Skip to contentSkip to conversation

openrisc: Call destructor during __pte_free_tlb

PATCH In review Stafford Horne 2 participants 2 messages Jun 17, 2018 Openrisc architecture

lore
Share & tools
lore

Stafford Horne

author maintainer
#0 lore raw

PATCH 1/1 openrisc: Call destructor during __pte_free_tlb

This fixes an issue uncovered when a recent change to add the "page table" flag was merged. During bootup we see many errors like the following:

    BUG: Bad page state in process mkdir  pfn:00bae
    page:c1ff15c0 count:0 mapcount:-1024 mapping:00000000 index:0x0
    flags: 0x0()
    raw: 00000000 00000000 00000000 fffffbff 00000000 00000100 00000200 00000000
    page dumped because: nonzero mapcount
    Modules linked in:
    CPU: 0 PID: 46 Comm: mkdir Tainted: G    B             4.17.0-simple-smp-07461-g1d40a5ea01d5-dirty #993
    Call trace:
    [<(ptrval)>] show_stack+0x44/0x54
    [<(ptrval)>] dump_stack+0xb0/0xe8
    [<(ptrval)>] bad_page+0x138/0x174
    [<(ptrval)>] ? cpumask_next+0x24/0x34
    [<(ptrval)>] free_pages_check_bad+0x6c/0xd0
    [<(ptrval)>] free_pcppages_bulk+0x174/0x42c
    [<(ptrval)>] free_unref_page_commit.isra.17+0xb8/0xc8
    [<(ptrval)>] free_unref_page_list+0x10c/0x190
    [<(ptrval)>] ? set_reset_devices+0x0/0x2c
    [<(ptrval)>] release_pages+0x3a0/0x414
    [<(ptrval)>] tlb_flush_mmu_free+0x5c/0x90
    [<(ptrval)>] tlb_flush_mmu+0x90/0xa4
    [<(ptrval)>] arch_tlb_finish_mmu+0x50/0x94
    [<(ptrval)>] tlb_finish_mmu+0x30/0x64
    [<(ptrval)>] exit_mmap+0x110/0x1e0
    [<(ptrval)>] mmput+0x50/0xf0
    [<(ptrval)>] do_exit+0x274/0xa94
    [<(ptrval)>] do_group_exit+0x50/0x110
    [<(ptrval)>] __wake_up_parent+0x0/0x38
    [<(ptrval)>] _syscall_return+0x0/0x4

During the __pte_free_tlb path openrisc fails to call the page destructor which would clear the new bits that were introduced. To fix this we are calling the destructor.

It seem openrisc was the only architecture missing this, all other architectures either call the destructor like we are doing here or use pte_free.

Fixes 1d40a5ea01d53 ("mm: mark pages in use for page tables")
Cc Matthew Wilcox <[email protected]>
---
 arch/openrisc/include/asm/pgalloc.h | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
arch/openrisc/include/asm/pgalloc.h +5 -1
index 3e1a46615120..8999b9226512 100644
@@ -98,8 +98,12 @@ static inline void pte_free(struct mm_struct *mm, struct page *pte)
 	__free_page(pte);
 }
 
+#define __pte_free_tlb(tlb, pte, addr)	\
+do {					\
+	pgtable_page_dtor(pte);		\
+	tlb_remove_page((tlb), (pte));	\
+} while (0)
 
-#define __pte_free_tlb(tlb, pte, addr) tlb_remove_page((tlb), (pte))
 #define pmd_pgtable(pmd) pmd_page(pmd)
 
 #define check_pgt_cache()          do { } while (0)
Signature · 1 line
2.17.0
#1 lore raw
Quoted 7 lines · Stafford Horne

During the __pte_free_tlb path openrisc fails to call the page destructor which would clear the new bits that were introduced. To fix this we are calling the destructor.

It seem openrisc was the only architecture missing this, all other architectures either call the destructor like we are doing here or use pte_free.

It may be worth noting that failing to call the destructor was also messing up the zone stats (and would be causing other problems if you were using SPLIT_PTE_PTLOCKS).

At any rate, Acked-by: Matthew Wilcox <[email protected]>