mbox series

[v7,0/6] arm64: ptdump: View the second stage page-tables

Message ID 20240621123230.1085265-1-sebastianene@google.com (mailing list archive)
Headers show
Series arm64: ptdump: View the second stage page-tables | expand

Message

Sebastian Ene June 21, 2024, 12:32 p.m. UTC 
Hi,


This series extends the ptdump support to allow dumping the guest
stage-2 pagetables. When CONFIG_PTDUMP_STAGE2_DEBUGFS is enabled, ptdump
registers '/sys/debug/kvm/<guest_id>/stage2_page_tables' entry with debugfs
upon guest creation. This allows userspace tools (eg. cat) to dump the
stage-2 pagetables by reading the registered file.

Reading the debugfs file shows stage-2 memory ranges in following format:
<IPA range> <size> <descriptor type> <access permissions> <mem_attributes>

Below is the output of a guest stage-2 pagetable dump running under Qemu:

---[ IPA bits 33 start lvl 2 ]---
0x0000000000000000-0x0000000080000000           2G PGD
0x0000000080000000-0x0000000080c00000          12M PGD      R W AF        BLK
0x0000000080c00000-0x0000000080e00000           2M PGD   XN R W AF        BLK
0x0000000080e00000-0x0000000081000000           2M PGD      R W AF        BLK
0x0000000081000000-0x0000000081400000           4M PGD   XN R W AF        BLK
0x0000000081400000-0x000000008fe00000         234M PGD
0x000000008fe00000-0x0000000090000000           2M PGD   XN R W AF        BLK
0x0000000090000000-0x00000000fa000000        1696M PGD
0x00000000fa000000-0x00000000fe000000          64M PGD   XN R W AF        BLK
0x00000000fe000000-0x0000000100000000          32M PGD
0x0000000100000000-0x0000000101c00000          28M PGD   XN R W AF        BLK
0x0000000101c00000-0x0000000102000000           4M PGD
0x0000000102000000-0x0000000102200000           2M PGD   XN R W AF        BLK
0x0000000102200000-0x000000017b000000        1934M PGD
0x000000017b000000-0x0000000180000000          80M PGD   XN R W AF        BLK

Link to v6:
https://lore.kernel.org/all/20240220151035.327199-1-sebastianene@google.com/

Link to v5:
https://lore.kernel.org/all/20240207144832.1017815-2-sebastianene@google.com/

Link to v4:
https://lore.kernel.org/all/20231218135859.2513568-2-sebastianene@google.com/

Link to v3:
https://lore.kernel.org/all/20231115171639.2852644-2-sebastianene@google.com/

Changelog:
 v6 -> v7:
 * Reworded commit for this patch : [PATCH v6 2/6] arm64: ptdump: Expose
   the attribute parsing functionality
 * fixed minor conflicts in the struct pg_state definition
 * moved the kvm_ptdump_guest_registration in the
 * kvm_arch_create_vm_debugfs
 * reset the parse state before walking the pagetables
 * copy the level name to the pg_level buffer


 v5 -> v6:
 * don't return an error if the kvm_arch_create_vm_debugfs fails to
   initialize (ref.
https://lore.kernel.org/all/20240216155941.2029458-1-oliver.upton@linux.dev/)  
 * fix use-after-free suggested by getting a reference to the
   KVM struct while manipulating the debugfs files
   and put the reference on the file close.
 * do all the allocations at once for the ptdump parser state tracking
   and simplify the initialization.
 * move the ptdump parser state initialization as part of the file_open
 * create separate files for printing the guest stage-2 pagetable
   configuration such as: the start level of the pagetable walk and the
   number of bits used for the IPA space representation.
 * fixed the wrong header format for the newly added file
 * include missing patch which hasn't been posted on the v5:
   "KVM-arm64-Move-pagetable-definitions-to-common-heade.patch" 


 v4 -> v5:
 * refactorization: split the series into two parts as per the feedback
   received from Oliver. Introduce the base support which allows dumping
   of the guest stage-2 pagetables.
 * removed the *ops* struct wrapper built on top of the file_ops and
   simplify the ptdump interface access.
 * keep the page table walker away from the ptdump specific code

  v3 -> current_version:
  * refactorization: moved all the **KVM** specific components under
    kvm/ as suggested by Oliver. Introduced a new file
    arm64/kvm/ptdump.c which handled the second stage translation.
    re-used only the display portion from mm/ptdump.c
  * pagetable snapshot creation now uses memory donated from the host.
    The memory is no longer shared with the host as this can pose a security
    risk if the host has access to manipulate the pagetable copy while
    the hypervisor iterates it.
  * fixed a memory leak: while memory was used from the memcache for
    building the snapshot pagetable, it was no longer giving back the
    pages to the host for freeing. A separate array was introduced to
    keep track of the pages allocated from the memcache.


  v2 -> v3:
  * register the stage-2 debugfs entry for the host under
    /sys/debug/kvm/host_stage2_page_tables and in
    /sys/debug/kvm/<guest_id>/stage2_page_tables for guests.
  * don't use a static array for parsing the attributes description,
    generate it dynamically based on the number of pagetable levels
  * remove the lock that was guarding the seq_file private inode data,
    and keep the data private to the open file session.
  * minor fixes & renaming of CONFIG_NVHE_EL2_PTDUMP_DEBUGFS to
    CONFIG_PTDUMP_STAGE2_DEBUGFS


  v1 -> v2:
  * use the stage-2 pagetable walker for dumping descriptors instead of
    the one provided by ptdump.
  * support for guests pagetables dumping under VHE/nVHE non-protected

Thanks,

Sebastian Ene (6):
  KVM: arm64: Move pagetable definitions to common header
  arm64: ptdump: Expose the attribute parsing functionality
  arm64: ptdump: Use the mask from the state structure
  KVM: arm64: Register ptdump with debugfs on guest creation
  KVM: arm64: Initialize the ptdump parser with stage-2 attributes
  KVM: arm64: Expose guest stage-2 pagetable config to debugfs

 arch/arm64/include/asm/kvm_pgtable.h |  42 +++++
 arch/arm64/include/asm/ptdump.h      |  42 ++++-
 arch/arm64/kvm/Kconfig               |  14 ++
 arch/arm64/kvm/Makefile              |   1 +
 arch/arm64/kvm/arm.c                 |   2 +
 arch/arm64/kvm/hyp/pgtable.c         |  42 -----
 arch/arm64/kvm/kvm_ptdump.h          |  20 ++
 arch/arm64/kvm/ptdump.c              | 272 +++++++++++++++++++++++++++
 arch/arm64/mm/ptdump.c               |  50 +----
 9 files changed, 402 insertions(+), 83 deletions(-)
 create mode 100644 arch/arm64/kvm/kvm_ptdump.h
 create mode 100644 arch/arm64/kvm/ptdump.c

Comments

Oliver Upton June 28, 2024, 9:25 p.m. UTC | #1 
Hi Seb,

I think we're getting closer on this series, thanks for reposting it. In
addition to polishing up the KVM side of things, I'd like to have acks
on the ptdump changes (patches 2-3).

Will + Catalin, any concerns?