.open = simple_open,
};
-static int check_memblock_online(struct memory_block *mem, void *arg)
-{
- if (mem->state != MEM_ONLINE)
- return -1;
-
- return 0;
-}
-
-static int change_memblock_state(struct memory_block *mem, void *arg)
-{
- unsigned long state = (unsigned long)arg;
-
- mem->state = state;
-
- return 0;
-}
-
static void memtrace_clear_range(unsigned long start_pfn,
unsigned long nr_pages)
{
unsigned long pfn;
- /*
- * As pages are offline, we cannot trust the memmap anymore. As HIGHMEM
- * does not apply, avoid passing around "struct page" and use
- * clear_page() instead directly.
- */
+ /* As HIGHMEM does not apply, use clear_page() directly. */
for (pfn = start_pfn; pfn < start_pfn + nr_pages; pfn++) {
if (IS_ALIGNED(pfn, PAGES_PER_SECTION))
cond_resched();
}
}
-/* called with device_hotplug_lock held */
-static bool memtrace_offline_pages(u32 nid, u64 start_pfn, u64 nr_pages)
-{
- const unsigned long start = PFN_PHYS(start_pfn);
- const unsigned long size = PFN_PHYS(nr_pages);
-
- if (walk_memory_blocks(start, size, NULL, check_memblock_online))
- return false;
-
- walk_memory_blocks(start, size, (void *)MEM_GOING_OFFLINE,
- change_memblock_state);
-
- if (offline_pages(start_pfn, nr_pages)) {
- walk_memory_blocks(start, size, (void *)MEM_ONLINE,
- change_memblock_state);
- return false;
- }
-
- walk_memory_blocks(start, size, (void *)MEM_OFFLINE,
- change_memblock_state);
-
-
- return true;
-}
-
static u64 memtrace_alloc_node(u32 nid, u64 size)
{
- u64 start_pfn, end_pfn, nr_pages, pfn;
- u64 base_pfn;
- u64 bytes = memory_block_size_bytes();
+ const unsigned long nr_pages = PHYS_PFN(size);
+ unsigned long pfn, start_pfn;
+ struct page *page;
- if (!node_spanned_pages(nid))
+ /*
+ * Trace memory needs to be aligned to the size, which is guaranteed
+ * by alloc_contig_pages().
+ */
+ page = alloc_contig_pages(nr_pages, GFP_KERNEL | __GFP_THISNODE |
+ __GFP_NOWARN, nid, NULL);
+ if (!page)
return 0;
+ start_pfn = page_to_pfn(page);
- start_pfn = node_start_pfn(nid);
- end_pfn = node_end_pfn(nid);
- nr_pages = size >> PAGE_SHIFT;
-
- /* Trace memory needs to be aligned to the size */
- end_pfn = round_down(end_pfn - nr_pages, nr_pages);
-
- lock_device_hotplug();
- for (base_pfn = end_pfn; base_pfn > start_pfn; base_pfn -= nr_pages) {
- if (memtrace_offline_pages(nid, base_pfn, nr_pages) == true) {
- /*
- * Clear the range while we still have a linear
- * mapping.
- */
- memtrace_clear_range(base_pfn, nr_pages);
- /*
- * Remove memory in memory block size chunks so that
- * iomem resources are always split to the same size and
- * we never try to remove memory that spans two iomem
- * resources.
- */
- end_pfn = base_pfn + nr_pages;
- for (pfn = base_pfn; pfn < end_pfn; pfn += bytes>> PAGE_SHIFT) {
- __remove_memory(nid, pfn << PAGE_SHIFT, bytes);
- }
- unlock_device_hotplug();
- return base_pfn << PAGE_SHIFT;
- }
- }
- unlock_device_hotplug();
+ /*
+ * Clear the range while we still have a linear mapping.
+ *
+ * TODO: use __GFP_ZERO with alloc_contig_pages() once supported.
+ */
+ memtrace_clear_range(start_pfn, nr_pages);
- return 0;
+ /*
+ * Set pages PageOffline(), to indicate that nobody (e.g., hibernation,
+ * dumping, ...) should be touching these pages.
+ */
+ for (pfn = start_pfn; pfn < start_pfn + nr_pages; pfn++)
+ __SetPageOffline(pfn_to_page(pfn));
+
+ arch_remove_linear_mapping(PFN_PHYS(start_pfn), size);
+
+ return PFN_PHYS(start_pfn);
}
static int memtrace_init_regions_runtime(u64 size)
return ret;
}
-static int online_mem_block(struct memory_block *mem, void *arg)
+static int memtrace_free(int nid, u64 start, u64 size)
{
- return device_online(&mem->dev);
+ struct mhp_params params = { .pgprot = PAGE_KERNEL };
+ const unsigned long nr_pages = PHYS_PFN(size);
+ const unsigned long start_pfn = PHYS_PFN(start);
+ unsigned long pfn;
+ int ret;
+
+ ret = arch_create_linear_mapping(nid, start, size, ¶ms);
+ if (ret)
+ return ret;
+
+ for (pfn = start_pfn; pfn < start_pfn + nr_pages; pfn++)
+ __ClearPageOffline(pfn_to_page(pfn));
+
+ free_contig_range(start_pfn, nr_pages);
+ return 0;
}
/*
- * Iterate through the chunks of memory we have removed from the kernel
- * and attempt to add them back to the kernel.
+ * Iterate through the chunks of memory we allocated and attempt to expose
+ * them back to the kernel.
*/
-static int memtrace_online(void)
+static int memtrace_free_regions(void)
{
int i, ret = 0;
struct memtrace_entry *ent;
for (i = memtrace_array_nr - 1; i >= 0; i--) {
ent = &memtrace_array[i];
- /* We have onlined this chunk previously */
+ /* We have freed this chunk previously */
if (ent->nid == NUMA_NO_NODE)
continue;
ent->mem = 0;
}
- if (add_memory(ent->nid, ent->start, ent->size, MHP_NONE)) {
- pr_err("Failed to add trace memory to node %d\n",
+ if (memtrace_free(ent->nid, ent->start, ent->size)) {
+ pr_err("Failed to free trace memory on node %d\n",
ent->nid);
ret += 1;
continue;
}
- lock_device_hotplug();
- walk_memory_blocks(ent->start, ent->size, NULL,
- online_mem_block);
- unlock_device_hotplug();
-
/*
- * Memory was added successfully so clean up references to it
- * so on reentry we can tell that this chunk was added.
+ * Memory was freed successfully so clean up references to it
+ * so on reentry we can tell that this chunk was freed.
*/
debugfs_remove_recursive(ent->dir);
- pr_info("Added trace memory back to node %d\n", ent->nid);
+ pr_info("Freed trace memory back on node %d\n", ent->nid);
ent->size = ent->start = ent->nid = NUMA_NO_NODE;
}
if (ret)
return ret;
- /* If all chunks of memory were added successfully, reset globals */
+ /* If all chunks of memory were freed successfully, reset globals */
kfree(memtrace_array);
memtrace_array = NULL;
memtrace_size = 0;
mutex_lock(&memtrace_mutex);
- /* Re-add/online previously removed/offlined memory */
- if (memtrace_size) {
- if (memtrace_online())
- goto out_unlock;
- }
+ /* Free all previously allocated memory. */
+ if (memtrace_size && memtrace_free_regions())
+ goto out_unlock;
if (!val) {
rc = 0;
goto out_unlock;
}
- /* Offline and remove memory */
+ /* Allocate memory. */
if (memtrace_init_regions_runtime(val))
goto out_unlock;