From: Guo Ren <guoren@xxxxxxxxxxxxxxxxx>
When the system memory is exhausted, linux will trigger kswapd to
shrink memory page cache. We found the csky's .text file mapping
pages would be reclaimed earlier than arm's elf. Because csky
doesn't give _PAGE_ACCESSED for default pgprot and in zap_pte_range
if (pte_young(ptent) &&
likely(!(vma->vm_flags & VM_SEQ_READ)))
mark_page_accessed(page);
mark_page_accessed will put the pages into active lru list.
[ 3.652722] delete busybox page from inactive file list
Call Trace:
[<9012a376>] dump_stack+0xe/0x24
[<9012a370>] dump_stack+0x8/0x24
[<9005b780>] activate_page+0x2b4/0x2d4
[<90132502>] vsnprintf+0x2c6/0x374
[<9005b880>] mark_page_accessed+0xe0/0x150
[<9006903e>] unmap_page_range+0x166/0x33c
[<90021844>] get_signal+0x98/0x3b4
[<90069232>] unmap_single_vma+0x1e/0x24
[<90069462>] unmap_vmas+0x26/0x40
[<9006d3d8>] exit_mmap+0x60/0xbc
[<9006a140>] handle_mm_fault+0x700/0xcec
[<900426b2>] ktime_get_with_offset+0x86/0x130
[<90017566>] mmput+0x2e/0x90
[<9001a30a>] do_exit+0x13e/0x6f0
[<90015448>] page_fault_end+0x14/0x74
[<9001b4bc>] SyS_exit_group+0x0/0xc
[<9001b47c>] do_group_exit+0x2c/0x6c
[<9001b4c8>] __wake_up_parent+0x0/0x20
[<9001399e>] csky_systemcall+0x6e/0x72
csky will throw the pages at first and keep them in active lru
list later after real accessed, but arm would keep them in active
lru list at the beginning.
The following are statistics of different architecture styles:
Default _PAGE_ACCESSED: alpha, arm, arm64, ia64, m68k, microblaze,
openrisc, powerpc, riscv, sh, um, x86,
xtensa
Not def _PAGE_ACCESSED: arc, c6x, h8300, hexgon, mips, s390, nds32,
nios2, parisc, sparc
Signed-off-by: Guo Ren <guoren@xxxxxxxxxxxxxxxxx>
Co-developed-by: Xu Kai <xukai@xxxxxxxxxxxxxxxx>
Signed-off-by: Xu Kai <xukai@xxxxxxxxxxxxxxxx>
---
arch/csky/include/asm/pgtable.h | 16 +++++++++-------
1 file changed, 9 insertions(+), 7 deletions(-)
diff --git a/arch/csky/include/asm/pgtable.h b/arch/csky/include/asm/pgtable.h
index 2485db84dba8..429d96f66620 100644
--- a/arch/csky/include/asm/pgtable.h
+++ b/arch/csky/include/asm/pgtable.h
@@ -63,18 +63,20 @@
* read. Also, write permissions imply read permissions. This is the closest
* we can get by reasonable means..
*/
-#define PAGE_NONE __pgprot(_PAGE_PRESENT | _CACHE_CACHED)
-#define PAGE_SHARED __pgprot(_PAGE_PRESENT | _PAGE_READ | _PAGE_WRITE | \
+#define _PAGE_BASE (_PAGE_PRESENT | _PAGE_ACCESSED)
+
+#define PAGE_NONE __pgprot(_PAGE_BASE | _CACHE_CACHED)
+#define PAGE_SHARED __pgprot(_PAGE_BASE | _PAGE_READ | _PAGE_WRITE | \
_CACHE_CACHED)
-#define PAGE_COPY __pgprot(_PAGE_PRESENT | _PAGE_READ | _CACHE_CACHED)
-#define PAGE_READONLY __pgprot(_PAGE_PRESENT | _PAGE_READ | _CACHE_CACHED)
-#define PAGE_KERNEL __pgprot(_PAGE_PRESENT | __READABLE | __WRITEABLE | \
+#define PAGE_COPY __pgprot(_PAGE_BASE | _PAGE_READ | _CACHE_CACHED)
+#define PAGE_READONLY __pgprot(_PAGE_BASE | _PAGE_READ | _CACHE_CACHED)
+#define PAGE_KERNEL __pgprot(_PAGE_BASE | __READABLE | __WRITEABLE | \
_PAGE_GLOBAL | _CACHE_CACHED)
-#define PAGE_USERIO __pgprot(_PAGE_PRESENT | _PAGE_READ | _PAGE_WRITE | \
+#define PAGE_USERIO __pgprot(_PAGE_BASE | _PAGE_READ | _PAGE_WRITE | \
_CACHE_CACHED)
#define _PAGE_IOREMAP \
- (_PAGE_PRESENT | __READABLE | __WRITEABLE | _PAGE_GLOBAL | \
+ (_PAGE_BASE | __READABLE | __WRITEABLE | _PAGE_GLOBAL | \
_CACHE_UNCACHED | _PAGE_SO)
#define __P000 PAGE_NONE
--
2.17.1
