`pagewalk` refactoring (#3317)

* refactoring pagewalk + printing pagetable idxes

* added idx mask

* docs

* added caching

* idxlen

* linting

* improved error handling

docs/commands/kernel/kbase.md

@@ -2,7 +2,7 @@
 # kbase
 
 ```text
-usage: kbase [-h] [-r]
+usage: kbase [-h] [-r] [-v]
 
 ```
 
@@ -13,6 +13,7 @@ Finds the kernel virtual base address.
 | :--- | :--- | :--- |
 |-h|--help|show this help message and exit|
 |-r|--rebase|rebase loaded symbol file|
+|-v|--verbose|show more information relevant to the kbase (e.g. phys addr)|
 
 <!-- END OF AUTOGENERATED PART. Do not modify this line or the line below, they mark the end of the auto-generated part of the file. If you want to extend the documentation in a way which cannot easily be done by adding to the command help description, write below the following line. -->
 <!-- ------------\>8---- ----\>8---- ----\>8------------ -->

pwndbg/aglib/kernel/__init__.py

@@ -28,6 +28,7 @@ import pwndbg.lib.kernel.structs
 import pwndbg.lib.memory
 import pwndbg.search
 from pwndbg.aglib.kernel.paging import ArchPagingInfo
+from pwndbg.aglib.kernel.paging import PageTableLevel
 from pwndbg.lib.regs import BitFlags
 
 _kconfig: pwndbg.lib.kernel.kconfig.Kconfig | None = None
@@ -366,12 +367,12 @@ class x86_64Ops(x86Ops):
         return pwndbg.dbg.selected_inferior().create_value(per_cpu_addr)
 
     def virt_to_phys(self, virt: int) -> int:
-        if pwndbg.aglib.memory.is_kernel(virt) and virt < arch_paginginfo().vmalloc:
-            return virt - self.page_offset
-        res = pagewalk(virt)[0][1]
-        if res is None:
+        if not (pwndbg.aglib.memory.is_kernel(virt) and virt < arch_paginginfo().vmalloc):
+            # if not within physmap range, first find the physmap address
+            virt = pagewalk(virt)[0].virt
+        if virt is None:
             return None
-        return res - self.page_offset
+        return virt - self.page_offset
 
     def pfn_to_page(self, pfn: int) -> int:
         # assumption: SPARSEMEM_VMEMMAP memory model used
@@ -405,12 +406,12 @@ class Aarch64Ops(ArchOps):
         return pwndbg.dbg.selected_inferior().create_value(per_cpu_addr)
 
     def virt_to_phys(self, virt: int) -> int:
-        if pwndbg.aglib.memory.is_kernel(virt) and virt < arch_paginginfo().vmalloc:
-            return virt - self.page_offset + self.phys_offset
-        res = pagewalk(virt)[0][1]
-        if res is None:
+        if not (pwndbg.aglib.memory.is_kernel(virt) and virt < arch_paginginfo().vmalloc):
+            # if not within physmap range, first find the physmap address
+            virt = pagewalk(virt)[0].virt
+        if virt is None:
             return None
-        return res - self.page_offset
+        return virt - self.page_offset + self.phys_offset
 
     def phys_to_virt(self, phys: int) -> int:
         # https://elixir.bootlin.com/linux/v6.16.4/source/arch/arm64/include/asm/memory.h#L356
@@ -595,7 +596,8 @@ def kbase() -> int | None:
         raise NotImplementedError()
 
 
-def pagewalk(addr, entry=None):
+@pwndbg.lib.cache.cache_until("stop")
+def pagewalk(addr, entry=None) -> Tuple[PageTableLevel, ...]:
     pi = arch_paginginfo()
     if pi:
         return pi.pagewalk(addr, entry)

pwndbg/aglib/kernel/paging.py

@@ -2,8 +2,8 @@ from __future__ import annotations
 
 import math
 import re
+from dataclasses import dataclass
 from typing import Dict
-from typing import List
 from typing import Tuple
 
 import pwndbg
@@ -31,6 +31,14 @@ def first_kernel_page_start():
     return INVALID_ADDR
 
 
+@dataclass
+class PageTableLevel:
+    name: str
+    entry: int
+    virt: int  # within physmap
+    idx: int
+
+
 class ArchPagingInfo:
     USERLAND = "userland"
     KERNELLAND = "kernel [.text]"
@@ -50,6 +58,7 @@ class ArchPagingInfo:
     va_bits: int
     pagetable_cache: Dict[pwndbg.dbg_mod.Value, Dict[int, int]] = {}
     pagetableptr_cache: Dict[int, pwndbg.dbg_mod.Value] = {}
+    pagetable_level_names: Tuple[str, ...]
 
     @property
     @pwndbg.lib.cache.cache_until("objfile")
@@ -99,27 +108,26 @@ class ArchPagingInfo:
 
         return None
 
-    def pagewalk(
-        self, target, entry
-    ) -> Tuple[Tuple[str, ...], List[Tuple[int | None, int | None]]]:
+    def pagewalk(self, target, entry) -> Tuple[PageTableLevel, ...]:
         raise NotImplementedError()
 
-    def pagewalk_helper(self, target, entry) -> List[Tuple[int | None, int | None]]:
+    def pagewalk_helper(self, target, entry) -> Tuple[PageTableLevel, ...]:
         base = self.physmap
         if entry > base:
             # user inputted a physmap address as pointer to pgd
             entry -= base
         level = self.paging_level
-        result: List[Tuple[int | None, int | None]] = [(None, None)] * (level + 1)
+        result = [PageTableLevel(None, None, None, None)] * (level + 1)
         page_shift = self.page_shift
         ENTRYMASK = ~((1 << page_shift) - 1) & ((1 << self.va_bits) - 1)
+        IDXMASK = (1 << (page_shift - math.ceil(math.log2(pwndbg.aglib.arch.ptrsize)))) - 1
         for i in range(level, 0, -1):
             vaddr = (entry & ENTRYMASK) + base - self.phys_offset
             if self.should_stop_pagewalk(entry):
                 break
             shift = (i - 1) * (page_shift - 3) + page_shift
             offset = target & ((1 << shift) - 1)
-            idx = (target & (0x1FF << shift)) >> shift
+            idx = (target & (IDXMASK << shift)) >> shift
             entry = 0
             try:
                 # with this optimization, roughly x2 as fast on average
@@ -142,10 +150,15 @@ class ArchPagingInfo:
                 print(M.warn(f"Exception while page walking: {e}"))
                 entry = 0
             if entry == 0:
-                return result
-            result[i] = (entry, vaddr)
-        result[0] = (entry, (entry & ENTRYMASK) + base + offset - self.phys_offset)
-        return result
+                return tuple(result)
+            result[i] = PageTableLevel(self.pagetable_level_names[i], entry, vaddr, idx)
+        result[0] = PageTableLevel(
+            self.pagetable_level_names[0],
+            entry,
+            (entry & ENTRYMASK) + base + offset - self.phys_offset,
+            None,
+        )
+        return tuple(result)
 
     def pageentry_flags(self, level) -> BitFlags:
         raise NotImplementedError()
@@ -313,12 +326,10 @@ class x86_64PagingInfo(ArchPagingInfo):
             if pwndbg.aglib.regs[pwndbg.aglib.regs.stack] in page:
                 page.objfile = "kernel [stack]"
 
-    def pagewalk(
-        self, target, entry
-    ) -> Tuple[Tuple[str, ...], List[Tuple[int | None, int | None]]]:
+    def pagewalk(self, target, entry) -> Tuple[PageTableLevel, ...]:
         if entry is None:
             entry = pwndbg.aglib.regs["cr3"]
-        return self.pagetable_level_names, self.pagewalk_helper(target, entry)
+        return self.pagewalk_helper(target, entry)
 
     def pageentry_flags(self, is_last) -> BitFlags:
         return BitFlags([("NX", 63), ("PS", 7), ("A", 5), ("U", 2), ("W", 1), ("P", 0)])
@@ -606,16 +617,14 @@ class Aarch64PagingInfo(ArchPagingInfo):
             pass
         return 0x40000000  # default
 
-    def pagewalk(
-        self, target, entry
-    ) -> Tuple[Tuple[str, ...], List[Tuple[int | None, int | None]]]:
+    def pagewalk(self, target, entry) -> Tuple[PageTableLevel, ...]:
         if entry is None:
             if pwndbg.aglib.memory.is_kernel(target):
                 entry = pwndbg.aglib.regs.TTBR1_EL1
             else:
                 entry = pwndbg.aglib.regs.TTBR0_EL1
         self.entry = entry
-        return self.pagetable_level_names, self.pagewalk_helper(target, entry)
+        return self.pagewalk_helper(target, entry)
 
     def pageentry_flags(self, is_last) -> BitFlags:
         if is_last:

pwndbg/aglib/kernel/vmmap.py

@@ -424,8 +424,7 @@ def kernel_vmmap(process_pages=True) -> Tuple[pwndbg.lib.memory.Page, ...]:
             for page in pages:
                 if page.objfile == kv.pi.ESPSTACK:
                     continue
-                _, pgwalk_res = pwndbg.aglib.kernel.pagewalk(page.start)
-                entry, _ = pgwalk_res[0]
+                entry = pwndbg.aglib.kernel.pagewalk(page.start)[0].entry
                 if entry and entry >> 63 == 0:
                     page.flags |= 1
 

pwndbg/commands/kbase.py

@@ -12,12 +12,18 @@ from pwndbg.commands import CommandCategory
 parser = argparse.ArgumentParser(description="Finds the kernel virtual base address.")
 
 parser.add_argument("-r", "--rebase", action="store_true", help="rebase loaded symbol file")
+parser.add_argument(
+    "-v",
+    "--verbose",
+    action="store_true",
+    help="show more information relevant to the kbase (e.g. phys addr)",
+)
 
 
 @pwndbg.commands.Command(parser, category=CommandCategory.KERNEL)
 @pwndbg.commands.OnlyWhenQemuKernel
 @pwndbg.commands.OnlyWhenPagingEnabled
-def kbase(rebase=False) -> None:
+def kbase(rebase=False, verbose=False) -> None:
     if config.kernel_vmmap == "none":
         print(M.error("kbase does not work when kernel-vmmap is set to none"))
         return
@@ -30,6 +36,11 @@ def kbase(rebase=False) -> None:
 
     print(M.success(f"Found virtual text base address: {hex(base)}"))
 
+    if verbose:
+        phys = pwndbg.aglib.kernel.virt_to_phys(base)
+        if phys is not None:
+            print(M.success(f"corresponding physical address: {hex(phys)}"))
+
     if not rebase:
         return
 

pwndbg/commands/paging.py

@@ -1,12 +1,14 @@
 from __future__ import annotations
 
 import argparse
+import math
 
 import pwndbg.aglib.kernel
 import pwndbg.aglib.kernel.paging
 import pwndbg.aglib.regs
 import pwndbg.color as C
 import pwndbg.color.message as M
+from pwndbg.aglib.kernel.paging import PageTableLevel
 from pwndbg.commands import CommandCategory
 
 parser = argparse.ArgumentParser(description="Performs pagewalk.")
@@ -25,13 +27,20 @@ PAGETYPES = (
 )
 
 
-def print_pagetable_entry(name: str, paddr: int | None, vaddr: int, level: int, is_last: bool):
+def print_pagetable_entry(ptl: PageTableLevel, level: int, is_last: bool):
     pageflags = pwndbg.aglib.kernel.arch_paginginfo().pageentry_flags(is_last)
     flags = ""
     arrow_right = pwndbg.chain.c.arrow(f"{pwndbg.chain.config_arrow_right}")
-    if paddr is not None:
-        flags = f"{arrow_right} {name + 'e'}: {C.context.format_flags(paddr, pageflags, paddr)}"
-    print(f"{C.blue(name)} @ {C.yellow(hex(vaddr))} {flags}")
+    name, entry, vaddr, idx = ptl.name, ptl.entry, ptl.virt, ptl.idx
+    if pwndbg.aglib.arch.name == "x86-64":
+        name = name.ljust(3, " ")
+    nbits = pwndbg.aglib.kernel.arch_ops().page_shift - math.ceil(
+        math.log2(pwndbg.aglib.arch.ptrsize)
+    )  # each idx has that many bits
+    idxlen = len(str((1 << nbits) - 1))
+    if entry is not None:
+        flags = f"[{idx:0{idxlen}}] {arrow_right} {name + 'e'}: {C.context.format_flags(entry, pageflags, entry)}"
+    print(f"{C.blue(name)} @ {C.yellow(hex(vaddr))}{flags}")
 
 
 def page_type(page):
@@ -82,14 +91,14 @@ def pagewalk(vaddr, entry=None):
     if entry is not None:
         entry = int(pwndbg.dbg.selected_frame().evaluate_expression(entry))
     vaddr = int(pwndbg.dbg.selected_frame().evaluate_expression(vaddr))
-    names, entries = pwndbg.aglib.kernel.pagewalk(vaddr, entry)
-    for i in range(len(names) - 1, 0, -1):
-        entry, vaddr = entries[i]
-        next, _ = entries[i - 1]
-        if entry is None:
+    levels = pwndbg.aglib.kernel.pagewalk(vaddr, entry)
+    for i in range(len(levels) - 1, 0, -1):
+        curr = levels[i]
+        next = levels[i - 1]
+        if curr.entry is None:
             break
-        print_pagetable_entry(names[i], entry, vaddr, i, next is None or i == 1)
-    _, vaddr = entries[0]
+        print_pagetable_entry(curr, i, next.entry is None or i == 1)
+    vaddr = levels[0].virt
     if vaddr is None:
         print(M.warn("address is not mapped"))
         return
@@ -117,10 +126,13 @@ p2v_parser.add_argument("paddr", type=str, help="")
 @pwndbg.aglib.proc.OnlyWithArch(["x86-64", "aarch64"])
 def p2v(paddr):
     paddr = int(pwndbg.dbg.selected_frame().evaluate_expression(paddr))
-    vaddr = pwndbg.aglib.kernel.phys_to_virt(paddr)
-    paging_print_helper("Virtual address", vaddr)
-    page = pwndbg.aglib.kernel.virt_to_page(vaddr)
-    page_info(page)
+    try:
+        vaddr = pwndbg.aglib.kernel.phys_to_virt(paddr)
+        paging_print_helper("Virtual address", vaddr)
+        page = pwndbg.aglib.kernel.virt_to_page(vaddr)
+        page_info(page)
+    except Exception:
+        print(M.warn("physical to virtual address failed, invalid physical address?"))
 
 
 v2p_parser = argparse.ArgumentParser(
@@ -136,9 +148,10 @@ v2p_parser.add_argument("vaddr", type=str, help="")
 @pwndbg.aglib.proc.OnlyWithArch(["x86-64", "aarch64"])
 def v2p(vaddr):
     vaddr = int(pwndbg.dbg.selected_frame().evaluate_expression(vaddr))
-    entry, paddr = pwndbg.aglib.kernel.pagewalk(vaddr)[1][0]  # more accurate
+    level = pwndbg.aglib.kernel.pagewalk(vaddr)[0]  # more accurate
+    entry, paddr = level.entry, level.virt
     if not entry:
-        print(M.warn("virtual to page failed, unmapped virtual address?"))
+        print(M.warn("virtual to physical address failed, unmapped virtual address?"))
         return
     paging_print_helper("Physmap address", paddr)
     # paddr is the physmap address which is a virtual address
@@ -159,6 +172,9 @@ page_parser.add_argument("page", type=str, help="")
 @pwndbg.aglib.proc.OnlyWithArch(["x86-64", "aarch64"])
 def pageinfo(page):
     page = int(pwndbg.dbg.selected_frame().evaluate_expression(page))
-    vaddr = pwndbg.aglib.kernel.page_to_virt(page)
-    paging_print_helper("Virtual address", vaddr)
-    page_info(page)
+    try:
+        vaddr = pwndbg.aglib.kernel.page_to_virt(page)
+        paging_print_helper("Virtual address", vaddr)
+        page_info(page)
+    except Exception:
+        print(M.warn("invalid page struct pointer"))