Re: [PATCH 02/65] powerpc/mm: use _PAGE_READ to indicate Read access

From
Balbir Singh <bsingharora@gmail.com>
Date
2016-04-05 03:25:49
Reply to
Aneesh Kumar K.V, 2016-04-04

On 05/04/16 00:55, Aneesh Kumar K.V wrote:
Balbir Singh [off-list ref] writes:
quoted
[ text/plain ]


On 27/03/16 19:23, Aneesh Kumar K.V wrote:
quoted
This split _PAGE_RW bit to _PAGE_READ and _PAGE_WRITE. It also remove
the dependency on _PAGE_USER for implying read only. Few things to note
here is that, we have read implied with write and execute permission.
Hence we should always find _PAGE_READ set on hash pte fault.

We still can't switch PROT_NONE to !(_PAGE_RWX). Auto numa do depend
on marking a prot none pte _PAGE_WRITE. (For more details look at
b191f9b106ea "mm: numa: preserve PTE write permissions across a NUMA hinting fault")
......

quoted
quoted
diff --git a/arch/powerpc/mm/hash_utils_64.c b/arch/powerpc/mm/hash_utils_64.c
index 90dd9280894f..ea23403b3fc0 100644
--- a/arch/powerpc/mm/hash_utils_64.c
+++ b/arch/powerpc/mm/hash_utils_64.c
@@ -175,8 +175,9 @@ unsigned long htab_convert_pte_flags(unsigned long pteflags)
 	 * or PP=0x3 for read-only (including writeable but clean pages).
 	 */
 	if (pteflags & _PAGE_USER) {
-		rflags |= 0x2;
-		if (!((pteflags & _PAGE_RW) && (pteflags & _PAGE_DIRTY)))
+		if (pteflags & _PAGE_RWX)
Should this be pteflags & _PAGE_RW?
That will work, because we always have _PAGE_READ set for _PAGE_EXEC.
But what we really need to check here is READ/WRITE/EXEC.
Does READ/WRITE matter for pp bits?
quoted
quoted
+			rflags |= 0x2;
+		if (!((pteflags & _PAGE_WRITE) && (pteflags & _PAGE_DIRTY)))
 			rflags |= 0x1;
if pteflags == _PAGE_USER | _PAGE_WRITE | _PAGE_DIRTY, what is rflags set to?
rflags will be 0x2. and for readlyonly 0x3. That is documented above. 
Yep the comment says that, I  think the comment can enhanced to add more information

For example

pp = 0x3 for PAGE_READ | PAGE_EXEC (read-only)
pp = 0x2 if the page is writable (read-write)


quoted
quoted
 	}
 	/*
@@ -1205,7 +1206,7 @@ EXPORT_SYMBOL_GPL(hash_page);
 int __hash_page(unsigned long ea, unsigned long msr, unsigned long trap,
 		unsigned long dsisr)
 {
-	unsigned long access = _PAGE_PRESENT;
+	unsigned long access = _PAGE_PRESENT | _PAGE_READ;
 	unsigned long flags = 0;
 	struct mm_struct *mm = current->mm;
....
quoted
quoted
diff --git a/arch/powerpc/mm/pgtable.c b/arch/powerpc/mm/pgtable.c
index 83dfd7925c72..98b5c03e344d 100644
--- a/arch/powerpc/mm/pgtable.c
+++ b/arch/powerpc/mm/pgtable.c
@@ -177,8 +177,8 @@ void set_pte_at(struct mm_struct *mm, unsigned long addr, pte_t *ptep,
 	 * _PAGE_PRESENT, but we can be sure that it is not in hpte.
 	 * Hence we can use set_pte_at for them.
 	 */
-	VM_WARN_ON((pte_val(*ptep) & (_PAGE_PRESENT | _PAGE_USER)) ==
-		(_PAGE_PRESENT | _PAGE_USER));
+	VM_WARN_ON(pte_present(*ptep) && !pte_protnone(*ptep));
+
What was wrong with the previous check? The compiler will optimize it, but the new
check uses two pte_val() calls on *ptep
That was confusing, even though we documented it clearly above, it does
confuse, because checking for _PAGE_USER is an indirect hint for prot
numa ptes.

quoted
quoted
 	/*
 	 * Add the pte bit when tryint set a pte
 	 */
diff --git a/arch/powerpc/mm/pgtable_64.c b/arch/powerpc/mm/pgtable_64.c
index aa742aa35b64..00d8d985bba3 100644
--- a/arch/powerpc/mm/pgtable_64.c
+++ b/arch/powerpc/mm/pgtable_64.c
@@ -277,7 +277,7 @@ void __iomem * ioremap_prot(phys_addr_t addr, unsigned long size,
 	void *caller = __builtin_return_address(0);
 
 	/* writeable implies dirty for kernel addresses */
-	if (flags & _PAGE_RW)
+	if (flags & _PAGE_WRITE)
 		flags |= _PAGE_DIRTY;
 
 	/* we don't want to let _PAGE_USER and _PAGE_EXEC leak out */
@@ -681,8 +681,7 @@ void set_pmd_at(struct mm_struct *mm, unsigned long addr,
 		pmd_t *pmdp, pmd_t pmd)
 {
 #ifdef CONFIG_DEBUG_VM
-	WARN_ON((pmd_val(*pmdp) & (_PAGE_PRESENT | _PAGE_USER)) ==
-		(_PAGE_PRESENT | _PAGE_USER));
+	WARN_ON(pte_present(pmd_pte(*pmdp)) && !pte_protnone(pmd_pte(*pmdp)));
Same as above, plus I think we can move these to VM_WARN_ON just to be consistent
We already have #ifdef CONFIG_DEBUG_VM around, hence we can avoid using
VM_WARN_ON.
-just nit-picking
Balbir Singh
Thread (92 messages) 92 messages, 7 authors, last reply 2018-06-28
Keyboard shortcuts
hback out one level
jnext message in thread / next series-diff position
kprevious message in thread / previous series-diff position
ldrill in
Escclose help / fold thread tree
?toggle this help