel&rdquo;'s combination with Linux code</h3>
 
       <p>Our example begins with examination of the file
 
           called <code>vmkdrivers/src_92/vmklinux_92/vmware/linux_pci.c</code>,
 
           which can be found in the &ldquo;Open Source&rdquo; release for
 
           ESXi 5.5.0 Update 2 (5.5U2).  A small excerpt from that file, found in the
 
           function <code>LinuxPCIDeviceRemoved()</code>, reads as follows:</p>
 

	
 
<pre>
 
if (unlikely(
 
  /* NOTE: vmk_PCIGetDeviceName is defined in vmvisor64-vmkernel */
 
  vmk_PCIGetDeviceName(vmkDev, vmkDevName, sizeof(vmkDevName)-1) != VMK_OK))
 
#include <linux/pci.h>
 
[...]
 
/*
 
 * This function [...] is modelled after pci_remove_device, the function which would
 
 * be called in a linux system.
 
 */
 
static void
 
LinuxPCIDeviceRemoved(vmk_PCIDevice vmkDev)
 
{
 
    vmkDevName[0] = 0;
 
}
 
   LinuxPCIDevExt *pciDevExt;
 
   struct pci_dev *linuxDev;
 
[...]
 
  if (unlikely(
 
    vmk_PCIGetDeviceName(vmkDev, vmkDevName, sizeof(vmkDevName)-1) != VMK_OK))
 
  {
 
      vmkDevName[0] = 0;
 
  }
 
[...]
 
/* VMKAPI_MODULE_CALL_VOID is a macro calling driver's remove() here */
 
VMKAPI_MODULE_CALL_VOID(pciDevExt->moduleID,
 
                        linuxDev->driver->remove,
 
                        linuxDev);
 
</pre>        
 

	
 
<h4>Combination of &ldquo;vmkernel&rdquo; code with &ldquo;vmkdrivers&rdquo;</h4>
 

	
 
<p>The function, <code>vmk_PCIGetDeviceName()</code> must be defined, with an
 
      implementation, for this code above to work, or even compile.
 
      Inside <code>BLD/build/HEADERS/vmkapi-current-all-public/vmkernel64/release/device/vmkapi_pci_incompat.h</code>,
...
 
@@ -286,28 +297,131 @@ VMKAPI_MODULE_CALL_VOID(pciDevExt->moduleID,
 
<p>&hellip; which indicated these binary file contains the function body
 
for  <code>vmk_PCIGetDeviceName</code>.</p>
 

	
 
<p>Finally, after detailed searching, Conservancy found no evidence that any
 
<p>Furthermore, after detailed searching, Conservancy found no evidence that any
 
  other code (other than modified Linux code) makes calls
 
  to <code>vmk_PCIGetDeviceName</code>.  This provides a strong indication
 
  that this function's primary purpose is to combine Linux code with
 
  &ldquo;vmkernel&rdquo;.  Conservancy also found other functions where similar analysis
 
  yields similar results as above.</p>
 

	
 
<h4>Linux's <code>struct pci</code> combined with <code>LinuxPCIDeviceRemoved()</code></h4>
 

	
 
<p>Having established the direct and close combination
 
  of <code>vmk_PCIGetDeviceName</code>
 
  and <code>LinuxPCIDeviceRemoved()</code>,  focus now on the
 
  quoted code from <code>LinuxPCIDeviceRemoved()</code>.  That code, note
 
  that one of the local variables is <code>struct pci_dev *linuxDev;</code>.
 
  A definition of <code>pci_dev</code> is found in
 
  <code>vmkdrivers/src_92/include/linux/pci.h</code> (which
 
  is <code>#include</code>'d above) reads:
 
  <pre>
 
struct pci_dev {
 
[...]
 
       struct pci_driver *driver;      /* which driver has allocated this device */
 
[...]
 
truct pci_driver {
 
        char *name;
 
[...]
 
        void (*remove) (struct pci_dev *dev);   /* Device removed (NULL if not a hot-plug capable driver) */
 
[...]
 
#if defined(__VMKLNX__)
 
        /* 2008: Update from Linux source */
 
        u8              revision;       /* PCI revision, low byte of class word */
 
#endif /* defined(__VMKLNX__) */
 
  };
 
</pre></p>
 

	
 
<p>These structures, and based on those from Linux itself
 
  (<a href="http://lxr.free-electrons.com/source/include/linux/pci.h?v=2.6.24">a
 
    similar version of this file can be seen in Linux 2.6.24</a>), and as can
 
  be seen above, have been modified to work with &ldquo;vmkernel&rdquo;</p>
 

	
 
<p>In <code>LinuxPCIDeviceRemoved()</code>, we saw a macro called with a
 
  variable, <code>linuxDev</code> which was of type <code>struct pci</code>.
 
  Thus, the combination of code from Linux's <code>pci.h</code>
 
  and VMware's <code>vmware/linux_pci.c</code> is very tightly coupled and
 
  interdependent.</p>
 

	
 
<h4><code>VMKAPI_MODULE_CALL_VOID</code> macro calls driver's code</code></h4>
 

	
 
<p>The
 
  file <code>BLD/build/HEADERS/vmkapi-current-all-public/vmkernel64/release/base/vmkapi_module.h</code>
 
  contains the macro definition of  <code>VMKAPI_MODULE_CALL_VOID</code>,
 
  which is quoted below (with debug lines removed):
 
<pre>
 
#define VMKAPI_MODULE_CALL_VOID(moduleID, function, args...)  \
 
do {                                                    \
 
    vmk_ModInfoStack modStack;                          \
 
    vmk_ModulePushId(moduleID, function, &modStack);    \
 
    (function)(args);                                   \
 
    )                                                   \
 
    vmk_ModulePopId();                                  \
 
} while(0)
 
</pre>
 

	
 
<p>When the macro is expanded, it means that <code>(function)(args)</code> is
 
  actually expanded to <code>linuxDev->driver->remove(linuxDev)</code>.
 
  Therefore, we see <code>LinuxPCIDeviceRemoved()</code>, makes directs calls
 
  to a driver's remove() function, by combining with Linux's <code>struct
 
  pci</code>, and by VMware's introduction of this new calling code.
 
  Conservancy has confirmed many drivers from Linux are incorporated via
 
  these mechanisms; one specific example is discussed next.</p>
 

	
 
<h4>Combination of the tg3 driver with &ldquo;vmkernel&rdquo;</h4>
 

	
 
<p>VMware includes a file <code>vmkdrivers/src_9/drivers/net/tg3/tg3.c</code>
 
  in their source release.  This file appears to be Linux's tg3 driver.  It
 
  includes a definition of the <code>struct pci_dev</code> for this device,
 
  which reads:
 
<pre>
 
static struct pci_driver tg3_driver = {
 
[...]
 
        .remove         = __devexit_p(tg3_remove_one),
 
</pre>
 
</p>
 

	
 
<p>Therefore, when the code in <code>LinuxPCIDeviceRemoved()</code>
 
  calls <code>linuxDev->driver->remove(linuxDev)</code>, the code ultimately
 
  called (in the case where a tg3 card is driven by the kernel)
 
  is <code>tg3_remove_one()</code>, which is found in <code>tg3.c</code> and
 
  comes directly from Linux.</p>
 

	
 
<p>(Note: <code>__devexit_p</code> is a straightforward macro found
 
  in <code>vmkdrivers/src_92/include/linux/init.h</code> (which also comes
 
  from Linux) that will simply expand to its first argument in this
 
  case.)</p>
 

	
 
<h4>VMware distribution of binary version of <code>tg3.c</code></h4>
 

	
 
<p>VMware furthermore distributes a modified version of <code>tg.c</code> in
 
  binary form.  This can be found in <code>usr/lib/vmware/vmkmod/tg3</code>,
 
  which is extracted by un-vmtar'ing the file <code>net_tg3.v00</code> (found
 
  on the ESXi 5.5U2 installer ISO image).  Conservancy has confirmed that
 
  file is a compiled version of <code>tg3.c</code></p>
 

	
 
<h4>Conclusions</h4>
 

	
 
<p>Given this evidence and related contextual clues, the only logical
 
  conclusions are:</p>
 
    <ul><li><code>vmklinux_9</code>, as a binary object, dynamically links
 
 with <code>k.b00</code>, another binary object, to form a single running
 
 binary.</li>
 
      <li>That binary contains code licensed under the GPLv2, and can be
 
       distributed in binary form only under permissions provided under
 
        GPLv2 &mdash; in particular <a href="https://gnu.org/licenses/gpl-2.0.html#section2">GPLv2&sect;2</a> and <a href="https://gnu.org/licenses/gpl-2.0.html#section3">GPLv2&sect;3</a>.</li>
 
    <ul><li><code>vmklinux_9</code>, a binary object, dynamically links with
 
        the binary objects: <code>k.b00</code> and <code>tg3</code> (the
 
        driver built from <code>tg3.c</code>'s source).  These three binary
 
        objects together form a single running binary (likely along with many
 
        other binary objects as well).</li>
 
      <li>That single running binary contains code licensed under the GPLv2
 
       &mdash; namely the code derived from <code>tg3.c</code>
 
       and <code>pci.h</code>.  Thus, the single running binary may be
 
       distributed in binary form only under permissions provided under GPLv2
 
       &mdash; in
 
       particular <a href="https://gnu.org/licenses/gpl-2.0.html#section2">GPLv2&sect;2</a>
 
       and <a href="https://gnu.org/licenses/gpl-2.0.html#section3">GPLv2&sect;3</a>.</li>
 
      <li>GPLv2&sect;3(a&ndash;b) requires that <q>complete corresponding
 
          machine-readable source code</q> must accompany binary
 
          distributions such as these.  GPLv2&sect;3 further states
 
          that <q>for an executable work, complete source code means all the
 
          source code for all modules it contains</q>.</li>
 
      <li>The binary work in question contains modules from <code>k.b00</code> and
 
        <code>vmlinux_9</code>.</li>
 
      <li>The binary work in question contains modules from <code>k.b00</code>,
 
        <code>vmlinux_9</code> and <code>tg3</code>.</li>
 
      <li>VMware did not provide source code for any modules found in
 
        <code>k.b00</code>.</li>
 
      <li>Therefore, VMware failed to comply with the GPLv2, as such
...
 
@@ -315,7 +429,7 @@ for  <code>vmk_PCIGetDeviceName</code>.</p>
 
        in <code>k.b00</code>.</li>
 
    </ul>
 
<p>The above is but one piece of evidence among many, but hopefully it helps
 
  to explain the types of &ldquo;combined work&rdquo; violations found in
 
  to explain some of the &ldquo;combined work&rdquo; violations found in
 
  VMware's ESXi product.</p>
 

	
 
<dt id="verify">How can I verify Conservancy's technical findings above?</dt>