On 19.7.2021 17.12, Qi Zheng wrote:
On 7/19/21 9:55 PM, Mika Penttilä wrote:
On 19.7.2021 15.56, Qi Zheng wrote:
On 7/18/21 2:19 PM, Mika Penttilä wrote:
+
+/*
+ * returns true if the pmd has been populated with PTE page table,
+ * or false for all other cases.
+ */
+bool pte_install_try_get(struct mm_struct *mm, pmd_t *pmd,
pgtable_t *pte)
+{
+ spinlock_t *ptl;
+ bool retval = true;
+
+retry:
+ ptl = pmd_lock(mm, pmd);
+ if (likely(pmd_none(*pmd))) {
+ __pte_install(mm, pmd, pte);
+ } else if (pmd_leaf(*pmd) || !pmd_present(*pmd)) {
+ retval = false;
+ } else if (!pte_get_unless_zero(pmd)) {
+ spin_unlock(ptl);
+ goto retry;
+ }
+ spin_unlock(ptl);
+ return retval;
+}
+
Can pte_get_unless_zero() return true above? Can the pmd have been
by populated by others? In that case the ref count is wrongly
incremented.
Here we only have mmap_read_lock(mm), so the pmd can be populated with
other PTE page table page after a page fault in a different thread B
of this mm. In this case, thread B already hold a pte_refcount of
the PTE page table page populated in the pmd, so
pte_get_unless_zero() can
return true above.
Yes but if thread B populates the page table page and pte, then we
also increase the refcount with pte_get_unless_zero() , but dont
decrease it when notice !pte_none().
And in the pte_none() case, the refcount is increased again, so
double accounting. see finish_fault().
The semantics of this function is to hold the pte_refcount count when it
returns 1, and its caller is responsible for decrease the pte_refcount
by calling pte_put(), like the following pattern:
do_anonymous_page()
--> pte_alloc_try_get()
do something about pte
pte_put()
Similarly in finish_fault().
Thanks,
Qi
Ok I didn't notice the added pte_put()'s at the end of these funcs.
Thanks,
Mika
Similarly, if THP is enabled, the pmd also can be populated with a
THP page, we can see more detail in comment in handle_pte_fault(). The
pmd_leaf() above is to detect this situation.
