Depending on how the type unit was actually set up hierarchically,
we sometimes ended up with the unit set to the wrong actual type
object. Instead, we now simply let the unit parse, and then look up
the entry by the offset specified in the header after all is said
and done. Fixes class/struct objects sometimes winding up resolving
to one of their members instead.
Some problems still remain with resolving variable values for such
types though, needs further investigation.
The DIEs generated by gcc sometimes pass a modified type for
DIEPointerToMember's containing type. As such, we need to resolve
the passed in type down to its base type before verifying that it's
indeed a compound type.
Type units need to be finished before compilation units to ensure that e.g.
modified types have their base types already set by the time we get to
compilation unit attributes.
- Add DwarfFile::_GetContainingCompilationUnit() to request a binary
search for the compilation unit containing the passed in global offset.
- DwarfFile::_ResolveReference(): handle global references by first
attempting to find a containing compilation unit, and then seeing if we
can find an entry for the corresponding relative offset within it.
Get rid of AC_EXPRESSION. DW_FORM_exprloc is for all practical intents
and purposes identical to DW_FORM_block except with a more explicit
intent. As such, simply treat it the same as a block attribute. Gets
value resolution working on DWARF4.
* push event requests on the event queue, handles them.
* activate feature VIRTIO_SCSI_F_CHANGE.
* when a capacity data changed event is received, schedule a rescan of the scsi device node
on the scsi dpc queue. we find out unique child node and call the rescan_node() hook.
* Haiku only handles media changed events on removable devices, so the feature
only actually works when the device is defined as such, aka for QEMU the following option:
-device scsi-hd,drive=hd,removable=true
* the processing of requests in drivers is eased a bit with this change, but
this could be improved for instance by enabling a driver to dequeue items
in a service thread instead of the interrupt handler.
* made a few methods const.
- Refactor ExpressionEvaluationContext::GetCallTarget() and
DwarfFile::_ResolveReference() to use an enum value for the type of
reference being passed in, as we now need to support signature
references in addition to local/global. Adjust callers.
- Factor out an _ParseDebugInfoSection() and add a corresponding
_ParseTypesSection() for the top level loops that handle .debug_info and
.debug_types respectively.
- Refactor _ParseDebugInfoEntry(), _ParseEntryAttributes() and
_ResolveReference to take a pointer to the relevant BaseUnit in order
to allow sharing between parsing of the types and info sections.
- Implement support for DW_FORM_ref_sig8. In order to handle this, we
use a hash table that maps from the corresponding 64-bit DW_AT_signature
values to their respective type units.
- Adjust debug output to more closely match offsets and values used by
objdump for ease of troubleshooting parsing issues.
Combined, these changes allow us to load DWARF 4 binaries sufficiently
to get source information and unwind call frame, though support is still
incomplete. In particular, global references aren't yet implemented,
and neither are location expression attributes, which are now used
quite heavily, and consequently we can't yet resolve variable values
and a few other things properly.
- Pull common base class BaseUnit out of CompilationUnit and adjust the
latter to inherit.
- Add TypeUnit to represent the top level units for .debug_types.
This form implicitly indicates a flag value of true without a
corresponding byte in the datastream. As such, we were introducing off
by one errors when parsing one.
- Add hook to FunctionDebugInfo to determine if the represented function
is the main function of the program. Adjust
{Basic,Dwarf}FunctionDebugInfo accordingly.
- {Team,Image}DebugInfo: When loading, see if the current image contains
the main function. If so, cache it for later retrieval.
It is possible, through the right combination of steps, for this method to
be run on startup before the view is attached to the window causing a crash
because Window() returns NULL. This fix returns false in this case instead.
Later, when an application such as Tracker is started and added to Deskbar
this method will be run again with the view attached to the window.
Rene Gollent