Volshell - A CLI tool for working with memory
Volshell is a utility to access the volatility framework interactively with a specific memory image. It allows for direct introspection and access to all features of the volatility library from within a command line environment.
Volshell is started in much the same way as volatility. Rather than providing a plugin, you just specify the file. If the operating system of the memory image is known, a flag can be provided allowing additional methods for the specific operating system.
$ volshell.py -f <path-to-memory-image> [-w|-m|-l]
The flags to specify a known operating system are -w for windows, -m for mac and -l for linux. Volshell will run through the usual automagic, trying to load the memory image. If no operating system is specified, all automagic will be run.
When volshell starts, it will show the version of volshell, a brief message indicating how to get more help, the current operating system mode for volshell, and the current layer available for use.
Volshell (Volatility 3 Framework) 2.0.2 Readline imported successfully PDB scanning finished Call help() to see available functions Volshell mode : Generic Current Layer : primary Current Symbol Table : None Current Kernel Name : None (primary) >>>
Volshell itself in essentially a plugin, but an interactive one. As such, most values are accessed through self although there is also a context object whenever a context must be provided.
The prompt for the tool will indicate the name of the current layer (which can be accessed as self.current_layer from within the tool).
The generic mode is quite limited, won’t have any symbols loaded and therefore won’t be able to display much information. When an operating system is chosen, the appropriate symbols should be loaded and additional functions become available. The mode cannot easily be changed once the tool has started.
All operating systems come with their equivalent of a process list, aliased to the function ps(). Running this will provide a list of volatility objects, based on the operating system in question. We will use these objects to run our examples against.
We’ll start by creating a process variable, and putting the first result from ps() in it. Since the shell is a python environment, we can do the following:
(layer_name) >>> proc = ps() (layer_name) >>> proc <EPROCESS symbol_table_name1!_EPROCESS: layer_name @ 0xe08ff2459040 #1968>
When printing a volatility structure, various information is output, in this case the type_name, the layer and offset that it’s been constructed on, and the size of the structure.
We can directly access the volatility information about a structure, using the .vol attribute, which contains basic information such as structure size, type_name, and the list of members amongst others. However, volshell has a built-in mechanism for providing more information about a structure, called display_type or dt. This can be given either a type name (which if not prefixed with symbol table name, will use the kernel symbol table identified by the automagic).
(layer_name) >>> dt('_EPROCESS') symbol_table_name1!_EPROCESS (1968 bytes) 0x0 : Pcb symbol_table_name1!_KPROCESS 0x2d8 : ProcessLock symbol_table_name1!_EX_PUSH_LOCK 0x2e0 : RundownProtect symbol_table_name1!_EX_RUNDOWN_REF 0x2e8 : UniqueProcessId symbol_table_name1!pointer ...
It can also be provided with an object and will interpret the data for each in the process:
(layer_name) >>> dt(proc) symbol_table_name1!_EPROCESS (1968 bytes) 0x0 : Pcb symbol_table_name1!_KPROCESS 0xe08ff2459040 0x2d8 : ProcessLock symbol_table_name1!_EX_PUSH_LOCK 0xe08ff2459318 0x2e0 : RundownProtect symbol_table_name1!_EX_RUNDOWN_REF 0xe08ff2459320 0x2e8 : UniqueProcessId symbol_table_name1!pointer 4 ...
These values can be accessed directory as attributes
(layer_name) >>> proc.UniqueProcessId 356
Pointer structures contain the value they point to, but attributes accessed are forwarded to the object they point to. This means that pointers do not need to be explicitly dereferenced to access underling objects.
(layer_name) >>> proc.Pcb.DirectoryTableBase 4355817472
It’s possible to run any plugin by importing it appropriately and passing it to the display_plugin_output or dpo method. In the following example we’ll provide no additional parameters. Volatility will show us which parameters were required:
(layer_name) >>> from volatility3.plugins.windows import pslist (layer_name) >>> display_plugin_output(pslist.PsList) Unable to validate the plugin requirements: ['plugins.Volshell.VH3FSA1JBG0QP9E62Z8OT5UCIMLNYKW4.PsList.kernel']
We can see that it’s made a temporary configuration path for the plugin, and that the kernel requirement was not fulfilled.
We can see all the options that the plugin can accept by access the get_requirements() method of the plugin. This is a classmethod, so can be called on an uninstantiated copy of the plugin.
(layer_name) >>> pslist.PsList.get_requirements() [<ModuleRequirement: kernel>, <BooleanRequirement: physical>, <ListRequirement: pid>, <BooleanRequirement: dump>]
We can provide arguments via the dpo method call:
(layer_name) >>> display_plugin_output(pslist.PsList, kernel = self.config['kernel']) PID PPID ImageFileName Offset(V) Threads Handles SessionId Wow64 CreateTime ExitTime File output 4 0 System 0x8c0bcac87040 143 - N/A False 2021-03-13 17:25:33.000000 N/A Disabled 92 4 Registry 0x8c0bcac5d080 4 - N/A False 2021-03-13 17:25:28.000000 N/A Disabled 356 4 smss.exe 0x8c0bccf8d040 3 - N/A False 2021-03-13 17:25:33.000000 N/A Disabled ...
Here’s we’ve provided the kernel name that was requested by the volshell plugin itself (the generic volshell does not load a kernel module, and instead only has a TranslationLayerRequirement). A different module could be created and provided instead. The context used by the dpo method is always context.
Instead of print the results directly to screen, they can be gathered into a TreeGrid objects for direct access by using the generate_treegrid or gt command.
(layer_name) >>> treegrid = gt(pslist.PsList, kernel = self.config['kernel']) (layer_name) >>> treegrid.populate()
Treegrids must be populated before the data in them can be accessed. This is where the plugin actually runs and produces data.
It might be beneficial to code up a small snippet of code, and execute that on a memory image, rather than writing a full plugin.
The snippet should be lines that will be executed within the volshell context (as such they can immediately access self and context, for example). These can be executed using the run_script or rs command, or by providing the file on the command line with –script.
For example, to load a layer and extract bytes from a particular offset into a new file, the following snippet could be used:
import volatility3.framework.layers.mynewlayer as mynewlayer layer = cc(mynewlayer.MyNewLayer, on_top_of = 'primary', other_parameter = 'important') with open('output.dmp', 'wb') as fp: for i in range(0, 1073741824, 0x1000): data = layer.read(i, 0x1000, pad = True) fp.write(data)
As this demonstrates, all of the python is accessible, as are the volshell built in functions (such as cc which creates a constructable, like a layer or a symbol table).
Files can be loaded as physical layers using the load_file or lf command, which takes a filename or a URI. This will be added to context.layers and can be accessed by the name returned by lf.