mm/memory_hotplug: skip adjust_managed_page_count() for PageOffline() pages when...

We currently have a hack for virtio-mem in place to handle memory
offlining with PageOffline pages for which we already adjusted the managed
page count.

Let's enlighten memory offlining code so we can get rid of that hack, and
document the situation.

Link: https://lkml.kernel.org/r/20240607090939.89524-4-david@redhat.com
Signed-off-by: David Hildenbrand <david@redhat.com>
Acked-by: Oscar Salvador <osalvador@suse.de>
Cc: Alexander Potapenko <glider@google.com>
Cc: Dexuan Cui <decui@microsoft.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Eugenio Pérez <eperezma@redhat.com>
Cc: Haiyang Zhang <haiyangz@microsoft.com>
Cc: Jason Wang <jasowang@redhat.com>
Cc: Juergen Gross <jgross@suse.com>
Cc: "K. Y. Srinivasan" <kys@microsoft.com>
Cc: Marco Elver <elver@google.com>
Cc: Michael S. Tsirkin <mst@redhat.com>
Cc: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Oleksandr Tyshchenko <oleksandr_tyshchenko@epam.com>
Cc: Stefano Stabellini <sstabellini@kernel.org>
Cc: Wei Liu <wei.liu@kernel.org>
Cc: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

diff --git a/drivers/virtio/virtio_mem.c b/drivers/virtio/virtio_mem.c
index b90df29621c81ec09565d715b0f845196c458533..b0b8714415783a5b4a461a947c52af839c7065f0 100644 (file)
--- a/drivers/virtio/virtio_mem.c
+++ b/drivers/virtio/virtio_mem.c
@@ -1269,12 +1269,6 @@ static void virtio_mem_fake_offline_going_offline(unsigned long pfn,
        struct page *page;
        unsigned long i;
 
-       /*
-        * Drop our reference to the pages so the memory can get offlined
-        * and add the unplugged pages to the managed page counters (so
-        * offlining code can correctly subtract them again).
-        */
-       adjust_managed_page_count(pfn_to_page(pfn), nr_pages);
        /* Drop our reference to the pages so the memory can get offlined. */
        for (i = 0; i < nr_pages; i++) {
                page = pfn_to_page(pfn + i);
@@ -1293,10 +1287,9 @@ static void virtio_mem_fake_offline_cancel_offline(unsigned long pfn,
        unsigned long i;
 
        /*
-        * Get the reference we dropped when going offline and subtract the
-        * unplugged pages from the managed page counters.
+        * Get the reference again that we dropped via page_ref_dec_and_test()
+        * when going offline.
         */
-       adjust_managed_page_count(pfn_to_page(pfn), -nr_pages);
        for (i = 0; i < nr_pages; i++)
                page_ref_inc(pfn_to_page(pfn + i));
 }

diff --git a/include/linux/memory_hotplug.h b/include/linux/memory_hotplug.h
index 7a9ff464608d702fa6e8c1f173e627f07d437bac..ebe876930e7821505805e38bc12bc8abd6c25163 100644 (file)
--- a/include/linux/memory_hotplug.h
+++ b/include/linux/memory_hotplug.h
@@ -175,8 +175,8 @@ extern int mhp_init_memmap_on_memory(unsigned long pfn, unsigned long nr_pages,
 extern void mhp_deinit_memmap_on_memory(unsigned long pfn, unsigned long nr_pages);
 extern int online_pages(unsigned long pfn, unsigned long nr_pages,
                        struct zone *zone, struct memory_group *group);
-extern void __offline_isolated_pages(unsigned long start_pfn,
-                                    unsigned long end_pfn);
+extern unsigned long __offline_isolated_pages(unsigned long start_pfn,
+               unsigned long end_pfn);
 
 typedef void (*online_page_callback_t)(struct page *page, unsigned int order);
 

diff --git a/include/linux/page-flags.h b/include/linux/page-flags.h
index b23772b08cc1f881c974f6f444a587f12cee9af7..1d441908b7266587b81619bcb0da4921a0301fb1 100644 (file)
--- a/include/linux/page-flags.h
+++ b/include/linux/page-flags.h
@@ -1023,11 +1023,15 @@ PAGE_TYPE_OPS(Buddy, buddy, buddy)
  * putting them back to the buddy, it can do so via the memory notifier by
  * decrementing the reference count in MEM_GOING_OFFLINE and incrementing the
  * reference count in MEM_CANCEL_OFFLINE. When offlining, the PageOffline()
- * pages (now with a reference count of zero) are treated like free pages,
- * allowing the containing memory block to get offlined. A driver that
+ * pages (now with a reference count of zero) are treated like free (unmanaged)
+ * pages, allowing the containing memory block to get offlined. A driver that
  * relies on this feature is aware that re-onlining the memory block will
  * require not giving them to the buddy via generic_online_page().
  *
+ * Memory offlining code will not adjust the managed page count for any
+ * PageOffline() pages, treating them like they were never exposed to the
+ * buddy using generic_online_page().
+ *
  * There are drivers that mark a page PageOffline() and expect there won't be
  * any further access to page content. PFN walkers that read content of random
  * pages should check PageOffline() and synchronize with such drivers using

diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
index f9dc9c2c9469723b4e25260aade0985776ab1b9b..fae1da27f0d7491b1d8018987642e6d1b45316ae 100644 (file)
--- a/mm/memory_hotplug.c
+++ b/mm/memory_hotplug.c
@@ -1942,7 +1942,7 @@ int __ref offline_pages(unsigned long start_pfn, unsigned long nr_pages,
                        struct zone *zone, struct memory_group *group)
 {
        const unsigned long end_pfn = start_pfn + nr_pages;
-       unsigned long pfn, system_ram_pages = 0;
+       unsigned long pfn, managed_pages, system_ram_pages = 0;
        const int node = zone_to_nid(zone);
        unsigned long flags;
        struct memory_notify arg;
@@ -2063,7 +2063,7 @@ int __ref offline_pages(unsigned long start_pfn, unsigned long nr_pages,
        } while (ret);
 
        /* Mark all sections offline and remove free pages from the buddy. */
-       __offline_isolated_pages(start_pfn, end_pfn);
+       managed_pages = __offline_isolated_pages(start_pfn, end_pfn);
        pr_debug("Offlined Pages %ld\n", nr_pages);
 
        /*
@@ -2079,7 +2079,7 @@ int __ref offline_pages(unsigned long start_pfn, unsigned long nr_pages,
        zone_pcp_enable(zone);
 
        /* removal success */
-       adjust_managed_page_count(pfn_to_page(start_pfn), -nr_pages);
+       adjust_managed_page_count(pfn_to_page(start_pfn), -managed_pages);
        adjust_present_page_count(pfn_to_page(start_pfn), group, -nr_pages);
 
        /* reinitialise watermarks and update pcp limits */

diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 7bc43cf24e3d9a715fa276b11ebe4a59210311b2..ee3b370f85c58498cabbf699bce3dae7d969bfc3 100644 (file)
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -6716,14 +6716,19 @@ void zone_pcp_reset(struct zone *zone)
 /*
  * All pages in the range must be in a single zone, must not contain holes,
  * must span full sections, and must be isolated before calling this function.
+ *
+ * Returns the number of managed (non-PageOffline()) pages in the range: the
+ * number of pages for which memory offlining code must adjust managed page
+ * counters using adjust_managed_page_count().
  */
-void __offline_isolated_pages(unsigned long start_pfn, unsigned long end_pfn)
+unsigned long __offline_isolated_pages(unsigned long start_pfn,
+               unsigned long end_pfn)
 {
+       unsigned long already_offline = 0, flags;
        unsigned long pfn = start_pfn;
        struct page *page;
        struct zone *zone;
        unsigned int order;
-       unsigned long flags;
 
        offline_mem_sections(pfn, end_pfn);
        zone = page_zone(pfn_to_page(pfn));
@@ -6745,6 +6750,7 @@ void __offline_isolated_pages(unsigned long start_pfn, unsigned long end_pfn)
                if (PageOffline(page)) {
                        BUG_ON(page_count(page));
                        BUG_ON(PageBuddy(page));
+                       already_offline++;
                        pfn++;
                        continue;
                }
@@ -6757,6 +6763,8 @@ void __offline_isolated_pages(unsigned long start_pfn, unsigned long end_pfn)
                pfn += (1 << order);
        }
        spin_unlock_irqrestore(&zone->lock, flags);
+
+       return end_pfn - start_pfn - already_offline;
 }
 #endif