On 06/15/2018 03:24 AM, Christophe Leroy wrote:
On 05/30/2018 12:32 PM, Aneesh Kumar K.V wrote:
Fix the below crash on BookE 64. pgtable_page_dtor expects struct page
*arg.
Also call the destructor on non book3s platforms correctly. This free
up the
split ptl locks correctly if we had allocated them before.
Call Trace:
[c0000000f30c7520] [c00000000021eeec] .kmem_cache_free+0x9c/0x44c
(unreliable)
[c0000000f30c75c0] [c0000000001ee07c] .ptlock_free+0x1c/0x30
[c0000000f30c7630] [c0000000001ee260] .tlb_remove_table+0xdc/0x224
[c0000000f30c76c0] [c0000000001ee640] .free_pgd_range+0x298/0x500
[c0000000f30c77d0] [c000000000232afc] .shift_arg_pages+0x10c/0x1e0
[c0000000f30c7910] [c000000000232dd0] .setup_arg_pages+0x200/0x25c
[c0000000f30c79c0] [c0000000002ad4fc] .load_elf_binary+0x450/0x16c8
[c0000000f30c7b10] [c000000000234914]
.search_binary_handler.part.11+0x9c/0x248
[c0000000f30c7bb0] [c00000000023595c]
.do_execveat_common.isra.13+0x868/0xc18
[c0000000f30c7cb0] [c00000000000184c] .run_init_process+0x34/0x4c
[c0000000f30c7d30] [c000000000001880] .try_to_run_init_process+0x1c/0x68
[c0000000f30c7db0] [c000000000002bd8] .kernel_init+0xdc/0x130
[c0000000f30c7e30] [c0000000000009dc] .ret_from_kernel_thread+0x58/0x7c
Fixes: 702346768 ("powerpc/mm/nohash: Remove pte fragment dependency
from nohash")
Signed-off-by: Aneesh Kumar K.V <aneesh.ku...@linux.ibm.com>
---
arch/powerpc/include/asm/book3s/32/pgalloc.h | 1 +
arch/powerpc/include/asm/nohash/32/pgalloc.h | 1 +
arch/powerpc/include/asm/nohash/64/pgalloc.h | 2 +-
3 files changed, 3 insertions(+), 1 deletion(-)
diff --git a/arch/powerpc/include/asm/book3s/32/pgalloc.h
b/arch/powerpc/include/asm/book3s/32/pgalloc.h
index 5073cc75f1c8..6a6673907e45 100644
--- a/arch/powerpc/include/asm/book3s/32/pgalloc.h
+++ b/arch/powerpc/include/asm/book3s/32/pgalloc.h
@@ -99,6 +99,7 @@ static inline void pte_free(struct mm_struct *mm,
pgtable_t ptepage)
static inline void pgtable_free(void *table, unsigned index_size)
{
if (!index_size) {
+ pgtable_page_dtor(virt_to_page(table));
__pte_free_tlb() already calls pgtable_page_dtor(table) before calling
pgtable_free() via pgtable_free_tlb().
Is it normal to call it twice ?
No. We should call pgtable_page_dtor only in the rcu callback. So we
should remove it from __pte_free_tlb().
free_page((unsigned long)table);
} else {
BUG_ON(index_size > MAX_PGTABLE_INDEX_SIZE);
diff --git a/arch/powerpc/include/asm/nohash/32/pgalloc.h
b/arch/powerpc/include/asm/nohash/32/pgalloc.h
index 29d37bd1f3b3..1707781d2f20 100644
--- a/arch/powerpc/include/asm/nohash/32/pgalloc.h
+++ b/arch/powerpc/include/asm/nohash/32/pgalloc.h
@@ -100,6 +100,7 @@ static inline void pte_free(struct mm_struct *mm,
pgtable_t ptepage)
static inline void pgtable_free(void *table, unsigned index_size)
{
if (!index_size) {
+ pgtable_page_dtor(virt_to_page(table));
Same here
I don't have the facility to test anything other than book3s64. Can you
do a patch for this?
-aneesh
