powerpc/mm: add DEBUG_VM WARN for pmd_clear
authorAneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Fri, 16 Oct 2020 03:04:26 +0000 (20:04 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Fri, 16 Oct 2020 18:11:14 +0000 (11:11 -0700)
commit392b466981ac42f491173c37f6ed574841645bcc
tree29685ebc44716dc0cecfc540656edd1e21ac3a46
parenta455aa72f7c46b881721668b3ee810713adc7a5b
powerpc/mm: add DEBUG_VM WARN for pmd_clear

Patch series "mm/debug_vm_pgtable fixes", v4.

This patch series includes fixes for debug_vm_pgtable test code so that
they follow page table updates rules correctly.  The first two patches
introduce changes w.r.t ppc64.

Hugetlb test is disabled on ppc64 because that needs larger change to satisfy
page table update rules.

These tests are broken w.r.t page table update rules and results in kernel
crash as below.

[   21.083519] kernel BUG at arch/powerpc/mm/pgtable.c:304!
cpu 0x0: Vector: 700 (Program Check) at [c000000c6d1e76c0]
    pc: c00000000009a5ec: assert_pte_locked+0x14c/0x380
    lr: c0000000005eeeec: pte_update+0x11c/0x190
    sp: c000000c6d1e7950
   msr: 8000000002029033
  current = 0xc000000c6d172c80
  paca    = 0xc000000003ba0000   irqmask: 0x03   irq_happened: 0x01
    pid   = 1, comm = swapper/0
kernel BUG at arch/powerpc/mm/pgtable.c:304!
[link register   ] c0000000005eeeec pte_update+0x11c/0x190
[c000000c6d1e79500000000000000001 (unreliable)
[c000000c6d1e79b0c0000000005eee14 pte_update+0x44/0x190
[c000000c6d1e7a10c000000001a2ca9c pte_advanced_tests+0x160/0x3d8
[c000000c6d1e7ab0c000000001a2d4fc debug_vm_pgtable+0x7e8/0x1338
[c000000c6d1e7ba0c0000000000116ec do_one_initcall+0xac/0x5f0
[c000000c6d1e7c80c0000000019e4fac kernel_init_freeable+0x4dc/0x5a4
[c000000c6d1e7db0c000000000012474 kernel_init+0x24/0x160
[c000000c6d1e7e20c00000000000cbd0 ret_from_kernel_thread+0x5c/0x6c

With DEBUG_VM disabled

[   20.530152] BUG: Kernel NULL pointer dereference on read at 0x00000000
[   20.530183] Faulting instruction address: 0xc0000000000df330
cpu 0x33: Vector: 380 (Data SLB Access) at [c000000c6d19f700]
    pc: c0000000000df330: memset+0x68/0x104
    lr: c00000000009f6d8: hash__pmdp_huge_get_and_clear+0xe8/0x1b0
    sp: c000000c6d19f990
   msr: 8000000002009033
   dar: 0
  current = 0xc000000c6d177480
  paca    = 0xc00000001ec4f400   irqmask: 0x03   irq_happened: 0x01
    pid   = 1, comm = swapper/0
[link register   ] c00000000009f6d8 hash__pmdp_huge_get_and_clear+0xe8/0x1b0
[c000000c6d19f990c00000000009f748 hash__pmdp_huge_get_and_clear+0x158/0x1b0 (unreliable)
[c000000c6d19fa10c0000000019ebf30 pmd_advanced_tests+0x1f0/0x378
[c000000c6d19fab0c0000000019ed088 debug_vm_pgtable+0x79c/0x1244
[c000000c6d19fba0c0000000000116ec do_one_initcall+0xac/0x5f0
[c000000c6d19fc80c0000000019a4fac kernel_init_freeable+0x4dc/0x5a4
[c000000c6d19fdb0c000000000012474 kernel_init+0x24/0x160
[c000000c6d19fe20c00000000000cbd0 ret_from_kernel_thread+0x5c/0x6c

This patch (of 13):

With the hash page table, the kernel should not use pmd_clear for clearing
huge pte entries.  Add a DEBUG_VM WARN to catch the wrong usage.

Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Cc: Christophe Leroy <christophe.leroy@csgroup.eu>
Link: https://lkml.kernel.org/r/20200902114222.181353-1-aneesh.kumar@linux.ibm.com
Link: https://lkml.kernel.org/r/20200902114222.181353-2-aneesh.kumar@linux.ibm.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
arch/powerpc/include/asm/book3s/64/pgtable.h