You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Some clients may need the full path to each application module/library.
One use case is post-processing for symbols. And once we have issue #44 ,
DR itself will probably want to re-organize the module name struct
and priority and make the path a separate field rather than one in a series
of choices for name. The name part of the path should still be a choice
for name, to be consistent, but not the whole path, which should be
available separately.
On Linux with HAVE_PROC_MAPS we have the path available. On Windows we
only set if at_map && -track_module_filenames due to concerns over
walking the LDR_MODULE list and pulling out LDR_MODULE.FullDllName: we only
use it for debugging. We'll have to re-evaluate that.
on Windows we can do what GetMappedFileName does: call NtQueryVirtualMemory with
MemorySectionName which returns the file name backing the section.
we should use that instead of -track_module_filenames, which relies
on seeing the CreateSection and on it being right before the map
(and currently only saves the short name).
From [email protected] on May 01, 2009 09:24:41
Some clients may need the full path to each application module/library.
One use case is post-processing for symbols. And once we have issue #44 ,
DR itself will probably want to re-organize the module name struct
and priority and make the path a separate field rather than one in a series
of choices for name. The name part of the path should still be a choice
for name, to be consistent, but not the whole path, which should be
available separately.
On Linux with HAVE_PROC_MAPS we have the path available. On Windows we
only set if at_map && -track_module_filenames due to concerns over
walking the LDR_MODULE list and pulling out LDR_MODULE.FullDllName: we only
use it for debugging. We'll have to re-evaluate that.
Original issue: http://code.google.com/p/dynamorio/issues/detail?id=138
The text was updated successfully, but these errors were encountered: