[RT] Revert "ARM: Initialize split page table locks for vector page"

Commit Message

Sebastian Andrzej Siewior Oct. 14, 2019, 4:02 p.m. UTC

I'm dropping this patch, with its original description:

|ARM: Initialize split page table locks for vector page
|
|Without this patch, ARM can not use SPLIT_PTLOCK_CPUS if
|PREEMPT_RT_FULL=y because vectors_user_mapping() creates a
|VM_ALWAYSDUMP mapping of the vector page (address 0xffff0000), but no
|ptl->lock has been allocated for the page.  An attempt to coredump
|that page will result in a kernel NULL pointer dereference when
|follow_page() attempts to lock the page.
|
|The call tree to the NULL pointer dereference is:
|
|   do_notify_resume()
|      get_signal_to_deliver()
|         do_coredump()
|            elf_core_dump()
|               get_dump_page()
|                  __get_user_pages()
|                     follow_page()
|                        pte_offset_map_lock() <----- a #define
|                           ...
|                              rt_spin_lock()
|
|The underlying problem is exposed by mm-shrink-the-page-frame-to-rt-size.patch.

The patch named mm-shrink-the-page-frame-to-rt-size.patch was dropped
from the RT queue once the SPLIT_PTLOCK_CPUS feature (in a slightly
different shape) went upstream (somewhere between v3.12 and v3.14).

I can see that the patch still allocates a lock which wasn't there
before. However I can't trigger a kernel oops like described in the
patch by triggering a coredump.

---
 arch/arm/kernel/process.c | 24 ------------------------
 1 file changed, 24 deletions(-)

Comments

Matthew Wilcox Oct. 15, 2019, 4:26 p.m. UTC | #1

On Mon, Oct 14, 2019 at 06:02:38PM +0200, Sebastian Andrzej Siewior wrote:
> I'm dropping this patch, with its original description:
> 
> |ARM: Initialize split page table locks for vector page
> |
> |Without this patch, ARM can not use SPLIT_PTLOCK_CPUS if
> |PREEMPT_RT_FULL=y because vectors_user_mapping() creates a
> |VM_ALWAYSDUMP mapping of the vector page (address 0xffff0000), but no
> |ptl->lock has been allocated for the page.  An attempt to coredump
> |that page will result in a kernel NULL pointer dereference when
> |follow_page() attempts to lock the page.
> |
> |The call tree to the NULL pointer dereference is:
> |
> |   do_notify_resume()
> |      get_signal_to_deliver()
> |         do_coredump()
> |            elf_core_dump()
> |               get_dump_page()
> |                  __get_user_pages()
> |                     follow_page()
> |                        pte_offset_map_lock() <----- a #define
> |                           ...
> |                              rt_spin_lock()
> |
> |The underlying problem is exposed by mm-shrink-the-page-frame-to-rt-size.patch.
> 
> The patch named mm-shrink-the-page-frame-to-rt-size.patch was dropped
> from the RT queue once the SPLIT_PTLOCK_CPUS feature (in a slightly
> different shape) went upstream (somewhere between v3.12 and v3.14).
> 
> I can see that the patch still allocates a lock which wasn't there
> before. However I can't trigger a kernel oops like described in the
> patch by triggering a coredump.

Did your test build have ALLOC_SPLIT_PTLOCKS defined?

Sebastian Andrzej Siewior Oct. 15, 2019, 4:40 p.m. UTC | #2

On 2019-10-15 09:26:09 [-0700], Matthew Wilcox wrote:
> Did your test build have ALLOC_SPLIT_PTLOCKS defined?

I tried to explain that ptlock_ptr() returns NULL for the page before
the ctor invocation and non-NULL afterwards which means that
USE_SPLIT_PTE_PTLOCKS and ALLOC_SPLIT_PTLOCKS was defined.

Sebastian

Patch

diff --git a/arch/arm/kernel/process.c b/arch/arm/kernel/process.c
index 1041300022177..f934a6739fc05 100644
--- a/arch/arm/kernel/process.c
+++ b/arch/arm/kernel/process.c
@@ -325,30 +325,6 @@  unsigned long arch_randomize_brk(struct mm_struct *mm)
 }
 
 #ifdef CONFIG_MMU
-/*
- * CONFIG_SPLIT_PTLOCK_CPUS results in a page->ptl lock.  If the lock is not
- * initialized by pgtable_page_ctor() then a coredump of the vector page will
- * fail.
- */
-static int __init vectors_user_mapping_init_page(void)
-{
-	struct page *page;
-	unsigned long addr = 0xffff0000;
-	pgd_t *pgd;
-	pud_t *pud;
-	pmd_t *pmd;
-
-	pgd = pgd_offset_k(addr);
-	pud = pud_offset(pgd, addr);
-	pmd = pmd_offset(pud, addr);
-	page = pmd_page(*(pmd));
-
-	pgtable_page_ctor(page);
-
-	return 0;
-}
-late_initcall(vectors_user_mapping_init_page);
-
 #ifdef CONFIG_KUSER_HELPERS
 /*
  * The vectors page is always readable from user space for the