Port kernel vmmap to aglib #2634
Conversation
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## dev #2634 +/- ##
=========================================
- Coverage 10.60% 5.36% -5.25%
=========================================
Files 214 218 +4
Lines 23904 24115 +211
Branches 3617 3652 +35
=========================================
- Hits 2535 1293 -1242
- Misses 20917 22633 +1716
+ Partials 452 189 -263 ☔ View full report in Codecov by Sentry. |
patryk4815
force-pushed
the
kernel-part1
branch
from
72c3137 to
04c5203
Compare
| @@ -649,6 +650,10 @@ def paging_enabled() -> bool: | |||
| return x86_64Ops.paging_enabled() | |||
There was a problem hiding this comment.
Random note: we should reorder this to x86-64, aarch64, rv64, i386
| elif arch_name == "rv64": | ||
| # https://starfivetech.com/uploads/u74_core_complex_manual_21G1.pdf | ||
| # page 41, satp.MODE, bits: 60,61,62,63 | ||
| return int(pwndbg.aglib.regs.satp) & (BIT(60) | BIT(61) | BIT(62) | BIT(63)) != 0 |
pwndbg/aglib/kernel/vmmap.py
Outdated
| self.pid = QemuMachine.get_qemu_pid() | ||
| self.file = None | ||
| self.file = os.open(f"/proc/{self.pid}/mem", os.O_RDONLY) | ||
| self.mem_size = os.fstat(self.file).st_size |
There was a problem hiding this comment.
Does this make sense?
(.venv) root@pwndbg:~# ls -la /proc/self/mem
-rw------- 1 root root 0 Dec 18 13:00 /proc/self/mem
(.venv) root@pwndbg:~# stat /proc/self/mem
File: /proc/self/mem
Size: 0 Blocks: 0 IO Block: 1024 regular empty file
Device: 0,22 Inode: 29332658 Links: 1
Access: (0600/-rw-------) Uid: ( 0/ root) Gid: ( 0/ root)
Access: 2024-12-18 13:00:15.561108327 +0000
Modify: 2024-12-18 13:00:15.561108327 +0000
Change: 2024-12-18 13:00:15.561108327 +0000
Birth: -There was a problem hiding this comment.
In [4]: f=os.open('/proc/self/mem', os.O_RDONLY)
In [5]: os.fstat(f).st_size
Out[5]: 0cc: @martinradev
There was a problem hiding this comment.
| f"chardev-add file,id={chardev_id},path={tmpf.name}" | ||
| ) | ||
| pid_found = QemuMachine.search_pids_for_file(pids, tmpf.name) | ||
| pwndbg.dbg.selected_inferior().send_monitor(f"chardev-remove {chardev_id}") |
There was a problem hiding this comment.
Lol this is a smart hack to detect the given QEMU process PID.
Maybe we should implement a monitor command to just return QEMU pid? :D
| # Handle disabled PG | ||
| # This will prevent a crash on abstract architectures | ||
| if len(lines) == 1 and lines[0] == "PG disabled": | ||
| return () |
There was a problem hiding this comment.
What's PG disabled, is it paging? Is it only on one or all archs?
Any chance we can support physical addresses here too? or should we handle it elsewhere?
|
Is there an alternative to copy-pasting the gdb-pt-dump code, and potentially something I could assist with? If the code becomes duplicated, code maintainability for both projects will suffer:
What would be required from gdb-pt-dump side to have the code live in a single place, but have things bundled for a release? |
Changes:
kernel_vmmap_via_monitor_info_memtoaglib.kernel.vmmapkernel_vmmap_via_page_tablestoaglib.kernel.vmmapkernel_vmmap_via_page_tableskernel_vmmap_via_page_tables